Changelog now links to CLI and Python API in most recent entry

Refs #791, #792, #794, #795

.github/actions/setup-sqlite-version/action.yml

@@ -0,0 +1,39 @@
+name: "Setup SQLite version"
+description: "Build and activate a specific SQLite version from its amalgamation archive"
+inputs:
+  version:
+    description: "The SQLite version to install"
+    required: true
+  cflags:
+    description: "CFLAGS to use when compiling SQLite"
+    required: false
+    default: ""
+  skip-activate:
+    description: "Set to true to skip modifying the library path"
+    required: false
+    default: "false"
+  fallback-urls:
+    description: "Whitespace-separated fallback download URLs to try after sqlite.org"
+    required: false
+    default: ""
+outputs:
+  sqlite-location:
+    description: "Directory containing the compiled SQLite library"
+    value: ${{ steps.build.outputs.sqlite-location }}
+runs:
+  using: "composite"
+  steps:
+    - shell: bash
+      run: mkdir -p "$RUNNER_TEMP/sqlite-versions/downloads"
+    - uses: actions/cache@v6
+      with:
+        path: ${{ runner.temp }}/sqlite-versions/downloads
+        key: setup-sqlite-version-${{ inputs.version }}-amalgamation-v1
+    - id: build
+      shell: bash
+      run: bash "$GITHUB_ACTION_PATH/setup-sqlite-version.sh"
+      env:
+        SQLITE_VERSION: ${{ inputs.version }}
+        SQLITE_CFLAGS: ${{ inputs.cflags }}
+        SQLITE_SKIP_ACTIVATE: ${{ inputs.skip-activate }}
+        SQLITE_EXTRA_FALLBACK_URLS: ${{ inputs.fallback-urls }}

.github/actions/setup-sqlite-version/setup-sqlite-version.sh

@@ -0,0 +1,144 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+version_spec="${SQLITE_VERSION:?SQLITE_VERSION is required}"
+cflags="${SQLITE_CFLAGS:-}"
+skip_activate="${SQLITE_SKIP_ACTIVATE:-false}"
+extra_fallback_urls="${SQLITE_EXTRA_FALLBACK_URLS:-}"
+
+case "$version_spec" in
+  3.46 | 3.46.0)
+    sqlite_version="3.46.0"
+    sqlite_year="2024"
+    amalgamation_id="3460000"
+    builtin_fallback_urls="https://static.simonwillison.net/static/2026/sqlite-amalgamation-3460000.zip"
+    ;;
+  3.23.1)
+    sqlite_version="3.23.1"
+    sqlite_year="2018"
+    amalgamation_id="3230100"
+    builtin_fallback_urls="https://static.simonwillison.net/static/2026/sqlite-amalgamation-3230100.zip"
+    ;;
+  *)
+    echo "::error::Unsupported SQLite version '$version_spec'. Add its release year and amalgamation id to $GITHUB_ACTION_PATH/setup-sqlite-version.sh."
+    exit 1
+    ;;
+esac
+
+case "$(uname -s)" in
+  Linux)
+    library_name="libsqlite3.so.0"
+    library_path_var="LD_LIBRARY_PATH"
+    ;;
+  Darwin)
+    library_name="libsqlite3.dylib"
+    library_path_var="DYLD_LIBRARY_PATH"
+    ;;
+  *)
+    echo "::error::Unsupported platform $(uname -s)"
+    exit 1
+    ;;
+esac
+
+runner_temp="${RUNNER_TEMP:-}"
+if [ -z "$runner_temp" ]; then
+  runner_temp="$(mktemp -d)"
+fi
+
+filename="sqlite-amalgamation-${amalgamation_id}"
+official_url="https://www.sqlite.org/${sqlite_year}/${filename}.zip"
+download_dir="${runner_temp}/sqlite-versions/downloads"
+source_root="${runner_temp}/sqlite-versions/source"
+source_dir="${source_root}/${filename}"
+build_dir="${runner_temp}/sqlite-versions/build/${sqlite_version}"
+archive_path="${download_dir}/${filename}.zip"
+
+mkdir -p "$download_dir" "$source_root" "$build_dir"
+
+download_archive() {
+  local url
+  local candidate_path="${archive_path}.tmp"
+  local urls=("$official_url")
+
+  for url in $builtin_fallback_urls $extra_fallback_urls; do
+    urls+=("$url")
+  done
+
+  rm -f "$candidate_path"
+  for url in "${urls[@]}"; do
+    echo "Downloading SQLite ${sqlite_version} amalgamation from ${url}"
+    if curl \
+      --fail \
+      --location \
+      --show-error \
+      --retry 5 \
+      --retry-delay 2 \
+      --retry-max-time 180 \
+      --retry-all-errors \
+      --connect-timeout 20 \
+      --max-time 240 \
+      --output "$candidate_path" \
+      "$url"; then
+      mv "$candidate_path" "$archive_path"
+      return 0
+    fi
+
+    echo "::warning::Download failed from ${url}"
+    rm -f "$candidate_path"
+  done
+
+  echo "::error::Could not download SQLite ${sqlite_version} amalgamation"
+  return 1
+}
+
+if [ ! -f "${source_dir}/sqlite3.c" ]; then
+  if [ ! -f "$archive_path" ]; then
+    download_archive
+  fi
+
+  rm -rf "$source_dir"
+  unzip -q "$archive_path" -d "$source_root"
+fi
+
+if [ ! -f "${source_dir}/sqlite3.c" ]; then
+  echo "::error::Expected ${source_dir}/sqlite3.c after extracting ${archive_path}"
+  exit 1
+fi
+
+read -r -a cflag_args <<< "$cflags"
+
+echo "Compiling SQLite ${sqlite_version} to ${build_dir}/${library_name}"
+gcc \
+  -fPIC \
+  -shared \
+  "${cflag_args[@]}" \
+  "${source_dir}/sqlite3.c" \
+  "-I${source_dir}" \
+  -o "${build_dir}/${library_name}"
+
+if [ "$library_name" = "libsqlite3.so.0" ]; then
+  ln -sf "$library_name" "${build_dir}/libsqlite3.so"
+fi
+
+if [ -n "${GITHUB_OUTPUT:-}" ]; then
+  echo "sqlite-location=${build_dir}" >> "$GITHUB_OUTPUT"
+else
+  echo "sqlite-location=${build_dir}"
+fi
+
+case "$(printf '%s' "$skip_activate" | tr '[:upper:]' '[:lower:]')" in
+  true | 1 | yes)
+    echo "Skipping ${library_path_var} activation"
+    ;;
+  *)
+    existing_value="${!library_path_var:-}"
+    if [ -n "${GITHUB_ENV:-}" ]; then
+      if [ -n "$existing_value" ]; then
+        echo "${library_path_var}=${build_dir}:${existing_value}" >> "$GITHUB_ENV"
+      else
+        echo "${library_path_var}=${build_dir}" >> "$GITHUB_ENV"
+      fi
+    fi
+    echo "Added ${build_dir} to ${library_path_var}"
+    ;;
+esac

.github/workflows/publish.yml

@@ -9,19 +9,19 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ["3.10", "3.11", "3.12", "3.13"]
+        python-version: ["3.10", "3.11", "3.12", "3.13", "3.14"]
         os: [ubuntu-latest, windows-latest, macos-latest]
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v7
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
         python-version: ${{ matrix.python-version }}
         cache: pip
         cache-dependency-path: pyproject.toml
     - name: Install dependencies
       run: |
-        pip install -e '.[test]'
+        pip install . --group dev
     - name: Run tests
       run: |
         pytest
@@ -29,11 +29,11 @@ jobs:
     runs-on: ubuntu-latest
     needs: [test]
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v7
     - name: Set up Python
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
-        python-version: '3.13'
+        python-version: '3.14'
         cache: pip
         cache-dependency-path: pyproject.toml
     - name: Install dependencies

.github/workflows/spellcheck.yml

@@ -15,7 +15,7 @@ jobs:
         cache-dependency-path: pyproject.toml
     - name: Install dependencies
       run: |
-        pip install -e '.[docs]'
+        pip install . --group docs
     - name: Check spelling
       run: |
         codespell docs/*.rst --ignore-words docs/codespell-ignore-words.txt

.github/workflows/test-coverage.yml

@@ -12,9 +12,9 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - name: Check out repo
-      uses: actions/checkout@v4
+      uses: actions/checkout@v7
     - name: Set up Python
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
         python-version: "3.11"
         cache: pip
@@ -24,7 +24,7 @@ jobs:
     - name: Install Python dependencies
       run: |
         python -m pip install --upgrade pip
-        python -m pip install -e .[test]
+        python -m pip install . --group dev
         python -m pip install pytest-cov
     - name: Run tests
       run: |-

.github/workflows/test-sqlite-support.yml

@@ -18,23 +18,23 @@ jobs:
           "3.23.1", # 2018-04-10, before UPSERT
         ]
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v7
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
         python-version: ${{ matrix.python-version }}
         allow-prereleases: true
         cache: pip
         cache-dependency-path: pyproject.toml
     - name: Set up SQLite ${{ matrix.sqlite-version }}
-      uses: asg017/sqlite-versions@71ea0de37ae739c33e447af91ba71dda8fcf22e6
+      uses: ./.github/actions/setup-sqlite-version
       with:
         version: ${{ matrix.sqlite-version }}
         cflags: "-DSQLITE_ENABLE_DESERIALIZE -DSQLITE_ENABLE_FTS5 -DSQLITE_ENABLE_FTS4 -DSQLITE_ENABLE_FTS3_PARENTHESIS -DSQLITE_ENABLE_RTREE -DSQLITE_ENABLE_JSON1"
     - run: python3 -c "import sqlite3; print(sqlite3.sqlite_version)"
     - name: Install dependencies
       run: |
-        pip install -e '.[test]'
+        pip install . --group dev
         pip freeze
     - name: Run tests
       run: |

.github/workflows/test.yml

@@ -10,13 +10,13 @@ jobs:
     runs-on: ${{ matrix.os }}
     strategy:
       matrix:
-        python-version: ["3.10", "3.11", "3.12", "3.13", "3.14"]
+        python-version: ["3.10", "3.11", "3.12", "3.13", "3.14", "3.15-dev"]
         numpy: [0, 1]
         os: [ubuntu-latest, macos-latest, windows-latest, macos-14]
     steps:
-    - uses: actions/checkout@v4
+    - uses: actions/checkout@v7
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v5
+      uses: actions/setup-python@v6
       with:
         python-version: ${{ matrix.python-version }}
         allow-prereleases: true
@@ -24,18 +24,13 @@ jobs:
         cache-dependency-path: pyproject.toml
     - name: Install dependencies
       run: |
-        pip install -e '.[test,mypy,flake8]'
-    - name: Optionally install tui dependencies
-      run: pip install -e '.[tui]'
+        pip install . --group dev
     - name: Optionally install numpy
       if: matrix.numpy == 1
       run: pip install numpy
     - name: Install SpatiaLite
       if: matrix.os == 'ubuntu-latest'
       run: sudo apt-get install libsqlite3-mod-spatialite
-    - name: On macOS with Python 3.10 test with sqlean.py
-      if: matrix.os == 'macos-latest' && matrix.python-version == '3.10'
-      run: pip install sqlean.py sqlite-dump
     - name: Build extension for --load-extension test
       if: matrix.os == 'ubuntu-latest'
       run: |-
@@ -43,10 +38,18 @@ jobs:
     - name: Run tests
       run: |
         pytest -v
+    - name: Run autocommit tests just on 3.14/Ubuntu
+      if: matrix.os == 'ubuntu-latest' && matrix.python-version == '3.14'
+      run: pytest --sqlite-autocommit
     - name: run mypy
       run: mypy sqlite_utils tests
     - name: run flake8
       run: flake8
+    - name: run ty
+      if: matrix.os != 'windows-latest' && matrix.python-version == '3.14'
+      run: |
+        pip install uv
+        uv run ty check sqlite_utils
     - name: Check formatting
       run: black . --check
     - name: Check if cog needs to be run

.gitpod.yml

@@ -1,2 +0,0 @@
-tasks:
-- init: pip install '.[test]'

.readthedocs.yaml

@@ -4,16 +4,13 @@ sphinx:
   configuration: docs/conf.py
 
 build:
-  os: ubuntu-22.04
+  os: ubuntu-24.04
   tools:
-    python: "3.11"
-
-python:
-  install:
-  - method: pip
-    path: .
-    extra_requirements:
-    - docs
+    python: "3.13"
+  jobs:
+    install:
+    - pip install --upgrade pip
+    - pip install . --group docs
 
 formats:
 - pdf

Justfile

@@ -3,30 +3,31 @@
 
 # Run pytest with supplied options
 @test *options:
-  just run pytest {{options}}
+  uv run pytest {{options}}
 
 @run *options:
-  uv run --isolated --with-editable '.[test,mypy,flake8,docs]' -- {{options}}
+  uv run -- {{options}}
 
-# Run linters: black, flake8, mypy, cog
+# Run linters: black, flake8, mypy, ty, cog
 @lint:
   just run black . --check
-  just run flake8
-  just run mypy sqlite_utils tests
-  just run cog --check README.md docs/*.rst
-  just run codespell docs/*.rst --ignore-words docs/codespell-ignore-words.txt
+  uv run flake8
+  uv run mypy sqlite_utils tests
+  uv run ty check sqlite_utils
+  uv run cog --check README.md docs/*.rst
+  uv run --group docs codespell docs/*.rst --ignore-words docs/codespell-ignore-words.txt
 
 # Rebuild docs with cog
 @cog:
-  just run cog -r README.md docs/*.rst
+  uv run --group docs cog -r README.md docs/*.rst
 
 # Serve live docs on localhost:8000
 @docs: cog
   #!/usr/bin/env bash
   cd docs
-  uv run --isolated --with-editable '../.[test,docs]' make livehtml
+  uv run --group docs make livehtml
 
 
 # Apply Black
 @black:
-  just run black .
+  uv run black .

README.md

@@ -18,9 +18,12 @@ Python CLI utility and library for manipulating SQLite databases.
 - [Configure SQLite full-text search](https://sqlite-utils.datasette.io/en/stable/cli.html#configuring-full-text-search) against your database tables and run search queries against them, ordered by relevance
 - Run [transformations against your tables](https://sqlite-utils.datasette.io/en/stable/cli.html#transforming-tables) to make schema changes that SQLite `ALTER TABLE` does not directly support, such as changing the type of a column
 - [Extract columns](https://sqlite-utils.datasette.io/en/stable/cli.html#extracting-columns-into-a-separate-table) into separate tables to better normalize your existing data
+- [Manage database migrations](https://sqlite-utils.datasette.io/en/stable/migrations.html) using Python migration files and the `sqlite-utils migrate` command
 - [Install plugins](https://sqlite-utils.datasette.io/en/stable/plugins.html) to add custom SQL functions and additional features
 
-Read more on my blog, in this series of posts on [New features in sqlite-utils](https://simonwillison.net/series/sqlite-utils-features/) and other [entries tagged sqliteutils](https://simonwillison.net/tags/sqliteutils/).
+Upgrading from sqlite-utils 3.x? See the [4.0 upgrade guide](https://sqlite-utils.datasette.io/en/stable/upgrading.html#upgrading-from-3-x-to-4-0).
+
+Read more on my blog, in this series of posts on [New features in sqlite-utils](https://simonwillison.net/series/sqlite-utils-features/) and other [entries tagged sqlite-utils](https://simonwillison.net/tags/sqlite-utils/).
 
 ## Installation
 

docs/changelog.rst

@@ -4,6 +4,186 @@
  Changelog
 ===========
 
+.. _v4_1_1:
+
+4.1.1 (2026-07-12)
+------------------
+
+- ``table.transform()`` now raises a ``TransactionError`` if called while a transaction is open with ``PRAGMA foreign_keys`` enabled and the table is referenced by foreign keys with destructive ``ON DELETE`` actions - ``CASCADE``, ``SET NULL`` or ``SET DEFAULT``. The pragma cannot be changed inside a transaction, so previously dropping the old table as part of the transform could fire those actions and silently delete or modify referencing rows. See :ref:`python_api_transform_foreign_keys_transactions` for details and workarounds. (:issue:`794`)
+- The :ref:`CLI <cli>` and :ref:`Python API <python_api>` documentation now cross-reference each other: CLI sections link to the equivalent Python API functionality and Python API sections link back to the corresponding CLI command. (:issue:`791`)
+.. _v4_1:
+
+4.1 (2026-07-11)
+----------------
+
+- ``sqlite-utils insert`` and ``sqlite-utils upsert`` now accept a ``--code`` option for :ref:`providing a block of Python code <cli_insert_code>` (or a path to a ``.py`` file) that defines a ``rows()`` function or ``rows`` iterable of rows to insert, as an alternative to importing from a file. (:issue:`684`)
+- ``sqlite-utils insert`` and ``sqlite-utils upsert`` now accept ``--type column-name type`` to :ref:`override the type automatically chosen when the table is created <cli_insert_csv_tsv_column_types>`. This is useful for CSV or TSV columns such as ZIP codes that look like integers but should be stored as ``TEXT`` to preserve leading zeros. (:issue:`131`)
+- New ``table.drop_index(name)`` method and ``sqlite-utils drop-index`` command for dropping an index by name. Both accept ``ignore=True``/``--ignore`` to ignore a missing index. (:issue:`626`)
+- ``sqlite-utils query`` can now read the SQL query from standard input by passing ``-`` in place of the query, for example ``echo "select * from dogs" | sqlite-utils query dogs.db -``. (:issue:`765`)
+- ``sqlite-utils upsert`` can now infer the primary key of an existing table, so ``--pk`` can be omitted when upserting into a table that already has a primary key.
+- ``table.transform()`` and ``table.transform_sql()`` now accept ``strict=True`` or ``strict=False`` to change a table's `SQLite strict mode <https://www.sqlite.org/stricttables.html>`__. Omitting the option preserves the existing mode. (:issue:`787`)
+- The ``sqlite-utils transform`` command now accepts ``--strict`` and ``--no-strict`` to change a table's strict mode. (:issue:`787`)
+
+.. _v4_0:
+
+4.0 (2026-07-07)
+----------------
+
+The 4.0 release includes some minor backwards-incompatible fixes (hence the major version number bump) and introduces three major new features:
+
+- :ref:`Database migrations <migrations>`, providing a structured mechanism for evolving a project's schema over time. (:issue:`752`)
+- :ref:`Nested transaction support <python_api_atomic>` via ``db.atomic()``, plus numerous improvements to how transactions work across the library. (:issue:`755`)
+- Support for :ref:`compound foreign keys <python_api_compound_foreign_keys>`, including creation, transformation and introspection through :ref:`table.foreign_keys <python_api_introspection_foreign_keys>`. (:issue:`594`)
+
+Other notable changes include:
+
+- Upserts now use SQLite's ``INSERT ... ON CONFLICT ... DO UPDATE SET`` syntax, detect existing table primary keys automatically and reject records that are missing required primary key values. (:issue:`652`)
+- ``db.query()`` now executes immediately and rejects statements that do not return rows; use ``db.execute()`` for writes and DDL.
+- CSV and TSV imports now detect column types by default, while inserts into existing tables preserve those tables' column types. (:issue:`679`)
+- Foreign key handling now preserves ``ON DELETE``/``ON UPDATE`` actions during transforms and resolves referenced primary keys more accurately. (:issue:`530`)
+- Column names passed to Python API methods are now matched case-insensitively, mirroring SQLite's own identifier behavior. (:issue:`760`)
+- The command-line tool now emits UTF-8 JSON output by default, with ``--ascii`` available to restore escaped output. (:issue:`625`)
+- ``table.extract()`` and ``extracts=`` no longer create lookup table records for all-``null`` values. (:issue:`186`)
+
+See :ref:`upgrading_3_to_4` for details on backwards-incompatible changes.
+
+The detailed release notes for the features and fixes shipped during the 4.0 pre-release cycle are available in :ref:`4.0a0 <v4_0a0>`, :ref:`4.0a1 <v4_0a1>`, :ref:`4.0rc1 <v4_0rc1>`, :ref:`4.0rc2 <v4_0rc2>`, :ref:`4.0rc3 <v4_0rc3>` and :ref:`4.0rc4 <v4_0rc4>`.
+
+Bug fixes since 4.0rc4
+~~~~~~~~~~~~~~~~~~~~~~
+
+- Fixed 4.0 regressions in ``insert``/``upsert`` against tables that use SQLite's implicit ``rowid`` primary key. Passing ``pk="rowid"``, ``pk="_rowid_"`` or ``pk="oid"`` now works again for rowid tables, and ``last_pk`` is set correctly. (:issue:`781`)
+- Fixed ``insert(..., ignore=True)`` and ``insert_all(..., ignore=True)`` so an ignored insert that conflicts with an existing primary key row now reports that existing row in ``last_rowid`` and ``last_pk`` where possible. This also works for compound primary keys and list-mode inserts. (:issue:`783`)
+
+.. _v4_0rc4:
+
+4.0rc4 (2026-07-06)
+-------------------
+
+- **Breaking change**: ``table.extract()`` - and the ``sqlite-utils extract`` command - no longer extract rows where every extracted column is ``null``. Those rows now keep a ``null`` value in the new foreign key column instead of pointing at an all-``null`` record in the lookup table. When extracting multiple columns, rows are still extracted if at least one of the columns has a value. (:issue:`186`)
+- The ``extracts=`` option to ``table.insert()`` and friends no longer creates a lookup table record for ``None`` values - the column value stays ``null``. Previously every batch of inserted rows containing a ``None`` value would add a duplicate ``null`` record to the lookup table.
+- Fixed a bug where ``table.lookup()`` inserted a duplicate row on every call if any of the lookup values were ``None``. Lookup values are now compared using ``IS`` so that ``None`` values match existing rows correctly.
+- JSON output from the command-line tool no longer escapes non-ASCII characters, so ``sqlite-utils data.db "select '日本語' as text"`` now outputs ``[{"text": "日本語"}]``. This matches how values were already stored by ``insert`` and how CSV/TSV output already behaved. A new ``--ascii`` option restores the previous behavior of escaping non-ASCII characters, for output destinations that cannot handle UTF-8 - see :ref:`cli_query_json_ascii`. The option is available on the ``query``, ``rows``, ``search``, ``tables``, ``views``, ``triggers``, ``indexes`` and ``memory`` commands. The ``convert --multi --dry-run`` preview and ``plugins`` output also no longer escape non-ASCII characters. (:issue:`625`)
+- ``--no-headers`` now omits the header row from ``--fmt`` and ``--table`` output, not just CSV and TSV output. (:issue:`566`)
+- ``table.insert_all(..., pk=...)`` now raises ``InvalidColumns`` if ``pk=`` names columns that do not exist in an existing table. Previously this behaved inconsistently, with single-row inserts raising a ``KeyError`` while other row counts succeeded. (:issue:`732`)
+- Fixed an ``IndexError`` from ``table.insert(..., pk=..., ignore=True)`` when an ignored insert followed writes to another table on the same connection. ``last_pk`` is now populated from the explicit primary key value instead of looking up a stale ``lastrowid``. (:issue:`554`)
+- Fixed a bug where a failed write statement executed with ``db.execute()`` left the driver's implicit transaction open. Every subsequent write then joined that phantom transaction, which nothing committed, so their work was silently rolled back when the connection was closed. The implicit transaction opened by a failed statement is now rolled back before the exception is raised. A failed write inside a transaction opened with ``db.begin()`` or ``db.atomic()`` leaves that transaction open and untouched, as before.
+- Fixed a bug where transaction-control statements prefixed with an empty statement - ``db.query("; COMMIT")`` - or a UTF-8 byte order mark slipped past the check that rejects them, committing the caller's open transaction before raising a confusing ``OperationalError``. The keyword scanner used by ``db.query()`` and ``db.execute()`` now skips leading ``;`` and byte order marks, matching what the ``sqlite3`` driver tolerates before the first token, so these statements are rejected with a ``ValueError`` without being executed. The same fix means ``db.execute("; BEGIN")`` no longer auto-commits the transaction it just opened.
+- Documented a limitation of ``db.query()``: a ``PRAGMA`` statement that returns no rows raises a ``ValueError`` but still takes effect, because PRAGMA statements run outside the savepoint guard used to roll back other rejected statements. Use ``db.execute()`` for row-less PRAGMA statements.
+- Fixed exception masking when a statement destroys the enclosing transaction. An error such as a ``RAISE(ROLLBACK)`` trigger or ``INSERT OR ROLLBACK`` conflict rolls back the whole transaction, destroying every savepoint - the cleanup in ``db.atomic()`` and ``db.query()`` then failed with ``OperationalError: no such savepoint`` (or ``cannot rollback - no transaction is active``), hiding the original ``IntegrityError`` from code that tried to catch it. Cleanup now checks whether a transaction is still open first, so the original exception propagates.
+- ``sqlite-utils migrate --list`` is now read-only even when the migrations file uses the legacy ``sqlite_migrate.Migrations`` class, whose listing methods create the ``_sqlite_migrations`` table as a side effect. The listing now runs inside a transaction that is rolled back.
+- ``sqlite-utils insert ... --pk <missing column>`` and ``sqlite-utils extract <missing column>`` now show a clean ``Error:`` message instead of a raw Python traceback. The ``extract`` command also shows a clean error when pointed at a view.
+- Fixed a bug where running ``table.extract()`` more than once against the same lookup table inserted duplicate rows for values containing ``null`` - SQLite unique indexes treat ``NULL`` values as distinct, so ``INSERT OR IGNORE`` alone could not dedupe them. Each repeat extract added another copy that nothing referenced. The insert now uses an ``IS``-based ``NOT EXISTS`` guard so ``null``-containing rows match existing lookup rows.
+- ``db.add_foreign_keys()`` no longer silently ignores requested ``ON DELETE``/``ON UPDATE`` actions when a foreign key with the same columns already exists - it raises ``AlterError`` suggesting ``table.transform()``, since the actions of an existing foreign key cannot be changed in place. Exact duplicates, including actions, are still skipped so repeated calls stay idempotent. The method also now validates that compound foreign keys have the same number of columns on both sides, instead of silently discarding the extra columns.
+- ``db.ensure_autocommit_on()`` now raises ``TransactionError`` if called while a transaction is open. Assigning ``isolation_level`` commits any pending transaction as a side effect, so entering the block silently committed the caller's open transaction and made a later ``rollback()`` a no-op.
+- ``sqlite-utils migrate --stop-before`` now exits with an error if the named migration has already been applied. Previously the name passed validation but was only checked against pending migrations, so every migration after it was silently applied - the exact outcome ``--stop-before`` exists to prevent. ``Migrations.apply(db, stop_before=...)`` raises ``ValueError`` in the same situation, before applying anything.
+- Fixed a regression where ``table.insert(..., pk=..., alter=True)`` raised ``InvalidColumns`` if the primary key column did not exist in the table yet. With ``alter=True`` the check now waits until the record keys are known, so a pk column supplied by the records is added by the alter as it was in 3.x. A pk column found in neither the table nor the records still raises ``InvalidColumns``.
+- Fixed a bug where inserting CSV or TSV data into an existing table rewrote that table's column types to match the incoming file. Type detection is the default in 4.0, so ``sqlite-utils insert data.db places places.csv --csv`` against a table with a ``TEXT`` zip code column would convert the column to ``INTEGER`` and corrupt values with leading zeros - ``"01234"`` became ``1234``. Detected types are now only applied when the ``insert`` or ``upsert`` command creates the table.
+- Fixed ``pks_and_rows_where()`` raising ``AttributeError`` when called on a view, and no longer double-quotes the synthesized ``rowid`` column in its generated SQL - SQLite turns a double-quoted identifier that does not resolve into a string literal, which on a view produced a confusing ``KeyError`` instead of the ``OperationalError`` raised in 3.x. Compound primary keys returned by this method now follow ``PRIMARY KEY`` declaration order.
+- The ``foreign_keys=`` argument to ``create()`` and ``insert()`` accepts a mixed list of ``ForeignKey`` objects, tuples and column name strings again. In 4.0 pre-releases mixing ``ForeignKey`` objects with tuples raised a ``ValueError`` - a regression from 3.x, where ``ForeignKey`` was a ``namedtuple`` and passed the tuple checks.
+- ``ForeignKey`` objects are hashable again. The 4.0 change from ``namedtuple`` to dataclass accidentally made them unhashable, breaking patterns like ``set(table.foreign_keys)`` that worked in 3.x. ``ForeignKey`` is now a frozen dataclass - immutable and hashable, like the namedtuple was.
+- Fixed a bug where compound primary key columns were returned in table column order instead of ``PRIMARY KEY`` declaration order. For a table declared as ``CREATE TABLE other (b TEXT, a TEXT, PRIMARY KEY (a, b))`` an implicit ``FOREIGN KEY (x, y) REFERENCES other`` was introspected as referencing ``(b, a)`` when SQLite resolves it as ``(a, b)`` - running ``transform()`` on such a table then rewrote the schema with the inverted column order, silently reversing the meaning of the constraint and causing foreign key errors on valid data. ``table.pks``, compound foreign key guessing and ``transform()`` now all use the primary key declaration order, and ``transform()`` no longer reorders a compound ``PRIMARY KEY (b, a)`` into table column order.
+
+.. _v4_0rc3:
+
+4.0rc3 (2026-07-05)
+-------------------
+
+Breaking changes
+~~~~~~~~~~~~~~~~
+
+- :ref:`table.foreign_keys <python_api_introspection_foreign_keys>` now returns ``ForeignKey`` objects that are dataclasses rather than ``namedtuple`` instances, so they can no longer be unpacked or indexed as ``(table, column, other_table, other_column)`` tuples - access their fields by name instead. Compound (multi-column) foreign keys are now represented as a single ``ForeignKey`` with ``is_compound=True`` and populated ``columns``/``other_columns`` tuples, where ``column`` and ``other_column`` are ``None``. Previously they were returned as one ``ForeignKey`` per column, misleadingly suggesting several independent foreign keys. See :ref:`upgrading_3_to_4` for details. (:issue:`594`)
+- Removed support for using ``sqlean.py`` as a drop-in replacement for the Python standard library ``sqlite3`` module. ``sqlite-utils`` will now use ``pysqlite3`` if it is installed, otherwise it will use ``sqlite3`` from the standard library.
+- The ``db.ensure_autocommit_off()`` context manager has been renamed to ``db.ensure_autocommit_on()``, because the old name described the opposite of what it did. The method temporarily puts the connection into driver-level autocommit mode - by setting ``isolation_level = None`` - so that statements such as ``PRAGMA journal_mode=wal`` can run outside of an implicit transaction. (:issue:`705`)
+
+Compound foreign key support
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- Tables can now be created with :ref:`compound foreign keys <python_api_compound_foreign_keys>`, by passing tuples of column names in ``foreign_keys=``: ``foreign_keys=[(("campus_name", "dept_code"), "departments")]``. The referenced columns default to the compound primary key of the other table. Compound keys are rendered as table-level ``FOREIGN KEY`` constraints in the generated schema.
+- ``table.transform()`` now preserves compound foreign keys, applying any column renames to them. Dropping a column that is part of a compound foreign key drops the whole constraint, matching the existing single-column behavior. ``drop_foreign_keys=`` accepts a bare column name - dropping any foreign key that column participates in - or a tuple of columns to target a compound key precisely.
+- ``table.add_foreign_key()`` and ``db.add_foreign_keys()`` accept tuples of column names to add a compound foreign key to an existing table.
+- ``db.index_foreign_keys()`` creates a single composite index for a compound foreign key.
+
+Other foreign key improvements
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+- ``ForeignKey`` now exposes ``on_delete`` and ``on_update`` fields reflecting the foreign key's ``ON DELETE``/``ON UPDATE`` actions, and ``table.transform()`` preserves those actions. Previously a transform silently stripped clauses such as ``ON DELETE CASCADE`` from the table schema.
+- ``table.add_foreign_key()`` accepts new ``on_delete=`` and ``on_update=`` parameters for creating foreign keys with actions, e.g. ``table.add_foreign_key("author_id", "authors", "id", on_delete="CASCADE")``. (:issue:`530`)
+- Foreign keys declared as ``REFERENCES other_table`` with no explicit column are now resolved to the other table's primary key by ``table.foreign_keys``, instead of reporting ``other_column=None``.
+- Fixed a ``TypeError`` when sorting ``ForeignKey`` objects where some were compound.
+
+Case-insensitive column matching
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+Column names passed to Python API methods are now matched against the table schema case-insensitively, mirroring how SQLite itself treats identifiers. Previously many methods accepted mixed-case identifiers in the SQL they generated but then failed - or silently did nothing - when performing Python-side comparisons against the schema. (:issue:`760`) Fixes include:
+
+- ``table.insert()`` and ``table.upsert()`` now populate ``table.last_pk`` correctly when the ``pk=`` argument uses different casing to the table schema or the record keys - previously this raised a ``KeyError`` after the row had already been written.
+- Upserts no longer raise or misbehave when the casing of ``pk=`` differs from the casing of the record keys. The primary key columns are correctly excluded from the generated ``DO UPDATE SET`` clause.
+- ``table.transform()`` arguments ``types=``, ``rename=``, ``drop=``, ``pk=``, ``not_null=``, ``defaults=``, ``column_order=`` and ``drop_foreign_keys=`` all resolve column names case-insensitively. Previously options like ``rename={"name": "title"}`` against a column called ``Name`` were silently ignored.
+- ``db.create_table(..., transform=True)`` now recognizes existing columns that differ only by case, instead of attempting to add them again and failing with ``duplicate column name``. The casing used in the existing schema is preserved.
+- ``table.lookup()`` returns the primary key value even if ``pk=`` casing differs from the schema, and recognizes existing unique indexes case-insensitively instead of creating redundant ones.
+- ``table.extract()`` and ``table.convert()`` - including ``multi=True`` and ``output=`` - accept column names in any casing.
+- Foreign key columns are validated and recorded using the casing of the actual schema columns, in ``foreign_keys=`` when creating tables, ``db.add_foreign_keys()``, ``table.add_foreign_key()`` and ``table.add_column(fk_col=...)``. Duplicate foreign key detection is also case-insensitive.
+- ``table.create()`` with ``pk=``, ``not_null=``, ``defaults=`` or ``column_order=`` referencing columns using different casing no longer creates an unwanted extra primary key column or raises a ``ValueError``.
+
+Everything else
+~~~~~~~~~~~~~~~
+
+- Fixed a bug where ``table.transform()`` could convert ``DEFAULT TRUE``, ``DEFAULT FALSE`` and ``DEFAULT NULL`` column defaults into quoted string defaults when rebuilding a table. Thanks, `Vincent Gao <https://github.com/gaoflow>`__. (`#764 <https://github.com/simonw/sqlite-utils/pull/764>`__)
+
+.. _v4_0rc2:
+
+4.0rc2 (2026-07-04)
+-------------------
+
+Breaking changes:
+
+- Write statements executed with ``db.execute()`` are now committed automatically, unless a transaction is already open in which case they join it. Previously they opened an implicit transaction that stayed open until something committed it - writes appeared to work when read on the same connection but were silently rolled back when the connection closed. Code that relied on rolling back uncommitted ``db.execute()`` writes should use the new ``db.begin()`` method to open an explicit transaction first. The transaction model is documented in full at :ref:`python_api_transactions`.
+- ``db.query()`` now executes its SQL as soon as it is called, rather than waiting until the returned generator is first iterated. Rows are still fetched lazily during iteration. SQL errors are now raised at the call site, statements such as ``INSERT ... RETURNING`` are executed and committed immediately without needing to iterate over their results, and passing a statement that returns no rows - previously a silent no-op - now raises a ``ValueError`` recommending ``db.execute()`` instead. A statement rejected this way is rolled back before the error is raised, so it has no effect on the database.
+- Python API validation errors now raise ``ValueError`` instead of ``AssertionError``. Previously invalid arguments - such as ``create_table()`` with no columns, ``transform()`` on a table that does not exist, or passing both ``ignore=True`` and ``replace=True`` - were rejected using bare ``assert`` statements, which are silently skipped when Python runs with the ``-O`` flag. Code that caught ``AssertionError`` for these cases should catch ``ValueError`` instead.
+- ``table.upsert()`` and ``table.upsert_all()`` now raise ``PrimaryKeyRequired`` if a record is missing a value for any primary key column, or has a value of ``None`` for one. Previously such records - which can never match an existing row - were quietly inserted as brand new rows, or triggered a confusing ``KeyError`` after the insert had already taken place.
+- ``db.enable_wal()`` and ``db.disable_wal()`` now raise a ``sqlite_utils.db.TransactionError`` if called while a transaction is open. Previously they would silently commit the open transaction as a side effect of changing the journal mode, breaking the rollback guarantee of ``db.atomic()`` and of user-managed transactions.
+- The ``View`` class no longer has an ``enable_fts()`` method. It existed only to raise ``NotImplementedError``, since full-text search is not supported for views - calling it now raises ``AttributeError`` instead, and the method no longer appears in the API reference. The ``sqlite-utils enable-fts`` command shows a clean error when pointed at a view.
+- The no-op ``-d/--detect-types`` flag has been removed from the ``insert`` and ``upsert`` commands. Type detection has been the default for CSV/TSV data since 4.0a1, so the flag did nothing - invocations using it should simply drop it. ``--no-detect-types`` remains available to disable detection.
+- ``Database()`` now raises a ``sqlite_utils.db.TransactionError`` if passed a connection created with the Python 3.12+ ``sqlite3.connect(..., autocommit=True)`` or ``autocommit=False`` options. ``commit()`` and ``rollback()`` behave differently on those connections, which previously caused every write made by the library to be silently discarded when the connection closed.
+
+Everything else:
+
+- Fixed a bug where ``table.delete_where()``, ``table.optimize()`` and ``table.rebuild_fts()`` did not commit their changes, leaving the connection inside an open transaction. Their work - and any subsequent writes - could then be silently rolled back when the connection was closed. All three now use ``db.atomic()``, consistent with the other write methods.
+- The ``sqlite-utils drop-table`` command now refuses to drop a view, and ``drop-view`` refuses to drop a table. Previously each would silently drop the wrong type of object if the name matched. Both now exit with an error suggesting the correct command to use.
+- Migrations applied by the new :ref:`migrations system <migrations>` now run inside a transaction, together with the record of the migration having been applied. If a migration raises an exception its changes are rolled back and it stays pending, so it can be safely re-applied after the error is fixed. Migrations that cannot run inside a transaction, such as those executing ``VACUUM``, can opt out using ``@migrations(transactional=False)`` - see :ref:`migrations_transactions`.
+- ``table.upsert()`` and ``table.upsert_all()`` now detect the primary key or compound primary key of an existing table, so the ``pk=`` argument is no longer required when upserting into a table that already has a primary key.
+- ``db.table(table_name).insert({})`` can now be used to insert a row consisting entirely of default values into an existing table, using ``INSERT INTO ... DEFAULT VALUES``. (:issue:`759`)
+- Improvements to the ``sqlite-utils migrate`` command: ``--stop-before`` values that do not match any known migration are now an error instead of being silently ignored, ``--stop-before`` now works correctly with migration files that still use the older ``sqlite_migrate.Migrations`` class, and ``--list`` is now a read-only operation that no longer creates the database file or the migrations tracking table. ``migrations.applied()`` now returns migrations in the order they were applied.
+- New ``db.begin()``, ``db.commit()`` and ``db.rollback()`` methods for taking manual control of transactions, as an alternative to the ``db.atomic()`` context manager.
+- New documentation: :ref:`python_api_transactions` describes how transactions work and when changes are committed, and a new :ref:`upgrading` page details the changes needed to move between major versions.
+
+.. _v4_0rc1:
+
+4.0rc1 (2026-06-21)
+-------------------
+
+- New :ref:`database migrations system <migrations>`, incorporating functionality that was previously provided by the separate `sqlite-migrate <https://github.com/simonw/sqlite-migrate>`__ plugin. Define migration sets using the new :class:`sqlite_utils.Migrations` class and apply them using the ``sqlite-utils migrate`` command or the :ref:`migrations Python API <migrations_python>`. (:issue:`752`)
+- New ``db.atomic()`` :ref:`context manager providing nested transaction support <python_api_atomic>` using SQLite transactions and savepoints. Internal multi-step operations such as ``table.transform()`` now use this mechanism to avoid unexpectedly committing an existing transaction. (:issue:`755`)
+- ``Database`` objects can now be :ref:`used as context managers <python_api_close>`, automatically closing the connection when the ``with`` block exits. The CLI also now closes database and file handles more reliably, resolving a number of ``ResourceWarning`` warnings. (:issue:`692`)
+- The ``sqlite-utils convert`` command can now accept a direct callable reference such as ``r.parsedate`` or ``json.loads --import json`` as the conversion code, as an alternative to calling it explicitly with ``r.parsedate(value)``. (:issue:`686`)
+- Fixed a bug where CSV or TSV files with only a header row could crash ``sqlite-utils insert`` and ``sqlite-utils memory`` when type detection was enabled. Thanks, `Rami Abdelrazzaq <https://github.com/RamiNoodle733>`__. (:issue:`702`, `#707 <https://github.com/simonw/sqlite-utils/pull/707>`__)
+- Fixed a bug where installed plugins could be loaded while running the test suite, despite the test-mode safeguard that disables plugin loading. Thanks, `Rami Abdelrazzaq <https://github.com/RamiNoodle733>`__. (:issue:`713`, `#719 <https://github.com/simonw/sqlite-utils/pull/719>`__)
+- ``table.detect_fts()`` now recognizes legacy FTS virtual tables that quote the ``content=`` table name using square brackets, allowing ``table.enable_fts(..., replace=True)`` to replace them correctly. (:issue:`694`)
+- Now depends on Click 8.3.1 or later, removing compatibility workarounds for Click's ``Sentinel`` default values. (:issue:`666`)
+- Improved type annotations throughout the package, with ``ty`` now run in CI. (:issue:`697`)
+- Development tooling now uses ``uv`` dependency groups, with separate ``dev`` and ``docs`` groups. (:issue:`691`)
+- The test suite now runs against Python 3.15-dev. (:issue:`738`)
+
+.. _v3_39:
+
+3.39 (2025-11-24)
+-----------------
+
+- Fixed a bug with ``sqlite-utils install`` when the tool had been installed using ``uv``. (:issue:`687`)
+- The ``--functions`` argument now optionally accepts a path to a Python file as an alternative to a string full of code, and can be specified multiple times - see :ref:`cli_query_functions`. (:issue:`659`)
+- ``sqlite-utils`` now requires Python 3.10 or higher.
+
 .. _v4_0a1:
 
 4.0a1 (2025-11-23)
@@ -173,7 +353,7 @@ This release introduces a new :ref:`plugin system <plugins>`. Read more about th
 - Conversion functions passed to :ref:`table.convert(...) <python_api_convert>` can now return lists or dictionaries, which will be inserted into the database as JSON strings. (:issue:`495`)
 - ``sqlite-utils install`` and ``sqlite-utils uninstall`` commands for installing packages into the same virtual environment as ``sqlite-utils``, :ref:`described here <cli_install>`. (:issue:`483`)
 - New :ref:`sqlite_utils.utils.flatten() <reference_utils_flatten>` utility function. (:issue:`500`)
-- Documentation on :ref:`using Just <contributing_just>` to run tests, linters and build documentation. 
+- Documentation on :ref:`using Just <contributing_just>` to run tests, linters and build documentation.
 - Documentation now covers the :ref:`release_process` for this package.
 
 .. _v3_29:

docs/cli-reference.rst

@@ -19,8 +19,8 @@ This page lists the ``--help`` for every ``sqlite-utils`` CLI sub-command.
     go_first = [
         "query", "memory", "insert", "upsert", "bulk", "search", "transform", "extract",
         "schema", "insert-files", "analyze-tables", "convert", "tables", "views", "rows",
-        "triggers", "indexes", "create-database", "create-table", "create-index",
-        "enable-fts", "populate-fts", "rebuild-fts", "disable-fts"
+        "triggers", "indexes", "create-database", "create-table", "create-index", "drop-index",
+        "migrate", "enable-fts", "populate-fts", "rebuild-fts", "disable-fts"
     ]
     refs = {
         "query": "cli_query",
@@ -46,9 +46,11 @@ This page lists the ``--help`` for every ``sqlite-utils`` CLI sub-command.
         "add-foreign-keys": "cli_add_foreign_keys",
         "index-foreign-keys": "cli_index_foreign_keys",
         "create-index": "cli_create_index",
+        "drop-index": "cli_drop_index",
         "enable-wal": "cli_wal",
         "enable-counts": "cli_enable_counts",
         "bulk": "cli_bulk",
+        "migrate": "cli_migrate",
         "create-database": "cli_create_database",
         "create-table": "cli_create_table",
         "drop-table": "cli_drop_table",
@@ -108,6 +110,10 @@ See :ref:`cli_query`.
               "select * from chickens where age > :age" \
               -p age 1
 
+      Pass "-" as the SQL to read the query from standard input:
+
+          echo "select * from chickens" | sqlite-utils data.db -
+
     Options:
       --attach <TEXT FILE>...     Additional databases to attach - specify alias and
                                   filepath
@@ -115,24 +121,26 @@ See :ref:`cli_query`.
       --arrays                    Output rows as arrays instead of objects
       --csv                       Output CSV
       --tsv                       Output TSV
-      --no-headers                Omit CSV headers
+      --no-headers                Omit headers from CSV/TSV and table/--fmt output
       -t, --table                 Output as a formatted table
-      --fmt TEXT                  Table format - one of asciidoc, double_grid,
-                                  double_outline, fancy_grid, fancy_outline, github,
-                                  grid, heavy_grid, heavy_outline, html, jira,
-                                  latex, latex_booktabs, latex_longtable, latex_raw,
-                                  mediawiki, mixed_grid, mixed_outline, moinmoin,
-                                  orgtbl, outline, pipe, plain, presto, pretty,
-                                  psql, rounded_grid, rounded_outline, rst, simple,
-                                  simple_grid, simple_outline, textile, tsv,
-                                  unsafehtml, youtrack
+      --fmt TEXT                  Table format - one of asciidoc, colon_grid,
+                                  double_grid, double_outline, fancy_grid,
+                                  fancy_outline, github, grid, heavy_grid,
+                                  heavy_outline, html, jira, latex, latex_booktabs,
+                                  latex_longtable, latex_raw, mediawiki, mixed_grid,
+                                  mixed_outline, moinmoin, orgtbl, outline, pipe,
+                                  plain, presto, pretty, psql, rounded_grid,
+                                  rounded_outline, rst, simple, simple_grid,
+                                  simple_outline, textile, tsv, unsafehtml, youtrack
       --json-cols                 Detect JSON cols and output them as JSON, not
                                   escaped strings
+      --ascii                     Escape non-ASCII characters in JSON output as
+                                  \uXXXX
       -r, --raw                   Raw output, first column of first row
       --raw-lines                 Raw output, first column of each row
       -p, --param <TEXT TEXT>...  Named :parameters for SQL query
-      --functions TEXT            Python code or file path defining custom SQL
-                                  functions
+      --functions TEXT            Python code or a file path defining custom SQL
+                                  functions; can be used multiple times
       --load-extension TEXT       Path to SQLite extension, with optional
                                   :entrypoint
       -h, --help                  Show this message and exit.
@@ -174,8 +182,8 @@ See :ref:`cli_memory`.
           sqlite-utils memory animals.csv --schema
 
     Options:
-      --functions TEXT            Python code or file path defining custom SQL
-                                  functions
+      --functions TEXT            Python code or a file path defining custom SQL
+                                  functions; can be used multiple times
       --attach <TEXT FILE>...     Additional databases to attach - specify alias and
                                   filepath
       --flatten                   Flatten nested JSON objects, so {"foo": {"bar":
@@ -184,19 +192,21 @@ See :ref:`cli_memory`.
       --arrays                    Output rows as arrays instead of objects
       --csv                       Output CSV
       --tsv                       Output TSV
-      --no-headers                Omit CSV headers
+      --no-headers                Omit headers from CSV/TSV and table/--fmt output
       -t, --table                 Output as a formatted table
-      --fmt TEXT                  Table format - one of asciidoc, double_grid,
-                                  double_outline, fancy_grid, fancy_outline, github,
-                                  grid, heavy_grid, heavy_outline, html, jira,
-                                  latex, latex_booktabs, latex_longtable, latex_raw,
-                                  mediawiki, mixed_grid, mixed_outline, moinmoin,
-                                  orgtbl, outline, pipe, plain, presto, pretty,
-                                  psql, rounded_grid, rounded_outline, rst, simple,
-                                  simple_grid, simple_outline, textile, tsv,
-                                  unsafehtml, youtrack
+      --fmt TEXT                  Table format - one of asciidoc, colon_grid,
+                                  double_grid, double_outline, fancy_grid,
+                                  fancy_outline, github, grid, heavy_grid,
+                                  heavy_outline, html, jira, latex, latex_booktabs,
+                                  latex_longtable, latex_raw, mediawiki, mixed_grid,
+                                  mixed_outline, moinmoin, orgtbl, outline, pipe,
+                                  plain, presto, pretty, psql, rounded_grid,
+                                  rounded_outline, rst, simple, simple_grid,
+                                  simple_outline, textile, tsv, unsafehtml, youtrack
       --json-cols                 Detect JSON cols and output them as JSON, not
                                   escaped strings
+      --ascii                     Escape non-ASCII characters in JSON output as
+                                  \uXXXX
       -r, --raw                   Raw output, first column of first row
       --raw-lines                 Raw output, first column of each row
       -p, --param <TEXT TEXT>...  Named :parameters for SQL query
@@ -221,7 +231,7 @@ See :ref:`cli_inserting_data`, :ref:`cli_insert_csv_tsv`, :ref:`cli_insert_unstr
 
 ::
 
-    Usage: sqlite-utils insert [OPTIONS] PATH TABLE FILE
+    Usage: sqlite-utils insert [OPTIONS] PATH TABLE [FILE]
 
       Insert records from FILE into a table, creating the table if it does not
       already exist.
@@ -237,6 +247,9 @@ See :ref:`cli_inserting_data`, :ref:`cli_insert_csv_tsv`, :ref:`cli_insert_unstr
       - Use --lines to write each incoming line to a column called "line"
       - Use --text to write the entire input to a column called "text"
 
+      Use --type column-name type to override the type automatically chosen when the
+      table is created.
+
       You can also use --convert to pass a fragment of Python code that will be used
       to convert each input.
 
@@ -263,8 +276,20 @@ See :ref:`cli_inserting_data`, :ref:`cli_insert_csv_tsv`, :ref:`cli_insert_unstr
           echo 'A bunch of words' | sqlite-utils insert words.db words - \
             --text --convert '({"word": w} for w in text.split())'
 
+      Instead of a FILE you can use --code to provide a block of Python code that
+      defines the rows to insert, as either a rows() function that yields
+      dictionaries or a "rows" iterable. --code can also be a path to a .py file:
+
+          sqlite-utils insert data.db creatures --code '
+          def rows():
+              yield {"id": 1, "name": "Cleo"}
+              yield {"id": 2, "name": "Suna"}
+          ' --pk id
+
     Options:
       --pk TEXT                 Columns to use as the primary key, e.g. id
+      --code TEXT               Python code defining a rows() function or iterable
+                                of rows to insert
       --flatten                 Flatten nested JSON objects, so {"a": {"b": 1}}
                                 becomes {"a_b": 1}
       --nl                      Expect newline-delimited JSON
@@ -285,7 +310,7 @@ See :ref:`cli_inserting_data`, :ref:`cli_insert_csv_tsv`, :ref:`cli_insert_unstr
       --alter                   Alter existing table to add any missing columns
       --not-null TEXT           Columns that should be created as NOT NULL
       --default <TEXT TEXT>...  Default value that should be set for a column
-      -d, --detect-types        Detect types for columns in CSV/TSV data (default)
+      --type <TEXT CHOICE>...   Column types to use when creating the table
       --no-detect-types         Treat all CSV/TSV columns as TEXT
       --analyze                 Run ANALYZE at the end of this operation
       --load-extension TEXT     Path to SQLite extension, with optional :entrypoint
@@ -307,12 +332,17 @@ See :ref:`cli_upsert`.
 
 ::
 
-    Usage: sqlite-utils upsert [OPTIONS] PATH TABLE FILE
+    Usage: sqlite-utils upsert [OPTIONS] PATH TABLE [FILE]
 
       Upsert records based on their primary key. Works like 'insert' but if an
       incoming record has a primary key that matches an existing record the existing
       record will be updated.
 
+      If the table already exists and has a primary key, --pk can be omitted.
+
+      Use --type column-name type to override the type automatically chosen when the
+      table is created.
+
       Example:
 
           echo '[
@@ -322,7 +352,8 @@ See :ref:`cli_upsert`.
 
     Options:
       --pk TEXT                 Columns to use as the primary key, e.g. id
-                                [required]
+      --code TEXT               Python code defining a rows() function or iterable
+                                of rows to insert
       --flatten                 Flatten nested JSON objects, so {"a": {"b": 1}}
                                 becomes {"a_b": 1}
       --nl                      Expect newline-delimited JSON
@@ -343,7 +374,7 @@ See :ref:`cli_upsert`.
       --alter                   Alter existing table to add any missing columns
       --not-null TEXT           Columns that should be created as NOT NULL
       --default <TEXT TEXT>...  Default value that should be set for a column
-      -d, --detect-types        Detect types for columns in CSV/TSV data (default)
+      --type <TEXT CHOICE>...   Column types to use when creating the table
       --no-detect-types         Treat all CSV/TSV columns as TEXT
       --analyze                 Run ANALYZE at the end of this operation
       --load-extension TEXT     Path to SQLite extension, with optional :entrypoint
@@ -376,7 +407,8 @@ See :ref:`cli_bulk`.
 
     Options:
       --batch-size INTEGER   Commit every X records
-      --functions TEXT       Python code or file path defining custom SQL functions
+      --functions TEXT       Python code or a file path defining custom SQL
+                             functions; can be used multiple times
       --flatten              Flatten nested JSON objects, so {"a": {"b": 1}} becomes
                              {"a_b": 1}
       --nl                   Expect newline-delimited JSON
@@ -423,18 +455,20 @@ See :ref:`cli_search`.
       --arrays               Output rows as arrays instead of objects
       --csv                  Output CSV
       --tsv                  Output TSV
-      --no-headers           Omit CSV headers
+      --no-headers           Omit headers from CSV/TSV and table/--fmt output
       -t, --table            Output as a formatted table
-      --fmt TEXT             Table format - one of asciidoc, double_grid,
-                             double_outline, fancy_grid, fancy_outline, github,
-                             grid, heavy_grid, heavy_outline, html, jira, latex,
-                             latex_booktabs, latex_longtable, latex_raw, mediawiki,
-                             mixed_grid, mixed_outline, moinmoin, orgtbl, outline,
-                             pipe, plain, presto, pretty, psql, rounded_grid,
-                             rounded_outline, rst, simple, simple_grid,
-                             simple_outline, textile, tsv, unsafehtml, youtrack
+      --fmt TEXT             Table format - one of asciidoc, colon_grid,
+                             double_grid, double_outline, fancy_grid, fancy_outline,
+                             github, grid, heavy_grid, heavy_outline, html, jira,
+                             latex, latex_booktabs, latex_longtable, latex_raw,
+                             mediawiki, mixed_grid, mixed_outline, moinmoin, orgtbl,
+                             outline, pipe, plain, presto, pretty, psql,
+                             rounded_grid, rounded_outline, rst, simple,
+                             simple_grid, simple_outline, textile, tsv, unsafehtml,
+                             youtrack
       --json-cols            Detect JSON cols and output them as JSON, not escaped
                              strings
+      --ascii                Escape non-ASCII characters in JSON output as \uXXXX
       --load-extension TEXT  Path to SQLite extension, with optional :entrypoint
       -h, --help             Show this message and exit.
 
@@ -474,6 +508,8 @@ See :ref:`cli_transform_table`.
                                       Add a foreign key constraint from a column to
                                       another table with another column
       --drop-foreign-key TEXT         Drop foreign key constraint for this column
+      --strict / --no-strict          Enable or disable STRICT mode (default:
+                                      preserve current mode)
       --sql                           Output SQL without executing it
       --load-extension TEXT           Path to SQLite extension, with optional
                                       :entrypoint
@@ -611,27 +647,33 @@ See :ref:`cli_convert`.
 
       "value" is a variable with the column value to be converted.
 
+      CODE can also be a reference to a callable that takes the value, for example:
+
+      sqlite-utils convert my.db mytable date r.parsedate
+      sqlite-utils convert my.db mytable data json.loads --import json
+
       Use "-" for CODE to read Python code from standard input.
 
       The following common operations are available as recipe functions:
 
-      r.jsonsplit(value, delimiter=',', type=<class 'str'>)
+      r.jsonsplit(value: 'str', delimiter: 'str' = ',', type: 'Callable[[str],
+      object]' = <class 'str'>) -> 'str'
 
       Convert a string like a,b,c into a JSON array ["a", "b", "c"]
 
-      r.parsedate(value, dayfirst=False, yearfirst=False, errors=None)
+      r.parsedate(value: 'str', dayfirst: 'bool' = False, yearfirst: 'bool' = False,
+      errors: 'Optional[object]' = None) -> 'Optional[str]'
 
       Parse a date and convert it to ISO date format: yyyy-mm-dd
-  
       - dayfirst=True: treat xx as the day in xx/yy/zz
       - yearfirst=True: treat xx as the year in xx/yy/zz
       - errors=r.IGNORE to ignore values that cannot be parsed
       - errors=r.SET_NULL to set values that cannot be parsed to null
 
-      r.parsedatetime(value, dayfirst=False, yearfirst=False, errors=None)
+      r.parsedatetime(value: 'str', dayfirst: 'bool' = False, yearfirst: 'bool' =
+      False, errors: 'Optional[object]' = None) -> 'Optional[str]'
 
       Parse a datetime and convert it to ISO datetime format: yyyy-mm-ddTHH:MM:SS
-  
       - dayfirst=True: treat xx as the day in xx/yy/zz
       - yearfirst=True: treat xx as the year in xx/yy/zz
       - errors=r.IGNORE to ignore values that cannot be parsed
@@ -685,18 +727,20 @@ See :ref:`cli_tables`.
       --arrays               Output rows as arrays instead of objects
       --csv                  Output CSV
       --tsv                  Output TSV
-      --no-headers           Omit CSV headers
+      --no-headers           Omit headers from CSV/TSV and table/--fmt output
       -t, --table            Output as a formatted table
-      --fmt TEXT             Table format - one of asciidoc, double_grid,
-                             double_outline, fancy_grid, fancy_outline, github,
-                             grid, heavy_grid, heavy_outline, html, jira, latex,
-                             latex_booktabs, latex_longtable, latex_raw, mediawiki,
-                             mixed_grid, mixed_outline, moinmoin, orgtbl, outline,
-                             pipe, plain, presto, pretty, psql, rounded_grid,
-                             rounded_outline, rst, simple, simple_grid,
-                             simple_outline, textile, tsv, unsafehtml, youtrack
+      --fmt TEXT             Table format - one of asciidoc, colon_grid,
+                             double_grid, double_outline, fancy_grid, fancy_outline,
+                             github, grid, heavy_grid, heavy_outline, html, jira,
+                             latex, latex_booktabs, latex_longtable, latex_raw,
+                             mediawiki, mixed_grid, mixed_outline, moinmoin, orgtbl,
+                             outline, pipe, plain, presto, pretty, psql,
+                             rounded_grid, rounded_outline, rst, simple,
+                             simple_grid, simple_outline, textile, tsv, unsafehtml,
+                             youtrack
       --json-cols            Detect JSON cols and output them as JSON, not escaped
                              strings
+      --ascii                Escape non-ASCII characters in JSON output as \uXXXX
       --columns              Include list of columns for each table
       --schema               Include schema for each table
       --load-extension TEXT  Path to SQLite extension, with optional :entrypoint
@@ -726,18 +770,20 @@ See :ref:`cli_views`.
       --arrays               Output rows as arrays instead of objects
       --csv                  Output CSV
       --tsv                  Output TSV
-      --no-headers           Omit CSV headers
+      --no-headers           Omit headers from CSV/TSV and table/--fmt output
       -t, --table            Output as a formatted table
-      --fmt TEXT             Table format - one of asciidoc, double_grid,
-                             double_outline, fancy_grid, fancy_outline, github,
-                             grid, heavy_grid, heavy_outline, html, jira, latex,
-                             latex_booktabs, latex_longtable, latex_raw, mediawiki,
-                             mixed_grid, mixed_outline, moinmoin, orgtbl, outline,
-                             pipe, plain, presto, pretty, psql, rounded_grid,
-                             rounded_outline, rst, simple, simple_grid,
-                             simple_outline, textile, tsv, unsafehtml, youtrack
+      --fmt TEXT             Table format - one of asciidoc, colon_grid,
+                             double_grid, double_outline, fancy_grid, fancy_outline,
+                             github, grid, heavy_grid, heavy_outline, html, jira,
+                             latex, latex_booktabs, latex_longtable, latex_raw,
+                             mediawiki, mixed_grid, mixed_outline, moinmoin, orgtbl,
+                             outline, pipe, plain, presto, pretty, psql,
+                             rounded_grid, rounded_outline, rst, simple,
+                             simple_grid, simple_outline, textile, tsv, unsafehtml,
+                             youtrack
       --json-cols            Detect JSON cols and output them as JSON, not escaped
                              strings
+      --ascii                Escape non-ASCII characters in JSON output as \uXXXX
       --columns              Include list of columns for each view
       --schema               Include schema for each view
       --load-extension TEXT  Path to SQLite extension, with optional :entrypoint
@@ -772,19 +818,21 @@ See :ref:`cli_rows`.
       --arrays                    Output rows as arrays instead of objects
       --csv                       Output CSV
       --tsv                       Output TSV
-      --no-headers                Omit CSV headers
+      --no-headers                Omit headers from CSV/TSV and table/--fmt output
       -t, --table                 Output as a formatted table
-      --fmt TEXT                  Table format - one of asciidoc, double_grid,
-                                  double_outline, fancy_grid, fancy_outline, github,
-                                  grid, heavy_grid, heavy_outline, html, jira,
-                                  latex, latex_booktabs, latex_longtable, latex_raw,
-                                  mediawiki, mixed_grid, mixed_outline, moinmoin,
-                                  orgtbl, outline, pipe, plain, presto, pretty,
-                                  psql, rounded_grid, rounded_outline, rst, simple,
-                                  simple_grid, simple_outline, textile, tsv,
-                                  unsafehtml, youtrack
+      --fmt TEXT                  Table format - one of asciidoc, colon_grid,
+                                  double_grid, double_outline, fancy_grid,
+                                  fancy_outline, github, grid, heavy_grid,
+                                  heavy_outline, html, jira, latex, latex_booktabs,
+                                  latex_longtable, latex_raw, mediawiki, mixed_grid,
+                                  mixed_outline, moinmoin, orgtbl, outline, pipe,
+                                  plain, presto, pretty, psql, rounded_grid,
+                                  rounded_outline, rst, simple, simple_grid,
+                                  simple_outline, textile, tsv, unsafehtml, youtrack
       --json-cols                 Detect JSON cols and output them as JSON, not
                                   escaped strings
+      --ascii                     Escape non-ASCII characters in JSON output as
+                                  \uXXXX
       --load-extension TEXT       Path to SQLite extension, with optional
                                   :entrypoint
       -h, --help                  Show this message and exit.
@@ -812,18 +860,20 @@ See :ref:`cli_triggers`.
       --arrays               Output rows as arrays instead of objects
       --csv                  Output CSV
       --tsv                  Output TSV
-      --no-headers           Omit CSV headers
+      --no-headers           Omit headers from CSV/TSV and table/--fmt output
       -t, --table            Output as a formatted table
-      --fmt TEXT             Table format - one of asciidoc, double_grid,
-                             double_outline, fancy_grid, fancy_outline, github,
-                             grid, heavy_grid, heavy_outline, html, jira, latex,
-                             latex_booktabs, latex_longtable, latex_raw, mediawiki,
-                             mixed_grid, mixed_outline, moinmoin, orgtbl, outline,
-                             pipe, plain, presto, pretty, psql, rounded_grid,
-                             rounded_outline, rst, simple, simple_grid,
-                             simple_outline, textile, tsv, unsafehtml, youtrack
+      --fmt TEXT             Table format - one of asciidoc, colon_grid,
+                             double_grid, double_outline, fancy_grid, fancy_outline,
+                             github, grid, heavy_grid, heavy_outline, html, jira,
+                             latex, latex_booktabs, latex_longtable, latex_raw,
+                             mediawiki, mixed_grid, mixed_outline, moinmoin, orgtbl,
+                             outline, pipe, plain, presto, pretty, psql,
+                             rounded_grid, rounded_outline, rst, simple,
+                             simple_grid, simple_outline, textile, tsv, unsafehtml,
+                             youtrack
       --json-cols            Detect JSON cols and output them as JSON, not escaped
                              strings
+      --ascii                Escape non-ASCII characters in JSON output as \uXXXX
       --load-extension TEXT  Path to SQLite extension, with optional :entrypoint
       -h, --help             Show this message and exit.
 
@@ -851,18 +901,20 @@ See :ref:`cli_indexes`.
       --arrays               Output rows as arrays instead of objects
       --csv                  Output CSV
       --tsv                  Output TSV
-      --no-headers           Omit CSV headers
+      --no-headers           Omit headers from CSV/TSV and table/--fmt output
       -t, --table            Output as a formatted table
-      --fmt TEXT             Table format - one of asciidoc, double_grid,
-                             double_outline, fancy_grid, fancy_outline, github,
-                             grid, heavy_grid, heavy_outline, html, jira, latex,
-                             latex_booktabs, latex_longtable, latex_raw, mediawiki,
-                             mixed_grid, mixed_outline, moinmoin, orgtbl, outline,
-                             pipe, plain, presto, pretty, psql, rounded_grid,
-                             rounded_outline, rst, simple, simple_grid,
-                             simple_outline, textile, tsv, unsafehtml, youtrack
+      --fmt TEXT             Table format - one of asciidoc, colon_grid,
+                             double_grid, double_outline, fancy_grid, fancy_outline,
+                             github, grid, heavy_grid, heavy_outline, html, jira,
+                             latex, latex_booktabs, latex_longtable, latex_raw,
+                             mediawiki, mixed_grid, mixed_outline, moinmoin, orgtbl,
+                             outline, pipe, plain, presto, pretty, psql,
+                             rounded_grid, rounded_outline, rst, simple,
+                             simple_grid, simple_outline, textile, tsv, unsafehtml,
+                             youtrack
       --json-cols            Detect JSON cols and output them as JSON, not escaped
                              strings
+      --ascii                Escape non-ASCII characters in JSON output as \uXXXX
       --load-extension TEXT  Path to SQLite extension, with optional :entrypoint
       -h, --help             Show this message and exit.
 
@@ -908,10 +960,10 @@ See :ref:`cli_create_table`.
           sqlite-utils create-table my.db people \
               id integer \
               name text \
-              height float \
+              height real \
               photo blob --pk id
 
-      Valid column types are text, integer, float and blob.
+      Valid column types are text, integer, real, float and blob.
 
     Options:
       --pk TEXT                 Column to use as primary key
@@ -957,6 +1009,67 @@ See :ref:`cli_create_index`.
       -h, --help                 Show this message and exit.
 
 
+.. _cli_ref_drop_index:
+
+drop-index
+==========
+
+See :ref:`cli_drop_index`.
+
+::
+
+    Usage: sqlite-utils drop-index [OPTIONS] PATH TABLE INDEX
+
+      Drop an index by index name from the specified table
+
+      Example:
+
+          sqlite-utils drop-index chickens.db chickens idx_chickens_name
+
+    Options:
+      --ignore               Ignore if index does not exist
+      --load-extension TEXT  Path to SQLite extension, with optional :entrypoint
+      -h, --help             Show this message and exit.
+
+
+.. _cli_ref_migrate:
+
+migrate
+=======
+
+See :ref:`cli_migrate`.
+
+::
+
+    Usage: sqlite-utils migrate [OPTIONS] DB_PATH [MIGRATIONS]...
+
+      Apply pending database migrations.
+
+      Usage:
+
+          sqlite-utils migrate database.db
+
+      This will find the migrations.py file in the current directory or
+      subdirectories and apply any pending migrations.
+
+      Or pass paths to one or more migrations.py files directly:
+
+          sqlite-utils migrate database.db path/to/migrations.py
+
+      Pass --list to see a list of applied and pending migrations without applying
+      them.
+
+      Use --stop-before migration_set:name to stop before a migration. This option
+      can be used multiple times.
+
+    Options:
+      --stop-before TEXT  Stop before applying this migration. Use set:name to
+                          target a migration set.
+      --list              List migrations without running them
+      -v, --verbose       Show verbose output
+      -h, --help          Show this message and exit.
+
+
 .. _cli_ref_enable_fts:
 
 enable-fts
@@ -968,7 +1081,7 @@ See :ref:`cli_fts`.
 
     Usage: sqlite-utils enable-fts [OPTIONS] PATH TABLE COLUMN...
 
-      Enable full-text search for specific table and columns"
+      Enable full-text search for specific table and columns
 
       Example:
 
@@ -1144,7 +1257,7 @@ See :ref:`cli_add_column`.
 ::
 
     Usage: sqlite-utils add-column [OPTIONS] PATH TABLE COL_NAME
-                          [[integer|int|float|real|text|str|blob|bytes]]
+                          [integer|int|float|real|text|str|blob|bytes]
 
       Add a column to the specified table
 

docs/cli.rst

@@ -29,6 +29,16 @@ The ``sqlite-utils query`` command lets you run queries directly against a SQLit
 .. note::
     In Python: :ref:`db.query() <python_api_query>`  CLI reference: :ref:`sqlite-utils query <cli_ref_query>`
 
+Pass ``-`` as the SQL query to read the query from standard input. This is useful for longer queries that would otherwise require careful shell escaping, or for piping in SQL generated by another tool:
+
+.. code-block:: bash
+
+    echo "select * from dogs" | sqlite-utils query dogs.db -
+
+.. code-block:: bash
+
+    sqlite-utils query dogs.db - < query.sql
+
 .. _cli_query_json:
 
 Returning JSON
@@ -45,6 +55,8 @@ The default format returned for queries is JSON:
     [{"id": 1, "age": 4, "name": "Cleo"},
      {"id": 2, "age": 2, "name": "Pancakes"}]
 
+If the query returns more than one column with the same name, later occurrences are renamed with a numeric suffix - ``select 1 as id, 2 as id`` returns ``[{"id": 1, "id_2": 2}]``. This only applies to JSON output: :ref:`CSV and TSV <cli_query_csv>` and :ref:`table <cli_query_table>` output keep the duplicate column headers unchanged.
+
 .. _cli_query_nl:
 
 Newline-delimited JSON
@@ -109,6 +121,33 @@ If you want to pretty-print the output further, you can pipe it through ``python
         }
     ]
 
+.. _cli_query_json_ascii:
+
+Unicode characters in JSON
+~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+JSON output includes unicode characters directly, without escaping them:
+
+.. code-block:: bash
+
+    sqlite-utils dogs.db "select '日本語' as text"
+
+.. code-block:: output
+
+    [{"text": "日本語"}]
+
+Use ``--ascii`` to escape non-ASCII characters as ``\uXXXX`` sequences instead:
+
+.. code-block:: bash
+
+    sqlite-utils dogs.db "select '日本語' as text" --ascii
+
+.. code-block:: output
+
+    [{"text": "\u65e5\u672c\u8a9e"}]
+
+The ``--ascii`` option can help on systems that cannot display or process UTF-8, such as Windows consoles using a legacy code page. On Windows, setting the ``PYTHONUTF8=1`` environment variable is an alternative fix for ``UnicodeEncodeError`` crashes when redirecting output to a file.
+
 .. _cli_query_binary_json:
 
 Binary data in JSON
@@ -257,6 +296,7 @@ Available ``--fmt`` options are:
 .. ]]]
 
 - ``asciidoc``
+- ``colon_grid``
 - ``double_grid``
 - ``double_outline``
 - ``fancy_grid``
@@ -321,7 +361,7 @@ To return the first column of each result as raw data, separated by newlines, us
 Using named parameters
 ----------------------
 
-You can pass named parameters to the query using ``-p``:
+You can pass named parameters to the query using ``-p name value``:
 
 .. code-block:: bash
 
@@ -384,6 +424,9 @@ The ``--functions`` option can be used multiple times to load functions from mul
         from urllib.parse import urlparse
         return urlparse(url).path'
 
+.. note::
+    In Python: :ref:`db.register_function() <python_api_register_function>`
+
 .. _cli_query_extensions:
 
 SQLite extensions
@@ -984,6 +1027,9 @@ To show more than 10 common values, use ``--common-limit 20``.  To skip the most
 
     sqlite-utils analyze-tables github.db tags --common-limit 20 --no-least
 
+.. note::
+    In Python: :ref:`table.analyze_column() <python_api_analyze_column>`  CLI reference: :ref:`sqlite-utils analyze-tables <cli_ref_analyze_tables>`
+
 .. _cli_analyze_tables_save:
 
 Saving the analyzed table details
@@ -1057,6 +1103,37 @@ That will look for SpatiaLite in a set of predictable locations. To load it from
 
     sqlite-utils create-database empty.db --init-spatialite --load-extension /path/to/spatialite.so
 
+.. _cli_migrate:
+
+Running migrations
+==================
+
+The ``migrate`` command applies pending Python migrations to a database. For the full migration file format and Python API, see :ref:`migrations`.
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db path/to/migrations.py
+
+If you omit the migration path it will search the current directory and subdirectories for files called ``migrations.py``:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db
+
+Use ``--list`` to list applied and pending migrations without running them:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db --list
+
+Use ``--stop-before`` to stop before a named migration. The option can be passed more than once, and can target a specific migration set using ``migration_set:migration_name``:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db path/to/migrations.py \
+      --stop-before creatures:add_weight \
+      --stop-before sales:drop_index
+
 .. _cli_inserting_data:
 
 Inserting JSON data
@@ -1120,6 +1197,9 @@ You can delete all the existing rows in the table before inserting the new recor
 
 You can add the ``--analyze`` option to run ``ANALYZE`` against the table after the rows have been inserted.
 
+.. note::
+    In Python: :ref:`table.insert_all() <python_api_bulk_inserts>`  CLI reference: :ref:`sqlite-utils insert <cli_ref_insert>`
+
 .. _cli_inserting_data_binary:
 
 Inserting binary data
@@ -1265,6 +1345,8 @@ A progress bar is displayed when inserting data from a file. You can hide the pr
 
 By default, column types are automatically detected for CSV or TSV files - resulting in a mix of ``TEXT``, ``INTEGER`` and ``REAL`` columns. To disable type detection and treat all columns as ``TEXT``, use the ``--no-detect-types`` option.
 
+Detected types are only applied when the table is created by the command. Inserting CSV or TSV data into a table that already exists leaves the existing column types unchanged - values are inserted using the table's existing schema.
+
 For example, given a ``creatures.csv`` file containing this:
 
 .. code-block::
@@ -1293,6 +1375,25 @@ Will produce this schema with automatically detected types:
        "weight" REAL
     );
 
+.. _cli_insert_csv_tsv_column_types:
+
+Overriding column types
+-----------------------
+
+Use ``--type column-name type`` to override the type automatically chosen when the table is created. This option can be used more than once, and works with both ``insert`` and ``upsert``:
+
+.. code-block:: bash
+
+    sqlite-utils insert places.db places places.csv --csv \
+        --type zipcode text \
+        --type score real
+
+This is useful for values such as ZIP codes, which may look like integers but should be stored as ``TEXT`` to preserve leading zeros.
+
+The column type should be one of ``TEXT``, ``INTEGER``, ``FLOAT``, ``REAL`` or ``BLOB``. Column types are matched case-insensitively.
+
+As with detected column types, ``--type`` only affects tables created by the command. If the table already exists, its existing column types are left unchanged.
+
 To disable type detection and treat all columns as TEXT, use ``--no-detect-types``:
 
 .. code-block:: bash
@@ -1488,6 +1589,27 @@ The result looks like this:
     COMMIT;
 
 
+.. _cli_insert_code:
+
+Inserting rows generated by Python code
+=======================================
+
+Instead of providing a ``FILE`` to import, you can use the ``--code`` option to pass a block of Python code that generates the rows to insert. This is the command-line equivalent of calling ``db["creatures"].insert_all(rows())`` from the :ref:`Python API <python_api>`.
+
+Your code should define either a ``rows()`` function that returns or yields dictionaries, or a ``rows`` iterable such as a list of dictionaries:
+
+.. code-block:: bash
+
+    sqlite-utils insert data.db creatures --code '
+    def rows():
+        yield {"id": 1, "name": "Cleo"}
+        yield {"id": 2, "name": "Suna"}
+    ' --pk id
+
+``--code`` can also be given a path to a Python ``.py`` file.
+
+The ``--code`` option works with both ``sqlite-utils insert`` and ``sqlite-utils upsert``, and composes with table options such as ``--pk``, ``--replace``, ``--alter``, ``--not-null`` and ``--default``. It cannot be combined with a ``FILE`` argument or with input format options such as ``--csv`` or ``--convert``.
+
 .. _cli_insert_replace:
 
 Insert-replacing data
@@ -1502,6 +1624,9 @@ To replace a dog with in ID of 2 with a new record, run the following:
     echo '{"id": 2, "name": "Pancakes", "age": 3}' | \
         sqlite-utils insert dogs.db dogs - --pk=id --replace
 
+.. note::
+    In Python: :ref:`table.insert(..., replace=True) <python_api_insert_replace>`  CLI reference: :ref:`sqlite-utils insert <cli_ref_insert>`
+
 .. _cli_upsert:
 
 Upserting data
@@ -1520,12 +1645,17 @@ For example:
 
 This will update the dog with an ID of 2 to have an age of 4, creating a new record (with a null name) if one does not exist. If a row DOES exist the name will be left as-is.
 
+If the table already exists and has a primary key, you can omit the ``--pk`` option and ``sqlite-utils`` will use that existing primary key.
+
 The command will fail if you reference columns that do not exist on the table. To automatically create missing columns, use the ``--alter`` option.
 
 .. note::
     ``upsert`` in sqlite-utils 1.x worked like ``insert ... --replace`` does in 2.x. See `issue #66 <https://github.com/simonw/sqlite-utils/issues/66>`__ for details of this change.
 
 
+.. note::
+    In Python: :ref:`table.upsert() <python_api_upsert>`  CLI reference: :ref:`sqlite-utils upsert <cli_ref_upsert>`
+
 .. _cli_bulk:
 
 Executing SQL in bulk
@@ -1727,6 +1857,9 @@ You can include named parameters in your where clause and populate them using on
 
 The ``--dry-run`` option will output a preview of the conversion against the first ten rows, without modifying the database.
 
+.. note::
+    In Python: :ref:`table.convert() <python_api_convert>`  CLI reference: :ref:`sqlite-utils convert <cli_ref_convert>`
+
 .. _cli_convert_import:
 
 Importing additional modules
@@ -1849,7 +1982,19 @@ These recipes can be used in the code passed to ``sqlite-utils convert`` like th
     sqlite-utils convert my.db mytable mycolumn \
       'r.jsonsplit(value)'
 
-To use any of the documented parameters, do this:
+You can also pass the recipe function directly without the ``(value)`` part - sqlite-utils will detect that it is a callable and use it automatically:
+
+.. code-block:: bash
+
+    sqlite-utils convert my.db mytable mycolumn r.parsedate
+
+This shorter syntax works for any callable, including functions from imported modules:
+
+.. code-block:: bash
+
+    sqlite-utils convert my.db mytable mycolumn json.loads --import json
+
+To use any of the documented parameters, use the full function call syntax:
 
 .. code-block:: bash
 
@@ -2013,6 +2158,9 @@ If a table with the same name already exists, you will get an error. You can cho
 
 You can also pass ``--transform`` to transform the existing table to match the new schema. See :ref:`python_api_explicit_create` in the Python library documentation for details of how this option works.
 
+.. note::
+    In Python: :ref:`table.create() <python_api_explicit_create>`  CLI reference: :ref:`sqlite-utils create-table <cli_ref_create_table>`
+
 .. _cli_renaming_tables:
 
 Renaming a table
@@ -2026,6 +2174,9 @@ Yo ucan rename a table using the ``rename-table`` command:
 
 Pass ``--ignore`` to ignore any errors caused by the table not existing, or the new name already being in use.
 
+.. note::
+    In Python: :ref:`db.rename_table() <python_api_rename_table>`  CLI reference: :ref:`sqlite-utils rename-table <cli_ref_rename_table>`
+
 .. _cli_duplicate_table:
 
 Duplicating tables
@@ -2037,6 +2188,9 @@ The ``duplicate`` command duplicates a table - creating a new table with the sam
 
     sqlite-utils duplicate books.db authors authors_copy
 
+.. note::
+    In Python: :ref:`table.duplicate() <python_api_duplicate>`  CLI reference: :ref:`sqlite-utils duplicate <cli_ref_duplicate>`
+
 .. _cli_drop_table:
 
 Dropping tables
@@ -2050,12 +2204,15 @@ You can drop a table using the ``drop-table`` command:
 
 Use ``--ignore`` to ignore the error if the table does not exist.
 
+.. note::
+    In Python: :ref:`table.drop() <python_api_drop>`  CLI reference: :ref:`sqlite-utils drop-table <cli_ref_drop_table>`
+
 .. _cli_transform_table:
 
 Transforming tables
 ===================
 
-The ``transform`` command allows you to apply complex transformations to a table that cannot be implemented using a regular SQLite ``ALTER TABLE`` command. See :ref:`python_api_transform` for details of how this works. The ``transform`` command preserves a table's ``STRICT`` mode.
+The ``transform`` command allows you to apply complex transformations to a table that cannot be implemented using a regular SQLite ``ALTER TABLE`` command. See :ref:`python_api_transform` for details of how this works. By default, the ``transform`` command preserves a table's ``STRICT`` mode.
 
 .. code-block:: bash
 
@@ -2101,6 +2258,12 @@ Every option for this table (with the exception of ``--pk-none``) can be specifi
 ``--add-foreign-key column other_table other_column``
     Add a foreign key constraint to ``column`` pointing to ``other_table.other_column``.
 
+``--strict``
+    Convert the table to a `SQLite STRICT table <https://www.sqlite.org/stricttables.html>`__. The command fails if the available SQLite version does not support strict tables. If existing rows contain values that are incompatible with their declared column types the transformation fails and the original table is left unchanged.
+
+``--no-strict``
+    Convert a strict table back to a regular non-strict table.
+
 If you want to see the SQL that will be executed to make the change without actually executing it, add the ``--sql`` flag. For example:
 
 .. code-block:: bash
@@ -2127,6 +2290,9 @@ If you want to see the SQL that will be executed to make the change without actu
     DROP TABLE "roadside_attractions";
     ALTER TABLE "roadside_attractions_new_4033a60276b9" RENAME TO "roadside_attractions";
 
+.. note::
+    In Python: :ref:`table.transform() <python_api_transform>`  CLI reference: :ref:`sqlite-utils transform <cli_ref_transform>`
+
 .. _cli_transform_table_add_primary_key_to_rowid:
 
 Adding a primary key to a rowid table
@@ -2213,6 +2379,8 @@ The ``sqlite-utils extract`` command can be used to extract specified columns in
 
 Take a look at the Python API documentation for :ref:`python_api_extract` for a detailed description of how this works, including examples of table schemas before and after running an extraction operation.
 
+Rows where every extracted column is ``null`` are not extracted - those rows get a ``null`` value in their new foreign key column and no record is created for them in the lookup table.
+
 The command takes a database, table and one or more columns that should be extracted. To extract the ``species`` column from the ``trees`` table you would run:
 
 .. code-block:: bash
@@ -2302,6 +2470,9 @@ After running the above, the command ``sqlite-utils schema global.db`` reveals t
     CREATE UNIQUE INDEX "idx_countries_country_name"
         ON "countries" ("country", "name");
 
+.. note::
+    In Python: :ref:`table.extract() <python_api_extract>`  CLI reference: :ref:`sqlite-utils extract <cli_ref_extract>`
+
 .. _cli_create_view:
 
 Creating views
@@ -2323,6 +2494,9 @@ You can create a view using the ``create-view`` command:
 
 Use ``--replace`` to replace an existing view of the same name, and ``--ignore`` to do nothing if a view already exists.
 
+.. note::
+    In Python: :ref:`db.create_view() <python_api_create_view>`  CLI reference: :ref:`sqlite-utils create-view <cli_ref_create_view>`
+
 .. _cli_drop_view:
 
 Dropping views
@@ -2336,6 +2510,9 @@ You can drop a view using the ``drop-view`` command:
 
 Use ``--ignore`` to ignore the error if the view does not exist.
 
+.. note::
+    In Python: :ref:`view.drop() <python_api_drop>`  CLI reference: :ref:`sqlite-utils drop-view <cli_ref_drop_view>`
+
 .. _cli_add_column:
 
 Adding columns
@@ -2376,6 +2553,9 @@ You can set a ``NOT NULL DEFAULT 'x'`` constraint on the new column using ``--no
 
     sqlite-utils add-column mydb.db dogs friends_count integer --not-null-default 0
 
+.. note::
+    In Python: :ref:`table.add_column() <python_api_add_column>`  CLI reference: :ref:`sqlite-utils add-column <cli_ref_add_column>`
+
 .. _cli_add_column_alter:
 
 Adding columns automatically on insert/update
@@ -2387,6 +2567,9 @@ You can use the ``--alter`` option to automatically add new columns if the data
 
     sqlite-utils insert dogs.db dogs new-dogs.json --pk=id --alter
 
+.. note::
+    In Python: :ref:`table.insert(..., alter=True) <python_api_add_column_alter>`
+
 .. _cli_add_foreign_key:
 
 Adding foreign key constraints
@@ -2414,6 +2597,9 @@ Add ``--ignore`` to ignore an existing foreign key (as opposed to returning an e
 
 See :ref:`python_api_add_foreign_key` in the Python API documentation for further details, including how the automatic table guessing mechanism works.
 
+.. note::
+    In Python: :ref:`table.add_foreign_key() <python_api_add_foreign_key>`  CLI reference: :ref:`sqlite-utils add-foreign-key <cli_ref_add_foreign_key>`
+
 .. _cli_add_foreign_keys:
 
 Adding multiple foreign keys at once
@@ -2429,6 +2615,9 @@ Adding a foreign key requires a ``VACUUM``. On large databases this can be an ex
 
 When you are using this command each foreign key needs to be defined in full, as four arguments - the table, column, other table and other column.
 
+.. note::
+    In Python: :ref:`db.add_foreign_keys() <python_api_add_foreign_keys>`  CLI reference: :ref:`sqlite-utils add-foreign-keys <cli_ref_add_foreign_keys>`
+
 .. _cli_index_foreign_keys:
 
 Adding indexes for all foreign keys
@@ -2440,6 +2629,9 @@ If you want to ensure that every foreign key column in your database has a corre
 
     sqlite-utils index-foreign-keys books.db
 
+.. note::
+    In Python: :ref:`db.index_foreign_keys() <python_api_index_foreign_keys>`  CLI reference: :ref:`sqlite-utils index-foreign-keys <cli_ref_index_foreign_keys>`
+
 .. _cli_defaults_not_null:
 
 Setting defaults and not null constraints
@@ -2455,6 +2647,9 @@ You can use the ``--not-null`` and ``--default`` options (to both ``insert`` and
         --default age 2 \
         --default score 5
 
+.. note::
+    In Python: :ref:`not_null= and defaults= arguments <python_api_defaults_not_null>`
+
 .. _cli_create_index:
 
 Creating indexes
@@ -2486,6 +2681,25 @@ If your column names are already prefixed with a hyphen you'll need to manually
 
 Add the ``--analyze`` option to run ``ANALYZE`` against the index after it has been created.
 
+.. note::
+    In Python: :ref:`table.create_index() <python_api_create_index>`  CLI reference: :ref:`sqlite-utils create-index <cli_ref_create_index>`
+
+.. _cli_drop_index:
+
+Dropping indexes
+================
+
+You can drop an index from an existing table using the ``drop-index`` command:
+
+.. code-block:: bash
+
+    sqlite-utils drop-index mydb.db mytable idx_mytable_col1
+
+Use ``--ignore`` to ignore the error if the index does not exist on that table.
+
+.. note::
+    In Python: :ref:`table.drop_index() <python_api_create_index>`  CLI reference: :ref:`sqlite-utils drop-index <cli_ref_drop_index>`
+
 .. _cli_fts:
 
 Configuring full-text search
@@ -2539,6 +2753,9 @@ You can rebuild every FTS table by running ``rebuild-fts`` without passing any t
 
     sqlite-utils rebuild-fts mydb.db
 
+.. note::
+    In Python: :ref:`table.enable_fts() <python_api_fts_enable>`  CLI reference: :ref:`sqlite-utils enable-fts <cli_ref_enable_fts>`
+
 .. _cli_search:
 
 Executing searches
@@ -2595,6 +2812,9 @@ Use the ``--sql`` option to output the SQL that would be executed, rather than r
     order by
         "documents_fts".rank
 
+.. note::
+    In Python: :ref:`table.search() <python_api_fts_search>`  CLI reference: :ref:`sqlite-utils search <cli_ref_search>`
+
 .. _cli_enable_counts:
 
 Enabling cached counts
@@ -2618,6 +2838,9 @@ If the ``_counts`` table ever becomes out-of-sync with the actual table counts y
 
     sqlite-utils reset-counts mydb.db
 
+.. note::
+    In Python: :ref:`table.enable_counts() <python_api_cached_table_counts>`  CLI reference: :ref:`sqlite-utils enable-counts <cli_ref_enable_counts>`
+
 .. _cli_analyze:
 
 Optimizing index usage with ANALYZE
@@ -2641,6 +2864,9 @@ You can run it against specific tables, or against specific named indexes, by pa
 
 You can also run ``ANALYZE`` as part of another command using the ``--analyze`` option. This is supported by the ``create-index``, ``insert`` and ``upsert`` commands.
 
+.. note::
+    In Python: :ref:`db.analyze() <python_api_analyze>`  CLI reference: :ref:`sqlite-utils analyze <cli_ref_analyze>`
+
 .. _cli_vacuum:
 
 Vacuum
@@ -2652,6 +2878,9 @@ You can run VACUUM to optimize your database like so:
 
     sqlite-utils vacuum mydb.db
 
+.. note::
+    In Python: :ref:`db.vacuum() <python_api_vacuum>`  CLI reference: :ref:`sqlite-utils vacuum <cli_ref_vacuum>`
+
 .. _cli_optimize:
 
 Optimize
@@ -2675,6 +2904,9 @@ To optimize specific tables rather than every FTS table, pass those tables as ex
 
     sqlite-utils optimize mydb.db table_1 table_2
 
+.. note::
+    In Python: :ref:`table.optimize() <python_api_fts_optimize>`  CLI reference: :ref:`sqlite-utils optimize <cli_ref_optimize>`
+
 .. _cli_wal:
 
 WAL mode
@@ -2694,6 +2926,9 @@ You can disable WAL mode using ``disable-wal``:
 
 Both of these commands accept one or more database files as arguments.
 
+.. note::
+    In Python: :ref:`db.enable_wal() and db.disable_wal() <python_api_wal>`  CLI reference: :ref:`sqlite-utils enable-wal <cli_ref_enable_wal>`
+
 .. _cli_dump:
 
 Dumping the database to SQL
@@ -2709,6 +2944,9 @@ The ``dump`` command outputs a SQL dump of the schema and full contents of the s
     ...
     COMMIT;
 
+.. note::
+    In Python: :ref:`db.iterdump() <python_api_itedump>`  CLI reference: :ref:`sqlite-utils dump <cli_ref_dump>`
+
 .. _cli_load_extension:
 
 Loading SQLite extensions
@@ -2756,6 +2994,9 @@ Eight (case-insensitive) types are allowed:
 * GEOMETRYCOLLECTION
 * GEOMETRY
 
+.. note::
+    In Python: :ref:`table.add_geometry_column() <python_api_gis_add_geometry_column>`  CLI reference: :ref:`sqlite-utils add-geometry-column <cli_ref_add_geometry_column>`
+
 .. _cli_spatialite_indexes:
 
 Adding spatial indexes
@@ -2769,6 +3010,9 @@ Once you have a geometry column, you can speed up bounding box queries by adding
 
 See this `SpatiaLite Cookbook recipe <http://www.gaia-gis.it/gaia-sins/spatialite-cookbook-5/cookbook_topics.03.html#topic_Wonderful_RTree_Spatial_Index>`__ for examples of how to use a spatial index.
 
+.. note::
+    In Python: :ref:`table.create_spatial_index() <python_api_gis_create_spatial_index>`  CLI reference: :ref:`sqlite-utils create-spatial-index <cli_ref_create_spatial_index>`
+
 .. _cli_install:
 
 Installing packages

docs/conf.py

@@ -1,8 +1,10 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 
-from subprocess import Popen, PIPE
-from beanbag_docutils.sphinx.ext.github import github_linkcode_resolve
+import inspect
+from pathlib import Path
+from subprocess import Popen, PIPE, check_output
+import sys
 
 # This file is execfile()d with the current directory set to its
 # containing dir.
@@ -45,14 +47,52 @@ extlinks = {
 }
 
 
+def _linkcode_git_ref():
+    try:
+        return check_output(["git", "rev-parse", "HEAD"]).decode("utf8").strip()
+    except Exception:
+        return "main"
+
+
 def linkcode_resolve(domain, info):
-    return github_linkcode_resolve(
-        domain=domain,
-        info=info,
-        allowed_module_names=["sqlite_utils"],
-        github_org_id="simonw",
-        github_repo_id="sqlite-utils",
-        branch="main",
+    if domain != "py":
+        return None
+
+    module_name = info.get("module")
+    if not module_name or module_name.split(".")[0] != "sqlite_utils":
+        return None
+
+    module = sys.modules.get(module_name)
+    if module is None:
+        return None
+
+    obj = module
+    for part in info.get("fullname", "").split("."):
+        obj = getattr(obj, part, None)
+        if obj is None:
+            return None
+
+    if isinstance(obj, property):
+        obj = obj.fget
+
+    try:
+        obj = inspect.unwrap(obj)
+        source_file = inspect.getsourcefile(obj)
+        _, line_number = inspect.getsourcelines(obj)
+    except Exception:
+        return None
+
+    if source_file is None:
+        return None
+
+    try:
+        filename = Path(source_file).resolve().relative_to(Path(__file__).parent.parent)
+    except ValueError:
+        return None
+
+    return (
+        "https://github.com/simonw/sqlite-utils/blob/"
+        f"{_linkcode_git_ref()}/{filename}#L{line_number}"
     )
 
 
@@ -79,7 +119,7 @@ author = "Simon Willison"
 #
 # The short X.Y version.
 pipe = Popen("git describe --tags --always", stdout=PIPE, shell=True)
-git_version = pipe.stdout.read().decode("utf8")
+git_version = pipe.stdout.read().decode("utf8") if pipe.stdout else ""
 
 if git_version:
     version = git_version.rsplit("-", 1)[0]

docs/contributing.rst

@@ -13,78 +13,61 @@ All improvements to the software should start with an issue. Read `How I build a
 Obtaining the code
 ==================
 
-To work on this library locally, first checkout the code. Then create a new virtual environment::
+To work on this library locally, first checkout the code::
 
     git clone git@github.com:simonw/sqlite-utils
     cd sqlite-utils
-    python3 -mvenv venv
-    source venv/bin/activate
 
-Or if you are using ``pipenv``::
+Use ``uv run`` to run the development version of the tool::
 
-    pipenv shell
-
-Within the virtual environment running ``sqlite-utils`` should run your locally editable version of the tool. You can use ``which sqlite-utils`` to confirm that you are running the version that lives in your virtual environment.
+    uv run sqlite-utils --help
 
 .. _contributing_tests:
 
 Running the tests
 =================
 
-To install the dependencies and test dependencies::
+Use ``uv run`` to run the tests::
 
-    pip install -e '.[test]'
-
-To run the tests::
-
-    pytest
+    uv run pytest
 
 .. _contributing_docs:
 
 Building the documentation
 ==========================
 
-To build the documentation, first install the documentation dependencies::
+To build the documentation run this command::
 
-    pip install -e '.[docs]'
+    uv run make livehtml --directory docs
 
-Then run ``make livehtml`` from the ``docs/`` directory to start a server on port 8000 that will serve the documentation and live-reload any time you make an edit to a ``.rst`` file::
-
-    cd docs
-    make livehtml
+This will start a server on port 8000 that will serve the documentation and live-reload any time you make an edit to a ``.rst`` file.
 
 The `cog <https://github.com/nedbat/cog>`__ tool is used to maintain portions of the documentation. You can run it like so::
 
-    cog -r docs/*.rst
+    uv run cog -r docs/*.rst
 
 .. _contributing_linting:
 
 Linting and formatting
 ======================
 
-``sqlite-utils`` uses `Black <https://black.readthedocs.io/>`__ for code formatting, and `flake8 <https://flake8.pycqa.org/>`__ and `mypy <https://mypy.readthedocs.io/>`__ for linting and type checking.
+``sqlite-utils`` uses `Black <https://black.readthedocs.io/>`__ for code formatting, and `flake8 <https://flake8.pycqa.org/>`__ and `mypy <https://mypy.readthedocs.io/>`__ for linting and type checking::
 
-Black is installed as part of ``pip install -e '.[test]'`` - you can then format your code by running it in the root of the project::
+    uv run black .
 
-    black .
+Linting tools can be run like this::
 
-To install ``mypy`` and ``flake8`` run the following::
-
-    pip install -e '.[flake8,mypy]'
-
-Both commands can then be run in the root of the project like this::
-
-    flake8
-    mypy sqlite_utils
+    uv run flake8
+    uv run mypy sqlite_utils
 
 All three of these tools are run by our CI mechanism against every commit and pull request.
 
 .. _contributing_just:
 
-Using Just and uv
-=================
+Using Just
+==========
 
-If you install `Just <https://github.com/casey/just>`__ and `uv <https://github.com/astral-sh/uv>`__ you can use them to manage your local development environment.
+If you install `Just <https://github.com/casey/just>`__ you can use it to manage your local development environment.
 
 To run all of the tests and linters::
 

docs/index.rst

@@ -36,8 +36,10 @@ Contents
    installation
    cli
    python-api
+   migrations
    plugins
    reference
    cli-reference
+   upgrading
    contributing
    changelog

docs/installation.rst

@@ -52,15 +52,15 @@ On some platforms the ability to load additional extensions (via ``conn.load_ext
 
 You may also see the error ``sqlite3.OperationalError: table sqlite_master may not be modified`` when trying to alter an existing table.
 
-You can work around these limitations by installing either the `pysqlite3 <https://pypi.org/project/pysqlite3/>`__ package or the `sqlean.py <https://pypi.org/project/sqlean.py/>`__ package, both of which provide drop-in replacements for the standard library ``sqlite3`` module but with a recent version of SQLite and full support for loading extensions.
+You can work around these limitations by installing the `pysqlite3 <https://pypi.org/project/pysqlite3/>`__ package, which provides a drop-in replacement for the standard library ``sqlite3`` module but with a recent version of SQLite and full support for loading extensions.
 
-To install ``sqlean.py`` (which has compiled binary wheels available for all major platforms) run the following:
+To install ``pysqlite3`` run the following:
 
 .. code-block:: bash
 
-    sqlite-utils install sqlean.py
+    sqlite-utils install pysqlite3
 
-``pysqlite3`` and ``sqlean.py`` do not provide implementations of the ``.iterdump()`` method. To use that method (see :ref:`python_api_itedump`) or the ``sqlite-utils dump`` command you should also install the ``sqlite-dump`` package:
+``pysqlite3`` does not provide an implementation of the ``.iterdump()`` method. To use that method (see :ref:`python_api_itedump`) or the ``sqlite-utils dump`` command you should also install the ``sqlite-dump`` package:
 
 .. code-block:: bash
 
@@ -87,4 +87,4 @@ For ``zsh``:
 
 Add this code to ``~/.zshrc`` or ``~/.bashrc`` to automatically run it when you start a new shell.
 
-See `the Click documentation <https://click.palletsprojects.com/en/8.1.x/shell-completion/>`__ for more details.
+See `the Click documentation <https://click.palletsprojects.com/en/8.1.x/shell-completion/>`__ for more details.

docs/migrations.rst

@@ -0,0 +1,194 @@
+.. _migrations:
+
+=====================
+ Database migrations
+=====================
+
+``sqlite-utils`` includes a migration system for applying repeatable changes to SQLite database files.
+
+A migration is a Python function that receives a :class:`sqlite_utils.Database` instance and then executes Python code to modify that database - creating or transforming tables, adding indexes, inserting rows, or any other operation supported by SQLite.
+
+Migrations are grouped into named sets using the :class:`sqlite_utils.Migrations` class, and each applied migration is recorded in the ``_sqlite_migrations`` table in that database.
+
+This means you can run the migrate operation multiple times and it will only apply migrations that have not previously been recorded.
+
+.. _migrations_define:
+
+Defining migrations
+===================
+
+Ordered migration sets are defined by first creating a :class:`sqlite_utils.Migrations` object.
+
+Individual migrations are Python functions that are then registered with that migration set. Each migration function is passed a single argument that is a :ref:`sqlite_utils.Database <reference_db_database>` instance.
+
+The name passed to ``Migrations("creatures")`` identifies that set of migrations. Use a name that is unique for your project, since multiple migration sets can be applied to the same database.
+
+Here is a simple example of a ``migrations.py`` file which creates a table, then adds an extra column to that table in a second migration:
+
+.. code-block:: python
+
+    from sqlite_utils import Migrations
+
+    migrations = Migrations("creatures")
+
+    @migrations()
+    def create_table(db):
+        db["creatures"].create(
+            {"id": int, "name": str, "species": str},
+            pk="id",
+        )
+
+    @migrations()
+    def add_weight(db):
+        db["creatures"].add_column("weight", float)
+
+.. _migrations_python:
+
+Applying migrations in Python
+=============================
+
+Once you have a ``Migrations(name)`` collection with one or more migrations registered to it, you can execute them in Python code like this:
+
+.. code-block:: python
+
+    from sqlite_utils import Database
+
+    db = Database("creatures.db")
+    migrations.apply(db)
+
+Running ``migrations.apply(db)`` repeatedly is safe. Migrations that already have a matching ``migration_set`` and ``name`` row in ``_sqlite_migrations`` will be skipped.
+
+Migration functions are applied in the order that they were registered. The function name is used as the migration name unless you pass one explicitly:
+
+.. code-block:: python
+
+    @migrations(name="001_create_table")
+    def create_table(db):
+        db["creatures"].create({"id": int, "name": str}, pk="id")
+
+When you apply a set of migrations you can stop part way through by specifying a ``stop_before=`` migration name:
+
+.. code-block:: python
+
+    migrations.apply(db, stop_before="add_weight")
+
+.. _migrations_transactions:
+
+Migrations and transactions
+===========================
+
+Each migration runs inside a transaction, together with the ``_sqlite_migrations`` record of it having been applied. If a migration function raises an exception, everything it did is rolled back, no record is written and the migration stays pending - so fixing the error and re-applying will run that migration again from a clean state. Migrations that completed earlier in the same ``apply()`` run stay applied.
+
+Some operations cannot run inside a transaction, for example ``VACUUM`` or changing the journal mode with ``db.enable_wal()``. Register migrations like these with ``transactional=False``:
+
+.. code-block:: python
+
+    @migrations(transactional=False)
+    def compact(db):
+        db.execute("VACUUM")
+
+A migration registered with ``transactional=False`` runs without a wrapping transaction, so if it fails part way through any changes it already made will not be rolled back, and re-applying will run the whole function again.
+
+Avoid calling ``db.commit()`` or otherwise managing transactions manually inside a transactional migration - register the migration with ``transactional=False`` if it needs to control its own transactions. Using ``with db.atomic():`` blocks inside a migration is fine: they nest as savepoints within the migration's transaction, so the migration as a whole still commits or rolls back as a single unit. See :ref:`python_api_transactions`.
+
+Applying migrations using the CLI
+=================================
+
+Run migrations using the ``sqlite-utils migrate`` command:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db path/to/migrations.py
+
+The first argument is the database file. The remaining arguments can be paths to migration files or directories containing migration files.
+
+If you omit migration paths, ``sqlite-utils`` searches the current directory and subdirectories for files called ``migrations.py``:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db
+
+You can also pass a directory. Every ``migrations.py`` file in that directory tree will be considered:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db path/to/project/
+
+Running the command repeatedly is safe. Migrations that already have a matching ``migration_set`` and ``name`` row in ``_sqlite_migrations`` will be skipped.
+
+Listing migrations
+==================
+
+Use ``--list`` to show applied and pending migrations without running them. This is a read-only operation - it will not create the database file or the ``_sqlite_migrations`` table:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db --list
+
+Example output:
+
+.. code-block:: output
+
+    Migrations for: creatures
+
+      Applied:
+        create_table - 2026-06-09 17:23:12.048092+00:00
+        add_weight - 2026-06-09 17:23:12.051249+00:00
+
+      Pending:
+        add_age
+
+Stopping before a migration
+===========================
+
+When applying migrations using the CLI, you can stop before a named migration:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db path/to/migrations.py --stop-before add_weight
+
+This applies any pending migrations before ``add_weight`` and leaves ``add_weight`` and later migrations pending. An unqualified migration name matches in any migration set.
+
+You can also target a specific migration set using ``migration_set:migration_name``. This is useful if a migrations file contains more than one migration set, or if multiple sets use the same migration name:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db path/to/migrations.py \
+      --stop-before creatures:add_weight \
+      --stop-before sales:drop_index
+
+The ``--stop-before`` option can be passed more than once.
+
+If a ``--stop-before`` value does not match any known migration the command exits with an error, rather than silently applying everything. Naming a migration that has already been applied is also an error - stopping before it is impossible to honor - and no pending migrations are applied.
+
+Verbose output
+==============
+
+Use ``--verbose`` or ``-v`` to show the schema before and after migrations are applied, plus a unified diff when the schema changes:
+
+.. code-block:: bash
+
+    sqlite-utils migrate creatures.db --verbose
+
+Migrating from sqlite-migrate
+=============================
+
+This system uses the same migration table format as the older `sqlite-migrate <https://github.com/simonw/sqlite-migrate>`__ package. To use existing migration files directly with ``sqlite-utils``, update their import from ``sqlite_migrate`` to ``sqlite_utils``:
+
+.. code-block:: python
+
+    from sqlite_utils import Migrations
+
+    migration = Migrations("creatures")
+
+    @migration()
+    def create_table(db):
+        db["creatures"].create({"id": int, "name": str}, pk="id")
+
+Python API
+==========
+
+.. autoclass:: sqlite_utils.migrations.Migrations
+   :members:
+   :undoc-members:
+   :exclude-members: _Migration, _AppliedMigration

docs/plugins.rst

@@ -140,9 +140,7 @@ As a special niche feature, if your plugin needs to import some files and then a
 prepare_connection(conn)
 ~~~~~~~~~~~~~~~~~~~~~~~~
 
-This hook is called when a new SQLite database connection is created. You can
-use it to `register custom SQL functions <https://docs.python.org/2/library/sqlite3.html#sqlite3.Connection.create_function>`_,
-aggregates and collations. For example:
+This hook is called when a new SQLite database connection is created. You can use it to `register custom SQL functions <https://docs.python.org/2/library/sqlite3.html#sqlite3.Connection.create_function>`_, aggregates and collations. For example:
 
 .. code-block:: python
 
@@ -154,8 +152,7 @@ aggregates and collations. For example:
             "hello", 1, lambda name: f"Hello, {name}!"
         )
 
-This registers a SQL function called ``hello`` which takes a single
-argument and can be called like this:
+This registers a SQL function called ``hello`` which takes a single argument and can be called like this:
 
 .. code-block:: sql
 

docs/python-api.rst

@@ -33,6 +33,8 @@ Here's how to create a new SQLite database file containing a new ``chickens`` ta
         "color": "black",
     }])
 
+The inserted rows are saved to the database file straight away - methods like ``insert_all()`` commit their own changes, so no ``commit()`` call is needed. See :ref:`python_api_transactions` for how this works.
+
 You can loop through those rows like this:
 
 .. code-block:: python
@@ -93,6 +95,8 @@ Instead of a file path you can pass in an existing SQLite connection:
 
     db = Database(sqlite3.connect("my_database.db"))
 
+The connection must use Python's default transaction handling. Connections created with the Python 3.12+ ``sqlite3.connect(..., autocommit=True)`` or ``autocommit=False`` options are rejected with a ``sqlite_utils.db.TransactionError`` - see :ref:`python_api_transactions_modes`.
+
 If you want to create an in-memory database, you can do so like this:
 
 .. code-block:: python
@@ -105,6 +109,14 @@ You can also create a named in-memory database. Unlike regular memory databases
 
     db = Database(memory_name="my_shared_database")
 
+After creating a ``Database`` you can use ``db.memory`` and ``db.memory_name`` to tell whether it is backed by an in-memory database and to read the shared cache name. ``db.memory`` is ``True`` for any in-memory database and ``db.memory_name`` holds the name passed to ``memory_name=``, or ``None`` otherwise.
+
+.. code-block:: python
+
+    db = Database(memory_name="shared")
+    db.memory       # True
+    db.memory_name  # "shared"
+
 Connections use ``PRAGMA recursive_triggers=on`` by default. If you don't want to use `recursive triggers <https://www.sqlite.org/pragma.html#pragma_recursive_triggers>`__ you can turn them off using:
 
 .. code-block:: python
@@ -123,6 +135,33 @@ You can pass ``strict=True`` to enable `SQLite STRICT mode <https://www.sqlite.o
 
     db = Database("my_database.db", strict=True)
 
+.. _python_api_close:
+
+Closing a database
+------------------
+
+Database objects maintain a connection to the underlying SQLite database. You can explicitly close this connection using the ``.close()`` method:
+
+.. code-block:: python
+
+    db = Database("my_database.db")
+    # ... use the database ...
+    db.close()
+
+The ``Database`` object also works as a context manager, which will automatically close the connection when the ``with`` block exits:
+
+.. code-block:: python
+
+    with Database("my_database.db") as db:
+        db["my_table"].insert({"name": "Example"})
+    # Connection is automatically closed here
+
+Exiting the block is equivalent to calling ``db.close()``: the connection is closed and any transaction still open at that point is rolled back. This matches SQLite's own behavior when a connection closes.
+
+This rarely matters in practice. Everything that writes to the database - including raw ``db.execute()`` statements - commits automatically, so a transaction can only be open here if you explicitly started one with ``db.begin()`` and have not yet committed it. In that case the decision to commit stays with you: committing automatically on exit could silently persist half-finished work, for example if your code returned early from the block. Call ``db.commit()`` when the work is complete.
+
+Note this differs from the ``sqlite3.Connection`` context manager in the standard library, which commits on success but does not close the connection. See :ref:`python_api_transactions` for the full transaction model.
+
 .. _python_api_attach:
 
 Attaching additional databases
@@ -145,6 +184,9 @@ You can attach an additional database using the ``.attach()`` method, providing
 
 You can reference tables in the attached database using the alias value you passed to ``db.attach(alias, filepath)`` as a prefix, for example the ``second.table_in_second`` reference in the SQL query above.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils --attach <cli_query_attach>`
+
 .. _python_api_tracing:
 
 Tracing queries
@@ -198,6 +240,26 @@ The ``db.query(sql)`` function executes a SQL query and returns an iterator over
     # {'name': 'Cleo'}
     # {'name': 'Pancakes'}
 
+The SQL query is executed as soon as ``db.query()`` is called. The resulting rows are fetched lazily as you iterate, so large result sets are not loaded into memory all at once. Because execution is immediate, an error in your SQL will raise an exception straight away, and a statement such as ``INSERT ... RETURNING`` will take effect - and be committed, unless a transaction is open - even if you do not iterate over its results.
+
+``db.query()`` can only be used with SQL that returns rows. Passing a statement that returns no rows - an ``INSERT`` or ``UPDATE`` without a ``RETURNING`` clause, for example - will raise a ``ValueError``. The rejected statement is rolled back, so it has no effect on the database. Use :ref:`db.execute() <python_api_execute>` for those statements instead.
+
+There is one exception to the rolled-back guarantee: a ``PRAGMA`` statement that returns no rows, such as ``PRAGMA user_version = 5``, still raises a ``ValueError`` but will already have taken effect. Some PRAGMA statements refuse to run inside a transaction, so PRAGMAs are executed outside the savepoint that is used to roll back other rejected statements. Use ``db.execute()`` for PRAGMA statements that do not return rows.
+
+If a query returns more than one column with the same name - a join between two tables that share column names, for example - later occurrences are renamed with a numeric suffix, so every value is included in the dictionary:
+
+.. code-block:: python
+
+    row = next(db.query("select 1 as id, 2 as id, 3 as id"))
+    print(row)
+    # Outputs:
+    # {'id': 1, 'id_2': 2, 'id_3': 3}
+
+A suffix that would collide with another column in the query is skipped - ``select 1 as id, 2 as id, 3 as id_2`` returns ``{'id': 1, 'id_3': 2, 'id_2': 3}``. The same renaming is applied by ``table.rows_where()`` and ``table.search()``.
+
+.. note::
+    In the CLI: :ref:`sqlite-utils query <cli_query>`
+
 .. _python_api_execute:
 
 db.execute(sql, params)
@@ -218,6 +280,9 @@ The ``db.execute()`` and ``db.executescript()`` methods provide wrappers around
 
 Other cursor methods such as ``.fetchone()`` and ``.fetchall()`` are also available, see the `standard library documentation <https://docs.python.org/3/library/sqlite3.html#sqlite3.Cursor>`__.
 
+.. note::
+    Write statements executed this way are committed automatically, unless a transaction is already open in which case they become part of it - see :ref:`python_api_transactions_execute`.
+
 .. _python_api_parameters:
 
 Passing parameters
@@ -246,6 +311,146 @@ Named parameters using ``:name`` can be filled using a dictionary:
 
 In this example ``next()`` is used to retrieve the first result in the iterator returned by the ``db.query()`` method.
 
+.. _python_api_transactions:
+
+Transactions and saving your changes
+====================================
+
+Every method in this library that writes to the database - ``insert()``, ``upsert()``, ``update()``, ``delete()``, ``delete_where()``, ``transform()``, ``create_table()``, ``create_index()``, ``enable_fts()`` and the rest - runs inside its own transaction and commits it before returning. Your changes are saved to disk as soon as the method call finishes:
+
+.. code-block:: python
+
+    db = Database("data.db")
+    db.table("news").insert({"headline": "Dog wins award"})
+    # The new row is already saved - no commit() required
+
+The same applies to raw SQL executed with :ref:`db.execute() <python_api_transactions_execute>` - a write statement is committed as soon as it has run.
+
+Another way to think about this is that each sqlite-utils method call is its own unit of work. If several method calls must either all succeed or all fail, use ``db.atomic()`` to turn them into a single unit of work.
+
+You never need to call ``commit()``, and you do not need to close the database to persist your changes. There are exactly two situations where you need to think about transactions:
+
+1. You want to group several write operations together, so they either all succeed or all fail - use :ref:`db.atomic() <python_api_atomic>`.
+2. You are :ref:`managing a transaction yourself <python_api_transactions_manual>` with ``db.begin()``, in which case nothing is committed until you commit - the library will never commit a transaction you opened.
+
+``with Database(...) as db:`` is not a transaction block. It manages the lifetime of the database connection and closes it on exit. Use ``with db.atomic():`` for a transaction.
+
+.. _python_api_atomic:
+
+Grouping changes with db.atomic()
+---------------------------------
+
+Use ``db.atomic()`` to group multiple operations in a single transaction:
+
+.. code-block:: python
+
+    with db.atomic():
+        db.table("dogs").insert({"id": 1, "name": "Cleo"}, pk="id")
+        db.table("dogs").insert({"id": 2, "name": "Pancakes"})
+
+The transaction commits when the block exits. If an exception is raised, changes made inside the block will be rolled back.
+
+This matters when several operations represent a single logical change. Without ``db.atomic()``, an earlier method call remains committed if a later one fails:
+
+.. code-block:: python
+
+    # These are two separate transactions
+    db.table("accounts").update(1, {"balance": 90})
+    db.table("accounts").update(2, {"balance": 110})
+
+    # These updates either both succeed or both fail
+    with db.atomic():
+        db.table("accounts").update(1, {"balance": 90})
+        db.table("accounts").update(2, {"balance": 110})
+
+Transactions can also improve performance. Calling ``insert()`` repeatedly outside ``db.atomic()`` creates and commits a separate transaction for every call. For bulk inserts, prefer :ref:`insert_all() <python_api_bulk_inserts>`. If you need to call several different methods in a loop, wrap the loop in ``db.atomic()``:
+
+.. code-block:: python
+
+    with db.atomic():
+        for row in rows:
+            db.table("events").insert(row)
+
+``db.atomic()`` can be nested. Nested blocks use SQLite savepoints, so an exception in an inner block can roll back to that savepoint without rolling back the entire outer transaction:
+
+.. code-block:: python
+
+    with db.atomic():
+        db.table("dogs").insert({"id": 1, "name": "Cleo"}, pk="id")
+        try:
+            with db.atomic():
+                db.table("dogs").insert({"id": 2, "name": "Pancakes"})
+                raise ValueError("skip this one")
+        except ValueError:
+            pass
+        db.table("dogs").insert({"id": 3, "name": "Marnie"})
+
+The transaction is opened with a deferred ``BEGIN`` - SQLite takes the necessary locks when the first statement inside the block runs.
+
+.. _python_api_transactions_execute:
+
+Raw SQL writes with db.execute()
+--------------------------------
+
+Write statements executed with :ref:`db.execute() <python_api_execute>` follow the same rule as everything else: they are committed automatically as soon as they have run.
+
+.. code-block:: python
+
+    db.execute("insert into news (headline) values (?)", ["Dog wins award"])
+    # Already committed
+
+``db.execute()`` participates in sqlite-utils transaction handling. Calling ``db.conn.execute()`` directly bypasses that policy and leaves transaction handling to Python's underlying ``sqlite3.Connection``. Prefer ``db.execute()`` unless you deliberately need the lower-level API.
+
+If a transaction is open - because the call happens inside a ``db.atomic()`` block, or after ``db.begin()`` - the statement becomes part of that transaction instead, and commits when the transaction commits:
+
+.. code-block:: python
+
+    with db.atomic():
+        db.execute("insert into news (headline) values (?)", ["Dog wins award"])
+        db.execute("insert into news (headline) values (?)", ["Cat unimpressed"])
+    # Both rows committed together
+
+One corner case: a row-returning write such as ``INSERT ... RETURNING`` executed through ``db.execute()`` cannot be auto-committed, because its rows have not been read yet - call ``db.commit()`` after fetching them, or use :ref:`db.query() <python_api_query>` for those statements, which executes the write and commits it immediately.
+
+.. _python_api_transactions_manual:
+
+Managing transactions yourself
+------------------------------
+
+You can take full manual control using the ``db.begin()``, ``db.commit()`` and ``db.rollback()`` methods:
+
+.. code-block:: python
+
+    db.begin()
+    db.table("news").insert({"headline": "Dog wins award"})
+    if all_looks_good:
+        db.commit()
+    else:
+        db.rollback()
+
+``db.begin()`` raises ``sqlite3.OperationalError`` if a transaction is already open. ``db.commit()`` and ``db.rollback()`` do nothing if there is no open transaction.
+
+The library will never commit a transaction you opened. If you call write methods such as ``insert()`` - or use ``db.atomic()`` - while your transaction is open, they participate in it using SQLite savepoints instead of committing: exiting an ``atomic()`` block releases its savepoint, but nothing is saved to disk until you commit the outer transaction yourself. If you roll back, their changes are rolled back too.
+
+Prefer ``db.atomic()`` or ``db.begin()``, ``db.commit()`` and ``db.rollback()`` over mixing sqlite-utils transaction methods with calls to ``db.conn.commit()``, ``db.conn.rollback()`` or raw transaction-control SQL. Mixing the two layers makes it much harder to tell which layer owns the current transaction.
+
+Some related safeguards to be aware of:
+
+- ``db.enable_wal()`` and ``db.disable_wal()`` raise a ``sqlite_utils.db.TransactionError`` if called while a transaction is open, because changing the journal mode would commit it as a side effect.
+- ``table.transform()`` raises a ``sqlite_utils.db.TransactionError`` if called while a transaction is open with ``PRAGMA foreign_keys`` enabled and the table is referenced by foreign keys with destructive ``ON DELETE`` actions, because the pragma cannot be turned off mid-transaction to protect those referencing rows - see :ref:`python_api_transform_foreign_keys_transactions`.
+- Closing the database - explicitly with ``db.close()``, or by exiting a ``with Database(...) as db:`` block - rolls back any transaction that is still open, see :ref:`python_api_close`.
+
+.. _python_api_transactions_modes:
+
+Supported connection modes
+--------------------------
+
+``db.atomic()`` and the automatic per-method transactions currently require a connection using Python's legacy transaction control mode (``sqlite3.LEGACY_TRANSACTION_CONTROL`` on Python 3.12 and later). Passing a connection created with the Python 3.12+ ``sqlite3.connect(..., autocommit=True)`` or ``autocommit=False`` options to ``Database()`` raises a ``sqlite_utils.db.TransactionError``.
+
+Connections using ``autocommit=False`` are not supported because Python keeps a transaction open continuously. sqlite-utils uses ``Connection.in_transaction`` to distinguish its own transactions from transactions opened by its caller, and that distinction is not available in this mode.
+
+Connections using ``autocommit=True`` are also currently rejected because sqlite-utils has not formally exposed that as a supported configuration.
+
 .. _python_api_table:
 
 Accessing tables
@@ -299,6 +504,9 @@ You can also iterate through the table objects themselves using the ``.tables``
     >>> db.tables
     [<Table dogs>]
 
+.. note::
+    In the CLI: :ref:`sqlite-utils tables <cli_tables>`
+
 .. _python_api_views:
 
 Listing views
@@ -324,6 +532,9 @@ View objects are similar to Table objects, except that any attempts to insert or
 * ``rows_where(where, where_args, order_by, select)``
 * ``drop()``
 
+.. note::
+    In the CLI: :ref:`sqlite-utils views <cli_views>`
+
 .. _python_api_rows:
 
 Listing rows
@@ -379,6 +590,9 @@ This method also accepts ``offset=`` and ``limit=`` arguments, for specifying an
     ...     print(row)
     {'id': 1, 'age': 4, 'name': 'Cleo'}
 
+.. note::
+    In the CLI: :ref:`sqlite-utils rows <cli_rows>`
+
 .. _python_api_rows_count_where:
 
 Counting rows
@@ -463,6 +677,9 @@ The ``db.schema`` property returns the full SQL schema for the database as a str
         "name" TEXT
     );
 
+.. note::
+    In the CLI: :ref:`sqlite-utils schema <cli_schema>`
+
 .. _python_api_creating_tables:
 
 Creating tables
@@ -611,6 +828,9 @@ You can pass ``strict=True`` to create a table in ``STRICT`` mode:
         "name": str,
     }, strict=True)
 
+.. note::
+    In the CLI: :ref:`sqlite-utils create-table <cli_create_table>`
+
 .. _python_api_compound_primary_keys:
 
 Compound primary keys
@@ -671,6 +891,61 @@ You can leave off the third item in the tuple to have the referenced column auto
         ("author_id", "authors")
     ])
 
+.. _python_api_compound_foreign_keys:
+
+Compound foreign keys
+~~~~~~~~~~~~~~~~~~~~~
+
+To create a compound (multi-column) foreign key, use tuples of column names in place of the single column names:
+
+.. code-block:: python
+
+    db.table("courses").create({
+        "course_code": str,
+        "campus_name": str,
+        "dept_code": str,
+    }, pk="course_code", foreign_keys=[
+        (("campus_name", "dept_code"), "departments", ("campus_name", "dept_code"))
+    ])
+
+This creates a table-level constraint:
+
+.. code-block:: sql
+
+    CREATE TABLE "courses" (
+       "course_code" TEXT PRIMARY KEY,
+       "campus_name" TEXT,
+       "dept_code" TEXT,
+       FOREIGN KEY ("campus_name", "dept_code") REFERENCES "departments"("campus_name", "dept_code")
+    )
+
+As with single columns, you can leave off the tuple of other columns to reference the compound primary key of the other table:
+
+.. code-block:: python
+
+    foreign_keys=[
+        (("campus_name", "dept_code"), "departments")
+    ]
+
+To specify ``ON DELETE`` or ``ON UPDATE`` actions, pass ``ForeignKey`` objects instead:
+
+.. code-block:: python
+
+    from sqlite_utils.db import ForeignKey
+
+    db.table("books").create({
+        "id": int,
+        "author_id": int,
+    }, pk="id", foreign_keys=[
+        ForeignKey(
+            table="books", column="author_id",
+            other_table="authors", other_column="id",
+            on_delete="CASCADE",
+        )
+    ])
+
+Foreign key actions are preserved by :ref:`table.transform() <python_api_transform>` - prior to sqlite-utils 4.0 they were silently dropped when a table was transformed.
+
 .. _python_api_table_configuration:
 
 Table configuration options
@@ -736,6 +1011,9 @@ Here's an example that uses these features:
     # )
 
 
+.. note::
+    In the CLI: :ref:`sqlite-utils insert --not-null and --default <cli_defaults_not_null>`
+
 .. _python_api_rename_table:
 
 Renaming a table
@@ -753,6 +1031,9 @@ This executes the following SQL:
 
     ALTER TABLE [my_table] RENAME TO [new_name_for_my_table]
 
+.. note::
+    In the CLI: :ref:`sqlite-utils rename-table <cli_renaming_tables>`
+
 .. _python_api_duplicate:
 
 Duplicating tables
@@ -768,6 +1049,9 @@ The new ``authors_copy`` table will now contain a duplicate copy of the data fro
 
 This method raises ``sqlite_utils.db.NoTable`` if the table does not exist.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils duplicate <cli_duplicate_table>`
+
 .. _python_api_bulk_inserts:
 
 Bulk inserts
@@ -810,6 +1094,9 @@ You can delete all the existing rows in the table before inserting the new recor
 
 Pass ``analyze=True`` to run ``ANALYZE`` against the table after inserting the new records.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils insert <cli_inserting_data>`
+
 .. _python_api_insert_lists:
 
 Inserting data from a list or tuple iterator
@@ -887,6 +1174,9 @@ To replace any existing records that have a matching primary key, use the ``repl
 .. note::
     Prior to sqlite-utils 2.0 the ``.upsert()`` and ``.upsert_all()`` methods worked the same way as ``.insert(replace=True)`` does today. See :ref:`python_api_upsert` for the new behaviour of those methods introduced in 2.0.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils insert --replace <cli_insert_replace>`
+
 .. _python_api_update:
 
 Updating a specific record
@@ -934,8 +1224,7 @@ You can delete all records in a table that match a specific WHERE statement usin
 
     >>> db = sqlite_utils.Database("dogs.db")
     >>> # Delete every dog with age less than 3
-    >>> with db.conn:
-    >>>     db.table("dogs").delete_where("age < ?", [3])
+    >>> db.table("dogs").delete_where("age < ?", [3])
 
 Calling ``table.delete_where()`` with no other arguments will delete every row in the table.
 
@@ -968,9 +1257,14 @@ Note that the ``pk`` and ``column_order`` parameters here are optional if you ar
 
 An ``upsert_all()`` method is also available, which behaves like ``insert_all()`` but performs upserts instead.
 
+Every record passed to ``upsert()`` or ``upsert_all()`` must include a value for each primary key column - a record without one could never match an existing row, so a ``sqlite_utils.db.PrimaryKeyRequired`` exception is raised instead of quietly inserting a new row.
+
 .. note::
     ``.upsert()`` and ``.upsert_all()`` in sqlite-utils 1.x worked like ``.insert(..., replace=True)`` and ``.insert_all(..., replace=True)`` do in 2.x. See `issue #66 <https://github.com/simonw/sqlite-utils/issues/66>`__ for details of this change.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils upsert <cli_upsert>`
+
 .. _python_api_old_upsert:
 
 Alternative upserts using INSERT OR IGNORE
@@ -1066,6 +1360,8 @@ To create a species record with a note on when it was first seen, you can use th
 
 The first time this is called the record will be created for ``name="Palm"``. Any subsequent calls with that name will ignore the second argument, even if it includes different values.
 
+``None`` values are matched correctly: calling ``.lookup()`` a second time with the same values will return the primary key of the existing row even if some of those values are ``None``.
+
 ``.lookup()`` also accepts keyword arguments, which are passed through to the :ref:`insert() method <python_api_creating_tables>` and can be used to influence the shape of the created table. Supported parameters are:
 
 - ``pk`` - which defaults to ``id``
@@ -1111,6 +1407,8 @@ To extract the ``species`` column out to a separate ``Species`` table, you can d
         "species": "Common Juniper"
     }, extracts={"species": "Species"})
 
+``None`` values are not extracted: no record is created for them in the lookup table and the column value stays ``null``.
+
 .. _python_api_m2m:
 
 Working with many-to-many relationships
@@ -1318,6 +1616,9 @@ You can set a ``NOT NULL DEFAULT 'x'`` constraint on the new column using ``not_
 
     db.table("dogs").add_column("friends_count", int, not_null_default=0)
 
+.. note::
+    In the CLI: :ref:`sqlite-utils add-column <cli_add_column>`
+
 .. _python_api_add_column_alter:
 
 Adding columns automatically on insert/update
@@ -1341,6 +1642,9 @@ You can insert or update data that includes new columns and have the table autom
     new_table = db.table("new_table", alter=True)
     new_table.insert({"name": "Gareth", "age": 32, "shoe_size": 11})
 
+.. note::
+    In the CLI: :ref:`sqlite-utils insert --alter <cli_add_column_alter>`
+
 .. _python_api_add_foreign_key:
 
 Adding foreign key constraints
@@ -1379,6 +1683,29 @@ To ignore the case where the key already exists, use ``ignore=True``:
 
     db.table("books").add_foreign_key("author_id", "authors", "id", ignore=True)
 
+To add a compound foreign key, pass tuples of columns:
+
+.. code-block:: python
+
+    db.table("courses").add_foreign_key(
+        ("campus_name", "dept_code"), "departments", ("campus_name", "dept_code")
+    )
+
+As with single columns, omitting the other columns will use the compound primary key of the other table. ``other_table`` must always be specified for a compound foreign key.
+
+Use ``on_delete=`` and ``on_update=`` to specify ``ON DELETE`` and ``ON UPDATE`` actions for the foreign key:
+
+.. code-block:: python
+
+    db.table("books").add_foreign_key(
+        "author_id", "authors", "id", on_delete="CASCADE"
+    )
+
+This creates a foreign key with an ``ON DELETE CASCADE`` clause, so deleting an author will also delete their books (provided foreign key enforcement is enabled with ``PRAGMA foreign_keys = ON``). Valid actions are ``"SET NULL"``, ``"SET DEFAULT"``, ``"CASCADE"``, ``"RESTRICT"`` and the default ``"NO ACTION"``.
+
+.. note::
+    In the CLI: :ref:`sqlite-utils add-foreign-key <cli_add_foreign_key>`
+
 .. _python_api_add_foreign_keys:
 
 Adding multiple foreign key constraints at once
@@ -1397,6 +1724,11 @@ Here's an example adding two foreign keys at once:
 
 This method runs the same checks as ``.add_foreign_keys()`` and will raise ``sqlite_utils.db.AlterError`` if those checks fail.
 
+Foreign keys that already exist are silently skipped, so repeated calls are idempotent - but only if they match exactly. Requesting a foreign key that exists with different ``ON DELETE``/``ON UPDATE`` actions raises ``AlterError``: use ``table.transform()`` to change the actions of an existing foreign key.
+
+.. note::
+    In the CLI: :ref:`sqlite-utils add-foreign-keys <cli_add_foreign_keys>`
+
 .. _python_api_index_foreign_keys:
 
 Adding indexes for all foreign keys
@@ -1408,6 +1740,11 @@ If you want to ensure that every foreign key column in your database has a corre
 
     db.index_foreign_keys()
 
+Compound foreign keys get a single composite index across their columns.
+
+.. note::
+    In the CLI: :ref:`sqlite-utils index-foreign-keys <cli_index_foreign_keys>`
+
 .. _python_api_drop:
 
 Dropping a table or view
@@ -1429,6 +1766,9 @@ Pass ``ignore=True`` if you want to ignore the error caused by the table or view
 
     db.table("my_table").drop(ignore=True)
 
+.. note::
+    In the CLI: :ref:`sqlite-utils drop-table <cli_drop_table>` and :ref:`sqlite-utils drop-view <cli_drop_view>`
+
 .. _python_api_transform:
 
 Transforming a table
@@ -1457,6 +1797,9 @@ To keep the original table around instead of dropping it, pass the ``keep_table=
 
 This method raises a ``sqlite_utils.db.TransformError`` exception if the table cannot be transformed, usually because there are existing constraints or indexes that are incompatible with modifications to the columns.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils transform <cli_transform_table>`
+
 .. _python_api_transform_alter_column_types:
 
 Altering column types
@@ -1471,6 +1814,29 @@ To alter the type of a column, use the ``types=`` argument:
 
 See :ref:`python_api_add_column` for a list of available types.
 
+.. _python_api_transform_strict:
+
+Changing strict mode
+--------------------
+
+The optional ``strict=`` parameter can change whether a table uses `SQLite STRICT mode <https://www.sqlite.org/stricttables.html>`__. Pass ``strict=True`` to convert a regular table to a strict table:
+
+.. code-block:: python
+
+    table.transform(strict=True)
+
+Pass ``strict=False`` to convert a strict table back to a regular non-strict table:
+
+.. code-block:: python
+
+    table.transform(strict=False)
+
+The default is ``strict=None``, which preserves the table's existing strict mode.
+
+Passing ``strict=True`` raises ``sqlite_utils.db.TransformError`` if the available SQLite version does not support strict tables.
+
+Converting to a strict table validates all existing rows as they are copied into the replacement table. If a value is incompatible with its declared column type, SQLite raises ``sqlite3.IntegrityError`` and the transformation is rolled back, leaving the original table and its data unchanged.
+
 .. _python_api_transform_rename_columns:
 
 Renaming columns
@@ -1610,6 +1976,16 @@ This example drops two foreign keys - the one from ``places.country`` to ``count
         drop_foreign_keys=("country", "continent")
     )
 
+A bare column name drops any foreign key that column participates in, including compound foreign keys. To target a compound foreign key precisely, pass a tuple of its columns:
+
+.. code-block:: python
+
+    db.table("courses").transform(
+        drop_foreign_keys=[("campus_name", "dept_code")]
+    )
+
+Renaming a column with ``rename=`` updates any foreign keys that use it, and dropping a column with ``drop=`` also drops any foreign keys it participates in - for a compound foreign key this removes the whole constraint.
+
 .. _python_api_transform_sql:
 
 Custom transformations with .transform_sql()
@@ -1621,6 +1997,36 @@ If you want to do something more advanced, you can call the ``table.transform_sq
 
 This method will return a list of SQL statements that should be executed to implement the change. You can then make modifications to that SQL - or add additional SQL statements - before executing it yourself.
 
+.. _python_api_transform_foreign_keys_transactions:
+
+Foreign keys and transactions
+-----------------------------
+
+Because ``.transform()`` drops the old table, running it with ``PRAGMA foreign_keys`` enabled could fire ``ON DELETE`` actions on any tables that reference it - an inbound ``ON DELETE CASCADE`` foreign key would silently delete those referencing rows. To prevent this, ``.transform()`` turns ``PRAGMA foreign_keys`` off for the duration of the operation and restores it afterwards, running ``PRAGMA foreign_key_check`` before committing.
+
+``PRAGMA foreign_keys`` cannot be changed inside a transaction, so this protection is impossible if you call ``.transform()`` while a transaction is already open - for example inside a ``with db.atomic():`` block or after ``db.begin()``. If ``PRAGMA foreign_keys`` is on and another table references the table being transformed with a destructive ``ON DELETE`` action - ``CASCADE``, ``SET NULL`` or ``SET DEFAULT`` - the method will refuse to run and raise a ``sqlite_utils.db.TransactionError``:
+
+.. code-block:: python
+
+    from sqlite_utils.db import TransactionError
+
+    try:
+        with db.atomic():
+            db["authors"].transform(types={"id": str})
+    except TransactionError as ex:
+        print("Could not transform in transaction:", ex)
+
+To transform such a table either call ``.transform()`` outside of the transaction, or execute ``PRAGMA foreign_keys = off`` before opening it:
+
+.. code-block:: python
+
+    db.execute("PRAGMA foreign_keys = off")
+    with db.atomic():
+        db["authors"].transform(types={"id": str})
+    db.execute("PRAGMA foreign_keys = on")
+
+Tables referenced by foreign keys without a destructive action (the default ``NO ACTION``, or ``RESTRICT``) can still be transformed inside a transaction - sqlite-utils uses ``PRAGMA defer_foreign_keys`` to postpone the foreign key checks until the transaction commits.
+
 .. _python_api_extract:
 
 Extracting columns into a separate table
@@ -1777,6 +2183,11 @@ This produces a lookup table like so:
         "latin" TEXT
     )
 
+Rows where every extracted column is ``null`` are not extracted: no record is created for them in the lookup table and their foreign key column is left as ``null``. When extracting multiple columns, rows where at least one of the extracted columns has a value will be extracted as usual.
+
+.. note::
+    In the CLI: :ref:`sqlite-utils extract <cli_extract>`
+
 .. _python_api_hash:
 
 Setting an ID based on the hash of the row contents
@@ -1838,6 +2249,9 @@ You can pass ``ignore=True`` to silently ignore an existing view and do nothing,
         select * from dogs where is_good_dog = 1
     """, replace=True)
 
+.. note::
+    In the CLI: :ref:`sqlite-utils create-view <cli_create_view>`
+
 Storing JSON
 ============
 
@@ -1950,12 +2364,15 @@ The ``db.iterdump()`` method returns a sequence of SQL strings representing a co
 
 This uses the `sqlite3.Connection.iterdump() <https://docs.python.org/3/library/sqlite3.html#sqlite3.Connection.iterdump>`__ method.
 
-If you are using ``pysqlite3`` or ``sqlean.py`` the underlying method may be missing. If you install the `sqlite-dump <https://pypi.org/project/sqlite-dump/>`__ package then the ``db.iterdump()`` method will use that implementation instead:
+If you are using ``pysqlite3`` the underlying method may be missing. If you install the `sqlite-dump <https://pypi.org/project/sqlite-dump/>`__ package then the ``db.iterdump()`` method will use that implementation instead:
 
 .. code-block:: bash
 
     pip install sqlite-dump
 
+.. note::
+    In the CLI: :ref:`sqlite-utils dump <cli_dump>`
+
 .. _python_api_introspection:
 
 Introspecting tables and views
@@ -2057,17 +2474,39 @@ Almost all SQLite tables have a ``rowid`` column, but a table with no explicitly
 .foreign_keys
 -------------
 
-The ``.foreign_keys`` property returns any foreign key relationships for the table, as a list of ``ForeignKey(table, column, other_table, other_column)`` named tuples. It is not available on views.
+The ``.foreign_keys`` property returns any foreign key relationships for the table, as a list of ``ForeignKey`` objects. It is not available on views.
+
+Each ``ForeignKey`` has the following attributes:
+
+``table``
+    The table the foreign key is defined on.
+``column``
+    The column on this table, or ``None`` for a compound foreign key.
+``other_table``
+    The table being referenced.
+``other_column``
+    The referenced column, or ``None`` for a compound foreign key.
+``columns``
+    A tuple of the columns on this table, always populated (a one-item tuple for single-column foreign keys).
+``other_columns``
+    A tuple of the referenced columns.
+``is_compound``
+    ``True`` if this is a compound (multi-column) foreign key.
+``on_delete``
+    The ``ON DELETE`` action, e.g. ``"CASCADE"`` - ``"NO ACTION"`` if not set.
+``on_update``
+    The ``ON UPDATE`` action - ``"NO ACTION"`` if not set.
+
+``ForeignKey`` was a ``namedtuple`` prior to sqlite-utils 4.0. It is now a dataclass and can no longer be unpacked or indexed as a tuple - access its fields by name instead. See :ref:`upgrading_3_to_4` for details.
 
 ::
 
     >>> db.table("Street_Tree_List").foreign_keys
-    [ForeignKey(table='Street_Tree_List', column='qLegalStatus', other_table='qLegalStatus', other_column='id'),
-     ForeignKey(table='Street_Tree_List', column='qCareAssistant', other_table='qCareAssistant', other_column='id'),
-     ForeignKey(table='Street_Tree_List', column='qSiteInfo', other_table='qSiteInfo', other_column='id'),
-     ForeignKey(table='Street_Tree_List', column='qSpecies', other_table='qSpecies', other_column='id'),
-     ForeignKey(table='Street_Tree_List', column='qCaretaker', other_table='qCaretaker', other_column='id'),
-     ForeignKey(table='Street_Tree_List', column='PlantType', other_table='PlantType', other_column='id')]
+    [ForeignKey(table='Street_Tree_List', column='qLegalStatus', other_table='qLegalStatus', other_column='id', columns=('qLegalStatus',), other_columns=('id',), is_compound=False, on_delete='NO ACTION', on_update='NO ACTION'),
+     ForeignKey(table='Street_Tree_List', column='qCareAssistant', other_table='qCareAssistant', other_column='id', columns=('qCareAssistant',), other_columns=('id',), is_compound=False, on_delete='NO ACTION', on_update='NO ACTION'),
+     ...]
+
+Compound foreign keys - defined with ``FOREIGN KEY (col_a, col_b) REFERENCES other(col_a, col_b)`` - are returned as a single ``ForeignKey`` with ``is_compound=True``, ``column`` and ``other_column`` set to ``None``, and the participating columns available in the ``columns`` and ``other_columns`` tuples.
 
 .. _python_api_introspection_schema:
 
@@ -2133,6 +2572,9 @@ The ``.indexes`` property returns all indexes created for a table, as a list of
      Index(seq=4, name='"Street_Tree_List_qCaretaker"', unique=0, origin='c', partial=0, columns=['qCaretaker']),
      Index(seq=5, name='"Street_Tree_List_PlantType"', unique=0, origin='c', partial=0, columns=['PlantType'])]
 
+.. note::
+    In the CLI: :ref:`sqlite-utils indexes <cli_indexes>`
+
 .. _python_api_introspection_xindexes:
 
 .xindexes
@@ -2176,6 +2618,9 @@ The ``.triggers`` property lists database triggers. It can be used on both datab
     >>> db.triggers
     ... similar output to db.table("authors").triggers
 
+.. note::
+    In the CLI: :ref:`sqlite-utils triggers <cli_triggers>`
+
 .. _python_api_introspection_triggers_dict:
 
 .triggers_dict
@@ -2324,6 +2769,9 @@ To remove the FTS tables and triggers you created, use the ``disable_fts()`` tab
 
     db.table("dogs").disable_fts()
 
+.. note::
+    In the CLI: :ref:`sqlite-utils enable-fts <cli_fts>`
+
 .. _python_api_quote_fts:
 
 Quoting characters for use in search
@@ -2388,6 +2836,9 @@ To return just the title and published columns for three matches for ``"dog"`` w
     ):
         print(article)
 
+.. note::
+    In the CLI: :ref:`sqlite-utils search <cli_search>`
+
 .. _python_api_fts_search_sql:
 
 Building SQL queries with table.search_sql()
@@ -2472,6 +2923,9 @@ This runs the following SQL::
 
     INSERT INTO dogs_fts (dogs_fts) VALUES ("rebuild");
 
+.. note::
+    In the CLI: :ref:`sqlite-utils rebuild-fts <cli_fts>`
+
 .. _python_api_fts_optimize:
 
 Optimizing a full-text search table
@@ -2487,6 +2941,9 @@ This runs the following SQL::
 
     INSERT INTO dogs_fts (dogs_fts) VALUES ("optimize");
 
+.. note::
+    In the CLI: :ref:`sqlite-utils optimize <cli_optimize>`
+
 .. _python_api_cached_table_counts:
 
 Cached table counts using triggers
@@ -2549,6 +3006,9 @@ If the ``_counts`` table ever becomes out-of-sync with the actual table counts y
 
     db.reset_counts()
 
+.. note::
+    In the CLI: :ref:`sqlite-utils enable-counts <cli_enable_counts>`
+
 .. _python_api_create_index:
 
 Creating indexes
@@ -2592,6 +3052,17 @@ Use ``if_not_exists=True`` to do nothing if an index with that name already exis
 
 Pass ``analyze=True`` to run ``ANALYZE`` against the new index after creating it.
 
+You can drop an index from a table using ``.drop_index(index_name)``:
+
+.. code-block:: python
+
+    db.table("dogs").drop_index("idx_dogs_name")
+
+Use ``ignore=True`` to ignore the error if the index does not exist.
+
+.. note::
+    In the CLI: :ref:`sqlite-utils create-index <cli_create_index>` and :ref:`sqlite-utils drop-index <cli_drop_index>`
+
 .. _python_api_analyze:
 
 Optimizing index usage with ANALYZE
@@ -2619,6 +3090,9 @@ To run against all indexes attached to a specific table, you can either pass the
 
     db.table("dogs").analyze()
 
+.. note::
+    In the CLI: :ref:`sqlite-utils analyze <cli_analyze>`
+
 .. _python_api_vacuum:
 
 Vacuum
@@ -2630,6 +3104,9 @@ You can optimize your database by running VACUUM against it like so:
 
     Database("my_database.db").vacuum()
 
+.. note::
+    In the CLI: :ref:`sqlite-utils vacuum <cli_vacuum>`
+
 .. _python_api_wal:
 
 WAL mode
@@ -2647,6 +3124,8 @@ You can disable WAL mode using ``.disable_wal()``:
 
     Database("my_database.db").disable_wal()
 
+The journal mode can only be changed outside of a transaction. Calling either method while a transaction is open - inside a ``db.atomic()`` block, for example - raises a ``sqlite_utils.db.TransactionError``, unless the database is already in the requested mode in which case the call is a no-op.
+
 You can check the current journal mode for a database using the ``journal_mode`` property:
 
 .. code-block:: python
@@ -2655,6 +3134,9 @@ You can check the current journal mode for a database using the ``journal_mode``
 
 This will usually be ``wal`` or ``delete`` (meaning WAL is disabled), but can have other values - see the `PRAGMA journal_mode <https://www.sqlite.org/pragma.html#pragma_journal_mode>`__ documentation.
 
+.. note::
+    In the CLI: :ref:`sqlite-utils enable-wal and disable-wal <cli_wal>`
+
 .. _python_api_suggest_column_types:
 
 Suggesting column types
@@ -2795,6 +3277,9 @@ You can cause ``sqlite3`` to return more useful errors, including the traceback
 
     sqlite3.enable_callback_tracebacks(True)
 
+.. note::
+    In the CLI: :ref:`sqlite-utils query --functions <cli_query_functions>`
+
 .. _python_api_quote:
 
 Quoting strings for use in SQL
@@ -2927,6 +3412,9 @@ Initialize SpatiaLite
 .. automethod:: sqlite_utils.db.Database.init_spatialite
    :noindex:
 
+.. note::
+    In the CLI: :ref:`sqlite-utils create-database --init-spatialite <cli_create_database>`
+
 .. _python_api_gis_find_spatialite:
 
 Finding SpatiaLite
@@ -2942,6 +3430,9 @@ Adding geometry columns
 .. automethod:: sqlite_utils.db.Table.add_geometry_column
    :noindex:
 
+.. note::
+    In the CLI: :ref:`sqlite-utils add-geometry-column <cli_spatialite>`
+
 .. _python_api_gis_create_spatial_index:
 
 Creating a spatial index
@@ -2949,3 +3440,6 @@ Creating a spatial index
 
 .. automethod:: sqlite_utils.db.Table.create_spatial_index
    :noindex:
+
+.. note::
+    In the CLI: :ref:`sqlite-utils create-spatial-index <cli_spatialite_indexes>`

docs/reference.rst

@@ -70,6 +70,13 @@ sqlite_utils.db.ColumnDetails
 
 .. autoclass:: sqlite_utils.db.ColumnDetails
 
+.. _reference_db_other_foreign_key:
+
+sqlite_utils.db.ForeignKey
+--------------------------
+
+.. autoclass:: sqlite_utils.db.ForeignKey
+
 sqlite_utils.utils
 ==================
 

docs/upgrading.rst

@@ -0,0 +1,146 @@
+.. _upgrading:
+
+===========
+ Upgrading
+===========
+
+This page describes the changes you may need to make to your own code or scripts when upgrading between major versions of ``sqlite-utils``.
+
+For the full list of changes in every release see the :ref:`changelog`.
+
+.. _upgrading_3_to_4:
+
+Upgrading from 3.x to 4.0
+=========================
+
+Requirements
+------------
+
+- Python 3.10 or higher is required.
+- The ``click`` dependency must be version 8.3.1 or later.
+
+Command-line changes
+--------------------
+
+**Type detection is now the default for CSV and TSV imports.** ``sqlite-utils insert`` and ``sqlite-utils upsert`` now detect column types when importing CSV or TSV data - previously every column was created as ``TEXT`` unless you passed ``--detect-types``. To restore the old behavior pass the new ``--no-detect-types`` flag:
+
+.. code-block:: bash
+
+    sqlite-utils insert data.db rows data.csv --csv --no-detect-types
+
+Two related things have been removed:
+
+- The ``SQLITE_UTILS_DETECT_TYPES`` environment variable.
+- The old ``-d/--detect-types`` flag itself. Since detection is now the default the flag did nothing - remove it from any scripts that used it.
+
+**The convert command no longer skips falsey values.** ``sqlite-utils convert`` previously skipped values that evaluated to ``False`` (empty strings, ``0``) unless you passed ``--no-skip-false``. All values are now converted and the ``--no-skip-false`` flag has been removed.
+
+**drop-table and drop-view check the object type.** ``sqlite-utils drop-table`` now refuses to drop a view, and ``drop-view`` refuses to drop a table. Previously each would silently drop the wrong type of object if the name matched. If you relied on that (unlikely), use the matching command instead.
+
+**sqlite-utils tui has moved to a plugin.** The optional terminal interface is now provided by the `sqlite-utils-tui <https://github.com/simonw/sqlite-utils-tui>`__ plugin:
+
+.. code-block:: bash
+
+    sqlite-utils install sqlite-utils-tui
+
+Python API changes
+------------------
+
+**db.query() now rejects SQL that does not return rows.** This is likely the most common change you will need to make to existing code. ``db.query()`` used to accept any SQL statement - passing one that returns no rows, such as an ``INSERT`` or ``UPDATE`` without a ``RETURNING`` clause or a ``CREATE TABLE``, did nothing at all, silently. Those statements now raise a ``ValueError``, and are rolled back so they have no effect on the database. Transaction control statements (``BEGIN``, ``COMMIT``, ``END``, ``ROLLBACK``, ``SAVEPOINT``, ``RELEASE``) plus ``VACUUM``, ``ATTACH`` and ``DETACH`` are also rejected with a ``ValueError``, without being executed at all. Use ``db.execute()`` for statements that do not return rows:
+
+.. code-block:: python
+
+    # 3.x accepted this but silently did nothing:
+    db.query("update dogs set name = 'Cleopaws'")
+
+    # In 4.0 use execute() for SQL that does not return rows:
+    db.execute("update dogs set name = 'Cleopaws'")
+
+**db.query() executes immediately.** ``db.query(sql)`` previously returned a generator that did not execute the SQL until you started iterating over it. The SQL now runs as soon as the method is called - rows are still fetched lazily, but errors in your SQL raise at the ``db.query()`` call site rather than on first iteration, and a write with a ``RETURNING`` clause takes effect even if you never iterate over its results.
+
+**db.table() no longer returns views.** ``db.table(name)`` now raises a ``sqlite_utils.db.NoTable`` exception if ``name`` is a SQL view. Use the new ``db.view(name)`` method for views:
+
+.. code-block:: python
+
+    table = db.table("my_table")
+    view = db.view("my_view")
+
+``db["name"]`` still returns either a ``Table`` or a ``View`` depending on what exists in the database.
+
+**Upserts use INSERT ... ON CONFLICT.** Upsert operations now use SQLite's ``INSERT ... ON CONFLICT SET`` syntax rather than the previous ``INSERT OR IGNORE`` followed by ``UPDATE``. If your code depends on the old behavior, pass ``use_old_upsert=True`` to the ``Database()`` constructor - see :ref:`python_api_old_upsert`.
+
+**Upsert records must include their primary keys.** ``table.upsert()`` and ``table.upsert_all()`` now raise ``sqlite_utils.db.PrimaryKeyRequired`` if a record is missing a value for any primary key column (or has ``None`` for one). Previously such records were quietly inserted as new rows. Relatedly, ``pk=`` is now optional when the table already exists with a primary key - it is detected automatically.
+
+**Floating point columns are now REAL.** Auto-detected floating point columns are created with the correct SQLite type ``REAL`` instead of ``FLOAT``. Code that inspects column types should expect ``REAL``.
+
+**Generated schemas use double quotes.** Tables created by this library now wrap table and column names in ``"double-quotes"`` where they previously used ``[square-braces]``. If you compare ``table.schema`` strings against expected values you will need to update them.
+
+**table.convert() no longer skips falsey values.** Matching the CLI change above, ``table.convert()`` now converts every value. The ``skip_false`` parameter has been removed - previously it defaulted to ``True``, skipping empty strings and other falsey values.
+
+**Null values are no longer extracted into lookup tables.** ``table.extract()`` and the ``sqlite-utils extract`` command leave rows alone if every extracted column is ``null`` - the new foreign key column is left as ``null`` instead of pointing at an all-``null`` record in the lookup table. The ``extracts=`` insert option similarly keeps ``None`` values as ``null``. Relatedly, ``table.lookup()`` now compares values using ``IS`` so that looking up a value containing ``None`` returns the existing matching row - previously it inserted a duplicate row on every call.
+
+**ensure_autocommit_off() is now ensure_autocommit_on().** The ``db.ensure_autocommit_off()`` context manager has been renamed to ``db.ensure_autocommit_on()``. The old name described the opposite of what the method did: it temporarily puts the connection into driver-level autocommit mode (by setting ``isolation_level = None``), so that statements such as ``PRAGMA journal_mode=wal`` can run outside of an implicit transaction. The behavior is unchanged - update any calls to use the new name.
+
+**View.enable_fts() has been removed.** The ``View`` class previously had an ``enable_fts()`` method that existed only to raise ``NotImplementedError`` - full-text search is not supported for views. Calling it now raises ``AttributeError`` like any other missing method.
+
+**ForeignKey is now a dataclass, not a namedtuple.** The ``ForeignKey`` objects returned by ``table.foreign_keys`` gained new fields - ``columns``, ``other_columns``, ``is_compound``, ``on_delete`` and ``on_update`` - so that compound (multi-column) foreign keys and foreign key actions can be represented. To make room for those fields cleanly ``ForeignKey`` is now a dataclass rather than a ``namedtuple``, so it can no longer be unpacked or indexed as a tuple. Access its fields by name instead:
+
+.. code-block:: python
+
+    # 3.x - tuple unpacking, no longer works:
+    for table, column, other_table, other_column in db["courses"].foreign_keys:
+        ...
+
+    # 4.0 - access fields by name:
+    for fk in db["courses"].foreign_keys:
+        fk.table, fk.column, fk.other_table, fk.other_column
+
+Attempting the old unpacking or ``fk[0]`` indexing now raises ``TypeError``, so any code using those patterns will fail loudly rather than silently misbehave. Like the old namedtuple, ``ForeignKey`` instances are immutable and hashable - they can be collected into sets and used as dictionary keys. Note that equality now includes the ``on_delete`` and ``on_update`` actions: a ``ForeignKey`` with ``ON DELETE CASCADE`` is not equal to one without.
+
+Compound foreign keys - previously returned as one ``ForeignKey`` per column, misleadingly suggesting several independent single-column keys - are now returned as a single ``ForeignKey`` with ``is_compound=True``. For these the scalar ``column`` and ``other_column`` fields are ``None``; use the ``columns`` and ``other_columns`` tuples instead. Single-column foreign keys are unaffected apart from the class change: ``column``/``other_column`` behave as before and ``columns``/``other_columns`` are one-item tuples.
+
+Two related behavior changes to ``table.transform()``: compound foreign keys now survive a transform (previously they were split into separate single-column keys), and ``ON DELETE``/``ON UPDATE`` actions such as ``ON DELETE CASCADE`` are now preserved (previously they were silently stripped from the schema).
+
+**Validation errors raise ValueError.** Invalid arguments to Python API methods - for example ``create_table()`` with no columns, or ``ignore=True`` together with ``replace=True`` - now raise ``ValueError``. They previously raised ``AssertionError`` from bare ``assert`` statements, which were silently skipped under ``python -O``.
+
+**Transaction behavior is now well-defined.** 4.0 introduces the :ref:`db.atomic() <python_api_atomic>` context manager and uses it consistently for every write operation - the full model is described in :ref:`python_api_transactions`. Changes you may notice:
+
+- Write statements executed with raw ``db.execute()`` calls now commit automatically, unless a transaction is already open in which case they join it. Previously they opened an implicit transaction that nothing committed - if your code used ``db.execute()`` for writes and relied on ``db.conn.rollback()`` to undo them, open an explicit transaction with the new ``db.begin()`` method first.
+- Multi-step operations such as ``table.transform()`` no longer commit an existing transaction you have open - they use savepoints inside it instead.
+- ``db.enable_wal()`` and ``db.disable_wal()`` raise a ``sqlite_utils.db.TransactionError`` if called while a transaction is open, instead of silently committing it.
+- Using ``Database`` as a context manager (``with Database(path) as db:``) closes the connection on exit *without* committing - a transaction you explicitly opened with ``db.begin()`` and did not commit is rolled back.
+- ``Database()`` rejects connections created with the Python 3.12+ ``sqlite3.connect(..., autocommit=True)`` or ``autocommit=False`` options, raising ``sqlite_utils.db.TransactionError``. On those connections every write the library made was silently discarded when the connection closed.
+
+Packaging changes
+-----------------
+
+- ``sqlite-utils`` now uses ``pyproject.toml`` in place of ``setup.py``.
+- ``pip`` is now a runtime dependency, used by the ``sqlite-utils install`` and ``uninstall`` commands.
+
+New features to be aware of
+---------------------------
+
+Not breaking changes, but new in 4.0 and worth knowing about when you upgrade:
+
+- A :ref:`database migrations system <migrations>`, incorporating the functionality of the ``sqlite-migrate`` plugin. If you used that plugin, the built-in system reads the same ``_sqlite_migrations`` table - your applied migrations will not run again. Update your migration files to use ``from sqlite_utils import Migrations``.
+- :ref:`db.atomic() <python_api_atomic>` for nested transaction support.
+- ``table.insert_all()`` and ``table.upsert_all()`` accept an iterator of lists or tuples as an alternative to dictionaries - see :ref:`python_api_insert_lists`.
+
+.. _upgrading_2_to_3:
+
+Upgrading from 2.x to 3.0
+=========================
+
+The 3.0 release redesigned search. The breaking changes were minor:
+
+- ``table.search()`` returns a generator of dictionaries, sorted by relevance. It previously returned a list of tuples sorted by ``rowid``.
+- The ``-c`` shortcut for ``--csv`` and the ``-f`` shortcut for ``--fmt`` were removed from the CLI - use the full option names.
+
+.. _upgrading_1_to_2:
+
+Upgrading from 1.x to 2.0
+=========================
+
+The 2.0 release changed the meaning of *upsert*. In 1.x, ``table.upsert()`` and ``table.upsert_all()`` actually performed ``INSERT OR REPLACE`` operations - entirely replacing the existing row. Since 2.0 an upsert updates only the columns you provide, leaving other columns untouched.
+
+If you want the 1.x behavior, use ``table.insert(..., replace=True)`` or ``table.insert_all(..., replace=True)`` instead.

mypy.ini

@@ -1,4 +1,32 @@
 [mypy]
+python_version = 3.10
+warn_return_any = False
+warn_unused_configs = True
+warn_redundant_casts = False
+warn_unused_ignores = False
+check_untyped_defs = True
+disallow_untyped_defs = False
+disallow_incomplete_defs = False
+no_implicit_optional = True
+strict_equality = True
 
-[mypy-pysqlite3,sqlean,sqlite_dump]
-ignore_missing_imports = True
+[mypy-sqlite_utils.cli]
+ignore_errors = True
+
+[mypy-pysqlite3.*]
+ignore_missing_imports = True
+
+[mypy-sqlite_dump.*]
+ignore_missing_imports = True
+
+[mypy-sqlite_fts4.*]
+ignore_missing_imports = True
+
+[mypy-pandas.*]
+ignore_missing_imports = True
+
+[mypy-numpy.*]
+ignore_missing_imports = True
+
+[mypy-tests.*]
+ignore_errors = True

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "sqlite-utils"
-version = "4.0a1"
+version = "4.1.1"
 description = "CLI tool and Python library for manipulating SQLite databases"
 readme = { file = "README.md", content-type = "text/markdown" }
 authors = [
@@ -22,12 +22,42 @@ classifiers = [
 ]
 
 dependencies = [
-    "click",
+    "click>=8.3.1",
     "click-default-group>=1.2.3",
     "pluggy",
     "python-dateutil",
     "sqlite-fts4",
     "tabulate",
+    "pip",
+]
+
+[dependency-groups]
+dev = [
+    "black>=26.3.1",
+    "click>=8.4.2",
+    "cogapp",
+    "hypothesis",
+    "pytest",
+    # mypy
+    "data-science-types",
+    "mypy",
+    "types-click",
+    "types-pluggy",
+    "types-python-dateutil",
+    "types-tabulate",
+    # flake8
+    "flake8",
+    "flake8-pyproject",
+    "ty>=0.0.37",
+    # For stable cog:
+    "tabulate>=0.10.0",
+]
+docs = [
+    "codespell",
+    "furo",
+    "pygments-csv-lexer",
+    "sphinx-autobuild",
+    "sphinx-copybutton",
 ]
 
 [project.urls]
@@ -40,36 +70,9 @@ CI = "https://github.com/simonw/sqlite-utils/actions"
 [project.scripts]
 sqlite-utils = "sqlite_utils.cli:cli"
 
-[project.optional-dependencies]
-test = [
-    "black>=24.1.1",
-    "cogapp",
-    "hypothesis",
-    "pytest",
-]
-docs = [
-    "beanbag-docutils>=2.0",
-    "codespell",
-    "furo",
-    "pygments-csv-lexer",
-    "sphinx-autobuild",
-    "sphinx-copybutton",
-]
-mypy = [
-    "data-science-types",
-    "mypy",
-    "types-click",
-    "types-pluggy",
-    "types-python-dateutil",
-    "types-tabulate",
-]
-flake8 = [
-    "flake8",
-    "flake8-pyproject",
-]
-
 [build-system]
-requires = ["setuptools"]
+# setuptools 77+ is needed for the PEP 639 license = "Apache-2.0" expression
+requires = ["setuptools>=77"]
 build-backend = "setuptools.build_meta"
 
 [tool.flake8]

sqlite_utils/__init__.py

@@ -2,5 +2,6 @@ from .utils import suggest_column_types
 from .hookspecs import hookimpl
 from .hookspecs import hookspec
 from .db import Database
+from .migrations import Migrations
 
-__all__ = ["Database", "suggest_column_types", "hookimpl", "hookspec"]
+__all__ = ["Database", "Migrations", "suggest_column_types", "hookimpl", "hookspec"]

sqlite_utils/hookspecs.py

@@ -1,3 +1,6 @@
+import sqlite3
+
+import click
 from pluggy import HookimplMarker
 from pluggy import HookspecMarker
 
@@ -6,10 +9,10 @@ hookimpl = HookimplMarker("sqlite_utils")
 
 
 @hookspec
-def register_commands(cli):
+def register_commands(cli: click.Group) -> None:
     """Register additional CLI commands, e.g. 'sqlite-utils mycommand ...'"""
 
 
 @hookspec
-def prepare_connection(conn):
+def prepare_connection(conn: sqlite3.Connection) -> None:
     """Modify SQLite connection in some way e.g. register custom SQL functions"""

sqlite_utils/migrations.py

@@ -0,0 +1,177 @@
+from collections.abc import Iterable
+from dataclasses import dataclass
+import datetime
+from typing import Callable, cast, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from sqlite_utils.db import Database, Table
+
+
+class Migrations:
+    migrations_table = "_sqlite_migrations"
+
+    @dataclass
+    class _Migration:
+        name: str
+        fn: Callable
+        transactional: bool = True
+
+    @dataclass
+    class _AppliedMigration:
+        name: str
+        # A string timestamp such as "2026-07-04 12:00:00.000000+00:00" -
+        # stored as TEXT in the _sqlite_migrations table
+        applied_at: str
+
+    def __init__(self, name: str):
+        """
+        :param name: The name of the migration set. This should be unique.
+        """
+        self.name = name
+        self._migrations: list[Migrations._Migration] = []
+
+    def __call__(
+        self, *, name: str | None = None, transactional: bool = True
+    ) -> Callable:
+        """
+        :param name: The name to use for this migration - if not provided,
+          the name of the function will be used.
+        :param transactional: If ``True`` (the default) the migration and the
+          record of it having been applied are wrapped in a transaction, which
+          will be rolled back if the migration raises an exception. Pass
+          ``False`` for migrations that cannot run inside a transaction, for
+          example those that execute ``VACUUM``.
+        """
+
+        def inner(func: Callable) -> Callable:
+            migration_name = name or getattr(func, "__name__")
+            if any(m.name == migration_name for m in self._migrations):
+                raise ValueError(
+                    "Migration '{}' is already registered in set '{}'".format(
+                        migration_name, self.name
+                    )
+                )
+            self._migrations.append(
+                self._Migration(migration_name, func, transactional)
+            )
+            return func
+
+        return inner
+
+    def pending(self, db: "Database") -> list["Migrations._Migration"]:
+        """
+        Return a list of pending migrations.
+
+        This is a read-only operation - it does not write to the database.
+        """
+        already_applied = {migration.name for migration in self.applied(db)}
+        return [
+            migration
+            for migration in self._migrations
+            if migration.name not in already_applied
+        ]
+
+    def applied(self, db: "Database") -> list["Migrations._AppliedMigration"]:
+        """
+        Return a list of applied migrations, in the order they were applied.
+
+        This is a read-only operation - it does not write to the database.
+        """
+        table = _table(db, self.migrations_table)
+        if not table.exists():
+            return []
+        return [
+            self._AppliedMigration(name=row["name"], applied_at=row["applied_at"])
+            for row in table.rows_where(
+                "migration_set = ?", [self.name], order_by="rowid"
+            )
+        ]
+
+    def apply(self, db: "Database", *, stop_before: str | Iterable[str] | None = None):
+        """
+        Apply any pending migrations to the database.
+
+        Each migration runs inside a transaction, together with the record of
+        it having been applied - if the migration raises an exception its
+        changes are rolled back, no record is written and the migration stays
+        pending. Migrations registered with ``transactional=False`` run
+        outside of a transaction.
+
+        :raises ValueError: if a ``stop_before`` name matches a migration in
+          this set that has already been applied - stopping before it is
+          impossible to honor, and no pending migrations are applied
+        """
+        if stop_before is None:
+            stop_before_names = set()
+        elif isinstance(stop_before, str):
+            stop_before_names = {stop_before}
+        else:
+            stop_before_names = set(stop_before)
+        # A stop_before naming an already-applied migration cannot be
+        # honored - error rather than applying everything after it. Names
+        # not in this set at all are ignored, because unqualified CLI
+        # values are offered to every migration set
+        already_applied = stop_before_names.intersection(
+            migration.name for migration in self.applied(db)
+        )
+        if already_applied:
+            raise ValueError(
+                "Cannot stop before migration{} {} in set '{}' - already "
+                "been applied".format(
+                    "s" if len(already_applied) > 1 else "",
+                    ", ".join(sorted(already_applied)),
+                    self.name,
+                )
+            )
+        self.ensure_migrations_table(db)
+        for migration in self.pending(db):
+            name = migration.name
+            if name in stop_before_names:
+                return
+            if migration.transactional:
+                with db.atomic():
+                    migration.fn(db)
+                    self._record_applied(db, name)
+            else:
+                migration.fn(db)
+                self._record_applied(db, name)
+
+    def _record_applied(self, db: "Database", name: str):
+        _table(db, self.migrations_table).insert(
+            {
+                "migration_set": self.name,
+                "name": name,
+                "applied_at": str(datetime.datetime.now(datetime.timezone.utc)),
+            }
+        )
+
+    def ensure_migrations_table(self, db: "Database"):
+        """
+        Ensure the _sqlite_migrations table exists and has the correct schema.
+        """
+        table = _table(db, self.migrations_table)
+        if not table.exists():
+            table.create(
+                {
+                    "id": int,
+                    "migration_set": str,
+                    "name": str,
+                    "applied_at": str,
+                },
+                pk="id",
+            )
+            table.create_index(["migration_set", "name"], unique=True)
+        elif table.pks != ["id"]:
+            table.transform(pk="id")
+            unique_indexes = {tuple(index.columns) for index in table.indexes}
+            if ("migration_set", "name") not in unique_indexes:
+                table.create_index(["migration_set", "name"], unique=True)
+
+    def __repr__(self):
+        return "<Migrations '{}': [{}]>".format(
+            self.name, ", ".join(m.name for m in self._migrations)
+        )
+
+
+def _table(db: "Database", name: str) -> "Table":
+    return cast("Table", db[name])

sqlite_utils/plugins.py

@@ -1,22 +1,31 @@
+from typing import Dict, List, Union
+
 import pluggy
 import sys
 from . import hookspecs
 
-pm = pluggy.PluginManager("sqlite_utils")
+pm: pluggy.PluginManager = pluggy.PluginManager("sqlite_utils")
 pm.add_hookspecs(hookspecs)
+_plugins_loaded = False
 
-if not getattr(sys, "_called_from_test", False):
-    # Only load plugins if not running tests
+
+def ensure_plugins_loaded() -> None:
+    global _plugins_loaded
+    if _plugins_loaded or getattr(sys, "_called_from_test", False):
+        return
     pm.load_setuptools_entrypoints("sqlite_utils")
+    _plugins_loaded = True
 
 
-def get_plugins():
-    plugins = []
+def get_plugins() -> List[Dict[str, Union[str, List[str]]]]:
+    ensure_plugins_loaded()
+    plugins: List[Dict[str, Union[str, List[str]]]] = []
     plugin_to_distinfo = dict(pm.list_plugin_distinfo())
     for plugin in pm.get_plugins():
-        plugin_info = {
+        hookcallers = pm.get_hookcallers(plugin) or []
+        plugin_info: Dict[str, Union[str, List[str]]] = {
             "name": plugin.__name__,
-            "hooks": [h.name for h in pm.get_hookcallers(plugin)],
+            "hooks": [h.name for h in hookcallers],
         }
         distinfo = plugin_to_distinfo.get(plugin)
         if distinfo:

sqlite_utils/recipes.py

@@ -1,11 +1,20 @@
+from __future__ import annotations
+
+from typing import Callable, Optional
+
 from dateutil import parser
 import json
 
-IGNORE = object()
-SET_NULL = object()
+IGNORE: object = object()
+SET_NULL: object = object()
 
 
-def parsedate(value, dayfirst=False, yearfirst=False, errors=None):
+def parsedate(
+    value: str,
+    dayfirst: bool = False,
+    yearfirst: bool = False,
+    errors: Optional[object] = None,
+) -> Optional[str]:
     """
     Parse a date and convert it to ISO date format: yyyy-mm-dd
     \b
@@ -31,7 +40,12 @@ def parsedate(value, dayfirst=False, yearfirst=False, errors=None):
             raise
 
 
-def parsedatetime(value, dayfirst=False, yearfirst=False, errors=None):
+def parsedatetime(
+    value: str,
+    dayfirst: bool = False,
+    yearfirst: bool = False,
+    errors: Optional[object] = None,
+) -> Optional[str]:
     """
     Parse a datetime and convert it to ISO datetime format: yyyy-mm-ddTHH:MM:SS
     \b
@@ -53,7 +67,9 @@ def parsedatetime(value, dayfirst=False, yearfirst=False, errors=None):
             raise
 
 
-def jsonsplit(value, delimiter=",", type=str):
+def jsonsplit(
+    value: str, delimiter: str = ",", type: Callable[[str], object] = str
+) -> str:
     """
     Convert a string like a,b,c into a JSON array ["a", "b", "c"]
     """

sqlite_utils/utils.py

@@ -3,26 +3,44 @@ import contextlib
 import csv
 import enum
 import hashlib
+import importlib
 import io
 import itertools
 import json
 import os
 import sys
-from . import recipes
-from typing import Dict, cast, BinaryIO, Iterable, Optional, Tuple, Type
+from typing import (
+    Any,
+    BinaryIO,
+    Callable,
+    Dict,
+    Generator,
+    Iterable,
+    Iterator,
+    List,
+    Optional,
+    Set,
+    Tuple,
+    Type,
+    TYPE_CHECKING,
+    TypeVar,
+    Union,
+    cast,
+)
 
 import click
 
-try:
-    import pysqlite3 as sqlite3  # noqa: F401
-    from pysqlite3 import dbapi2  # noqa: F401
+from . import recipes
+
+if TYPE_CHECKING:
+    import sqlite3  # noqa: F401
+    from sqlite3 import dbapi2  # noqa: F401
 
     OperationalError = dbapi2.OperationalError
-except ImportError:
+else:
     try:
-        import sqlean as sqlite3  # noqa: F401
-        from sqlean import dbapi2  # noqa: F401
-
+        sqlite3 = importlib.import_module("pysqlite3")
+        dbapi2 = importlib.import_module("pysqlite3.dbapi2")
         OperationalError = dbapi2.OperationalError
     except ImportError:
         import sqlite3  # noqa: F401
@@ -42,8 +60,35 @@ SPATIALITE_PATHS = (
 # Mainly so we can restore it if needed in the tests:
 ORIGINAL_CSV_FIELD_SIZE_LIMIT = csv.field_size_limit()
 
+# Type alias for row dictionaries - values can be various SQLite-compatible types
+RowValue = Union[None, int, float, str, bytes, bool, List[str]]
+Row = Dict[str, RowValue]
 
-def maximize_csv_field_size_limit():
+T = TypeVar("T")
+
+
+class _CloseableIterator(Iterator[Row]):
+    """Iterator wrapper that closes a file when iteration is complete."""
+
+    def __init__(self, iterator: Iterator[Row], closeable: io.IOBase) -> None:
+        self._iterator = iterator
+        self._closeable = closeable
+
+    def __iter__(self) -> "_CloseableIterator":
+        return self
+
+    def __next__(self) -> Row:
+        try:
+            return next(self._iterator)
+        except StopIteration:
+            self._closeable.close()
+            raise
+
+    def close(self) -> None:
+        self._closeable.close()
+
+
+def maximize_csv_field_size_limit() -> None:
     """
     Increase the CSV field size limit to the maximum possible.
     """
@@ -86,20 +131,25 @@ def find_spatialite() -> Optional[str]:
     return None
 
 
-def suggest_column_types(records):
-    all_column_types = {}
+def suggest_column_types(
+    records: Iterable[Dict[str, Any]],
+) -> Dict[str, type]:
+    all_column_types: Dict[str, Set[type]] = {}
     for record in records:
         for key, value in record.items():
             all_column_types.setdefault(key, set()).add(type(value))
     return types_for_column_types(all_column_types)
 
 
-def types_for_column_types(all_column_types):
-    column_types = {}
+def types_for_column_types(
+    all_column_types: Dict[str, Set[type]],
+) -> Dict[str, type]:
+    column_types: Dict[str, type] = {}
     for key, types in all_column_types.items():
         # Ignore null values if at least one other type present:
         if len(types) > 1:
             types.discard(None.__class__)
+        t: type
         if {None.__class__} == types:
             t = str
         elif len(types) == 1:
@@ -121,7 +171,7 @@ def types_for_column_types(all_column_types):
     return column_types
 
 
-def column_affinity(column_type):
+def column_affinity(column_type: str) -> type:
     # Implementation of SQLite affinity rules from
     # https://www.sqlite.org/datatype3.html#determination_of_column_affinity
     assert isinstance(column_type, str)
@@ -140,38 +190,42 @@ def column_affinity(column_type):
     return float
 
 
-def decode_base64_values(doc):
+def decode_base64_values(doc: Dict[str, Any]) -> Dict[str, Any]:
     # Looks for '{"$base64": true..., "encoded": ...}' values and decodes them
     to_fix = [
         k
         for k in doc
         if isinstance(doc[k], dict)
-        and doc[k].get("$base64") is True
-        and "encoded" in doc[k]
+        and cast(dict, doc[k]).get("$base64") is True
+        and "encoded" in cast(dict, doc[k])
     ]
     if not to_fix:
         return doc
-    return dict(doc, **{k: base64.b64decode(doc[k]["encoded"]) for k in to_fix})
+    return dict(
+        doc, **{k: base64.b64decode(cast(dict, doc[k])["encoded"]) for k in to_fix}
+    )
 
 
 class UpdateWrapper:
-    def __init__(self, wrapped, update):
+    def __init__(self, wrapped: io.IOBase, update: Callable[[int], None]) -> None:
         self._wrapped = wrapped
         self._update = update
 
-    def __iter__(self):
+    def __iter__(self) -> Iterator[bytes]:
         for line in self._wrapped:
             self._update(len(line))
             yield line
 
-    def read(self, size=-1):
+    def read(self, size: int = -1) -> bytes:
         data = self._wrapped.read(size)
         self._update(len(data))
         return data
 
 
 @contextlib.contextmanager
-def file_progress(file, silent=False, **kwargs):
+def file_progress(
+    file: io.IOBase, silent: bool = False, **kwargs: object
+) -> Generator[Union[io.IOBase, "UpdateWrapper"], None, None]:
     if silent:
         yield file
         return
@@ -184,8 +238,8 @@ def file_progress(file, silent=False, **kwargs):
     if fileno == 0:  # 0 means stdin
         yield file
     else:
-        file_length = os.path.getsize(file.name)
-        with click.progressbar(length=file_length, **kwargs) as bar:
+        file_length = os.path.getsize(file.name)  # type: ignore
+        with click.progressbar(length=file_length, **kwargs) as bar:  # type: ignore
             yield UpdateWrapper(file, bar.update)
 
 
@@ -209,28 +263,30 @@ class RowError(Exception):
 
 
 def _extra_key_strategy(
-    reader: Iterable[dict],
+    reader: Iterable[Dict[Optional[str], object]],
     ignore_extras: Optional[bool] = False,
     extras_key: Optional[str] = None,
-) -> Iterable[dict]:
+) -> Iterable[Row]:
     # Logic for handling CSV rows with more values than there are headings
     for row in reader:
         # DictReader adds a 'None' key with extra row values
         if None not in row:
-            yield row
+            yield cast(Row, row)
         elif ignore_extras:
             # ignoring row.pop(none) because of this issue:
             # https://github.com/simonw/sqlite-utils/issues/440#issuecomment-1155358637
-            row.pop(None)  # type: ignore
-            yield row
+            row.pop(None)
+            yield cast(Row, row)
         elif not extras_key:
-            extras = row.pop(None)  # type: ignore
+            extras = row.pop(None)
             raise RowError(
                 "Row {} contained these extra values: {}".format(row, extras)
             )
         else:
-            row[extras_key] = row.pop(None)  # type: ignore
-            yield row
+            extras_value = row.pop(None)
+            row_out = cast(Row, row)
+            row_out[extras_key] = cast(RowValue, extras_value)
+            yield row_out
 
 
 def rows_from_file(
@@ -240,7 +296,7 @@ def rows_from_file(
     encoding: Optional[str] = None,
     ignore_extras: Optional[bool] = False,
     extras_key: Optional[str] = None,
-) -> Tuple[Iterable[dict], Format]:
+) -> Tuple[Iterable[Row], Format]:
     """
     Load a sequence of dictionaries from a file-like object containing one of four different formats.
 
@@ -299,12 +355,20 @@ def rows_from_file(
             reader = csv.DictReader(decoded_fp, dialect=dialect)
         else:
             reader = csv.DictReader(decoded_fp)
-        return _extra_key_strategy(reader, ignore_extras, extras_key), Format.CSV
+        rows = _extra_key_strategy(reader, ignore_extras, extras_key)
+        return _CloseableIterator(iter(rows), decoded_fp), Format.CSV
     elif format == Format.TSV:
-        rows = rows_from_file(
+        rows, _ = rows_from_file(
             fp, format=Format.CSV, dialect=csv.excel_tab, encoding=encoding
-        )[0]
-        return _extra_key_strategy(rows, ignore_extras, extras_key), Format.TSV
+        )
+        return (
+            _extra_key_strategy(
+                cast(Iterable[Dict[Optional[str], object]], rows),
+                ignore_extras,
+                extras_key,
+            ),
+            Format.TSV,
+        )
     elif format is None:
         # Detect the format, then call this recursively
         buffered = io.BufferedReader(cast(io.RawIOBase, fp), buffer_size=4096)
@@ -326,8 +390,15 @@ def rows_from_file(
                 buffered, format=Format.CSV, dialect=dialect, encoding=encoding
             )
             # Make sure we return the format we detected
-            format = Format.TSV if dialect.delimiter == "\t" else Format.CSV
-            return _extra_key_strategy(rows, ignore_extras, extras_key), format
+            detected_format = Format.TSV if dialect.delimiter == "\t" else Format.CSV
+            return (
+                _extra_key_strategy(
+                    cast(Iterable[Dict[Optional[str], object]], rows),
+                    ignore_extras,
+                    extras_key,
+                ),
+                detected_format,
+            )
     else:
         raise RowsFromFileError("Bad format")
 
@@ -353,10 +424,10 @@ class TypeTracker:
         db["creatures"].transform(types=tracker.types)
     """
 
-    def __init__(self):
-        self.trackers = {}
+    def __init__(self) -> None:
+        self.trackers: Dict[str, "ValueTracker"] = {}
 
-    def wrap(self, iterator: Iterable[dict]) -> Iterable[dict]:
+    def wrap(self, iterator: Iterable[Dict[str, Any]]) -> Iterable[Dict[str, Any]]:
         """
         Use this to loop through an existing iterator, tracking the column types
         as part of the iteration.
@@ -379,27 +450,29 @@ class TypeTracker:
 
 
 class ValueTracker:
-    def __init__(self):
+    couldbe: Dict[str, Callable[[object], bool]]
+
+    def __init__(self) -> None:
         self.couldbe = {key: getattr(self, "test_" + key) for key in self.get_tests()}
 
     @classmethod
-    def get_tests(cls):
+    def get_tests(cls) -> List[str]:
         return [
             key.split("test_")[-1]
             for key in cls.__dict__.keys()
             if key.startswith("test_")
         ]
 
-    def test_integer(self, value):
+    def test_integer(self, value: object) -> bool:
         try:
-            int(value)
+            int(cast(Any, value))
             return True
         except (ValueError, TypeError):
             return False
 
-    def test_float(self, value):
+    def test_float(self, value: object) -> bool:
         try:
-            float(value)
+            float(cast(Any, value))
             return True
         except (ValueError, TypeError):
             return False
@@ -408,7 +481,7 @@ class ValueTracker:
         return self.guessed_type + ": possibilities = " + repr(self.couldbe)
 
     @property
-    def guessed_type(self):
+    def guessed_type(self) -> str:
         options = set(self.couldbe.keys())
         # Return based on precedence
         for key in self.get_tests():
@@ -416,10 +489,10 @@ class ValueTracker:
                 return key
         return "text"
 
-    def evaluate(self, value):
+    def evaluate(self, value: object) -> None:
         if not value or not self.couldbe:
             return
-        not_these = []
+        not_these: List[str] = []
         for name, test in self.couldbe.items():
             if not test(value):
                 not_these.append(name)
@@ -428,35 +501,50 @@ class ValueTracker:
 
 
 class NullProgressBar:
-    def __init__(self, *args):
+    def __init__(self, *args: Iterable[T]) -> None:
         self.args = args
 
-    def __iter__(self):
-        yield from self.args[0]
+    def __iter__(self) -> Iterator[T]:
+        yield from self.args[0]  # type: ignore
 
-    def update(self, value):
+    def update(self, value: int) -> None:
         pass
 
 
 @contextlib.contextmanager
-def progressbar(*args, **kwargs):
+def progressbar(*args: Iterable[T], **kwargs: Any) -> Generator[Any, None, None]:
     silent = kwargs.pop("silent")
     if silent:
         yield NullProgressBar(*args)
     else:
-        with click.progressbar(*args, **kwargs) as bar:
+        with click.progressbar(*args, **kwargs) as bar:  # type: ignore
             yield bar
 
 
-def _compile_code(code, imports, variable="value"):
-    globals = {"r": recipes, "recipes": recipes}
+def _compile_code(
+    code: str, imports: Iterable[str], variable: str = "value"
+) -> Callable[..., Any]:
+    globals_dict: Dict[str, Any] = {"r": recipes, "recipes": recipes}
+    # Handle imports first so they're available for all approaches
+    for import_ in imports:
+        globals_dict[import_.split(".")[0]] = __import__(import_)
+
     # If user defined a convert() function, return that
     try:
-        exec(code, globals)
-        return globals["convert"]
+        exec(code, globals_dict)
+        return cast(Callable[..., object], globals_dict["convert"])
     except (AttributeError, SyntaxError, NameError, KeyError, TypeError):
         pass
 
+    # Check if code is a direct callable reference
+    # e.g. "r.parsedate" instead of "r.parsedate(value)"
+    try:
+        fn = eval(code, globals_dict)
+        if callable(fn):
+            return cast(Callable[..., object], fn)
+    except Exception:
+        pass
+
     # Try compiling their code as a function instead
     body_variants = [code]
     # If single line and no 'return', try adding the return
@@ -478,13 +566,11 @@ def _compile_code(code, imports, variable="value"):
     if code_o is None:
         raise SyntaxError("Could not compile code")
 
-    for import_ in imports:
-        globals[import_.split(".")[0]] = __import__(import_)
-    exec(code_o, globals)
-    return globals["fn"]
+    exec(code_o, globals_dict)
+    return cast(Callable[..., object], globals_dict["fn"])
 
 
-def chunks(sequence: Iterable, size: int) -> Iterable[Iterable]:
+def chunks(sequence: Iterable[T], size: int) -> Iterable[Iterable[T]]:
     """
     Iterate over chunks of the sequence of the given size.
 
@@ -496,7 +582,7 @@ def chunks(sequence: Iterable, size: int) -> Iterable[Iterable]:
         yield itertools.chain([item], itertools.islice(iterator, size - 1))
 
 
-def hash_record(record: Dict, keys: Optional[Iterable[str]] = None):
+def hash_record(record: Dict[str, Any], keys: Optional[Iterable[str]] = None) -> str:
     """
     ``record`` should be a Python dictionary. Returns a sha1 hash of the
     keys and values in that record.
@@ -517,7 +603,7 @@ def hash_record(record: Dict, keys: Optional[Iterable[str]] = None):
     :param record: Record to generate a hash for
     :param keys: Subset of keys to use for that hash
     """
-    to_hash = record
+    to_hash: Dict[str, Any] = record
     if keys is not None:
         to_hash = {key: record[key] for key in keys}
     return hashlib.sha1(
@@ -527,7 +613,38 @@ def hash_record(record: Dict, keys: Optional[Iterable[str]] = None):
     ).hexdigest()
 
 
-def _flatten(d):
+def dedupe_keys(keys: Iterable[str]) -> List[str]:
+    """
+    Rename duplicates in a list of column names so every name is unique,
+    by appending ``_2``, ``_3``... to later occurrences - skipping any
+    suffix that would collide with another column in the list.
+
+    Used when converting SQL query rows to dictionaries, where duplicate
+    column names would otherwise silently overwrite each other.
+
+    :param keys: List of column names, possibly containing duplicates
+    """
+    keys = list(keys)
+    taken = set(keys)
+    if len(taken) == len(keys):
+        # No duplicates - the common case
+        return keys
+    seen: set = set()
+    result = []
+    for key in keys:
+        if key in seen:
+            new_key = key
+            suffix = 2
+            while new_key in seen or new_key in taken:
+                new_key = "{}_{}".format(key, suffix)
+                suffix += 1
+            key = new_key
+        seen.add(key)
+        result.append(key)
+    return result
+
+
+def _flatten(d: Dict[str, Any]) -> Generator[Tuple[str, Any], None, None]:
     for key, value in d.items():
         if isinstance(value, dict):
             for key2, value2 in _flatten(value):
@@ -536,7 +653,7 @@ def _flatten(d):
             yield key, value
 
 
-def flatten(row: dict) -> dict:
+def flatten(row: Dict[str, Any]) -> Dict[str, Any]:
     """
     Turn a nested dict e.g. ``{"a": {"b": 1}}`` into a flat dict: ``{"a_b": 1}``
 

tests/conftest.py

@@ -8,10 +8,55 @@ create table Gosh2 (c1 text, c2 text, c3 text);
 """
 
 
+def pytest_addoption(parser):
+    parser.addoption(
+        "--sqlite-autocommit",
+        action="store_true",
+        default=False,
+        help=(
+            "Run every test against connections created with the Python 3.12+ "
+            "sqlite3.connect(autocommit=True) mode"
+        ),
+    )
+
+
 def pytest_configure(config):
     import sys
 
-    sys._called_from_test = True
+    sys._called_from_test = True  # type: ignore[attr-defined]
+
+    if config.getoption("--sqlite-autocommit"):
+        if sys.version_info < (3, 12):
+            raise pytest.UsageError(
+                "--sqlite-autocommit requires Python 3.12 or higher"
+            )
+        real_connect = sqlite3.connect
+
+        def autocommit_connect(*args, **kwargs):
+            kwargs.setdefault("autocommit", True)
+            return real_connect(*args, **kwargs)
+
+        sqlite3.connect = autocommit_connect
+
+
+@pytest.fixture(autouse=True)
+def close_all_databases():
+    """Automatically close all Database objects created during a test."""
+    databases = []
+    original_init = Database.__init__
+
+    def tracking_init(self, *args, **kwargs):
+        original_init(self, *args, **kwargs)
+        databases.append(self)
+
+    Database.__init__ = tracking_init  # type: ignore[method-assign]
+    yield
+    Database.__init__ = original_init  # type: ignore[method-assign]
+    for db in databases:
+        try:
+            db.close()
+        except Exception:
+            pass
 
 
 @pytest.fixture
@@ -22,14 +67,12 @@ def fresh_db():
 @pytest.fixture
 def existing_db():
     database = Database(memory=True)
-    database.executescript(
-        """
+    database.executescript("""
         CREATE TABLE foo (text TEXT);
         INSERT INTO foo (text) values ("one");
         INSERT INTO foo (text) values ("two");
         INSERT INTO foo (text) values ("three");
-    """
-    )
+    """)
     return database
 
 
@@ -38,4 +81,5 @@ def db_path(tmpdir):
     path = str(tmpdir / "test.db")
     db = sqlite3.connect(path)
     db.executescript(CREATE_TABLES)
+    db.close()
     return path

tests/test_analyze_tables.py

@@ -137,15 +137,13 @@ def db_to_analyze_path(db_to_analyze, tmpdir):
     db = sqlite3.connect(path)
     sql = "\n".join(db_to_analyze.iterdump())
     db.executescript(sql)
+    db.close()
     return path
 
 
 def test_analyze_table(db_to_analyze_path):
     result = CliRunner().invoke(cli.cli, ["analyze-tables", db_to_analyze_path])
-    assert (
-        result.output.strip()
-        == (
-            """
+    assert result.output.strip() == ("""
 stuff.id: (1/3)
 
   Total rows: 8
@@ -178,9 +176,7 @@ stuff.size: (3/3)
 
   Most common:
     5: 5
-    3: 4"""
-        ).strip()
-    )
+    3: 4""").strip()
 
 
 def test_analyze_table_save(db_to_analyze_path):

tests/test_atomic.py

@@ -0,0 +1,382 @@
+import pytest
+
+from sqlite_utils.db import Database, _iter_complete_sql_statements
+from sqlite_utils.utils import sqlite3
+
+
+@pytest.mark.parametrize(
+    "sql,expected",
+    (
+        (
+            "CREATE TABLE t(id); INSERT INTO t VALUES (1)",
+            ["CREATE TABLE t(id);", "INSERT INTO t VALUES (1)"],
+        ),
+        (
+            "INSERT INTO t VALUES ('a;b');",
+            ["INSERT INTO t VALUES ('a;b');"],
+        ),
+        (
+            "-- comment;\nCREATE TABLE t(id);",
+            ["-- comment;\nCREATE TABLE t(id);"],
+        ),
+        (
+            """
+            CREATE TRIGGER t_ai AFTER INSERT ON t
+            BEGIN
+                UPDATE t SET value = 'a;b' WHERE id = new.id;
+                INSERT INTO log VALUES ('x;y');
+            END;
+            """,
+            [
+                "CREATE TRIGGER t_ai AFTER INSERT ON t\n"
+                "            BEGIN\n"
+                "                UPDATE t SET value = 'a;b' WHERE id = new.id;\n"
+                "                INSERT INTO log VALUES ('x;y');\n"
+                "            END;"
+            ],
+        ),
+    ),
+)
+def test_iter_complete_sql_statements(sql, expected):
+    assert list(_iter_complete_sql_statements(sql)) == expected
+
+
+def test_atomic_commits(fresh_db):
+    with fresh_db.atomic():
+        fresh_db["dogs"].insert({"id": 1, "name": "Cleo"}, pk="id")
+
+    assert list(fresh_db["dogs"].rows) == [{"id": 1, "name": "Cleo"}]
+
+
+def test_atomic_rolls_back(fresh_db):
+    with pytest.raises(RuntimeError):
+        with fresh_db.atomic():
+            fresh_db["dogs"].insert({"id": 1, "name": "Cleo"}, pk="id")
+            raise RuntimeError("boom")
+
+    assert not fresh_db["dogs"].exists()
+
+
+def test_nested_atomic_rolls_back_to_savepoint(fresh_db):
+    fresh_db["dogs"].create({"id": int, "name": str}, pk="id")
+
+    with fresh_db.atomic():
+        fresh_db["dogs"].insert({"id": 1, "name": "Cleo"})
+        with pytest.raises(RuntimeError):
+            with fresh_db.atomic():
+                fresh_db["dogs"].insert({"id": 2, "name": "Pancakes"})
+                raise RuntimeError("boom")
+        fresh_db["dogs"].insert({"id": 3, "name": "Marnie"})
+
+    assert list(fresh_db["dogs"].rows) == [
+        {"id": 1, "name": "Cleo"},
+        {"id": 3, "name": "Marnie"},
+    ]
+
+
+def test_outer_atomic_rolls_back_released_savepoint(fresh_db):
+    with pytest.raises(RuntimeError):
+        with fresh_db.atomic():
+            fresh_db["dogs"].insert({"id": 1, "name": "Cleo"}, pk="id")
+            with fresh_db.atomic():
+                fresh_db["dogs"].insert({"id": 2, "name": "Pancakes"})
+            raise RuntimeError("boom")
+
+    assert not fresh_db["dogs"].exists()
+
+
+def test_executescript_does_not_commit_open_atomic_block(fresh_db):
+    with pytest.raises(RuntimeError):
+        with fresh_db.atomic():
+            fresh_db.executescript("""
+                CREATE TABLE dogs(id INTEGER PRIMARY KEY, name TEXT);
+                CREATE TRIGGER dogs_ai AFTER INSERT ON dogs
+                BEGIN
+                    UPDATE dogs SET name = upper(new.name) || '; updated' WHERE id = new.id;
+                END;
+                -- This comment has a semicolon;
+                INSERT INTO dogs VALUES (1, 'Cleo; the first');
+            """)
+            raise RuntimeError("boom")
+
+    assert not fresh_db["dogs"].exists()
+
+
+def test_transform_does_not_commit_open_atomic_block(fresh_db):
+    fresh_db["dogs"].insert({"id": 1, "name": "Cleo", "age": "5"}, pk="id")
+
+    with pytest.raises(RuntimeError):
+        with fresh_db.atomic():
+            fresh_db["dogs"].insert({"id": 2, "name": "Pancakes", "age": "6"})
+            fresh_db["dogs"].transform(rename={"age": "dog_age"})
+            raise RuntimeError("boom")
+
+    assert (
+        fresh_db["dogs"].schema
+        == 'CREATE TABLE "dogs" (\n   "id" INTEGER PRIMARY KEY,\n   "name" TEXT,\n   "age" TEXT\n)'
+    )
+    assert list(fresh_db["dogs"].rows) == [
+        {"id": 1, "name": "Cleo", "age": "5"},
+    ]
+
+
+def test_transform_parent_table_with_foreign_keys_in_atomic(fresh_db):
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db["authors"].insert({"id": 1, "name": "Tina"}, pk="id")
+    fresh_db["books"].insert(
+        {"id": 1, "title": "Book", "author_id": 1},
+        pk="id",
+        foreign_keys={"author_id"},
+    )
+
+    with fresh_db.atomic():
+        fresh_db["authors"].transform(rename={"name": "full_name"})
+        assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
+
+    assert (
+        fresh_db["authors"].schema
+        == 'CREATE TABLE "authors" (\n   "id" INTEGER PRIMARY KEY,\n   "full_name" TEXT\n)'
+    )
+    assert fresh_db.execute("PRAGMA foreign_key_check").fetchall() == []
+
+
+def test_transform_parent_table_with_foreign_keys_rolls_back(fresh_db):
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db["authors"].insert({"id": 1, "name": "Tina"}, pk="id")
+    fresh_db["books"].insert(
+        {"id": 1, "title": "Book", "author_id": 1},
+        pk="id",
+        foreign_keys={"author_id"},
+    )
+
+    with pytest.raises(RuntimeError):
+        with fresh_db.atomic():
+            fresh_db["authors"].transform(rename={"name": "full_name"})
+            raise RuntimeError("boom")
+
+    assert (
+        fresh_db["authors"].schema
+        == 'CREATE TABLE "authors" (\n   "id" INTEGER PRIMARY KEY,\n   "name" TEXT\n)'
+    )
+    assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
+    assert fresh_db.execute("PRAGMA foreign_key_check").fetchall() == []
+
+
+def test_transform_detects_foreign_key_check_violations(fresh_db):
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db["authors"].insert({"id": 1, "name": "Tina"}, pk="id")
+    fresh_db["books"].insert({"id": 1, "author_id": 2}, pk="id")
+
+    with pytest.raises(sqlite3.IntegrityError):
+        fresh_db["books"].transform(add_foreign_keys=(("author_id", "authors", "id"),))
+
+    assert fresh_db["books"].foreign_keys == []
+    assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
+
+
+def test_atomic_inside_manual_transaction_uses_savepoint(fresh_db):
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    fresh_db.execute("begin")
+    with fresh_db.atomic():
+        fresh_db["t"].insert({"id": 2}, pk="id")
+    # Nothing is committed until the user's own transaction commits
+    assert fresh_db.conn.in_transaction
+    fresh_db.rollback()
+    assert [r["id"] for r in fresh_db["t"].rows] == [1]
+    # And with a commit instead, the atomic block's writes persist
+    fresh_db.execute("begin")
+    with fresh_db.atomic():
+        fresh_db["t"].insert({"id": 3}, pk="id")
+    fresh_db.commit()
+    assert [r["id"] for r in fresh_db["t"].rows] == [1, 3]
+
+
+def test_begin_commit_rollback(tmpdir):
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    db["t"].insert({"id": 1}, pk="id")
+    db.begin()
+    db["t"].insert({"id": 2}, pk="id")
+    assert db.conn.in_transaction
+    db.rollback()
+    assert not db.conn.in_transaction
+    assert [r["id"] for r in db["t"].rows] == [1]
+    db.begin()
+    db["t"].insert({"id": 3}, pk="id")
+    db.commit()
+    db.close()
+    db2 = Database(path)
+    assert [r["id"] for r in db2["t"].rows] == [1, 3]
+    db2.close()
+
+
+def test_begin_inside_transaction_errors(fresh_db):
+    fresh_db.begin()
+    with pytest.raises(sqlite3.OperationalError):
+        fresh_db.begin()
+    fresh_db.rollback()
+
+
+def test_commit_and_rollback_without_transaction_are_noops(fresh_db):
+    fresh_db.commit()
+    fresh_db.rollback()
+    assert not fresh_db.conn.in_transaction
+
+
+def test_execute_write_commits_immediately(tmpdir):
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    db["t"].insert({"id": 1}, pk="id")
+    db.execute("insert into t (id) values (2)")
+    # No implicit transaction is left open
+    assert not db.conn.in_transaction
+    # A completely separate connection sees the row straight away
+    other = sqlite3.connect(path)
+    assert other.execute("select count(*) from t").fetchone()[0] == 2
+    other.close()
+    db.close()
+
+
+def test_execute_write_respects_explicit_transaction(fresh_db):
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    fresh_db.begin()
+    fresh_db.execute("insert into t (id) values (2)")
+    # Still inside the explicit transaction - not committed
+    assert fresh_db.conn.in_transaction
+    fresh_db.rollback()
+    assert [r["id"] for r in fresh_db["t"].rows] == [1]
+
+
+def test_execute_comment_prefixed_begin_leaves_transaction_open(fresh_db):
+    # A BEGIN hidden behind a leading comment must not be auto-committed
+    # out from under the caller
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    fresh_db.execute("-- start a transaction\nbegin")
+    assert fresh_db.conn.in_transaction
+    fresh_db.execute("insert into t (id) values (2)")
+    fresh_db.rollback()
+    assert [r["id"] for r in fresh_db["t"].rows] == [1]
+
+
+def _sqlite_accepts_bom():
+    try:
+        sqlite3.connect(":memory:").execute("\ufeffselect 1")
+        return True
+    except sqlite3.OperationalError:
+        return False
+
+
+@pytest.mark.parametrize("begin_sql", ["; begin", "\ufeffbegin"])
+def test_execute_prefixed_begin_leaves_transaction_open(fresh_db, begin_sql):
+    # sqlite3 tolerates empty statements and a UTF-8 BOM before the first
+    # real token, so a BEGIN behind either must not be auto-committed
+    # out from under the caller
+    if begin_sql.startswith("\ufeff") and not _sqlite_accepts_bom():
+        pytest.skip("This SQLite version rejects a leading byte order mark")
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    fresh_db.execute(begin_sql)
+    assert fresh_db.conn.in_transaction
+    fresh_db.execute("insert into t (id) values (2)")
+    fresh_db.rollback()
+    assert [r["id"] for r in fresh_db["t"].rows] == [1]
+
+
+def test_execute_failed_write_rolls_back_implicit_transaction(tmpdir):
+    # A failed write must not leave the driver's implicit transaction open -
+    # that would silently disable auto-commit for every subsequent write
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    db["t"].insert({"id": 1}, pk="id")
+    with pytest.raises(sqlite3.IntegrityError):
+        db.execute("insert into t (id) values (1)")
+    assert not db.conn.in_transaction
+    # Subsequent writes commit as normal and survive closing the connection
+    db["other"].insert({"id": 2})
+    db.close()
+    db2 = Database(path)
+    assert db2["other"].exists()
+    db2.close()
+
+
+def test_execute_failed_write_preserves_explicit_transaction(fresh_db):
+    # A failed write inside an explicit transaction must not roll back
+    # the caller's earlier work - only the caller decides that
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    fresh_db.begin()
+    fresh_db.execute("insert into t (id) values (2)")
+    with pytest.raises(sqlite3.IntegrityError):
+        fresh_db.execute("insert into t (id) values (1)")
+    assert fresh_db.conn.in_transaction
+    fresh_db.commit()
+    assert [r["id"] for r in fresh_db["t"].rows] == [1, 2]
+
+
+def test_execute_failed_write_inside_atomic_preserves_block(fresh_db):
+    # A caught failure inside an atomic() block must leave the block's
+    # transaction open so its other work still commits
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    with fresh_db.atomic():
+        fresh_db.execute("insert into t (id) values (2)")
+        with pytest.raises(sqlite3.IntegrityError):
+            fresh_db.execute("insert into t (id) values (1)")
+    assert [r["id"] for r in fresh_db["t"].rows] == [1, 2]
+
+
+def test_query_returning_commits_after_iteration(tmpdir):
+    if sqlite3.sqlite_version_info < (3, 35, 0):
+        import pytest as _pytest
+
+        _pytest.skip("RETURNING requires SQLite 3.35.0 or higher")
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    db["t"].insert({"id": 1}, pk="id")
+    rows = list(db.query("insert into t (id) values (2) returning id"))
+    assert rows == [{"id": 2}]
+    assert not db.conn.in_transaction
+    other = sqlite3.connect(path)
+    assert other.execute("select count(*) from t").fetchone()[0] == 2
+    other.close()
+    db.close()
+
+
+TRIGGER_SQL = """
+create trigger no_bad before insert on t
+when new.v = 'bad'
+begin
+    select raise(rollback, 'trigger says no');
+end
+"""
+
+
+def test_atomic_preserves_error_from_transaction_destroying_trigger(fresh_db):
+    # RAISE(ROLLBACK) rolls back the whole transaction and destroys every
+    # savepoint - atomic()'s cleanup must not mask the IntegrityError
+    # with "cannot rollback - no transaction is active"
+    fresh_db.execute("create table t (id integer primary key, v text)")
+    fresh_db.execute(TRIGGER_SQL)
+    with pytest.raises(sqlite3.IntegrityError, match="trigger says no"):
+        with fresh_db.atomic():
+            fresh_db.execute("insert into t (v) values ('bad')")
+    assert not fresh_db.conn.in_transaction
+
+
+def test_nested_atomic_preserves_error_from_transaction_destroying_trigger(
+    fresh_db,
+):
+    # The nested savepoint branch previously raised
+    # "no such savepoint" from ROLLBACK TO SAVEPOINT
+    fresh_db.execute("create table t (id integer primary key, v text)")
+    fresh_db.execute(TRIGGER_SQL)
+    with pytest.raises(sqlite3.IntegrityError, match="trigger says no"):
+        with fresh_db.atomic():
+            with fresh_db.atomic():
+                fresh_db.execute("insert into t (v) values ('bad')")
+    assert not fresh_db.conn.in_transaction
+
+
+def test_atomic_preserves_error_from_insert_or_rollback(fresh_db):
+    fresh_db["t"].insert({"id": 1}, pk="id")
+    with pytest.raises(sqlite3.IntegrityError):
+        with fresh_db.atomic():
+            fresh_db.execute("insert or rollback into t (id) values (1)")
+    assert not fresh_db.conn.in_transaction

tests/test_cli.py

@@ -3,6 +3,7 @@ from sqlite_utils.db import Index, ForeignKey
 from click.testing import CliRunner
 from pathlib import Path
 import subprocess
+import sqlite3
 import sys
 import json
 import os
@@ -19,9 +20,11 @@ def _supports_pragma_function_list():
     db = Database(memory=True)
     try:
         db.execute("select * from pragma_function_list()")
+        return True
     except Exception:
         return False
-    return True
+    finally:
+        db.close()
 
 
 def _has_compiled_ext():
@@ -193,6 +196,50 @@ def test_output_table(db_path, options, expected):
     assert expected == result.output.strip()
 
 
+@pytest.mark.parametrize(
+    "fmt_option", [["--fmt", "simple"], ["-t"], ["--fmt", "github"]]
+)
+def test_output_table_no_headers(db_path, fmt_option):
+    # --no-headers should omit the header row from --fmt/--table output too, not
+    # just from --csv/--tsv (#566). Previously the flag was silently ignored for
+    # tabulate formats and the column names were always printed.
+    db = Database(db_path)
+    with db.conn:
+        db["dogs"].insert_all(
+            [
+                {"id": 1, "name": "Cleo", "age": 4},
+                {"id": 2, "name": "Pancakes", "age": 2},
+            ]
+        )
+    sql = "select id, name, age from dogs order by id"
+
+    with_headers = CliRunner().invoke(cli.cli, ["query", db_path, sql] + fmt_option)
+    without_headers = CliRunner().invoke(
+        cli.cli, ["query", db_path, sql] + fmt_option + ["--no-headers"]
+    )
+    assert with_headers.exit_code == 0
+    assert without_headers.exit_code == 0
+
+    # The column names appear when headers are shown, and must not appear at all
+    # once --no-headers is passed.
+    assert "name" in with_headers.output
+    for header in ("id", "name", "age"):
+        assert (
+            header not in without_headers.output
+        ), f"header {header!r} leaked into --no-headers output"
+    # The data is still all present.
+    for value in ("Cleo", "Pancakes", "1", "2", "4"):
+        assert value in without_headers.output
+
+    # The rows command shares the same code path.
+    rows_no_headers = CliRunner().invoke(
+        cli.cli, ["rows", db_path, "dogs"] + fmt_option + ["--no-headers"]
+    )
+    assert rows_no_headers.exit_code == 0
+    assert "name" not in rows_no_headers.output
+    assert "Cleo" in rows_no_headers.output
+
+
 def test_create_index(db_path):
     db = Database(db_path)
     assert [] == db["Gosh"].indexes
@@ -245,6 +292,24 @@ def test_create_index(db_path):
         )
 
 
+def test_drop_index(db_path):
+    db = Database(db_path)
+    db["Gosh"].create_index(["c1"])
+    assert [index.name for index in db["Gosh"].indexes] == ["idx_Gosh_c1"]
+    result = CliRunner().invoke(cli.cli, ["drop-index", db_path, "Gosh", "idx_Gosh_c1"])
+    assert result.exit_code == 0
+    assert db["Gosh"].indexes == []
+
+    result = CliRunner().invoke(cli.cli, ["drop-index", db_path, "Gosh", "idx_Gosh_c1"])
+    assert result.exit_code == 1
+    assert "No index named idx_Gosh_c1" in result.output
+
+    result = CliRunner().invoke(
+        cli.cli, ["drop-index", db_path, "Gosh", "idx_Gosh_c1", "--ignore"]
+    )
+    assert result.exit_code == 0
+
+
 def test_create_index_analyze(db_path):
     db = Database(db_path)
     assert "sqlite_stat1" not in db.table_names()
@@ -736,6 +801,25 @@ def test_query_json(db_path, sql, args, expected):
     assert expected == result.output.strip()
 
 
+def test_query_sql_from_stdin(db_path):
+    # https://github.com/simonw/sqlite-utils/issues/765
+    db = Database(db_path)
+    with db.conn:
+        db["dogs"].insert_all(
+            [
+                {"id": 1, "age": 4, "name": "Cleo"},
+                {"id": 2, "age": 2, "name": "Pancakes"},
+            ]
+        )
+    result = CliRunner().invoke(
+        cli.cli,
+        ["query", db_path, "-"],
+        input="select name from dogs order by name",
+    )
+    assert result.exit_code == 0, result.output
+    assert json.loads(result.output) == [{"name": "Cleo"}, {"name": "Pancakes"}]
+
+
 def test_query_json_empty(db_path):
     result = CliRunner().invoke(
         cli.cli,
@@ -744,6 +828,26 @@ def test_query_json_empty(db_path):
     assert result.output.strip() == "[]"
 
 
+def test_query_json_duplicate_columns_are_deduped(db_path):
+    # https://github.com/simonw/sqlite-utils/issues/624
+    result = CliRunner().invoke(
+        cli.cli,
+        [db_path, "select 1 as id, 2 as id, 'x' as value, 'y' as value"],
+    )
+    assert result.output.strip() == (
+        '[{"id": 1, "id_2": 2, "value": "x", "value_2": "y"}]'
+    )
+
+
+def test_query_csv_duplicate_columns_are_preserved(db_path):
+    # CSV output should keep the duplicate headers, not rename them
+    result = CliRunner().invoke(
+        cli.cli,
+        [db_path, "select 1 as id, 2 as id", "--csv"],
+    )
+    assert result.output.replace("\r", "").strip() == "id,id\n1,2"
+
+
 def test_query_invalid_function(db_path):
     result = CliRunner().invoke(
         cli.cli, [db_path, "select bad()", "--functions", "def invalid_python"]
@@ -965,12 +1069,9 @@ def test_query_json_with_json_cols(db_path):
     result = CliRunner().invoke(
         cli.cli, [db_path, "select id, name, friends from dogs"]
     )
-    assert (
-        r"""
+    assert r"""
     [{"id": 1, "name": "Cleo", "friends": "[{\"name\": \"Pancakes\"}, {\"name\": \"Bailey\"}]"}]
-    """.strip()
-        == result.output.strip()
-    )
+    """.strip() == result.output.strip()
     # With --json-cols:
     result = CliRunner().invoke(
         cli.cli, [db_path, "select id, name, friends from dogs", "--json-cols"]
@@ -984,6 +1085,34 @@ def test_query_json_with_json_cols(db_path):
     assert expected == result_rows.output.strip()
 
 
+def test_query_json_unicode_not_escaped_by_default(db_path):
+    db = Database(db_path)
+    with db.conn:
+        db["text"].insert({"id": 1, "text": "Japanese 日本語"}, pk="id")
+    result = CliRunner().invoke(cli.cli, [db_path, "select id, text from text"])
+    assert result.exit_code == 0
+    assert result.output.strip() == '[{"id": 1, "text": "Japanese 日本語"}]'
+    # Same for --nl
+    result = CliRunner().invoke(cli.cli, [db_path, "select id, text from text", "--nl"])
+    assert result.exit_code == 0
+    assert result.output.strip() == '{"id": 1, "text": "Japanese 日本語"}'
+
+
+@pytest.mark.parametrize("command", ["query", "rows"])
+def test_query_json_ascii_option(db_path, command):
+    db = Database(db_path)
+    with db.conn:
+        db["text"].insert({"id": 1, "text": "Japanese 日本語"}, pk="id")
+    if command == "query":
+        args = [db_path, "select id, text from text", "--ascii"]
+    else:
+        args = ["rows", db_path, "text", "--ascii"]
+    result = CliRunner().invoke(cli.cli, args)
+    assert result.exit_code == 0
+    expected = '[{"id": 1, "text": "Japanese ' + "\\u65e5\\u672c\\u8a9e" + '"}]'
+    assert result.output.strip() == expected
+
+
 @pytest.mark.parametrize(
     "content,is_binary",
     [(b"\x00\x0fbinary", True), ("this is text", False), (1, False), (1.5, False)],
@@ -1125,20 +1254,38 @@ def test_upsert(db_path, tmpdir):
     ]
 
 
-def test_upsert_pk_required(db_path, tmpdir):
+def test_upsert_pk_inferred_from_existing_table(db_path, tmpdir):
     json_path = str(tmpdir / "dogs.json")
+    db = Database(db_path)
     insert_dogs = [
         {"id": 1, "name": "Cleo", "age": 4},
         {"id": 2, "name": "Nixie", "age": 4},
     ]
     write_json(json_path, insert_dogs)
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "dogs", json_path, "--pk", "id"],
+        catch_exceptions=False,
+    )
+    assert result.exit_code == 0, result.output
+
+    write_json(
+        json_path,
+        [
+            {"id": 1, "age": 5},
+            {"id": 2, "age": 5},
+        ],
+    )
     result = CliRunner().invoke(
         cli.cli,
         ["upsert", db_path, "dogs", json_path],
         catch_exceptions=False,
     )
-    assert result.exit_code == 2
-    assert "Error: Missing option '--pk'" in result.output
+    assert result.exit_code == 0, result.output
+    assert list(db.query("select * from dogs order by id")) == [
+        {"id": 1, "name": "Cleo", "age": 5},
+        {"id": 2, "name": "Nixie", "age": 5},
+    ]
 
 
 def test_upsert_analyze(db_path, tmpdir):
@@ -1468,6 +1615,24 @@ def test_drop_table_error():
         assert result.exit_code == 0
 
 
+def test_drop_table_on_view_errors():
+    runner = CliRunner()
+    with runner.isolated_filesystem():
+        db = Database("test.db")
+        db["t"].insert({"id": 1})
+        db.create_view("v", "select * from t")
+        result = runner.invoke(cli.cli, ["drop-table", "test.db", "v"])
+        assert result.exit_code == 1
+        assert 'Error: "v" is a view, not a table - use drop-view to drop it' == (
+            result.output.strip()
+        )
+        assert "v" in db.view_names()
+        # --ignore exits cleanly but must still not drop the view
+        result = runner.invoke(cli.cli, ["drop-table", "test.db", "v", "--ignore"])
+        assert result.exit_code == 0
+        assert "v" in db.view_names()
+
+
 def test_drop_view():
     runner = CliRunner()
     with runner.isolated_filesystem():
@@ -1486,6 +1651,23 @@ def test_drop_view():
         assert "hello" not in db.view_names()
 
 
+def test_drop_view_on_table_errors():
+    runner = CliRunner()
+    with runner.isolated_filesystem():
+        db = Database("test.db")
+        db["t"].insert({"id": 1})
+        result = runner.invoke(cli.cli, ["drop-view", "test.db", "t"])
+        assert result.exit_code == 1
+        assert 'Error: "t" is a table, not a view - use drop-table to drop it' == (
+            result.output.strip()
+        )
+        assert "t" in db.table_names()
+        # --ignore exits cleanly but must still not drop the table
+        result = runner.invoke(cli.cli, ["drop-view", "test.db", "t", "--ignore"])
+        assert result.exit_code == 0
+        assert "t" in db.table_names()
+
+
 def test_drop_view_error():
     runner = CliRunner()
     with runner.isolated_filesystem():
@@ -1735,6 +1917,87 @@ def test_transform(db_path, args, expected_schema):
     assert schema == expected_schema
 
 
+def test_transform_sql(db_path):
+    db = Database(db_path)
+    with db.conn:
+        db["dogs"].insert(
+            {"id": 1, "age": 4, "name": "Cleo"},
+            not_null={"age"},
+            defaults={"age": 1},
+            pk="id",
+        )
+    original_schema = db["dogs"].schema
+
+    result = CliRunner().invoke(
+        cli.cli, ["transform", db_path, "dogs", "--drop", "name", "--sql"]
+    )
+
+    assert result.exit_code == 0, result.output
+    assert 'CREATE TABLE "dogs_new_' in result.output
+    assert '"age" INTEGER NOT NULL DEFAULT' in result.output
+    assert 'DROP TABLE "dogs";' in result.output
+    assert 'ALTER TABLE "dogs_new_' in result.output
+    assert db["dogs"].schema == original_schema
+
+
+@pytest.mark.parametrize(
+    "initial_strict,args,expected_strict",
+    (
+        (False, [], False),
+        (True, [], True),
+        (False, ["--strict"], True),
+        (True, ["--no-strict"], False),
+    ),
+)
+def test_transform_strict_option(db_path, initial_strict, args, expected_strict):
+    db = Database(db_path)
+    if not db.supports_strict:
+        pytest.skip("SQLite version does not support strict tables")
+    db["dogs"].create({"id": int}, strict=initial_strict)
+
+    result = CliRunner().invoke(cli.cli, ["transform", db_path, "dogs"] + args)
+
+    assert result.exit_code == 0, result.output
+    assert db["dogs"].strict is expected_strict
+
+
+@pytest.mark.parametrize(
+    "initial_strict,flag,sql_is_strict",
+    (
+        (False, "--strict", True),
+        (True, "--no-strict", False),
+    ),
+)
+def test_transform_strict_option_sql(db_path, initial_strict, flag, sql_is_strict):
+    db = Database(db_path)
+    if not db.supports_strict:
+        pytest.skip("SQLite version does not support strict tables")
+    db["dogs"].create({"id": int}, strict=initial_strict)
+
+    result = CliRunner().invoke(cli.cli, ["transform", db_path, "dogs", flag, "--sql"])
+
+    assert result.exit_code == 0, result.output
+    assert (") STRICT;" in result.output) is sql_is_strict
+    assert db["dogs"].strict is initial_strict
+
+
+def test_transform_strict_option_with_invalid_data(db_path):
+    db = Database(db_path)
+    if not db.supports_strict:
+        pytest.skip("SQLite version does not support strict tables")
+    dogs = db["dogs"]
+    dogs.create({"id": int})
+    dogs.insert({"id": "not-an-integer"})
+
+    result = CliRunner().invoke(cli.cli, ["transform", db_path, "dogs", "--strict"])
+
+    assert result.exit_code == 1
+    assert isinstance(result.exception, sqlite3.IntegrityError)
+    assert dogs.strict is False
+    assert list(dogs.rows) == [{"id": "not-an-integer"}]
+    assert not any(name.startswith("dogs_new_") for name in db.table_names())
+
+
 @pytest.mark.parametrize(
     "extra_args,expected_schema",
     (
@@ -1996,12 +2259,10 @@ def test_search_quote(tmpdir):
 def test_indexes(tmpdir):
     db_path = str(tmpdir / "test.db")
     db = Database(db_path)
-    db.conn.executescript(
-        """
+    db.conn.executescript("""
         create table Gosh (c1 text, c2 text, c3 text);
         create index Gosh_idx on Gosh(c2, c3 desc);
-    """
-    )
+    """)
     result = CliRunner().invoke(
         cli.cli,
         ["indexes", str(db_path)],
@@ -2092,16 +2353,12 @@ def test_triggers(tmpdir, extra_args, expected):
         pk="id",
     )
     db["counter"].insert({"count": 1})
-    db.conn.execute(
-        textwrap.dedent(
-            """
+    db.conn.execute(textwrap.dedent("""
         CREATE TRIGGER blah AFTER INSERT ON articles
         BEGIN
             UPDATE counter SET count = count + 1;
         END
-    """
-        )
-    )
+    """))
     args = ["triggers", db_path]
     if extra_args:
         args.extend(extra_args)
@@ -2280,18 +2537,14 @@ def test_csv_insert_bom(tmpdir):
     ]
 
 
-@pytest.mark.parametrize("option", (None, "-d", "--detect-types"))
-def test_insert_detect_types(tmpdir, option):
-    """Test that type detection is now the default behavior"""
+def test_insert_detect_types(tmpdir):
+    """Test that type detection is the default behavior"""
     db_path = str(tmpdir / "test.db")
     data = "name,age,weight\nCleo,6,45.5\nDori,1,3.5"
-    extra = []
-    if option:
-        extra = [option]
 
     result = CliRunner().invoke(
         cli.cli,
-        ["insert", db_path, "creatures", "-", "--csv"] + extra,
+        ["insert", db_path, "creatures", "-", "--csv"],
         catch_exceptions=False,
         input=data,
     )
@@ -2303,17 +2556,27 @@ def test_insert_detect_types(tmpdir, option):
     ]
 
 
-@pytest.mark.parametrize("option", (None, "-d", "--detect-types"))
-def test_upsert_detect_types(tmpdir, option):
-    """Test that type detection is now the default behavior for upsert"""
+@pytest.mark.parametrize("command", ("insert", "upsert"))
+@pytest.mark.parametrize("option", ("-d", "--detect-types"))
+def test_detect_types_flag_removed(tmpdir, command, option):
+    # The old no-op flag was removed in 4.0 - it should now error
     db_path = str(tmpdir / "test.db")
-    data = "id,name,age,weight\n1,Cleo,6,45.5\n2,Dori,1,3.5"
-    extra = []
-    if option:
-        extra = [option]
     result = CliRunner().invoke(
         cli.cli,
-        ["upsert", db_path, "creatures", "-", "--csv", "--pk", "id"] + extra,
+        [command, db_path, "creatures", "-", "--csv", "--pk", "id", option],
+        input="id,name\n1,Cleo",
+    )
+    assert result.exit_code == 2
+    assert "No such option" in result.output
+
+
+def test_upsert_detect_types(tmpdir):
+    """Test that type detection is the default behavior for upsert"""
+    db_path = str(tmpdir / "test.db")
+    data = "id,name,age,weight\n1,Cleo,6,45.5\n2,Dori,1,3.5"
+    result = CliRunner().invoke(
+        cli.cli,
+        ["upsert", db_path, "creatures", "-", "--csv", "--pk", "id"],
         catch_exceptions=False,
         input=data,
     )
@@ -2609,3 +2872,22 @@ def test_insert_upsert_strict(tmpdir, method, strict):
     assert result.exit_code == 0
     db = Database(db_path)
     assert db["items"].strict == strict or not db.supports_strict
+
+
+def test_extract_bad_column_clean_error(db_path):
+    db = Database(db_path)
+    db["trees"].insert({"id": 1, "species": "Palm"}, pk="id")
+    result = CliRunner().invoke(cli.cli, ["extract", db_path, "trees", "nope"])
+    assert result.exit_code == 1
+    assert result.exception is None or isinstance(result.exception, SystemExit)
+    assert result.output.startswith("Error: Invalid columns")
+
+
+def test_extract_view_clean_error(db_path):
+    db = Database(db_path)
+    db["trees"].insert({"id": 1, "species": "Palm"}, pk="id")
+    db.create_view("v", "select * from trees")
+    result = CliRunner().invoke(cli.cli, ["extract", db_path, "v", "species"])
+    assert result.exit_code == 1
+    assert result.exception is None or isinstance(result.exception, SystemExit)
+    assert result.output.startswith("Error:")

tests/test_cli_convert.py

@@ -215,6 +215,25 @@ def test_convert_multi_dryrun(test_db_and_path):
     )
 
 
+def test_convert_multi_dryrun_unicode_not_escaped(test_db_and_path):
+    db_path = test_db_and_path[1]
+    result = CliRunner().invoke(
+        cli.cli,
+        [
+            "convert",
+            db_path,
+            "example",
+            "dt",
+            "{'text': 'Japanese 日本語'}",
+            "--dry-run",
+            "--multi",
+        ],
+    )
+    assert result.exit_code == 0
+    # Preview should match what jsonify_if_needed() would actually store
+    assert '{"text": "Japanese 日本語"}' in result.output
+
+
 @pytest.mark.parametrize("drop", (True, False))
 def test_convert_output_column(test_db_and_path, drop):
     db, db_path = test_db_and_path
@@ -371,16 +390,14 @@ def test_convert_multi_complex_column_types(fresh_db_and_path):
         ],
         pk="id",
     )
-    code = textwrap.dedent(
-        """
+    code = textwrap.dedent("""
     if value == 1:
         return {"is_str": "", "is_float": 1.2, "is_int": None}
     elif value == 2:
         return {"is_float": 1, "is_int": 12}
     elif value == 3:
         return {"is_bytes": b"blah"}
-    """
-    )
+    """)
     result = CliRunner().invoke(
         cli.cli,
         [
@@ -535,7 +552,7 @@ def test_convert_where(test_db_and_path):
             "id = :id",
             "-p",
             "id",
-            2,
+            "2",
         ],
     )
     assert result.exit_code == 0, result.output
@@ -564,7 +581,7 @@ def test_convert_where_multi(fresh_db_and_path):
             "id = :id",
             "-p",
             "id",
-            2,
+            "2",
             "--multi",
         ],
     )
@@ -645,3 +662,51 @@ def test_convert_handles_falsey_values(fresh_db_and_path):
     assert result.exit_code == 0, result.output
     assert db["t"].get(1)["x"] == 1
     assert db["t"].get(2)["x"] == 2
+
+
+@pytest.mark.parametrize(
+    "code",
+    [
+        # Direct callable reference (issue #686)
+        "r.parsedate",
+        "recipes.parsedate",
+        # Traditional call syntax still works
+        "r.parsedate(value)",
+        "recipes.parsedate(value)",
+    ],
+)
+def test_convert_callable_reference(test_db_and_path, code):
+    """Test that callable references like r.parsedate work without (value)"""
+    db, db_path = test_db_and_path
+    result = CliRunner().invoke(
+        cli.cli, ["convert", db_path, "example", "dt", code], catch_exceptions=False
+    )
+    assert result.exit_code == 0, result.output
+    rows = list(db["example"].rows)
+    assert rows[0]["dt"] == "2019-10-05"
+    assert rows[1]["dt"] == "2019-10-06"
+    assert rows[2]["dt"] == ""
+    assert rows[3]["dt"] is None
+
+
+def test_convert_callable_reference_with_import(fresh_db_and_path):
+    """Test callable reference from an imported module"""
+    db, db_path = fresh_db_and_path
+    db["example"].insert({"id": 1, "data": '{"name": "test"}'})
+    result = CliRunner().invoke(
+        cli.cli,
+        [
+            "convert",
+            db_path,
+            "example",
+            "data",
+            "json.loads",
+            "--import",
+            "json",
+        ],
+        catch_exceptions=False,
+    )
+    assert result.exit_code == 0, result.output
+    # json.loads returns a dict, which sqlite stores as JSON string
+    row = db["example"].get(1)
+    assert row["data"] == '{"name": "test"}'

tests/test_cli_insert.py

@@ -597,3 +597,294 @@ def test_insert_streaming_batch_size_1(db_path):
     proc.stdin.close()
     proc.wait()
     assert proc.returncode == 0
+
+
+def test_insert_csv_headers_only(tmpdir):
+    """Test that CSV with only header row (no data) works with --detect-types (issue #702)"""
+    db_path = str(tmpdir / "test.db")
+    csv_path = str(tmpdir / "headers_only.csv")
+    with open(csv_path, "w") as fp:
+        fp.write("id,name,age\n")
+    # Should not crash with --detect-types (which is now the default)
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "data", csv_path, "--csv"],
+        catch_exceptions=False,
+    )
+    assert result.exit_code == 0
+    # Table should not exist since there were no data rows
+    db = Database(db_path)
+    assert not db["data"].exists()
+
+
+def test_insert_into_view_errors(tmpdir):
+    db_path = str(tmpdir / "test.db")
+    db = Database(db_path)
+    db["t"].insert({"id": 1})
+    db.create_view("v", "select * from t")
+    db.close()
+    result = CliRunner().invoke(
+        cli.cli, ["insert", db_path, "v", "-"], input='{"id": 2}'
+    )
+    assert result.exit_code == 1
+    assert result.output.strip() == "Error: Table v is actually a view"
+
+
+def test_insert_csv_detect_types_leaves_existing_table_alone(db_path):
+    # Type detection is the default for CSV/TSV inserts, but it must only
+    # apply to tables created by this command - transforming a pre-existing
+    # table would rewrite its column types and corrupt data such as
+    # TEXT zip codes with leading zeros
+    db = Database(db_path)
+    db["places"].insert({"name": "Boston", "zip": "01234"})
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "places", "-", "--csv"],
+        catch_exceptions=False,
+        input="name,zip\nSF,94107",
+    )
+    assert result.exit_code == 0, result.output
+    assert db["places"].columns_dict["zip"] is str
+    assert list(db["places"].rows) == [
+        {"name": "Boston", "zip": "01234"},
+        {"name": "SF", "zip": "94107"},
+    ]
+
+
+def test_insert_csv_detect_types_new_table(db_path):
+    # A table created by the insert still gets detected types
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "data", "-", "--csv"],
+        catch_exceptions=False,
+        input="name,age,weight\nCleo,5,12.5",
+    )
+    assert result.exit_code == 0, result.output
+    db = Database(db_path)
+    assert db["data"].columns_dict == {"name": str, "age": int, "weight": float}
+
+
+@pytest.mark.parametrize(
+    "command,extra_args,input_text,expected_row",
+    (
+        (
+            "insert",
+            [],
+            "zipcode,score\n01234,9.5\n",
+            {"zipcode": "01234", "score": 9.5},
+        ),
+        (
+            "upsert",
+            ["--pk", "id"],
+            "id,zipcode,score\n1,01234,9.5\n",
+            {"id": 1, "zipcode": "01234", "score": 9.5},
+        ),
+    ),
+)
+def test_insert_upsert_csv_type_overrides_detected_types(
+    db_path, command, extra_args, input_text, expected_row
+):
+    result = CliRunner().invoke(
+        cli.cli,
+        [
+            command,
+            db_path,
+            "places",
+            "-",
+            "--csv",
+        ]
+        + extra_args
+        + [
+            "--type",
+            "zipcode",
+            "text",
+        ],
+        catch_exceptions=False,
+        input=input_text,
+    )
+    assert result.exit_code == 0, result.output
+    db = Database(db_path)
+    expected_columns = {"zipcode": str, "score": float}
+    if command == "upsert":
+        expected_columns = {"id": int, **expected_columns}
+    assert db["places"].columns_dict == expected_columns
+    assert list(db["places"].rows) == [expected_row]
+
+
+def test_upsert_csv_detect_types_leaves_existing_table_alone(db_path):
+    db = Database(db_path)
+    db["places"].insert({"id": 1, "name": "Boston", "zip": "01234"}, pk="id")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["upsert", db_path, "places", "-", "--csv", "--pk", "id"],
+        catch_exceptions=False,
+        input="id,name,zip\n2,SF,94107",
+    )
+    assert result.exit_code == 0, result.output
+    assert db["places"].columns_dict["zip"] is str
+    assert db["places"].get(1)["zip"] == "01234"
+
+
+def test_insert_invalid_pk_clean_error(db_path):
+    # An invalid --pk against an existing table should be a clean CLI
+    # error, not a raw InvalidColumns traceback
+    db = Database(db_path)
+    db["t"].insert({"a": 1})
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "t", "-", "--pk", "badcol"],
+        input='{"a": 2}',
+    )
+    assert result.exit_code == 1
+    assert result.exception is None or isinstance(result.exception, SystemExit)
+    assert result.output.startswith("Error: Invalid primary key column")
+
+
+# --code tests, see https://github.com/simonw/sqlite-utils/issues/684
+CODE_ROWS_FUNCTION = """
+def rows():
+    yield {"id": 1, "name": "Cleo"}
+    yield {"id": 2, "name": "Suna"}
+"""
+
+CODE_ROWS_ITERABLE = """
+rows = [
+    {"id": 1, "name": "Cleo"},
+    {"id": 2, "name": "Suna"},
+]
+"""
+
+
+@pytest.mark.parametrize("code", (CODE_ROWS_FUNCTION, CODE_ROWS_ITERABLE))
+def test_insert_code(tmpdir, code):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", code, "--pk", "id"],
+    )
+    assert result.exit_code == 0, result.output
+    db = Database(db_path)
+    assert db["creatures"].pks == ["id"]
+    assert list(db["creatures"].rows) == [
+        {"id": 1, "name": "Cleo"},
+        {"id": 2, "name": "Suna"},
+    ]
+
+
+def test_insert_code_from_file(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    code_path = str(tmpdir / "gen.py")
+    with open(code_path, "w") as fp:
+        fp.write(CODE_ROWS_FUNCTION)
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", code_path],
+    )
+    assert result.exit_code == 0, result.output
+    assert list(Database(db_path)["creatures"].rows) == [
+        {"id": 1, "name": "Cleo"},
+        {"id": 2, "name": "Suna"},
+    ]
+
+
+def test_upsert_code(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    db = Database(db_path)
+    db["creatures"].insert_all(
+        [{"id": 1, "name": "old"}, {"id": 2, "name": "Suna"}], pk="id"
+    )
+    result = CliRunner().invoke(
+        cli.cli,
+        ["upsert", db_path, "creatures", "--code", CODE_ROWS_FUNCTION, "--pk", "id"],
+    )
+    assert result.exit_code == 0, result.output
+    assert list(db["creatures"].rows) == [
+        {"id": 1, "name": "Cleo"},
+        {"id": 2, "name": "Suna"},
+    ]
+
+
+def test_insert_code_requires_file_or_code(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(cli.cli, ["insert", db_path, "creatures"])
+    assert result.exit_code == 1
+    assert "Provide either a FILE argument or --code" in result.output
+
+
+def test_insert_code_mutually_exclusive_with_file(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "-", "--code", CODE_ROWS_FUNCTION],
+        input="{}",
+    )
+    assert result.exit_code == 1
+    assert "--code cannot be used with a FILE argument" in result.output
+
+
+def test_insert_code_rejects_input_format_options(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", CODE_ROWS_FUNCTION, "--csv"],
+    )
+    assert result.exit_code == 1
+    assert "--code cannot be used with input format options" in result.output
+
+
+def test_insert_code_missing_rows(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", "x = 1"],
+    )
+    assert result.exit_code == 1
+    assert "must define a 'rows' function or iterable" in result.output
+
+
+def test_insert_code_single_dict(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        [
+            "insert",
+            db_path,
+            "creatures",
+            "--code",
+            'rows = {"id": 1, "name": "Cleo"}',
+            "--pk",
+            "id",
+        ],
+    )
+    assert result.exit_code == 0, result.output
+    assert list(Database(db_path)["creatures"].rows) == [{"id": 1, "name": "Cleo"}]
+
+
+def test_insert_code_not_iterable(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", "rows = 5"],
+    )
+    assert result.exit_code == 1
+    assert "must define a 'rows' function or iterable" in result.output
+
+
+def test_insert_code_syntax_error(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", "def rows(:"],
+    )
+    assert result.exit_code == 1
+    assert "Error in --code" in result.output
+
+
+def test_insert_code_file_not_found(tmpdir):
+    db_path = str(tmpdir / "dogs.db")
+    result = CliRunner().invoke(
+        cli.cli,
+        ["insert", db_path, "creatures", "--code", "missing.py"],
+    )
+    assert result.exit_code == 1
+    assert "File not found: missing.py" in result.output

tests/test_cli_memory.py

@@ -156,6 +156,24 @@ def test_memory_csv_encoding(tmpdir, use_stdin):
     }
 
 
+def test_memory_csv_headers_only(tmpdir):
+    csv_path = str(tmpdir / "headers_only.csv")
+    with open(csv_path, "w") as fp:
+        fp.write("id,name,age\n")
+
+    result = CliRunner().invoke(
+        cli.cli,
+        ["memory", csv_path, "", "--schema"],
+        catch_exceptions=False,
+    )
+
+    assert result.exit_code == 0
+    assert result.output.strip() == (
+        'CREATE VIEW "t1" AS select * from "headers_only";\n'
+        'CREATE VIEW "t" AS select * from "headers_only";'
+    )
+
+
 @pytest.mark.parametrize("extra_args", ([], ["select 1"]))
 def test_memory_dump(extra_args):
     result = CliRunner().invoke(
@@ -328,9 +346,10 @@ def test_memory_return_db(tmpdir):
     from sqlite_utils.cli import cli
 
     path = str(tmpdir / "dogs.csv")
-    open(path, "w").write("id,name\n1,Cleo")
+    with open(path, "w") as f:
+        f.write("id,name\n1,Cleo")
 
-    with click.Context(cli) as ctx:
+    with click.Context(cli) as ctx:  # type: ignore[attr-defined]
         db = ctx.invoke(cli.commands["memory"], paths=(path,), return_db=True)
 
     assert db.table_names() == ["dogs"]

tests/test_cli_migrate.py

@@ -0,0 +1,507 @@
+import pathlib
+
+from click.testing import CliRunner
+import pytest
+import sqlite_utils
+import sqlite_utils.cli
+
+TWO_MIGRATIONS = """
+from sqlite_utils import Migrations
+
+m = Migrations("hello")
+
+@m()
+def foo(db):
+    db["foo"].insert({"hello": "world"})
+
+@m()
+def bar(db):
+    db["bar"].insert({"hello": "world"})
+"""
+
+
+@pytest.fixture
+def two_migrations(tmpdir):
+    path = pathlib.Path(tmpdir)
+    (path / "foo").mkdir()
+    migrations_py = path / "foo" / "migrations.py"
+    migrations_py.write_text(TWO_MIGRATIONS, "utf-8")
+    return path, migrations_py
+
+
+@pytest.fixture
+def two_sets_same_migration_name(tmpdir):
+    path = pathlib.Path(tmpdir)
+    migrations_py = path / "migrations.py"
+    migrations_py.write_text(
+        """
+from sqlite_utils import Migrations
+
+creatures = Migrations("creatures")
+
+@creatures()
+def create_table(db):
+    db["creatures"].insert({"name": "Cleo"})
+
+@creatures()
+def add_weight(db):
+    db["creature_weights"].insert({"weight": 4.2})
+
+sales = Migrations("sales")
+
+@sales()
+def create_table(db):
+    db["sales"].insert({"id": 1})
+
+@sales()
+def add_weight(db):
+    db["sales_weights"].insert({"weight": 10})
+""",
+        "utf-8",
+    )
+    return path, migrations_py
+
+
+@pytest.mark.parametrize("arg", ("TMPDIR", "TMPDIR/foo/migrations.py", "TMPDIR/foo/"))
+def test_basic(two_migrations, arg):
+    path, _ = two_migrations
+    db_path = str(path / "test.db")
+
+    runner = CliRunner()
+
+    def _list():
+        list_result = runner.invoke(
+            sqlite_utils.cli.cli,
+            ["migrate", db_path, "--list", arg.replace("TMPDIR", str(path))],
+        )
+        assert list_result.exit_code == 0
+        return list_result.output
+
+    assert _list() == (
+        "Migrations for: hello\n\n"
+        "  Applied:\n\n"
+        "  Pending:\n"
+        "    foo\n"
+        "    bar\n\n"
+    )
+
+    result = runner.invoke(
+        sqlite_utils.cli.cli, ["migrate", db_path, arg.replace("TMPDIR", str(path))]
+    )
+    assert result.exit_code == 0, result.output
+
+    list_output = _list()
+    assert "Migrations for: hello\n\n  Applied:\n    " in list_output
+    prior_to_pending = list_output.split(" Pending")[0]
+    assert "  foo" in prior_to_pending
+    assert "  bar" in prior_to_pending
+    assert " Pending:\n    (none)" in list_output
+
+    db = sqlite_utils.Database(db_path)
+    assert db["foo"].exists()
+    assert db["bar"].exists()
+    assert db["_sqlite_migrations"].exists()
+    rows = list(db["_sqlite_migrations"].rows)
+    assert len(rows) == 2
+    assert rows[0]["name"] == "foo"
+    assert rows[1]["name"] == "bar"
+
+
+def test_list_same_migration_names_in_different_sets(capsys):
+    applied = sqlite_utils.Migrations("applied")
+
+    @applied(name="foo")
+    def applied_foo(db):
+        db["applied"].insert({"hello": "world"})
+
+    pending = sqlite_utils.Migrations("pending")
+
+    @pending(name="foo")
+    def pending_foo(db):
+        db["pending"].insert({"hello": "world"})
+
+    db = sqlite_utils.Database(memory=True)
+    applied.apply(db)
+
+    sqlite_utils.cli._display_migration_list(db, [applied, pending])
+
+    output = capsys.readouterr().out
+    assert (
+        "Migrations for: pending\n\n" "  Applied:\n\n" "  Pending:\n" "    foo\n\n"
+    ) in output
+
+
+def test_verbose(tmpdir):
+    path = pathlib.Path(tmpdir)
+    (path / "foo").mkdir()
+    migrations_py = path / "foo" / "migrations.py"
+    migrations_py.write_text(
+        """
+from sqlite_utils import Migrations
+
+m = Migrations("hello")
+
+@m()
+def foo(db):
+    db["dogs"].insert({"id": 1, "name": "Cleo"})
+    """,
+        "utf-8",
+    )
+    db_path = str(path / "test.db")
+    runner = CliRunner()
+    result = runner.invoke(
+        sqlite_utils.cli.cli, ["migrate", db_path, str(migrations_py)]
+    )
+    assert result.exit_code == 0
+
+    result = runner.invoke(
+        sqlite_utils.cli.cli, ["migrate", db_path, str(migrations_py), "--verbose"]
+    )
+    assert result.exit_code == 0
+    expected = """
+Schema before:
+
+  CREATE TABLE "_sqlite_migrations" (
+     "id" INTEGER PRIMARY KEY,
+     "migration_set" TEXT,
+     "name" TEXT,
+     "applied_at" TEXT
+  );
+  CREATE UNIQUE INDEX "idx__sqlite_migrations_migration_set_name"
+      ON "_sqlite_migrations" ("migration_set", "name");
+  CREATE TABLE "dogs" (
+     "id" INTEGER,
+     "name" TEXT
+  );
+
+Schema after:
+
+  (unchanged)
+""".strip()
+    assert expected in result.output
+
+    new_migration = """
+@m()
+def bar(db):
+    db["dogs"].add_column("age", int)
+    db["dogs"].add_column("weight", float)
+    db["dogs"].transform()
+"""
+    migrations_py.write_text(migrations_py.read_text("utf-8") + new_migration)
+
+    result = runner.invoke(
+        sqlite_utils.cli.cli, ["migrate", db_path, str(migrations_py), "--verbose"]
+    )
+    assert result.exit_code == 0
+    expected_diff = """
+Schema diff:
+
+     ON "_sqlite_migrations" ("migration_set", "name");
+ CREATE TABLE "dogs" (
+    "id" INTEGER,
+-   "name" TEXT
++   "name" TEXT,
++   "age" INTEGER,
++   "weight" REAL
+ );
+""".strip()
+    assert expected_diff in result.output
+
+
+def test_stop_before(two_migrations):
+    path, _ = two_migrations
+    db_path = str(path / "test.db")
+    result = CliRunner().invoke(
+        sqlite_utils.cli.cli,
+        [
+            "migrate",
+            db_path,
+            str(path / "foo" / "migrations.py"),
+            "--stop-before",
+            "bar",
+        ],
+    )
+    assert result.exit_code == 0
+    db = sqlite_utils.Database(db_path)
+    assert db["foo"].exists()
+    assert not db["bar"].exists()
+
+
+def test_stop_before_multiple_sets_unqualified(two_migrations):
+    path, _ = two_migrations
+    db_path = str(path / "test.db")
+    (path / "foo" / "migrations2.py").write_text(
+        """
+from sqlite_utils import Migrations
+
+m = Migrations("hello2")
+
+@m()
+def foo(db):
+    db["foo"].insert({"hello": "world"})
+    """,
+        "utf-8",
+    )
+    result = CliRunner().invoke(
+        sqlite_utils.cli.cli,
+        [
+            "migrate",
+            db_path,
+            str(path / "foo" / "migrations.py"),
+            str(path / "foo" / "migrations2.py"),
+            "--stop-before",
+            "foo",
+        ],
+    )
+    assert result.exit_code == 0, result.output
+    db = sqlite_utils.Database(db_path)
+    assert db.table_names() == ["_sqlite_migrations"]
+    assert list(db["_sqlite_migrations"].rows) == []
+
+
+def test_stop_before_qualified_only_affects_named_set(two_sets_same_migration_name):
+    path, migrations_py = two_sets_same_migration_name
+    db_path = str(path / "test.db")
+    result = CliRunner().invoke(
+        sqlite_utils.cli.cli,
+        [
+            "migrate",
+            db_path,
+            str(migrations_py),
+            "--stop-before",
+            "creatures:add_weight",
+        ],
+    )
+    assert result.exit_code == 0, result.output
+    db = sqlite_utils.Database(db_path)
+    assert db["creatures"].exists()
+    assert not db["creature_weights"].exists()
+    assert db["sales"].exists()
+    assert db["sales_weights"].exists()
+
+
+def test_stop_before_multiple_qualified(two_sets_same_migration_name):
+    path, migrations_py = two_sets_same_migration_name
+    db_path = str(path / "test.db")
+    result = CliRunner().invoke(
+        sqlite_utils.cli.cli,
+        [
+            "migrate",
+            db_path,
+            str(migrations_py),
+            "--stop-before",
+            "creatures:add_weight",
+            "--stop-before",
+            "sales:add_weight",
+        ],
+    )
+    assert result.exit_code == 0, result.output
+    db = sqlite_utils.Database(db_path)
+    assert db["creatures"].exists()
+    assert not db["creature_weights"].exists()
+    assert db["sales"].exists()
+    assert not db["sales_weights"].exists()
+
+
+LEGACY_MIGRATIONS = """
+import datetime
+
+class _Migration:
+    def __init__(self, name, fn):
+        self.name = name
+        self.fn = fn
+
+class _Applied:
+    def __init__(self, name, applied_at):
+        self.name = name
+        self.applied_at = applied_at
+
+class LegacyMigrations:
+    # Mimics the sqlite-migrate 0.x Migrations class, in particular
+    # apply(db, stop_before=None) taking a single string
+    migrations_table = "_sqlite_migrations"
+
+    def __init__(self, name):
+        self.name = name
+        self._migrations = []
+
+    def __call__(self, fn):
+        self._migrations.append(_Migration(fn.__name__, fn))
+        return fn
+
+    def ensure_migrations_table(self, db):
+        db[self.migrations_table].create(
+            {"migration_set": str, "name": str, "applied_at": str},
+            pk=("migration_set", "name"),
+            if_not_exists=True,
+        )
+
+    def applied(self, db):
+        self.ensure_migrations_table(db)
+        return [
+            _Applied(row["name"], row["applied_at"])
+            for row in db[self.migrations_table].rows_where(
+                "migration_set = ?", [self.name]
+            )
+        ]
+
+    def pending(self, db):
+        applied = {m.name for m in self.applied(db)}
+        return [m for m in self._migrations if m.name not in applied]
+
+    def apply(self, db, stop_before=None):
+        for migration in self.pending(db):
+            if migration.name == stop_before:
+                return
+            migration.fn(db)
+            db[self.migrations_table].insert(
+                {
+                    "migration_set": self.name,
+                    "name": migration.name,
+                    "applied_at": str(
+                        datetime.datetime.now(datetime.timezone.utc)
+                    ),
+                }
+            )
+
+legacy = LegacyMigrations("legacy_set")
+
+@legacy
+def first(db):
+    db["first"].insert({"hello": "world"})
+
+@legacy
+def second(db):
+    db["second"].insert({"hello": "world"})
+"""
+
+
+def test_stop_before_unknown_name_errors(two_migrations):
+    path, _ = two_migrations
+    db_path = str(path / "test.db")
+    runner = CliRunner()
+    result = runner.invoke(
+        sqlite_utils.cli.cli,
+        ["migrate", db_path, str(path), "--stop-before", "fooo"],
+    )
+    assert result.exit_code == 1
+    assert "--stop-before did not match any migrations: fooo" in result.output
+    # Nothing should have been applied
+    db = sqlite_utils.Database(db_path)
+    assert "foo" not in db.table_names()
+    assert "bar" not in db.table_names()
+
+
+def test_stop_before_with_legacy_migrations_class(tmpdir):
+    path = pathlib.Path(tmpdir)
+    (path / "migrations.py").write_text(LEGACY_MIGRATIONS, "utf-8")
+    db_path = str(path / "test.db")
+    runner = CliRunner()
+    result = runner.invoke(
+        sqlite_utils.cli.cli,
+        ["migrate", db_path, str(path), "--stop-before", "second"],
+    )
+    assert result.exit_code == 0, result.output
+    db = sqlite_utils.Database(db_path)
+    assert "first" in db.table_names()
+    assert "second" not in db.table_names()
+
+
+def test_stop_before_multiple_values_for_legacy_set_errors(tmpdir):
+    path = pathlib.Path(tmpdir)
+    (path / "migrations.py").write_text(LEGACY_MIGRATIONS, "utf-8")
+    db_path = str(path / "test.db")
+    runner = CliRunner()
+    result = runner.invoke(
+        sqlite_utils.cli.cli,
+        [
+            "migrate",
+            db_path,
+            str(path),
+            "--stop-before",
+            "legacy_set:first",
+            "--stop-before",
+            "legacy_set:second",
+        ],
+    )
+    assert result.exit_code == 1
+    assert "single --stop-before" in result.output
+
+
+def test_list_does_not_create_database_file(two_migrations):
+    path, _ = two_migrations
+    db_path = path / "test.db"
+    runner = CliRunner()
+    result = runner.invoke(
+        sqlite_utils.cli.cli, ["migrate", str(db_path), str(path), "--list"]
+    )
+    assert result.exit_code == 0, result.output
+    assert "Pending:\n    foo\n    bar" in result.output
+    # Listing migrations must not create the database file
+    assert not db_path.exists()
+
+
+def test_list_does_not_upgrade_legacy_migrations_table(two_migrations):
+    path, _ = two_migrations
+    db_path = str(path / "test.db")
+    db = sqlite_utils.Database(db_path)
+    db["_sqlite_migrations"].create(
+        {"migration_set": str, "name": str, "applied_at": str},
+        pk=("migration_set", "name"),
+    )
+    db["_sqlite_migrations"].insert(
+        {"migration_set": "hello", "name": "foo", "applied_at": "x"}
+    )
+    db.close()
+    runner = CliRunner()
+    result = runner.invoke(
+        sqlite_utils.cli.cli, ["migrate", db_path, str(path), "--list"]
+    )
+    assert result.exit_code == 0, result.output
+    assert "foo - x" in result.output
+    # --list must not perform the one-way legacy schema upgrade
+    db2 = sqlite_utils.Database(db_path)
+    assert db2["_sqlite_migrations"].pks == ["migration_set", "name"]
+    db2.close()
+
+
+def test_stop_before_applied_migration_errors(two_migrations):
+    path, _ = two_migrations
+    db_path = str(path / "test.db")
+    migrations_path = str(path / "foo" / "migrations.py")
+    # Apply everything first
+    first = CliRunner().invoke(
+        sqlite_utils.cli.cli,
+        ["migrate", db_path, migrations_path, "--stop-before", "bar"],
+    )
+    assert first.exit_code == 0
+    # foo is now applied - stopping before it is an error, and bar
+    # must not be applied as a side effect
+    result = CliRunner().invoke(
+        sqlite_utils.cli.cli,
+        ["migrate", db_path, migrations_path, "--stop-before", "foo"],
+    )
+    assert result.exit_code != 0
+    assert "already been applied" in result.output
+    db = sqlite_utils.Database(db_path)
+    assert not db["bar"].exists()
+
+
+def test_list_with_legacy_class_is_read_only(tmpdir):
+    # Legacy sqlite-migrate classes create the _sqlite_migrations table
+    # from their pending()/applied() methods - --list must roll that
+    # back so it stays a read-only operation as documented
+    path = pathlib.Path(tmpdir)
+    (path / "migrations.py").write_text(LEGACY_MIGRATIONS, "utf-8")
+    db_path = str(path / "test.db")
+    db = sqlite_utils.Database(db_path)
+    db["existing"].insert({"id": 1})
+    db.close()
+    result = CliRunner().invoke(
+        sqlite_utils.cli.cli, ["migrate", db_path, str(path), "--list"]
+    )
+    assert result.exit_code == 0, result.output
+    assert "first" in result.output
+    db2 = sqlite_utils.Database(db_path)
+    assert "_sqlite_migrations" not in db2.table_names()
+    db2.close()

tests/test_column_casing.py

@@ -0,0 +1,233 @@
+"""
+SQLite treats column names as case-insensitive. These tests exercise the
+places where sqlite-utils performs Python-side lookups of column names
+provided by the caller, which should match the schema case-insensitively.
+
+https://github.com/simonw/sqlite-utils/issues/760
+"""
+
+import pytest
+
+from sqlite_utils import Database
+from sqlite_utils.db import ForeignKey
+
+
+def test_insert_populates_last_pk_case_insensitively(fresh_db):
+    books = fresh_db["books"]
+    books.create({"Id": int, "Title": str}, pk="Id")
+    books.insert({"Id": 1, "Title": "One"}, pk="id")
+    assert books.last_pk == 1
+
+
+def test_insert_populates_last_pk_compound_pk_case_insensitively(fresh_db):
+    books = fresh_db["books"]
+    books.create({"Author": str, "Position": int, "Title": str})
+    books.insert(
+        {"Author": "Sue", "Position": 1, "Title": "One"}, pk=("author", "position")
+    )
+    assert books.last_pk == ("Sue", 1)
+
+
+@pytest.mark.parametrize("use_old_upsert", (False, True))
+def test_upsert_pk_case_differs_from_schema(use_old_upsert):
+    db = Database(memory=True, use_old_upsert=use_old_upsert)
+    books = db["books"]
+    books.create({"Id": int, "Title": str}, pk="Id")
+    books.insert({"Id": 1, "Title": "One"})
+    books.upsert({"id": 1, "title": "Won"}, pk="id")
+    assert list(books.rows) == [{"Id": 1, "Title": "Won"}]
+    assert books.last_pk == 1
+
+
+@pytest.mark.parametrize("use_old_upsert", (False, True))
+def test_upsert_record_key_case_differs_from_pk(use_old_upsert):
+    # all_columns comes from the record keys, pk= from the caller
+    db = Database(memory=True, use_old_upsert=use_old_upsert)
+    books = db["books"]
+    books.create({"Id": int, "Title": str}, pk="Id")
+    books.upsert({"ID": 1, "Title": "One"}, pk="id")
+    assert list(books.rows) == [{"Id": 1, "Title": "One"}]
+    assert books.last_pk == 1
+
+
+def test_upsert_inferred_pk_case_differs_from_record_keys(fresh_db):
+    # pk is inferred from the existing schema as "Id", records use "id"
+    books = fresh_db["books"]
+    books.create({"Id": int, "Title": str}, pk="Id")
+    books.upsert({"id": 1, "title": "One"})
+    assert list(books.rows) == [{"Id": 1, "Title": "One"}]
+    assert books.last_pk == 1
+
+
+def test_upsert_list_mode_pk_case_insensitive(fresh_db):
+    books = fresh_db["books"]
+    books.create({"Id": int, "Title": str}, pk="Id")
+    books.upsert_all([["id", "title"], [1, "One"]], pk="Id")
+    assert list(books.rows) == [{"Id": 1, "Title": "One"}]
+    assert books.last_pk == 1
+
+
+def test_lookup_pk_case_insensitive(fresh_db):
+    fresh_db["species"].create({"ID": int, "Name": str}, pk="ID")
+    fresh_db["species"].insert({"ID": 5, "Name": "Palm"})
+    fresh_db["species"].create_index(["Name"], unique=True)
+    assert fresh_db["species"].lookup({"Name": "Palm"}, pk="id") == 5
+
+
+def test_lookup_does_not_create_redundant_index(fresh_db):
+    fresh_db["species"].create({"id": int, "Name": str}, pk="id")
+    fresh_db["species"].create_index(["Name"], unique=True)
+    fresh_db["species"].lookup({"name": "Palm"})
+    assert len(fresh_db["species"].indexes) == 1
+
+
+def test_create_table_transform_same_columns_different_case(fresh_db):
+    fresh_db["t"].create({"Name": str, "Age": int})
+    fresh_db["t"].insert({"Name": "Cleo", "Age": 5})
+    fresh_db.create_table("t", {"name": str, "age": int}, transform=True)
+    # Schema casing is preserved - SQLite considers these the same columns
+    assert fresh_db["t"].columns_dict == {"Name": str, "Age": int}
+    assert list(fresh_db["t"].rows) == [{"Name": "Cleo", "Age": 5}]
+
+
+def test_create_table_transform_case_insensitive_with_changes(fresh_db):
+    fresh_db["t"].create({"Name": str, "Age": int})
+    fresh_db.create_table("t", {"name": str, "age": str, "size": int}, transform=True)
+    # age changed type, size added, Name untouched
+    assert fresh_db["t"].columns_dict == {"Name": str, "Age": str, "size": int}
+
+
+def test_transform_types_case_insensitive(fresh_db):
+    fresh_db["t"].create({"Name": str, "Age": str})
+    fresh_db["t"].transform(types={"age": int})
+    assert fresh_db["t"].columns_dict == {"Name": str, "Age": int}
+
+
+def test_transform_rename_case_insensitive(fresh_db):
+    fresh_db["t"].create({"Name": str})
+    fresh_db["t"].transform(rename={"name": "title"})
+    assert fresh_db["t"].columns_dict == {"title": str}
+
+
+def test_transform_drop_case_insensitive(fresh_db):
+    fresh_db["t"].create({"Name": str, "Age": int})
+    fresh_db["t"].transform(drop=["name"])
+    assert fresh_db["t"].columns_dict == {"Age": int}
+
+
+def test_transform_not_null_and_defaults_case_insensitive(fresh_db):
+    fresh_db["t"].create({"Name": str, "Age": int})
+    fresh_db["t"].transform(not_null={"name"}, defaults={"age": 3})
+    columns = {c.name: c for c in fresh_db["t"].columns}
+    assert columns["Name"].notnull
+    assert fresh_db["t"].default_values == {"Age": 3}
+
+
+def test_transform_pk_case_insensitive(fresh_db):
+    fresh_db["t"].create({"Id": int, "Name": str})
+    fresh_db["t"].transform(pk="id")
+    assert fresh_db["t"].pks == ["Id"]
+    assert fresh_db["t"].columns_dict == {"Id": int, "Name": str}
+
+
+def test_transform_drop_foreign_keys_case_insensitive(fresh_db):
+    fresh_db["parent"].create({"Id": int}, pk="Id")
+    fresh_db["child"].create(
+        {"id": int, "Parent_ID": int},
+        pk="id",
+        foreign_keys=[("Parent_ID", "parent", "Id")],
+    )
+    fresh_db["child"].transform(drop_foreign_keys=["parent_id"])
+    assert fresh_db["child"].foreign_keys == []
+
+
+def test_add_foreign_key_case_insensitive(fresh_db):
+    fresh_db["parent"].create({"Id": int}, pk="Id")
+    fresh_db["child"].create({"id": int, "Parent_ID": int}, pk="id")
+    fresh_db["child"].add_foreign_key("parent_id", "parent", "id")
+    fks = fresh_db["child"].foreign_keys
+    assert len(fks) == 1
+    # The foreign key should use the schema casing of the columns
+    assert fks[0].column == "Parent_ID"
+    assert fks[0].other_column == "Id"
+
+
+def test_add_foreign_keys_case_insensitive(fresh_db):
+    fresh_db["parent"].create({"Id": int}, pk="Id")
+    fresh_db["child"].create({"id": int, "Parent_ID": int}, pk="id")
+    fresh_db.add_foreign_keys([("child", "parent_id", "parent", "id")])
+    fks = fresh_db["child"].foreign_keys
+    assert len(fks) == 1
+    assert fks[0].column == "Parent_ID"
+    assert fks[0].other_column == "Id"
+
+
+def test_add_foreign_key_detects_existing_case_insensitively(fresh_db):
+    fresh_db["parent"].create({"Id": int}, pk="Id")
+    fresh_db["child"].create(
+        {"id": int, "Parent_ID": int},
+        pk="id",
+        foreign_keys=[("Parent_ID", "parent", "Id")],
+    )
+    # ignore=True should treat this as already existing, not add a duplicate
+    fresh_db["child"].add_foreign_key("parent_id", "parent", "id", ignore=True)
+    assert len(fresh_db["child"].foreign_keys) == 1
+
+
+def test_add_column_fk_col_case_insensitive(fresh_db):
+    fresh_db["parent"].create({"Id": int}, pk="Id")
+    fresh_db["child"].create({"id": int}, pk="id")
+    fresh_db["child"].add_column("parent_id", int, fk="parent", fk_col="id")
+    fks = fresh_db["child"].foreign_keys
+    assert len(fks) == 1
+    assert fks[0].other_column == "Id"
+
+
+def test_extract_case_insensitive(fresh_db):
+    fresh_db["trees"].insert({"id": 1, "Species": "Palm"}, pk="id")
+    fresh_db["trees"].extract("species")
+    assert fresh_db["trees"].columns_dict == {"id": int, "Species_id": int}
+    assert list(fresh_db["Species"].rows) == [{"id": 1, "Species": "Palm"}]
+
+
+def test_convert_multi_case_insensitive(fresh_db):
+    fresh_db["t"].insert({"id": 1, "Name": "Cleo"}, pk="id")
+    fresh_db["t"].convert("name", lambda v: {"upper": v.upper()}, multi=True)
+    assert list(fresh_db["t"].rows) == [{"id": 1, "Name": "Cleo", "upper": "CLEO"}]
+
+
+def test_convert_output_case_insensitive(fresh_db):
+    fresh_db["t"].insert({"id": 1, "Name": "Cleo", "Upper": None}, pk="id")
+    fresh_db["t"].convert("name", lambda v: v.upper(), output="upper")
+    assert list(fresh_db["t"].rows) == [{"id": 1, "Name": "Cleo", "Upper": "CLEO"}]
+
+
+def test_create_table_sql_pk_case_insensitive(fresh_db):
+    fresh_db["t"].create({"Id": int, "Name": str}, pk="id")
+    # Should not have created an extra lowercase "id" column
+    assert fresh_db["t"].columns_dict == {"Id": int, "Name": str}
+    assert fresh_db["t"].pks == ["Id"]
+
+
+def test_create_table_not_null_and_defaults_case_insensitive(fresh_db):
+    fresh_db["t"].create(
+        {"Name": str, "Age": int}, not_null={"name"}, defaults={"age": 1}
+    )
+    columns = {c.name: c for c in fresh_db["t"].columns}
+    assert columns["Name"].notnull
+    assert fresh_db["t"].default_values == {"Age": 1}
+
+
+def test_create_table_foreign_keys_case_insensitive(fresh_db):
+    fresh_db["parent"].create({"Id": int}, pk="Id")
+    fresh_db["child"].create(
+        {"id": int, "Parent_ID": int},
+        pk="id",
+        foreign_keys=[("parent_id", "parent", "id")],
+    )
+    fks = fresh_db["child"].foreign_keys
+    assert fks == [
+        ForeignKey(
+            table="child", column="Parent_ID", other_table="parent", other_column="Id"
+        )
+    ]

tests/test_constructor.py

@@ -1,6 +1,8 @@
 from sqlite_utils import Database
+from sqlite_utils.db import TransactionError
 from sqlite_utils.utils import sqlite3
 import pytest
+import sys
 
 
 def test_recursive_triggers():
@@ -29,6 +31,24 @@ def test_sqlite_version():
     assert actual == as_string
 
 
+def test_database_context_manager(tmpdir):
+    path = str(tmpdir / "test.db")
+    with Database(path) as db:
+        db["t"].insert({"id": 1})
+        # Raw writes commit automatically too
+        db.execute("insert into t (id) values (2)")
+        # An explicitly opened transaction left uncommitted on purpose:
+        db.begin()
+        db.execute("insert into t (id) values (3)")
+    # The connection is closed...
+    with pytest.raises(sqlite3.ProgrammingError):
+        db.execute("select 1")
+    # ... and the open explicit transaction was rolled back, not committed
+    db2 = Database(path)
+    assert [r["id"] for r in db2["t"].rows] == [1, 2]
+    db2.close()
+
+
 @pytest.mark.parametrize("memory", [True, False])
 def test_database_close(tmpdir, memory):
     if memory:
@@ -39,3 +59,55 @@ def test_database_close(tmpdir, memory):
     db.close()
     with pytest.raises(sqlite3.ProgrammingError):
         db.execute("select 1 + 1")
+
+
+@pytest.mark.skipif(
+    sys.version_info < (3, 12),
+    reason="sqlite3.connect(autocommit=) requires Python 3.12",
+)
+@pytest.mark.parametrize("autocommit", [True, False])
+def test_autocommit_connections_are_rejected(tmpdir, autocommit):
+    # These connection modes break commit()/rollback() in ways that
+    # silently lose data, so the constructor refuses them
+    conn = sqlite3.connect(str(tmpdir / "test.db"), autocommit=autocommit)
+    with pytest.raises(TransactionError):
+        Database(conn)
+    conn.close()
+
+
+@pytest.mark.skipif(
+    sys.version_info < (3, 12),
+    reason="sqlite3.LEGACY_TRANSACTION_CONTROL requires Python 3.12",
+)
+def test_legacy_transaction_control_connection_is_accepted(tmpdir):
+    conn = sqlite3.connect(
+        str(tmpdir / "test.db"), autocommit=sqlite3.LEGACY_TRANSACTION_CONTROL
+    )
+    db = Database(conn)
+    db["t"].insert({"id": 1}, pk="id")
+    assert [r["id"] for r in db["t"].rows] == [1]
+    db.close()
+
+
+def test_memory_attribute_for_memory_true():
+    db = Database(memory=True)
+    assert db.memory is True
+    assert db.memory_name is None
+
+
+def test_memory_attribute_for_memory_name():
+    db = Database(memory_name="shared_attr")
+    assert db.memory is True
+    assert db.memory_name == "shared_attr"
+
+
+def test_memory_attribute_for_memory_string_path():
+    db = Database(":memory:")
+    assert db.memory is True
+    assert db.memory_name is None
+
+
+def test_memory_attribute_for_file_path(tmpdir):
+    db = Database(str(tmpdir / "file.db"))
+    assert db.memory is False
+    assert db.memory_name is None

tests/test_convert.py

@@ -77,7 +77,7 @@ def test_convert_output(fresh_db, drop, expected):
 
 def test_convert_output_multiple_column_error(fresh_db):
     table = fresh_db["table"]
-    with pytest.raises(AssertionError) as excinfo:
+    with pytest.raises(ValueError) as excinfo:
         table.convert(["title", "other"], lambda v: v, output="out")
         assert "output= can only be used with a single column" in str(excinfo.value)
 

tests/test_create.py

@@ -3,6 +3,7 @@ from sqlite_utils.db import (
     Database,
     DescIndex,
     AlterError,
+    InvalidColumns,
     NoObviousTable,
     OperationalError,
     ForeignKey,
@@ -20,7 +21,6 @@ import pathlib
 import pytest
 import uuid
 
-
 try:
     import pandas as pd  # type: ignore
 except ImportError:
@@ -109,7 +109,7 @@ def test_create_table_with_defaults(fresh_db):
 
 
 def test_create_table_with_bad_not_null(fresh_db):
-    with pytest.raises(AssertionError):
+    with pytest.raises(ValueError):
         fresh_db.create_table(
             "players", {"name": str, "score": int}, not_null={"mouse"}
         )
@@ -244,11 +244,11 @@ def test_create_table_column_order(fresh_db, use_table_factory):
         # If you specify a column that doesn't point to a table, you  get an error:
         (("one_id", "two_id", "three_id"), NoObviousTable),
         # Tuples of the wrong length get an error:
-        ((("one_id", "one", "id", "five"), ("two_id", "two", "id")), AssertionError),
+        ((("one_id", "one", "id", "five"), ("two_id", "two", "id")), ValueError),
         # Likewise a bad column:
         ((("one_id", "one", "id2"),), AlterError),
         # Or a list of dicts
-        (({"one_id": "one"},), AssertionError),
+        (({"one_id": "one"},), ValueError),
     ),
 )
 @pytest.mark.parametrize("use_table_factory", [True, False])
@@ -701,7 +701,7 @@ def test_bulk_insert_more_than_999_values(fresh_db):
 def test_error_if_more_than_999_columns(fresh_db, num_columns, should_error):
     record = dict([("c{}".format(i), i) for i in range(num_columns)])
     if should_error:
-        with pytest.raises(AssertionError):
+        with pytest.raises(ValueError):
             fresh_db["big"].insert(record)
     else:
         fresh_db["big"].insert(record)
@@ -809,6 +809,34 @@ def test_create_index_if_not_exists(fresh_db):
     dogs.create_index(["name"], if_not_exists=True)
 
 
+def test_drop_index(fresh_db):
+    dogs = fresh_db["dogs"]
+    dogs.insert({"name": "Cleo", "twitter": "cleopaws", "age": 3, "is_good_dog": True})
+    dogs.create_index(["name"])
+    assert [index.name for index in dogs.indexes] == ["idx_dogs_name"]
+    dogs.drop_index("idx_dogs_name")
+    assert dogs.indexes == []
+
+
+def test_drop_index_ignore(fresh_db):
+    dogs = fresh_db["dogs"]
+    dogs.insert({"name": "Cleo"})
+    with pytest.raises(OperationalError, match="No index named idx_dogs_name"):
+        dogs.drop_index("idx_dogs_name")
+    dogs.drop_index("idx_dogs_name", ignore=True)
+
+
+def test_drop_index_wrong_table(fresh_db):
+    dogs = fresh_db["dogs"]
+    cats = fresh_db["cats"]
+    dogs.insert({"name": "Cleo"})
+    cats.insert({"name": "Misty"})
+    dogs.create_index(["name"])
+    with pytest.raises(OperationalError, match="No index named idx_dogs_name"):
+        cats.drop_index("idx_dogs_name")
+    assert [index.name for index in dogs.indexes] == ["idx_dogs_name"]
+
+
 def test_create_index_desc(fresh_db):
     dogs = fresh_db["dogs"]
     dogs.insert({"name": "Cleo", "twitter": "cleopaws", "age": 3, "is good dog": True})
@@ -926,6 +954,58 @@ def test_insert_thousands_adds_extra_columns_after_first_100_with_alter(fresh_db
     assert rows == [{"i": 101, "word": None, "extra": "Should trigger ALTER"}]
 
 
+@pytest.mark.parametrize("num_rows", (0, 1, 2, 3, 10))
+def test_insert_all_pk_not_in_records_raises(fresh_db, num_rows):
+    # https://github.com/simonw/sqlite-utils/issues/732
+    fresh_db.conn.execute("CREATE TABLE t (a TEXT, b INT, PRIMARY KEY (a, b))")
+    rows = [{"a": "x{}".format(i), "b": i} for i in range(num_rows)]
+
+    with pytest.raises(InvalidColumns) as ex:
+        fresh_db["t"].insert_all(rows, pk="not_a_column")
+
+    assert ex.value.args == (
+        "Invalid primary key column ['not_a_column'] for table t with columns ['a', 'b']",
+    )
+    assert fresh_db["t"].count == 0
+
+
+@pytest.mark.parametrize("num_rows", (1, 2, 3, 10))
+def test_insert_all_pk_not_in_records_alter_raises(fresh_db, num_rows):
+    # With alter=True the check is deferred until the record keys are
+    # known - a pk column that is in neither the table nor the records
+    # still raises
+    fresh_db.conn.execute("CREATE TABLE t (a TEXT, b INT, PRIMARY KEY (a, b))")
+    rows = [{"a": "x{}".format(i), "b": i} for i in range(num_rows)]
+
+    with pytest.raises(InvalidColumns) as ex:
+        fresh_db["t"].insert_all(rows, pk="not_a_column", alter=True)
+
+    assert ex.value.args == (
+        "Invalid primary key column ['not_a_column'] for table t with columns ['a', 'b']",
+    )
+    assert fresh_db["t"].count == 0
+
+
+def test_insert_pk_in_records_with_alter_adds_column(fresh_db):
+    # 3.x allowed insert(pk=..., alter=True) to add the pk column from the
+    # records - the InvalidColumns check must not fire in that case
+    fresh_db["t"].insert({"a": 1})
+    fresh_db["t"].insert({"id": 5, "a": 2}, pk="id", alter=True)
+    assert fresh_db["t"].columns_dict.keys() == {"a", "id"}
+    assert list(fresh_db.query("select * from t order by a")) == [
+        {"a": 1, "id": None},
+        {"a": 2, "id": 5},
+    ]
+
+
+def test_insert_all_invalid_pk_alter_empty_records_is_noop(fresh_db):
+    # With alter=True the pk check needs record keys, so an empty iterator
+    # returns without error - matching the 3.x no-op for empty inserts
+    fresh_db.conn.execute("CREATE TABLE t (a TEXT)")
+    fresh_db["t"].insert_all([], pk="not_a_column", alter=True)
+    assert fresh_db["t"].count == 0
+
+
 def test_insert_ignore(fresh_db):
     fresh_db["test"].insert({"id": 1, "bar": 2}, pk="id")
     # Should raise an error if we try this again
@@ -938,6 +1018,114 @@ def test_insert_ignore(fresh_db):
     assert rows == [{"id": 1, "bar": 2}]
 
 
+def test_insert_ignore_reports_existing_row(fresh_db):
+    # An ignored insert (row already exists) should point last_rowid and
+    # last_pk at the existing conflicting row - see the Datasette insert API
+    fresh_db["docs"].insert({"id": 1, "title": "Exists"}, pk="id")
+    # Insert a conflicting row with ignore=True and no explicit pk=
+    table = fresh_db["docs"].insert({"id": 1, "title": "One"}, ignore=True)
+    assert table.last_rowid == 1
+    assert table.last_pk == 1
+    assert list(fresh_db["docs"].rows_where("rowid = ?", [table.last_rowid])) == [
+        {"id": 1, "title": "Exists"}
+    ]
+
+
+@pytest.mark.parametrize("rowid_alias", ("rowid", "_rowid_", "oid"))
+@pytest.mark.parametrize("method", ("upsert", "insert_replace", "insert_ignore"))
+def test_pk_rowid_alias_on_rowid_table(fresh_db, rowid_alias, method):
+    # rowid and its aliases are valid primary keys for a rowid table even
+    # though they are not listed among the table's columns - see the Datasette
+    # upsert API against tables without an explicit primary key
+    fresh_db["t"].insert({"title": "Hello"})
+    assert fresh_db["t"].pks == ["rowid"]
+    record = {rowid_alias: 1, "title": "Updated"}
+    if method == "upsert":
+        table = fresh_db["t"].upsert(record, pk=rowid_alias)
+    elif method == "insert_replace":
+        table = fresh_db["t"].insert(record, pk=rowid_alias, replace=True)
+    else:
+        table = fresh_db["t"].insert(record, pk=rowid_alias, ignore=True)
+    assert table.last_pk == 1
+    expected_title = "Hello" if method == "insert_ignore" else "Updated"
+    assert list(fresh_db["t"].rows) == [{"title": expected_title}]
+
+
+def test_insert_ignore_reports_existing_row_compound_pk(fresh_db):
+    # Compound primary key variant of the ignored-insert lookup
+    fresh_db["t"].insert_all([{"a": 1, "b": 2, "note": "first"}], pk=("a", "b"))
+    table = fresh_db["t"].insert(
+        {"a": 1, "b": 2, "note": "second"}, pk=("a", "b"), ignore=True
+    )
+    assert table.last_pk == (1, 2)
+    assert list(fresh_db["t"].rows_where("rowid = ?", [table.last_rowid])) == [
+        {"a": 1, "b": 2, "note": "first"}
+    ]
+
+
+def test_insert_ignore_reports_existing_row_list_mode(fresh_db):
+    # List-based iteration variant of the ignored-insert lookup
+    fresh_db["t"].insert_all([["id", "title"], [1, "first"]], pk="id")
+    table = fresh_db["t"].insert_all(
+        [["id", "title"], [1, "second"]], pk="id", ignore=True
+    )
+    assert table.last_pk == 1
+    assert table.last_rowid == 1
+    assert list(fresh_db["t"].rows) == [{"id": 1, "title": "first"}]
+
+
+def test_insert_ignore_hash_id_reports_pk(fresh_db):
+    # With hash_id the pk is the computed hash; the original record has no id
+    # column to look up so last_rowid is left unset
+    first = fresh_db["dogs"].insert({"name": "Cleo"}, hash_id="id")
+    table = fresh_db["dogs"].insert({"name": "Cleo"}, hash_id="id", ignore=True)
+    assert table.last_pk == first.last_pk
+    assert table.last_rowid is None
+    assert fresh_db["dogs"].count == 1
+
+
+def test_insert_ignore_unresolvable_conflict_leaves_pk_unset(fresh_db):
+    # When the conflict cannot be resolved to a primary key lookup, last_pk and
+    # last_rowid are left unset rather than reporting a misleading value
+
+    # rowid table with a UNIQUE column and no primary key: no pk to look up
+    fresh_db["u"].db.execute("create table u (title text unique)")
+    fresh_db["u"].insert({"title": "x"})
+    table = fresh_db["u"].insert({"title": "x"}, ignore=True)
+    assert table.last_pk is None
+    assert table.last_rowid is None
+    assert fresh_db["u"].count == 1
+
+    # Conflict on a UNIQUE column other than the primary key: the pk value from
+    # the record does not match the existing row, so the lookup finds nothing
+    fresh_db["docs"].db.execute(
+        "create table docs (id integer primary key, email text unique)"
+    )
+    fresh_db["docs"].insert({"id": 1, "email": "a"}, pk="id")
+    table = fresh_db["docs"].insert({"id": 2, "email": "a"}, ignore=True)
+    assert table.last_pk is None
+    assert table.last_rowid is None
+    assert fresh_db["docs"].count == 1
+
+
+def test_insert_ignore_with_pk_after_other_table_insert(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/554
+    user = {"id": "abc", "name": "david"}
+
+    fresh_db["users"].insert(user, pk="id")
+    fresh_db["comments"].insert_all(
+        [
+            {"id": "def", "text": "ok"},
+            {"id": "ghi", "text": "great"},
+        ],
+    )
+
+    table = fresh_db["users"].insert(user, pk="id", ignore=True)
+
+    assert table.last_pk == "abc"
+    assert list(fresh_db["users"].rows) == [user]
+
+
 def test_insert_hash_id(fresh_db):
     dogs = fresh_db["dogs"]
     id = dogs.insert({"name": "Cleo", "twitter": "cleopaws"}, hash_id="id").last_pk
@@ -1062,7 +1250,7 @@ def test_create_table_numpy(fresh_db):
 
 def test_cannot_provide_both_filename_and_memory():
     with pytest.raises(
-        AssertionError, match="Either specify a filename_or_conn or pass memory=True"
+        ValueError, match="Either specify a filename_or_conn or pass memory=True"
     ):
         Database("/tmp/foo.db", memory=True)
 
@@ -1215,7 +1403,7 @@ def test_create_if_not_exists(fresh_db):
 
 
 def test_create_if_no_columns(fresh_db):
-    with pytest.raises(AssertionError) as error:
+    with pytest.raises(ValueError) as error:
         fresh_db["t"].create({})
     assert error.value.args[0] == "Tables must have at least one column"
 
@@ -1383,7 +1571,10 @@ def test_bad_table_and_view_exceptions(fresh_db):
     assert ex.value.args[0] == "Table v is actually a view"
     with pytest.raises(NoView) as ex2:
         fresh_db.view("t")
-    assert ex2.value.args[0] == "View t does not exist"
+    assert ex2.value.args[0] == "View t does not exist - t is a table"
+    with pytest.raises(NoView) as ex3:
+        fresh_db.view("missing")
+    assert ex3.value.args[0] == "View missing does not exist"
 
 
 # Tests for issue #655: Table configuration should be stored in _defaults

tests/test_create_view.py

@@ -15,7 +15,7 @@ def test_create_view_error(fresh_db):
 
 
 def test_create_view_only_arrow_one_param(fresh_db):
-    with pytest.raises(AssertionError):
+    with pytest.raises(ValueError):
         fresh_db.create_view("bar", "select 1 + 2", ignore=True, replace=True)
 
 

tests/test_default_value.py

@@ -1,6 +1,5 @@
 import pytest
 
-
 EXAMPLES = [
     ("TEXT DEFAULT 'foo'", "'foo'", "'foo'"),
     ("TEXT DEFAULT 'foo)'", "'foo)'", "'foo)'"),
@@ -22,6 +21,11 @@ EXAMPLES = [
     # Strings
     ("TEXT DEFAULT 'CURRENT_TIMESTAMP'", "'CURRENT_TIMESTAMP'", "'CURRENT_TIMESTAMP'"),
     ('TEXT DEFAULT "CURRENT_TIMESTAMP"', '"CURRENT_TIMESTAMP"', '"CURRENT_TIMESTAMP"'),
+    # Boolean and null keyword literals must stay unquoted
+    ("INTEGER DEFAULT TRUE", "TRUE", "TRUE"),
+    ("INTEGER DEFAULT FALSE", "FALSE", "FALSE"),
+    ("INTEGER DEFAULT true", "true", "true"),
+    ("TEXT DEFAULT NULL", "NULL", "NULL"),
 ]
 
 
@@ -32,3 +36,24 @@ def test_quote_default_value(fresh_db, column_def, initial_value, expected_value
     assert expected_value == fresh_db.quote_default_value(
         fresh_db["foo"].columns[0].default_value
     )
+
+
+def test_insert_empty_record_uses_default_values(fresh_db):
+    fresh_db.execute("""
+        CREATE TABLE has_defaults (
+            id INTEGER PRIMARY KEY,
+            name TEXT,
+            timestamp TEXT DEFAULT CURRENT_TIMESTAMP,
+            is_active INTEGER NOT NULL DEFAULT 1
+        )
+        """)
+
+    table = fresh_db["has_defaults"]
+    table.insert({})
+
+    rows = list(table.rows)
+    assert len(rows) == 1
+    assert rows[0]["id"] == 1
+    assert rows[0]["name"] is None
+    assert rows[0]["timestamp"] is not None
+    assert rows[0]["is_active"] == 1

tests/test_delete.py

@@ -1,3 +1,6 @@
+import sqlite_utils
+
+
 def test_delete_rowid_table(fresh_db):
     table = fresh_db["table"]
     table.insert({"foo": 1}).last_pk
@@ -32,6 +35,21 @@ def test_delete_where_all(fresh_db):
     assert table.count == 0
 
 
+def test_delete_where_commits(tmpdir):
+    path = str(tmpdir / "test.db")
+    db = sqlite_utils.Database(path)
+    db["table"].insert_all([{"id": i} for i in range(5)], pk="id")
+    db["table"].delete_where("id > ?", [2])
+    # The connection must not be left inside an open transaction,
+    # otherwise subsequent atomic() blocks never commit either
+    assert not db.conn.in_transaction
+    db["table"].insert({"id": 100})
+    db.close()
+    db2 = sqlite_utils.Database(path)
+    assert [r["id"] for r in db2["table"].rows] == [0, 1, 2, 100]
+    db2.close()
+
+
 def test_delete_where_analyze(fresh_db):
     table = fresh_db["table"]
     table.insert_all(({"id": i, "i": i} for i in range(10)), pk="id")

tests/test_docs.py

@@ -41,9 +41,9 @@ def test_convert_help():
     result = CliRunner().invoke(cli.cli, ["convert", "--help"])
     assert result.exit_code == 0
     for expected in (
-        "r.jsonsplit(value, ",
-        "r.parsedate(value, ",
-        "r.parsedatetime(value, ",
+        "r.jsonsplit(value:",
+        "r.parsedate(value:",
+        "r.parsedatetime(value:",
     ):
         assert expected in result.output
 
@@ -54,7 +54,7 @@ def test_convert_help():
         n
         for n in dir(recipes)
         if not n.startswith("_")
-        and n not in ("json", "parser")
+        and n not in ("json", "parser", "Callable", "Optional")
         and callable(getattr(recipes, n))
     ],
 )

tests/test_duplicate.py

@@ -5,14 +5,12 @@ import pytest
 
 def test_duplicate(fresh_db):
     # Create table using native Sqlite statement:
-    fresh_db.execute(
-        """CREATE TABLE "table1" (
+    fresh_db.execute("""CREATE TABLE "table1" (
     "text_col" TEXT,
     "real_col" REAL,
     "int_col" INTEGER,
     "bool_col" INTEGER,
-    "datetime_col" TEXT)"""
-    )
+    "datetime_col" TEXT)""")
     # Insert one row of mock data:
     dt = datetime.datetime.now()
     data = {

tests/test_extract.py

@@ -126,9 +126,7 @@ def test_extract_rowid_table(fresh_db):
         '   "common_name_latin_name_id" INTEGER REFERENCES "common_name_latin_name"("id")\n'
         ")"
     )
-    assert (
-        fresh_db.execute(
-            """
+    assert fresh_db.execute("""
         select
             tree.name,
             common_name_latin_name.common_name,
@@ -136,10 +134,7 @@ def test_extract_rowid_table(fresh_db):
         from tree
             join common_name_latin_name
             on tree.common_name_latin_name_id = common_name_latin_name.id
-    """
-        ).fetchall()
-        == [("Tree 1", "Palm", "Arecaceae")]
-    )
+    """).fetchall() == [("Tree 1", "Palm", "Arecaceae")]
 
 
 def test_reuse_lookup_table(fresh_db):
@@ -194,9 +189,116 @@ def test_extract_works_with_null_values(fresh_db):
     )
     assert list(fresh_db["listens"].rows) == [
         {"id": 1, "track_title": "foo", "album_id": 1},
-        {"id": 2, "track_title": "baz", "album_id": 2},
+        {"id": 2, "track_title": "baz", "album_id": None},
     ]
     assert list(fresh_db["albums"].rows) == [
         {"id": 1, "album_title": "bar"},
-        {"id": 2, "album_title": None},
     ]
+
+
+def test_extract_null_values_single_column(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/186
+    fresh_db["species"].insert({"id": 1, "species": "Wolf"}, pk="id")
+    fresh_db["individuals"].insert_all(
+        [
+            {"id": 10, "name": "Terriana", "species": "Fox"},
+            {"id": 11, "name": "Spenidorm", "species": None},
+            {"id": 12, "name": "Grantheim", "species": "Wolf"},
+            {"id": 13, "name": "Turnutopia", "species": None},
+            {"id": 14, "name": "Wargal", "species": "Wolf"},
+        ],
+        pk="id",
+    )
+    fresh_db["individuals"].extract("species")
+    # No null row should have been added to species
+    assert list(fresh_db["species"].rows) == [
+        {"id": 1, "species": "Wolf"},
+        {"id": 2, "species": "Fox"},
+    ]
+    assert list(fresh_db["individuals"].rows) == [
+        {"id": 10, "name": "Terriana", "species_id": 2},
+        {"id": 11, "name": "Spenidorm", "species_id": None},
+        {"id": 12, "name": "Grantheim", "species_id": 1},
+        {"id": 13, "name": "Turnutopia", "species_id": None},
+        {"id": 14, "name": "Wargal", "species_id": 1},
+    ]
+
+
+def test_extract_null_values_multiple_columns(fresh_db):
+    # A row should be extracted if at least one column is not null -
+    # only rows where ALL extracted columns are null are left alone
+    fresh_db["circulation"].insert_all(
+        [
+            {"id": 1, "title": "title one", "creator": "creator one", "year": 2018},
+            {"id": 2, "title": "title two", "creator": None, "year": 2019},
+            {"id": 3, "title": None, "creator": None, "year": 2020},
+            {"id": 4, "title": None, "creator": None, "year": 2021},
+        ],
+        pk="id",
+    )
+    fresh_db["circulation"].extract(
+        ["title", "creator"], table="books", fk_column="book_id"
+    )
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "title": "title one", "creator": "creator one"},
+        {"id": 2, "title": "title two", "creator": None},
+    ]
+    assert list(fresh_db["circulation"].rows) == [
+        {"id": 1, "book_id": 1, "year": 2018},
+        {"id": 2, "book_id": 2, "year": 2019},
+        {"id": 3, "book_id": None, "year": 2020},
+        {"id": 4, "book_id": None, "year": 2021},
+    ]
+
+
+def test_extract_null_values_existing_lookup_table_with_null_row(fresh_db):
+    # Even if the lookup table already contains an all-null row, rows where
+    # every extracted column is null should keep a null foreign key
+    fresh_db["species"].insert({"id": 1, "species": None}, pk="id")
+    fresh_db["individuals"].insert_all(
+        [
+            {"id": 10, "name": "Terriana", "species": "Fox"},
+            {"id": 11, "name": "Spenidorm", "species": None},
+        ],
+        pk="id",
+    )
+    fresh_db["individuals"].extract("species")
+    assert list(fresh_db["species"].rows) == [
+        {"id": 1, "species": None},
+        {"id": 2, "species": "Fox"},
+    ]
+    assert list(fresh_db["individuals"].rows) == [
+        {"id": 10, "name": "Terriana", "species_id": 2},
+        {"id": 11, "name": "Spenidorm", "species_id": None},
+    ]
+
+
+def test_extract_repeated_into_shared_lookup_with_nulls(fresh_db):
+    # Unique indexes treat NULLs as distinct, so INSERT OR IGNORE alone
+    # cannot dedupe NULL-containing rows against the existing lookup
+    # table - extracting a second table into the same lookup previously
+    # inserted duplicate rows that nothing pointed to
+    fresh_db["t1"].insert_all(
+        [
+            {"id": 1, "species": None, "common": "X"},
+            {"id": 2, "species": "Oak", "common": "Oak"},
+        ],
+        pk="id",
+    )
+    fresh_db["t2"].insert_all([{"id": 1, "species": None, "common": "X"}], pk="id")
+    fresh_db["t1"].extract(["species", "common"], table="lk")
+    fresh_db["t2"].extract(["species", "common"], table="lk")
+    assert fresh_db["lk"].count == 2
+    # Both tables point at the same lookup row
+    t1_fk = fresh_db.execute("select lk_id from t1 where id = 1").fetchone()[0]
+    t2_fk = fresh_db.execute("select lk_id from t2 where id = 1").fetchone()[0]
+    assert t1_fk == t2_fk
+
+
+def test_extract_repeated_into_shared_lookup_no_nulls(fresh_db):
+    # Non-NULL rows were already deduped by the unique index - keep it so
+    fresh_db["t1"].insert_all([{"id": 1, "species": "Oak"}], pk="id")
+    fresh_db["t2"].insert_all([{"id": 1, "species": "Oak"}], pk="id")
+    fresh_db["t1"].extract(["species"], table="lk")
+    fresh_db["t2"].extract(["species"], table="lk")
+    assert fresh_db["lk"].count == 1

tests/test_extracts.py

@@ -67,3 +67,51 @@ def test_extracts(fresh_db, kwargs, expected_table, use_table_factory):
         {"id": 2, "species_id": 1},
         {"id": 3, "species_id": 2},
     ] == list(fresh_db["Trees"].rows)
+
+
+def test_extracts_null_values(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/186
+    # Null values should stay null, not be extracted into the lookup table
+    fresh_db["Trees"].insert_all(
+        [
+            {"id": 1, "species_id": "Oak"},
+            {"id": 2, "species_id": None},
+            {"id": 3, "species_id": "Palm"},
+            {"id": 4, "species_id": None},
+        ],
+        extracts={"species_id": "Species"},
+    )
+    assert list(fresh_db["Species"].rows) == [
+        {"id": 1, "value": "Oak"},
+        {"id": 2, "value": "Palm"},
+    ]
+    assert list(fresh_db["Trees"].rows) == [
+        {"id": 1, "species_id": 1},
+        {"id": 2, "species_id": None},
+        {"id": 3, "species_id": 2},
+        {"id": 4, "species_id": None},
+    ]
+
+
+def test_extracts_null_values_list_mode(fresh_db):
+    # Same as test_extracts_null_values but for list-based records
+    fresh_db["Trees"].insert_all(
+        [
+            ["id", "species_id"],
+            [1, "Oak"],
+            [2, None],
+            [3, "Palm"],
+            [4, None],
+        ],
+        extracts={"species_id": "Species"},
+    )
+    assert list(fresh_db["Species"].rows) == [
+        {"id": 1, "value": "Oak"},
+        {"id": 2, "value": "Palm"},
+    ]
+    assert list(fresh_db["Trees"].rows) == [
+        {"id": 1, "species_id": 1},
+        {"id": 2, "species_id": None},
+        {"id": 3, "species_id": 2},
+        {"id": 4, "species_id": None},
+    ]

tests/test_foreign_keys.py

@@ -0,0 +1,691 @@
+"""Tests for compound (multi-column) foreign keys - issue #594."""
+
+import pytest
+from sqlite_utils import Database
+from sqlite_utils.db import AlterError, ForeignKey
+from sqlite_utils.utils import sqlite3
+
+COMPOUND_SCHEMA = """
+CREATE TABLE departments (
+    campus_name TEXT NOT NULL,
+    dept_code TEXT NOT NULL,
+    dept_name TEXT,
+    PRIMARY KEY (campus_name, dept_code)
+);
+CREATE TABLE courses (
+    course_code TEXT PRIMARY KEY,
+    course_name TEXT,
+    campus_name TEXT NOT NULL,
+    dept_code TEXT NOT NULL,
+    FOREIGN KEY (campus_name, dept_code)
+    REFERENCES departments(campus_name, dept_code)
+);
+"""
+
+
+@pytest.fixture
+def compound_db():
+    db = Database(memory=True)
+    db.executescript(COMPOUND_SCHEMA)
+    return db
+
+
+def test_compound_foreign_key(compound_db):
+    fks = compound_db["courses"].foreign_keys
+    assert len(fks) == 1
+    fk = fks[0]
+    assert fk.is_compound is True
+    assert fk.table == "courses"
+    assert fk.other_table == "departments"
+    assert fk.columns == ("campus_name", "dept_code")
+    assert fk.other_columns == ("campus_name", "dept_code")
+    # Scalar column/other_column can't sensibly hold a compound key
+    assert fk.column is None
+    assert fk.other_column is None
+
+
+def test_single_foreign_key_gets_columns_fields(fresh_db):
+    fresh_db["authors"].insert({"id": 1, "name": "Sally"}, pk="id")
+    fresh_db["books"].insert({"title": "Hedgehogs", "author_id": 1})
+    fresh_db["books"].add_foreign_key("author_id", "authors", "id")
+    fk = fresh_db["books"].foreign_keys[0]
+    assert fk.is_compound is False
+    assert fk.column == "author_id"
+    assert fk.other_column == "id"
+    assert fk.columns == ("author_id",)
+    assert fk.other_columns == ("id",)
+
+
+def test_foreign_key_no_longer_unpacks_as_tuple(fresh_db):
+    # Clean break in 4.0: ForeignKey is a dataclass, not a namedtuple, so the
+    # old tuple unpacking and indexing patterns now fail hard.
+    fresh_db["authors"].insert({"id": 1, "name": "Sally"}, pk="id")
+    fresh_db["books"].insert({"title": "Hedgehogs", "author_id": 1})
+    fresh_db["books"].add_foreign_key("author_id", "authors", "id")
+    fk = fresh_db["books"].foreign_keys[0]
+    with pytest.raises(TypeError):
+        table, column, other_table, other_column = fk
+    with pytest.raises(TypeError):
+        fk[0]
+
+
+def test_foreign_keys_are_sortable(fresh_db):
+    fresh_db["authors"].insert({"id": 1, "name": "Sally"}, pk="id")
+    fresh_db["categories"].insert({"id": 1, "name": "Wildlife"}, pk="id")
+    fresh_db["books"].insert({"title": "Hedgehogs", "author_id": 1, "category_id": 1})
+    fresh_db.add_foreign_keys(
+        [
+            ("books", "author_id", "authors", "id"),
+            ("books", "category_id", "categories", "id"),
+        ]
+    )
+    fks = sorted(fresh_db["books"].foreign_keys)
+    assert fks[0].column == "author_id"
+    assert fks[1].column == "category_id"
+
+
+def test_mixed_compound_and_single_foreign_keys_are_sortable():
+    # compound FKs have column=None, which must not break sorting
+    # against single-column FKs (None < str raises TypeError)
+    db = Database(memory=True)
+    db.executescript("""
+    CREATE TABLE departments (
+        campus_name TEXT NOT NULL,
+        dept_code TEXT NOT NULL,
+        PRIMARY KEY (campus_name, dept_code)
+    );
+    CREATE TABLE accreditations (id INTEGER PRIMARY KEY);
+    CREATE TABLE courses (
+        course_code TEXT PRIMARY KEY,
+        campus_name TEXT NOT NULL,
+        dept_code TEXT NOT NULL,
+        accreditation_id INTEGER REFERENCES accreditations(id),
+        FOREIGN KEY (campus_name, dept_code)
+        REFERENCES departments(campus_name, dept_code)
+    );
+    """)
+    fks = db["courses"].foreign_keys
+    assert len(fks) == 2
+    assert {fk.is_compound for fk in fks} == {True, False}
+    fks_sorted = sorted(fks)
+    assert fks_sorted[0].other_table == "accreditations"
+    assert fks_sorted[1].other_table == "departments"
+
+
+@pytest.fixture
+def departments_db():
+    db = Database(memory=True)
+    db.create_table(
+        "departments",
+        {"campus_name": str, "dept_code": str, "dept_name": str},
+        pk=("campus_name", "dept_code"),
+    )
+    return db
+
+
+EXPECTED_COURSES_SCHEMA = (
+    'CREATE TABLE "courses" (\n'
+    '   "course_code" TEXT PRIMARY KEY,\n'
+    '   "campus_name" TEXT,\n'
+    '   "dept_code" TEXT,\n'
+    '   FOREIGN KEY ("campus_name", "dept_code") '
+    'REFERENCES "departments"("campus_name", "dept_code")\n'
+    ")"
+)
+
+
+@pytest.mark.parametrize(
+    "foreign_keys",
+    (
+        [
+            ForeignKey(
+                table="courses",
+                column=None,
+                other_table="departments",
+                other_column=None,
+                columns=("campus_name", "dept_code"),
+                other_columns=("campus_name", "dept_code"),
+                is_compound=True,
+            )
+        ],
+        [(("campus_name", "dept_code"), "departments", ("campus_name", "dept_code"))],
+        # Two-item form guesses the other table's primary key:
+        [(("campus_name", "dept_code"), "departments")],
+        # Lists work too, though tuples are the documented form:
+        [(["campus_name", "dept_code"], "departments", ["campus_name", "dept_code"])],
+    ),
+)
+def test_create_table_with_compound_foreign_key(departments_db, foreign_keys):
+    departments_db.create_table(
+        "courses",
+        {"course_code": str, "campus_name": str, "dept_code": str},
+        pk="course_code",
+        foreign_keys=foreign_keys,
+    )
+    assert departments_db["courses"].schema == EXPECTED_COURSES_SCHEMA
+    fks = departments_db["courses"].foreign_keys
+    assert len(fks) == 1
+    fk = fks[0]
+    assert fk.is_compound is True
+    assert fk.columns == ("campus_name", "dept_code")
+    assert fk.other_table == "departments"
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_create_table_compound_foreign_key_enforced(departments_db):
+    departments_db.execute("PRAGMA foreign_keys = ON")
+    departments_db.create_table(
+        "courses",
+        {"course_code": str, "campus_name": str, "dept_code": str},
+        pk="course_code",
+        foreign_keys=[(("campus_name", "dept_code"), "departments")],
+    )
+    departments_db["departments"].insert(
+        {"campus_name": "Berkeley", "dept_code": "CS", "dept_name": "Computer Science"}
+    )
+    departments_db["courses"].insert(
+        {"course_code": "CS101", "campus_name": "Berkeley", "dept_code": "CS"}
+    )
+    with pytest.raises(sqlite3.IntegrityError):
+        departments_db.execute(
+            "insert into courses (course_code, campus_name, dept_code) "
+            "values ('X1', 'Nowhere', 'NOPE')"
+        )
+
+
+def test_create_table_compound_foreign_key_missing_other_column(departments_db):
+    with pytest.raises(AlterError):
+        departments_db.create_table(
+            "courses",
+            {"course_code": str, "campus_name": str, "dept_code": str},
+            pk="course_code",
+            foreign_keys=[
+                (("campus_name", "dept_code"), "departments", ("campus_name", "nope"))
+            ],
+        )
+
+
+def test_transform_preserves_compound_foreign_key(compound_db):
+    compound_db["courses"].transform(rename={"course_name": "title"})
+    fks = compound_db["courses"].foreign_keys
+    assert len(fks) == 1
+    fk = fks[0]
+    assert fk.is_compound is True
+    assert fk.columns == ("campus_name", "dept_code")
+    assert fk.other_table == "departments"
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_transform_rename_member_column_updates_compound_foreign_key(compound_db):
+    compound_db["courses"].transform(rename={"campus_name": "campus"})
+    fks = compound_db["courses"].foreign_keys
+    assert len(fks) == 1
+    fk = fks[0]
+    assert fk.is_compound is True
+    assert fk.columns == ("campus", "dept_code")
+    # Referenced columns in the other table are unchanged
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_transform_drop_member_column_drops_compound_foreign_key(compound_db):
+    # Matches single-column behavior: dropping the column silently
+    # drops the foreign key that used it
+    compound_db["courses"].transform(drop={"dept_code"})
+    assert compound_db["courses"].foreign_keys == []
+    assert "FOREIGN KEY" not in compound_db["courses"].schema
+
+
+@pytest.mark.parametrize(
+    "drop_foreign_keys",
+    (
+        # A bare column name matches any foreign key it participates in:
+        ["campus_name"],
+        # A tuple must match the full compound key:
+        [("campus_name", "dept_code")],
+    ),
+)
+def test_transform_drop_compound_foreign_key(compound_db, drop_foreign_keys):
+    compound_db["courses"].transform(drop_foreign_keys=drop_foreign_keys)
+    assert compound_db["courses"].foreign_keys == []
+    # The columns themselves survive
+    assert {"campus_name", "dept_code"} <= set(
+        compound_db["courses"].columns_dict.keys()
+    )
+
+
+@pytest.fixture
+def courses_db(departments_db):
+    departments_db.create_table(
+        "courses",
+        {"course_code": str, "campus_name": str, "dept_code": str},
+        pk="course_code",
+    )
+    return departments_db
+
+
+def test_add_compound_foreign_key(courses_db):
+    t = courses_db["courses"].add_foreign_key(
+        ("campus_name", "dept_code"), "departments", ("campus_name", "dept_code")
+    )
+    # Returns self
+    assert t.name == "courses"
+    fks = courses_db["courses"].foreign_keys
+    assert len(fks) == 1
+    fk = fks[0]
+    assert fk.is_compound is True
+    assert fk.columns == ("campus_name", "dept_code")
+    assert fk.other_table == "departments"
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_add_compound_foreign_key_guesses_other_columns(courses_db):
+    # Lists work here too, though tuples are the documented form
+    courses_db["courses"].add_foreign_key(["campus_name", "dept_code"], "departments")
+    fk = courses_db["courses"].foreign_keys[0]
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_add_compound_foreign_key_error_if_already_exists(courses_db):
+    courses_db["courses"].add_foreign_key(("campus_name", "dept_code"), "departments")
+    with pytest.raises(AlterError) as ex:
+        courses_db["courses"].add_foreign_key(
+            ("campus_name", "dept_code"), "departments"
+        )
+    assert "already exists" in ex.value.args[0]
+    # ignore=True should not raise
+    courses_db["courses"].add_foreign_key(
+        ("campus_name", "dept_code"), "departments", ignore=True
+    )
+
+
+def test_add_compound_foreign_key_error_if_column_missing(courses_db):
+    with pytest.raises(AlterError):
+        courses_db["courses"].add_foreign_key(("campus_name", "nope"), "departments")
+
+
+def test_db_add_foreign_keys_compound(courses_db):
+    courses_db.add_foreign_keys(
+        [
+            (
+                "courses",
+                ("campus_name", "dept_code"),
+                "departments",
+                ("campus_name", "dept_code"),
+            )
+        ]
+    )
+    fk = courses_db["courses"].foreign_keys[0]
+    assert fk.is_compound is True
+    assert fk.columns == ("campus_name", "dept_code")
+
+
+def test_index_foreign_keys_compound_creates_composite_index(compound_db):
+    compound_db.index_foreign_keys()
+    index_columns = [i.columns for i in compound_db["courses"].indexes]
+    assert ["campus_name", "dept_code"] in index_columns
+    # No separate single-column indexes for the members
+    assert ["campus_name"] not in index_columns
+    assert ["dept_code"] not in index_columns
+
+
+def test_foreign_key_captures_on_delete_and_on_update():
+    db = Database(memory=True)
+    db.executescript("""
+    CREATE TABLE authors (id INTEGER PRIMARY KEY);
+    CREATE TABLE books (
+        id INTEGER PRIMARY KEY,
+        author_id INTEGER REFERENCES authors(id)
+            ON DELETE CASCADE ON UPDATE RESTRICT
+    );
+    """)
+    fk = db["books"].foreign_keys[0]
+    assert fk.on_delete == "CASCADE"
+    assert fk.on_update == "RESTRICT"
+
+
+def test_foreign_key_on_delete_defaults_to_no_action(fresh_db):
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["books"].insert({"id": 1, "author_id": 1}, pk="id")
+    fresh_db["books"].add_foreign_key("author_id", "authors", "id")
+    fk = fresh_db["books"].foreign_keys[0]
+    assert fk.on_delete == "NO ACTION"
+    assert fk.on_update == "NO ACTION"
+
+
+def test_create_table_foreign_key_with_on_delete(fresh_db):
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db.create_table(
+        "books",
+        {"id": int, "author_id": int},
+        pk="id",
+        foreign_keys=[
+            ForeignKey(
+                table="books",
+                column="author_id",
+                other_table="authors",
+                other_column="id",
+                on_delete="CASCADE",
+            )
+        ],
+    )
+    assert "ON DELETE CASCADE" in fresh_db["books"].schema
+    assert fresh_db["books"].foreign_keys[0].on_delete == "CASCADE"
+
+
+def test_transform_preserves_on_delete_cascade():
+    db = Database(memory=True)
+    db.executescript("""
+    CREATE TABLE authors (id INTEGER PRIMARY KEY);
+    CREATE TABLE books (
+        id INTEGER PRIMARY KEY,
+        title TEXT,
+        author_id INTEGER REFERENCES authors(id) ON DELETE CASCADE
+    );
+    """)
+    db["books"].transform(rename={"title": "book_title"})
+    fk = db["books"].foreign_keys[0]
+    assert fk.on_delete == "CASCADE"
+    assert fk.on_update == "NO ACTION"
+    assert "ON DELETE CASCADE" in db["books"].schema
+
+
+def test_transform_preserves_compound_foreign_key_on_delete():
+    db = Database(memory=True)
+    db.executescript("""
+    CREATE TABLE departments (
+        campus_name TEXT NOT NULL,
+        dept_code TEXT NOT NULL,
+        PRIMARY KEY (campus_name, dept_code)
+    );
+    CREATE TABLE courses (
+        course_code TEXT PRIMARY KEY,
+        campus_name TEXT NOT NULL,
+        dept_code TEXT NOT NULL,
+        FOREIGN KEY (campus_name, dept_code)
+        REFERENCES departments(campus_name, dept_code) ON DELETE CASCADE
+    );
+    """)
+    db["courses"].transform(rename={"course_code": "code"})
+    fk = db["courses"].foreign_keys[0]
+    assert fk.is_compound is True
+    assert fk.on_delete == "CASCADE"
+    assert "ON DELETE CASCADE" in db["courses"].schema
+
+
+def test_implicit_primary_key_reference_is_resolved():
+    # REFERENCES authors (no column) has "to" of None in the pragma -
+    # it should be resolved to the primary key of the other table
+    db = Database(memory=True)
+    db.executescript("""
+    CREATE TABLE authors (author_id INTEGER PRIMARY KEY);
+    CREATE TABLE books (
+        id INTEGER PRIMARY KEY,
+        author_id INTEGER REFERENCES authors
+    );
+    """)
+    fk = db["books"].foreign_keys[0]
+    assert fk.is_compound is False
+    assert fk.other_column == "author_id"
+    assert fk.other_columns == ("author_id",)
+
+
+def test_implicit_compound_primary_key_reference_is_resolved():
+    db = Database(memory=True)
+    db.executescript("""
+    CREATE TABLE departments (
+        campus_name TEXT NOT NULL,
+        dept_code TEXT NOT NULL,
+        PRIMARY KEY (campus_name, dept_code)
+    );
+    CREATE TABLE courses (
+        course_code TEXT PRIMARY KEY,
+        campus_name TEXT NOT NULL,
+        dept_code TEXT NOT NULL,
+        FOREIGN KEY (campus_name, dept_code) REFERENCES departments
+    );
+    """)
+    fk = db["courses"].foreign_keys[0]
+    assert fk.is_compound is True
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_foreign_key_normalizes_list_columns_to_tuples():
+    # Compound columns passed as lists are normalized to tuples, so they
+    # compare equal to introspected ForeignKeys
+    fk = ForeignKey(
+        table="courses",
+        column=None,
+        other_table="departments",
+        other_column=None,
+        columns=["campus_name", "dept_code"],
+        other_columns=["campus_name", "dept_code"],
+        is_compound=True,
+    )
+    assert fk.columns == ("campus_name", "dept_code")
+    assert fk.other_columns == ("campus_name", "dept_code")
+
+
+def test_add_foreign_keys_preserves_actions(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/594 review finding:
+    # ForeignKey objects passed to db.add_foreign_keys() were flattened
+    # to plain tuples, losing on_delete/on_update
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["books"].insert({"id": 1, "author_id": 1}, pk="id")
+    fresh_db.add_foreign_keys(
+        [ForeignKey("books", "author_id", "authors", "id", on_delete="CASCADE")]
+    )
+    fk = fresh_db["books"].foreign_keys[0]
+    assert fk.on_delete == "CASCADE"
+    assert "ON DELETE CASCADE" in fresh_db["books"].schema
+
+
+def test_add_foreign_keys_preserves_actions_compound(courses_db):
+    courses_db.add_foreign_keys(
+        [
+            ForeignKey(
+                table="courses",
+                column=None,
+                other_table="departments",
+                other_column=None,
+                columns=("campus_name", "dept_code"),
+                other_columns=("campus_name", "dept_code"),
+                is_compound=True,
+                on_delete="CASCADE",
+            )
+        ]
+    )
+    fk = courses_db["courses"].foreign_keys[0]
+    assert fk.is_compound is True
+    assert fk.on_delete == "CASCADE"
+    assert "ON DELETE CASCADE" in courses_db["courses"].schema
+
+
+def test_add_foreign_key_on_delete_on_update(fresh_db):
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["books"].insert({"id": 1, "author_id": 1}, pk="id")
+    fresh_db["books"].add_foreign_key(
+        "author_id", "authors", "id", on_delete="CASCADE", on_update="RESTRICT"
+    )
+    fk = fresh_db["books"].foreign_keys[0]
+    assert fk.on_delete == "CASCADE"
+    assert fk.on_update == "RESTRICT"
+    assert "ON UPDATE RESTRICT ON DELETE CASCADE" in fresh_db["books"].schema
+    # The cascade should actually fire
+    fresh_db.execute("PRAGMA foreign_keys = ON")
+    fresh_db.execute("delete from authors where id = 1")
+    assert fresh_db["books"].count == 0
+
+
+def test_add_compound_foreign_key_on_delete(courses_db):
+    courses_db["courses"].add_foreign_key(
+        ("campus_name", "dept_code"), "departments", on_delete="SET NULL"
+    )
+    fk = courses_db["courses"].foreign_keys[0]
+    assert fk.is_compound is True
+    assert fk.on_delete == "SET NULL"
+    assert "ON DELETE SET NULL" in courses_db["courses"].schema
+
+
+def test_implicit_compound_foreign_key_resolves_pk_declaration_order(fresh_db):
+    # The other table's PRIMARY KEY declares its columns in a different
+    # order to the table's column order. SQLite resolves the implicit
+    # "REFERENCES other" using PRIMARY KEY declaration order, so the
+    # introspected other_columns must too
+    fresh_db.execute("create table other (b text, a text, primary key (a, b))")
+    fresh_db.execute(
+        "create table child (x text, y text, foreign key (x, y) references other)"
+    )
+    fk = fresh_db["child"].foreign_keys[0]
+    assert fk.other_columns == ("a", "b")
+
+
+def test_transform_implicit_compound_foreign_key_stays_valid(fresh_db):
+    # transform() rewrites the implicit FK with explicit columns - they
+    # must be in PRIMARY KEY declaration order or valid data fails the
+    # foreign key check with an IntegrityError
+    fresh_db.execute("create table other (b text, a text, primary key (a, b))")
+    fresh_db.execute(
+        "create table child (x text, y text, foreign key (x, y) references other)"
+    )
+    fresh_db.execute("PRAGMA foreign_keys = ON")
+    fresh_db["other"].insert({"a": "A", "b": "B"})
+    fresh_db["child"].insert({"x": "A", "y": "B"})
+    fresh_db["child"].transform(types={"x": str})
+    assert fresh_db["child"].foreign_keys[0].other_columns == ("a", "b")
+    # The constraint still points the right way around
+    fresh_db["child"].insert({"x": "A", "y": "B"})
+    with pytest.raises(sqlite3.IntegrityError):
+        fresh_db["child"].insert({"x": "B", "y": "A"})
+
+
+def test_create_compound_foreign_key_guesses_pk_declaration_order(fresh_db):
+    fresh_db.execute("create table other (b text, a text, primary key (a, b))")
+    fresh_db["other"].insert({"a": "A", "b": "B"})
+    fresh_db["child"].create(
+        {"id": int, "x": str, "y": str},
+        pk="id",
+        foreign_keys=[(("x", "y"), "other")],
+    )
+    assert fresh_db["child"].foreign_keys[0].other_columns == ("a", "b")
+    fresh_db.execute("PRAGMA foreign_keys = ON")
+    fresh_db["child"].insert({"id": 1, "x": "A", "y": "B"})
+    with pytest.raises(sqlite3.IntegrityError):
+        fresh_db["child"].insert({"id": 2, "x": "B", "y": "A"})
+
+
+def test_add_compound_foreign_key_guesses_pk_declaration_order(fresh_db):
+    fresh_db.execute("create table other (b text, a text, primary key (a, b))")
+    fresh_db["child"].insert({"id": 1, "x": "A", "y": "B"}, pk="id")
+    fresh_db["child"].add_foreign_key(("x", "y"), "other")
+    assert fresh_db["child"].foreign_keys[0].other_columns == ("a", "b")
+
+
+def test_foreign_keys_are_hashable(fresh_db):
+    # set() over foreign_keys worked with the 3.x namedtuple and must
+    # keep working with the dataclass
+    fresh_db["p"].insert({"id": 1}, pk="id")
+    fresh_db["c"].insert(
+        {"id": 1, "pid": 1}, pk="id", foreign_keys=[("pid", "p", "id")]
+    )
+    fks = set(fresh_db["c"].foreign_keys)
+    assert len(fks) == 1
+    assert ForeignKey("c", "pid", "p", "id") in fks
+    # Usable as dict keys too
+    assert {fk: True for fk in fks}
+
+
+def test_foreign_key_is_immutable():
+    import dataclasses
+
+    fk = ForeignKey("c", "pid", "p", "id")
+    with pytest.raises(dataclasses.FrozenInstanceError):
+        fk.table = "other"
+
+
+def test_foreign_key_equality_and_hash_include_actions():
+    # Two foreign keys differing only in ON DELETE behavior are different
+    # constraints - they compare unequal and hash separately
+    plain = ForeignKey("c", "pid", "p", "id")
+    cascade = ForeignKey("c", "pid", "p", "id", on_delete="CASCADE")
+    assert plain != cascade
+    assert len({plain, cascade}) == 2
+    assert plain == ForeignKey("c", "pid", "p", "id")
+
+
+def test_create_table_mixed_foreign_keys_list(fresh_db):
+    # 3.x accepted a mix of ForeignKey objects, tuples and bare column
+    # strings in foreign_keys= (ForeignKey was a namedtuple, so it passed
+    # the tuple check) - keep accepting the mix
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["publishers"].insert({"id": 1}, pk="id")
+    fresh_db["books"].create(
+        {"id": int, "author_id": int, "publisher_id": int},
+        pk="id",
+        foreign_keys=[
+            ForeignKey("books", "author_id", "authors", "id"),
+            ("publisher_id", "publishers", "id"),
+        ],
+    )
+    fks = {fk.column: fk.other_table for fk in fresh_db["books"].foreign_keys}
+    assert fks == {"author_id": "authors", "publisher_id": "publishers"}
+
+
+def test_create_table_mixed_foreign_keys_with_string(fresh_db):
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["publishers"].insert({"id": 1}, pk="id")
+    fresh_db["books"].create(
+        {"id": int, "author_id": int, "publisher_id": int},
+        pk="id",
+        foreign_keys=[
+            "author_id",  # bare column, table and column guessed
+            ("publisher_id", "publishers", "id"),
+        ],
+    )
+    fks = {fk.column: fk.other_table for fk in fresh_db["books"].foreign_keys}
+    assert fks == {"author_id": "authors", "publisher_id": "publishers"}
+
+
+def test_add_foreign_keys_existing_with_different_actions_errors(fresh_db):
+    # Requesting an existing foreign key with different ON DELETE/ON UPDATE
+    # actions was silently skipped, dropping the requested change
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["books"].insert(
+        {"id": 1, "author_id": 1},
+        pk="id",
+        foreign_keys=[("author_id", "authors", "id")],
+    )
+    with pytest.raises(AlterError) as ex:
+        fresh_db.add_foreign_keys(
+            [ForeignKey("books", "author_id", "authors", "id", on_delete="CASCADE")]
+        )
+    assert "ON DELETE" in str(ex.value)
+    assert fresh_db["books"].foreign_keys[0].on_delete == "NO ACTION"
+
+
+def test_add_foreign_keys_identical_existing_is_noop(fresh_db):
+    # An exact match, including actions, is silently skipped so repeated
+    # calls stay idempotent
+    fresh_db["authors"].insert({"id": 1}, pk="id")
+    fresh_db["books"].insert({"id": 1, "author_id": 1}, pk="id")
+    fresh_db["books"].add_foreign_key("author_id", "authors", "id", on_delete="CASCADE")
+    fresh_db.add_foreign_keys(
+        [ForeignKey("books", "author_id", "authors", "id", on_delete="CASCADE")]
+    )
+    fks = fresh_db["books"].foreign_keys
+    assert len(fks) == 1
+    assert fks[0].on_delete == "CASCADE"
+
+
+def test_add_foreign_keys_compound_column_count_mismatch_errors(fresh_db):
+    # Previously the extra other-column was silently discarded, creating
+    # a single-column foreign key to just ("id")
+    fresh_db["departments"].insert(
+        {"campus": "north", "code": "cs"}, pk=("campus", "code")
+    )
+    fresh_db["courses"].insert({"id": 1, "campus": "north"}, pk="id")
+    with pytest.raises(ValueError) as ex:
+        fresh_db.add_foreign_keys(
+            [("courses", ("campus",), "departments", ("campus", "code"))]
+        )
+    assert "same number of columns" in str(ex.value)
+    assert fresh_db["courses"].foreign_keys == []

tests/test_fts.py

@@ -83,6 +83,20 @@ def test_enable_fts_escape_table_names(fresh_db):
     assert [] == list(table.search("bar"))
 
 
+def test_search_duplicate_columns_are_deduped(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/624
+    table = fresh_db["t"]
+    table.insert_all(search_records)
+    table.enable_fts(["text", "country"], fts_version="FTS4")
+    rows = list(table.search("tanuki", columns=["text", "text"]))
+    assert rows == [
+        {
+            "text": "tanuki are running tricksters",
+            "text_2": "tanuki are running tricksters",
+        }
+    ]
+
+
 def test_search_limit_offset(fresh_db):
     table = fresh_db["t"]
     table.insert_all(search_records)
@@ -336,6 +350,24 @@ def test_rebuild_fts(fresh_db):
     assert len(rows2) == 2
 
 
+@pytest.mark.parametrize("method", ["optimize", "rebuild_fts"])
+def test_optimize_and_rebuild_fts_commit(tmpdir, method):
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    table = db["searchable"]
+    table.insert(search_records[0])
+    table.enable_fts(["text", "country"])
+    getattr(table, method)()
+    # The connection must not be left inside an open transaction,
+    # otherwise this and all subsequent writes are lost on close
+    assert not db.conn.in_transaction
+    table.insert(search_records[1])
+    db.close()
+    db2 = Database(path)
+    assert db2["searchable"].count == 2
+    db2.close()
+
+
 @pytest.mark.parametrize("invalid_table", ["does_not_exist", "not_searchable"])
 def test_rebuild_fts_invalid(fresh_db, invalid_table):
     fresh_db["not_searchable"].insert({"foo": "bar"})
@@ -420,12 +452,35 @@ def test_enable_fts_replace_does_nothing_if_args_the_same():
     assert all(q[0].startswith("select ") for q in queries)
 
 
-def test_enable_fts_error_message_on_views():
+def test_enable_fts_replace_handles_legacy_bracket_quoted_content_table():
+    db = Database(memory=True)
+    db["books"].insert(
+        {
+            "id": 1,
+            "title": "Habits of Australian Marsupials",
+            "author": "Marlee Hawkins",
+        },
+        pk="id",
+    )
+    db.executescript("""
+        CREATE VIRTUAL TABLE [books_fts] USING FTS5 (
+            [title],
+            content=[books]
+        );
+    """)
+
+    db["books"].enable_fts(["title", "author"], replace=True)
+
+    assert db["books_fts"].columns_dict.keys() == {"title", "author"}
+    assert 'content="books"' in db["books_fts"].schema
+
+
+def test_view_has_no_enable_fts():
     db = Database(memory=True)
     db.create_view("hello", "select 1 + 1")
-    with pytest.raises(NotImplementedError) as e:
-        db["hello"].enable_fts()
-        assert e.value.args[0] == "enable_fts() is supported on tables but not on views"
+    # Views deliberately do not have an enable_fts() method
+    with pytest.raises(AttributeError):
+        db["hello"].enable_fts()  # type: ignore[union-attr]
 
 
 @pytest.mark.parametrize(
@@ -677,3 +732,17 @@ def test_search_quote(fresh_db):
         list(table.search(query))
     # No exception with quote=True
     list(table.search(query, quote=True))
+
+
+def test_enable_fts_cli_on_view_errors(tmpdir):
+    db_path = str(tmpdir / "test.db")
+    db = Database(db_path)
+    db["t"].insert({"text": "hello"})
+    db.create_view("v", "select * from t")
+    db.close()
+    from click.testing import CliRunner
+    from sqlite_utils import cli as cli_module
+
+    result = CliRunner().invoke(cli_module.cli, ["enable-fts", db_path, "v", "text"])
+    assert result.exit_code == 1
+    assert result.output.strip() == "Error: Table v is actually a view"

tests/test_gis.py

@@ -6,12 +6,6 @@ from sqlite_utils.cli import cli
 from sqlite_utils.db import Database
 from sqlite_utils.utils import find_spatialite, sqlite3
 
-try:
-    import sqlean
-except ImportError:
-    sqlean = None
-
-
 pytestmark = [
     pytest.mark.skipif(
         not find_spatialite(), reason="Could not find SpatiaLite extension"
@@ -20,9 +14,6 @@ pytestmark = [
         not hasattr(sqlite3.Connection, "enable_load_extension"),
         reason="sqlite3.Connection missing enable_load_extension",
     ),
-    pytest.mark.skipif(
-        sqlean is not None, reason="sqlean.py is not compatible with SpatiaLite"
-    ),
 ]
 
 
@@ -50,7 +41,7 @@ def test_add_geometry_column():
         column_name="geometry",
         geometry_type="Point",
         srid=4326,
-        coord_dimension=2,
+        coord_dimension="XY",
     )
 
     assert db["geometry_columns"].get(["locations", "geometry"]) == {

tests/test_introspect.py

@@ -2,6 +2,14 @@ from sqlite_utils.db import Index, View, Database, XIndex, XIndexColumn
 import pytest
 
 
+def _check_supports_strict():
+    """Check if SQLite supports strict tables without leaking the database."""
+    db = Database(memory=True)
+    result = db.supports_strict
+    db.close()
+    return result
+
+
 def test_table_names(existing_db):
     assert ["foo"] == existing_db.table_names()
 
@@ -101,13 +109,11 @@ def test_table_repr(fresh_db):
 
 
 def test_indexes(fresh_db):
-    fresh_db.executescript(
-        """
+    fresh_db.executescript("""
         create table Gosh (c1 text, c2 text, c3 text);
         create index Gosh_c1 on Gosh(c1);
         create index Gosh_c2c3 on Gosh(c2, c3);
-    """
-    )
+    """)
     assert [
         Index(
             seq=0,
@@ -122,13 +128,11 @@ def test_indexes(fresh_db):
 
 
 def test_xindexes(fresh_db):
-    fresh_db.executescript(
-        """
+    fresh_db.executescript("""
         create table Gosh (c1 text, c2 text, c3 text);
         create index Gosh_c1 on Gosh(c1);
         create index Gosh_c2c3 on Gosh(c2, c3 desc);
-    """
-    )
+    """)
     assert fresh_db["Gosh"].xindexes == [
         XIndex(
             name="Gosh_c2c3",
@@ -282,7 +286,7 @@ def test_use_rowid(fresh_db):
 
 
 @pytest.mark.skipif(
-    not Database(memory=True).supports_strict,
+    not _check_supports_strict(),
     reason="Needs SQLite version that supports strict",
 )
 @pytest.mark.parametrize(
@@ -317,3 +321,18 @@ def test_table_default_values(fresh_db, value):
     )
     default_values = fresh_db["default_values"].default_values
     assert default_values == {"value": value}
+
+
+def test_pks_use_primary_key_declaration_order(fresh_db):
+    # PRIMARY KEY (a, b) declared against columns stored in order (b, a) -
+    # pks must follow the declaration order, which is what SQLite uses to
+    # resolve implicit foreign key references and compound pk lookups
+    fresh_db.execute("create table t (b text, a text, primary key (a, b))")
+    assert fresh_db["t"].pks == ["a", "b"]
+
+
+def test_transform_preserves_compound_pk_declaration_order(fresh_db):
+    fresh_db.execute("create table t (a text, b text, c text, primary key (b, a))")
+    fresh_db["t"].transform(drop={"c"})
+    assert fresh_db["t"].pks == ["b", "a"]
+    assert 'PRIMARY KEY ("b", "a")' in fresh_db["t"].schema

tests/test_lookup.py

@@ -157,3 +157,27 @@ def test_lookup_with_extra_insert_parameters(fresh_db):
 def test_lookup_new_table_strict(fresh_db, strict):
     fresh_db["species"].lookup({"name": "Palm"}, strict=strict)
     assert fresh_db["species"].strict == strict or not fresh_db.supports_strict
+
+
+def test_lookup_null_value_idempotent(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/186
+    # Repeated lookups of a null value should return the same row,
+    # not insert a duplicate row each time
+    species = fresh_db["species"]
+    first_id = species.lookup({"name": None})
+    second_id = species.lookup({"name": None})
+    assert first_id == second_id
+    assert list(species.rows) == [{"id": first_id, "name": None}]
+
+
+def test_lookup_compound_key_with_null_idempotent(fresh_db):
+    species = fresh_db["species"]
+    palm_id = species.lookup({"name": "Palm", "type": None})
+    oak_id = species.lookup({"name": "Oak", "type": "Tree"})
+    assert palm_id == species.lookup({"name": "Palm", "type": None})
+    assert oak_id == species.lookup({"name": "Oak", "type": "Tree"})
+    assert palm_id != oak_id
+    assert list(species.rows) == [
+        {"id": palm_id, "name": "Palm", "type": None},
+        {"id": oak_id, "name": "Oak", "type": "Tree"},
+    ]

tests/test_m2m.py

@@ -139,9 +139,9 @@ def test_m2m_lookup(fresh_db):
 
 def test_m2m_requires_either_records_or_lookup(fresh_db):
     people = fresh_db.table("people", pk="id").insert({"name": "Wahyu"})
-    with pytest.raises(AssertionError):
+    with pytest.raises(ValueError):
         people.m2m("tags")
-    with pytest.raises(AssertionError):
+    with pytest.raises(ValueError):
         people.m2m("tags", {"tag": "hello"}, lookup={"foo": "bar"})
 
 

tests/test_migrations.py

@@ -0,0 +1,246 @@
+import pytest
+import sqlite_utils
+from sqlite_utils import Migrations
+
+
+@pytest.fixture
+def migrations():
+    migrations = Migrations("test")
+
+    @migrations()
+    def m001(db):
+        db["dogs"].insert({"name": "Cleo"})
+
+    @migrations()
+    def m002(db):
+        db["cats"].create({"name": str})
+        db.execute("insert into dogs (name) values ('Pancakes')")
+
+    return migrations
+
+
+@pytest.fixture
+def migrations_not_ordered_alphabetically():
+    # Names order alphabetically in the wrong direction but this
+    # should still be applied correctly.
+    migrations = Migrations("test")
+
+    @migrations()
+    def m002(db):
+        db["dogs"].insert({"name": "Cleo"})
+
+    @migrations()
+    def m001(db):
+        db["cats"].create({"name": str})
+        db.execute("insert into dogs (name) values ('Pancakes')")
+
+    return migrations
+
+
+@pytest.fixture
+def migrations2():
+    migrations = Migrations("test2")
+
+    @migrations()
+    def m001(db):
+        db["dogs2"].insert({"name": "Cleo"})
+
+    return migrations
+
+
+def test_basic(migrations):
+    db = sqlite_utils.Database(memory=True)
+    assert db.table_names() == []
+    migrations.apply(db)
+    assert set(db.table_names()) == {"_sqlite_migrations", "dogs", "cats"}
+
+
+def test_stop_before(migrations):
+    db = sqlite_utils.Database(memory=True)
+    assert db.table_names() == []
+    migrations.apply(db, stop_before="m002")
+    assert set(db.table_names()) == {"_sqlite_migrations", "dogs"}
+    migrations.apply(db)
+    assert set(db.table_names()) == {"_sqlite_migrations", "dogs", "cats"}
+
+
+def test_two_migration_sets(migrations, migrations2):
+    db = sqlite_utils.Database(memory=True)
+    assert db.table_names() == []
+    migrations.apply(db)
+    migrations2.apply(db)
+    assert set(db.table_names()) == {"_sqlite_migrations", "dogs", "cats", "dogs2"}
+
+
+def test_order_does_not_matter(migrations, migrations_not_ordered_alphabetically):
+    db1 = sqlite_utils.Database(memory=True)
+    db2 = sqlite_utils.Database(memory=True)
+    migrations.apply(db1)
+    migrations_not_ordered_alphabetically.apply(db2)
+    assert db1.schema == db2.schema
+
+
+def test_applied_at_is_a_string(migrations):
+    db = sqlite_utils.Database(memory=True)
+    migrations.apply(db)
+    applied = migrations.applied(db)
+    assert len(applied) == 2
+    for migration in applied:
+        # applied_at is the TEXT timestamp straight from the
+        # _sqlite_migrations table, e.g. "2026-07-04 12:00:00.000000+00:00"
+        assert isinstance(migration.applied_at, str)
+        assert migration.applied_at.endswith("+00:00")
+
+
+def test_failing_migration_rolls_back(migrations):
+    @migrations()
+    def m003(db):
+        db["birds"].create({"name": str})
+        db.execute("insert into dogs (name) values ('Dozer')")
+        raise ValueError("boom")
+
+    db = sqlite_utils.Database(memory=True)
+    with pytest.raises(ValueError):
+        migrations.apply(db)
+    # m001 and m002 committed before the failure and stay applied
+    assert set(db.table_names()) == {"_sqlite_migrations", "dogs", "cats"}
+    assert [r["name"] for r in db["dogs"].rows] == ["Cleo", "Pancakes"]
+    assert [m.name for m in migrations.applied(db)] == ["m001", "m002"]
+    # Everything m003 did was rolled back and it is still pending
+    assert [m.name for m in migrations.pending(db)] == ["m003"]
+
+
+def test_rerun_after_failure_applies_each_migration_once():
+    state = {"fail": True}
+    migrations = Migrations("test")
+
+    @migrations()
+    def m001(db):
+        db["dogs"].insert({"name": "Cleo"})
+
+    @migrations()
+    def m002(db):
+        db["dogs"].insert({"name": "Pancakes"})
+        if state["fail"]:
+            raise ValueError("boom")
+
+    db = sqlite_utils.Database(memory=True)
+    with pytest.raises(ValueError):
+        migrations.apply(db)
+    state["fail"] = False
+    migrations.apply(db)
+    # m001 must not have been re-applied, m002 applied exactly once
+    assert [r["name"] for r in db["dogs"].rows] == ["Cleo", "Pancakes"]
+    assert [m.name for m in migrations.applied(db)] == ["m001", "m002"]
+
+
+def test_non_transactional_migration_allows_vacuum(tmpdir):
+    path = str(tmpdir / "test.db")
+    db = sqlite_utils.Database(path)
+    migrations = Migrations("test")
+
+    @migrations()
+    def m001(db):
+        db["dogs"].insert({"name": "Cleo"})
+
+    @migrations(transactional=False)
+    def m002(db):
+        db.execute("VACUUM")
+
+    migrations.apply(db)
+    assert [m.name for m in migrations.applied(db)] == ["m001", "m002"]
+    db.close()
+
+
+def test_apply_composes_inside_outer_transaction(migrations):
+    db = sqlite_utils.Database(memory=True)
+    with pytest.raises(ZeroDivisionError):
+        with db.atomic():
+            migrations.apply(db)
+            raise ZeroDivisionError
+    # The outer transaction rolled back, taking the migrations with it
+    assert db.table_names() == []
+
+
+@pytest.mark.parametrize(
+    "create_table,pk",
+    (
+        (
+            {
+                "migration_set": str,
+                "name": str,
+                "applied_at": str,
+            },
+            "name",
+        ),
+        (
+            {
+                "migration_set": str,
+                "name": str,
+                "applied_at": str,
+            },
+            ("migration_set", "name"),
+        ),
+    ),
+)
+def test_upgrades_sqlite_migrations(migrations, create_table, pk):
+    db = sqlite_utils.Database(memory=True)
+    db["_sqlite_migrations"].create(create_table, pk=pk)
+    assert db.table_names() == ["_sqlite_migrations"]
+    assert db["_sqlite_migrations"].pks == ([pk] if isinstance(pk, str) else list(pk))
+    migrations.apply(db)
+    assert db["_sqlite_migrations"].pks == ["id"]
+
+
+def test_pending_and_applied_are_read_only(migrations):
+    db = sqlite_utils.Database(memory=True)
+    assert [m.name for m in migrations.pending(db)] == ["m001", "m002"]
+    assert migrations.applied(db) == []
+    # Neither call should have created the tracking table
+    assert db.table_names() == []
+
+
+def test_duplicate_migration_name_errors():
+    migrations = Migrations("test")
+
+    @migrations()
+    def m001(db):
+        pass
+
+    with pytest.raises(ValueError) as ex:
+
+        @migrations(name="m001")
+        def m001_again(db):
+            pass
+
+    assert "m001" in str(ex.value)
+
+
+def test_stop_before_applied_migration_errors(migrations):
+    # Stopping before a migration that has already been applied is
+    # impossible to honor - previously the stop name was only checked
+    # against pending migrations, so everything after it was applied
+    db = sqlite_utils.Database(memory=True)
+    migrations.apply(db, stop_before="m002")  # applies m001 only
+    with pytest.raises(ValueError) as ex:
+        migrations.apply(db, stop_before="m001")
+    assert "m001" in str(ex.value)
+    assert "already been applied" in str(ex.value)
+    # Nothing else was applied
+    assert not db["cats"].exists()
+
+
+def test_stop_before_applied_migration_errors_before_any_apply(migrations):
+    # The error fires before any pending migration runs, even those that
+    # come before the already-applied stop target in registration order
+    db = sqlite_utils.Database(memory=True)
+    only_second = Migrations("test")
+
+    @only_second()
+    def m002(db):
+        db["cats"].create({"name": str})
+
+    only_second.apply(db)  # m002 applied, m001 still pending
+    with pytest.raises(ValueError):
+        migrations.apply(db, stop_before="m002")
+    assert not db["dogs"].exists()

tests/test_plugins.py

@@ -2,6 +2,7 @@ from click.testing import CliRunner
 import click
 import importlib
 import pytest
+import sys
 from sqlite_utils import cli, Database, hookimpl, plugins
 
 
@@ -9,9 +10,41 @@ def _supports_pragma_function_list():
     db = Database(memory=True)
     try:
         db.execute("select * from pragma_function_list()")
+        return True
     except Exception:
         return False
-    return True
+    finally:
+        db.close()
+
+
+def test_get_plugins_loads_setuptools_entrypoints_once(monkeypatch):
+    calls = []
+    monkeypatch.delattr(sys, "_called_from_test", raising=False)
+    monkeypatch.setattr(plugins, "_plugins_loaded", False)
+    monkeypatch.setattr(
+        plugins.pm,
+        "load_setuptools_entrypoints",
+        lambda group: calls.append(group) or 0,
+    )
+
+    plugins.get_plugins()
+    plugins.get_plugins()
+
+    assert calls == ["sqlite_utils"]
+
+
+def test_get_plugins_does_not_load_setuptools_entrypoints_in_tests(monkeypatch):
+    calls = []
+    monkeypatch.setattr(sys, "_called_from_test", True, raising=False)
+    monkeypatch.setattr(plugins, "_plugins_loaded", False)
+    monkeypatch.setattr(
+        plugins.pm,
+        "load_setuptools_entrypoints",
+        lambda group: calls.append(group) or 0,
+    )
+
+    assert plugins.get_plugins() == []
+    assert calls == []
 
 
 def test_register_commands():

tests/test_query.py

@@ -1,5 +1,8 @@
+import pytest
 import types
 
+from sqlite_utils.utils import sqlite3
+
 
 def test_query(fresh_db):
     fresh_db["dogs"].insert_all([{"name": "Cleo"}, {"name": "Pancakes"}])
@@ -8,6 +11,268 @@ def test_query(fresh_db):
     assert list(results) == [{"name": "Pancakes"}, {"name": "Cleo"}]
 
 
+def test_query_executes_eagerly(fresh_db):
+    # The SQL runs when query() is called, not when the result is iterated,
+    # so errors are raised at the call site
+    with pytest.raises(sqlite3.OperationalError):
+        fresh_db.query("select * from missing_table")
+
+
+def test_query_rejects_statements_that_return_no_rows(fresh_db):
+    fresh_db["dogs"].insert({"name": "Cleo"})
+    with pytest.raises(ValueError) as ex:
+        fresh_db.query("update dogs set name = 'Cleopaws'")
+    assert "execute()" in str(ex.value)
+    # The rejected update was rolled back, and no transaction is left open
+    assert not fresh_db.conn.in_transaction
+    assert [row["name"] for row in fresh_db["dogs"].rows] == ["Cleo"]
+
+
+def test_query_rejected_ddl_is_rolled_back(fresh_db):
+    with pytest.raises(ValueError):
+        fresh_db.query("create table dogs (id integer primary key)")
+    assert not fresh_db.conn.in_transaction
+    assert fresh_db.table_names() == []
+
+
+def test_query_rejected_write_inside_transaction_is_rolled_back(fresh_db):
+    fresh_db["dogs"].insert({"name": "Cleo"})
+    fresh_db.begin()
+    fresh_db.execute("insert into dogs (name) values ('Pancakes')")
+    with pytest.raises(ValueError):
+        fresh_db.query("update dogs set name = 'Cleopaws'")
+    # The transaction is still open and the earlier insert is intact
+    assert fresh_db.conn.in_transaction
+    fresh_db.commit()
+    assert [row["name"] for row in fresh_db["dogs"].rows] == ["Cleo", "Pancakes"]
+
+
+@pytest.mark.parametrize(
+    "sql",
+    [
+        "begin",
+        "commit",
+        "rollback",
+        "vacuum",
+        "detach database foo",
+        "/* comment */ commit",
+        "-- comment\nbegin",
+        "/* multi\nline */ -- and another\n  vacuum",
+        "\t /* a */ /* b */ savepoint s1",
+        "; commit",
+        ";;\n ; rollback",
+        "; /* comment */ vacuum",
+        "\ufeffbegin",
+    ],
+)
+def test_query_rejects_transaction_control_and_vacuum(fresh_db, sql):
+    with pytest.raises(ValueError) as ex:
+        fresh_db.query(sql)
+    assert "execute()" in str(ex.value)
+    assert not fresh_db.conn.in_transaction
+
+
+def test_query_comment_prefixed_commit_does_not_commit_transaction(fresh_db):
+    # A COMMIT hidden behind a leading comment must not slip past the
+    # keyword check - previously it committed the caller's open
+    # transaction before the ValueError was raised
+    fresh_db["dogs"].insert({"name": "Cleo"})
+    fresh_db.begin()
+    fresh_db.execute("insert into dogs (name) values ('Pancakes')")
+    with pytest.raises(ValueError):
+        fresh_db.query("/* comment */ COMMIT")
+    # The explicit transaction is still open and can still be rolled back
+    assert fresh_db.conn.in_transaction
+    fresh_db.rollback()
+    assert [row["name"] for row in fresh_db["dogs"].rows] == ["Cleo"]
+
+
+@pytest.mark.parametrize("sql", ["; COMMIT", "\ufeffCOMMIT"])
+def test_query_prefixed_commit_does_not_commit_transaction(fresh_db, sql):
+    # sqlite3 tolerates empty statements and a UTF-8 BOM before the first
+    # real token, so the keyword scanner must skip them too - previously
+    # '; COMMIT' slipped past the check and committed the caller's open
+    # transaction before raising OperationalError
+    fresh_db["dogs"].insert({"name": "Cleo"})
+    fresh_db.begin()
+    fresh_db.execute("insert into dogs (name) values ('Pancakes')")
+    with pytest.raises(ValueError):
+        fresh_db.query(sql)
+    # The explicit transaction is still open and can still be rolled back
+    assert fresh_db.conn.in_transaction
+    fresh_db.rollback()
+    assert [row["name"] for row in fresh_db["dogs"].rows] == ["Cleo"]
+
+
+def test_query_error_leaves_no_transaction_open(fresh_db):
+    with pytest.raises(sqlite3.OperationalError):
+        fresh_db.query("select * from missing_table")
+    assert not fresh_db.conn.in_transaction
+
+
+def test_query_pragma(tmpdir):
+    from sqlite_utils import Database
+
+    db = Database(str(tmpdir / "test.db"))
+    # A row-returning PRAGMA works, including one that cannot run in a transaction
+    assert list(db.query("pragma journal_mode = wal")) == [{"journal_mode": "wal"}]
+    # A PRAGMA that returns no rows raises ValueError
+    with pytest.raises(ValueError):
+        db.query("pragma user_version = 5")
+    db.close()
+
+
+def test_query_rejected_pragma_still_takes_effect(fresh_db):
+    # Documented limitation: PRAGMAs run outside the savepoint guard,
+    # because some of them refuse to run inside a transaction - so a
+    # row-less PRAGMA takes effect even though it raises ValueError.
+    # If this test starts failing because the pragma was rolled back,
+    # the limitation has been fixed - update the docs in python-api.rst
+    # and the query() docstring to remove the carve-out
+    with pytest.raises(ValueError):
+        fresh_db.query("pragma user_version = 5")
+    assert fresh_db.execute("pragma user_version").fetchone()[0] == 5
+
+
+def test_query_comment_prefixed_pragma(tmpdir):
+    from sqlite_utils import Database
+
+    db = Database(str(tmpdir / "test.db"))
+    # A leading comment must not stop a PRAGMA being recognized as one -
+    # previously it was executed inside the savepoint guard, where
+    # journal mode changes are refused
+    assert list(db.query("-- set WAL mode\npragma journal_mode = wal")) == [
+        {"journal_mode": "wal"}
+    ]
+    db.close()
+
+
+def test_query_comment_prefixed_pragma_inside_transaction(fresh_db):
+    fresh_db.begin()
+    assert list(fresh_db.query("-- check version\npragma user_version")) == [
+        {"user_version": 0}
+    ]
+    assert fresh_db.conn.in_transaction
+    fresh_db.rollback()
+
+
+@pytest.mark.parametrize(
+    "sql,expected",
+    [
+        ("select 1", "SELECT"),
+        ("  \t\n select 1", "SELECT"),
+        ("-- comment\nbegin", "BEGIN"),
+        ("/* one */ /* two */ pragma user_version", "PRAGMA"),
+        ("/* multi\nline */vacuum", "VACUUM"),
+        ("insert into t values (1)", "INSERT"),
+        ("-- only a comment", ""),
+        ("/* unterminated", ""),
+        ("", ""),
+        ("   ", ""),
+        ("123", ""),
+        ("; commit", "COMMIT"),
+        (";;\n ; rollback", "ROLLBACK"),
+        ("; -- comment\n begin", "BEGIN"),
+        ("\ufeffcommit", "COMMIT"),
+        ("\ufeff ; select 1", "SELECT"),
+        (";", ""),
+    ],
+)
+def test_first_keyword(sql, expected):
+    from sqlite_utils.db import _first_keyword
+
+    assert _first_keyword(sql) == expected
+
+
+@pytest.mark.skipif(
+    sqlite3.sqlite_version_info < (3, 35, 0),
+    reason="RETURNING requires SQLite 3.35.0 or higher",
+)
+def test_query_insert_returning(fresh_db):
+    fresh_db["dogs"].insert({"name": "Cleo"})
+    rows = list(
+        fresh_db.query("insert into dogs (name) values ('Pancakes') returning name")
+    )
+    assert rows == [{"name": "Pancakes"}]
+    assert fresh_db["dogs"].count == 2
+
+
+@pytest.mark.skipif(
+    sqlite3.sqlite_version_info < (3, 35, 0),
+    reason="RETURNING requires SQLite 3.35.0 or higher",
+)
+def test_query_insert_returning_commits_without_iteration(tmpdir):
+    from sqlite_utils import Database
+
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    db["dogs"].insert({"name": "Cleo"})
+    # Never iterate over the results
+    db.query("insert into dogs (name) values ('Pancakes') returning name")
+    assert not db.conn.in_transaction
+    # A completely separate connection sees the new row straight away
+    other = sqlite3.connect(path)
+    assert other.execute("select count(*) from dogs").fetchone()[0] == 2
+    other.close()
+    db.close()
+
+
+@pytest.mark.skipif(
+    sqlite3.sqlite_version_info < (3, 35, 0),
+    reason="RETURNING requires SQLite 3.35.0 or higher",
+)
+def test_query_insert_returning_partial_iteration_still_commits(tmpdir):
+    from sqlite_utils import Database
+
+    path = str(tmpdir / "test.db")
+    db = Database(path)
+    db["dogs"].insert({"name": "Cleo"})
+    row = next(
+        db.query(
+            "insert into dogs (name) values ('Pancakes'), ('Marnie') returning name"
+        )
+    )
+    assert row == {"name": "Pancakes"}
+    assert not db.conn.in_transaction
+    other = sqlite3.connect(path)
+    assert other.execute("select count(*) from dogs").fetchone()[0] == 3
+    other.close()
+    db.close()
+
+
+@pytest.mark.skipif(
+    sqlite3.sqlite_version_info < (3, 35, 0),
+    reason="RETURNING requires SQLite 3.35.0 or higher",
+)
+def test_query_insert_returning_respects_explicit_transaction(fresh_db):
+    fresh_db["dogs"].insert({"name": "Cleo"})
+    fresh_db.begin()
+    rows = list(
+        fresh_db.query("insert into dogs (name) values ('Pancakes') returning name")
+    )
+    assert rows == [{"name": "Pancakes"}]
+    # Still inside the explicit transaction - not committed
+    assert fresh_db.conn.in_transaction
+    fresh_db.rollback()
+    assert [row["name"] for row in fresh_db["dogs"].rows] == ["Cleo"]
+
+
+def test_query_duplicate_column_names_are_deduped(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/624
+    fresh_db["one"].insert({"id": 1, "value": "left"})
+    fresh_db["two"].insert({"id": 2, "value": "right"})
+    rows = list(
+        fresh_db.query("select one.id, two.id, one.value, two.value from one, two")
+    )
+    assert rows == [{"id": 1, "id_2": 2, "value": "left", "value_2": "right"}]
+
+
+def test_query_deduped_column_avoids_existing_names(fresh_db):
+    # The renamed duplicate must not overwrite a real column called id_2
+    rows = list(fresh_db.query("select 1 as id, 2 as id, 3 as id_2"))
+    assert rows == [{"id": 1, "id_3": 2, "id_2": 3}]
+
+
 def test_execute_returning_dicts(fresh_db):
     # Like db.query() but returns a list, included for backwards compatibility
     # see https://github.com/simonw/sqlite-utils/issues/290
@@ -15,3 +280,24 @@ def test_execute_returning_dicts(fresh_db):
     assert fresh_db.execute_returning_dicts("select * from test") == [
         {"id": 1, "bar": 2}
     ]
+
+
+@pytest.mark.skipif(
+    sqlite3.sqlite_version_info < (3, 35, 0),
+    reason="RETURNING requires SQLite 3.35.0 or higher",
+)
+def test_query_preserves_error_from_transaction_destroying_trigger(fresh_db):
+    # RAISE(ROLLBACK) destroys the savepoint guard - the original
+    # IntegrityError must propagate, not "no such savepoint"
+    fresh_db.execute("create table t (id integer primary key, v text)")
+    fresh_db.execute("""
+        create trigger no_bad before insert on t
+        when new.v = 'bad'
+        begin
+            select raise(rollback, 'trigger says no');
+        end
+    """)
+    with pytest.raises(sqlite3.IntegrityError, match="trigger says no"):
+        fresh_db.query("insert into t (id, v) values (1, 'bad') returning id")
+    assert not fresh_db.conn.in_transaction
+    assert fresh_db.execute("select count(*) from t").fetchone()[0] == 0

tests/test_recreate.py

@@ -14,8 +14,11 @@ def test_recreate_ignored_for_in_memory():
 
 def test_recreate_not_allowed_for_connection():
     conn = sqlite3.connect(":memory:")
-    with pytest.raises(AssertionError):
-        Database(conn, recreate=True)
+    try:
+        with pytest.raises(ValueError):
+            Database(conn, recreate=True)
+    finally:
+        conn.close()
 
 
 @pytest.mark.parametrize(

tests/test_register_function.py

@@ -42,39 +42,46 @@ def test_register_function_deterministic(fresh_db):
 
 
 def test_register_function_deterministic_tries_again_if_exception_raised(fresh_db):
+    # Save the original connection so we can close it later
+    original_conn = fresh_db.conn
     fresh_db.conn = MagicMock()
     fresh_db.conn.create_function = MagicMock()
 
-    @fresh_db.register_function(deterministic=True)
-    def to_lower_2(s):
-        return s.lower()
+    try:
 
-    fresh_db.conn.create_function.assert_called_with(
-        "to_lower_2", 1, to_lower_2, deterministic=True
-    )
+        @fresh_db.register_function(deterministic=True)
+        def to_lower_2(s):
+            return s.lower()
 
-    first = True
+        fresh_db.conn.create_function.assert_called_with(
+            "to_lower_2", 1, to_lower_2, deterministic=True
+        )
 
-    def side_effect(*args, **kwargs):
-        # Raise exception only first time this is called
-        nonlocal first
-        if first:
-            first = False
-            raise sqlite3.NotSupportedError()
+        first = True
 
-    # But if sqlite3.NotSupportedError is raised, it tries again
-    fresh_db.conn.create_function.reset_mock()
-    fresh_db.conn.create_function.side_effect = side_effect
+        def side_effect(*args, **kwargs):
+            # Raise exception only first time this is called
+            nonlocal first
+            if first:
+                first = False
+                raise sqlite3.NotSupportedError()
 
-    @fresh_db.register_function(deterministic=True)
-    def to_lower_3(s):
-        return s.lower()
+        # But if sqlite3.NotSupportedError is raised, it tries again
+        fresh_db.conn.create_function.reset_mock()
+        fresh_db.conn.create_function.side_effect = side_effect
 
-    # Should have been called once with deterministic=True and once without
-    assert fresh_db.conn.create_function.call_args_list == [
-        call("to_lower_3", 1, to_lower_3, deterministic=True),
-        call("to_lower_3", 1, to_lower_3),
-    ]
+        @fresh_db.register_function(deterministic=True)
+        def to_lower_3(s):
+            return s.lower()
+
+        # Should have been called once with deterministic=True and once without
+        assert fresh_db.conn.create_function.call_args_list == [
+            call("to_lower_3", 1, to_lower_3, deterministic=True),
+            call("to_lower_3", 1, to_lower_3),
+        ]
+    finally:
+        # Close the original connection that was replaced with the mock
+        original_conn.close()
 
 
 def test_register_function_replace(fresh_db):

tests/test_rows.py

@@ -104,3 +104,37 @@ def test_pks_and_rows_where_compound_pk(fresh_db):
         (("number", 1), {"type": "number", "number": 1, "plusone": 2}),
         (("number", 2), {"type": "number", "number": 2, "plusone": 3}),
     ]
+
+
+def test_rows_where_duplicate_select_columns_are_deduped(fresh_db):
+    # https://github.com/simonw/sqlite-utils/issues/624
+    fresh_db["t"].insert({"id": 1, "name": "Cleo"})
+    rows = list(fresh_db["t"].rows_where(select="id, id, name"))
+    assert rows == [{"id": 1, "id_2": 1, "name": "Cleo"}]
+
+
+def test_pks_and_rows_where_view(fresh_db):
+    # pks_and_rows_where() lives on Queryable so views expose it, but
+    # SQLite views have no rowid. Modern SQLite (3.36+) raises an
+    # OperationalError from the generated SQL; older versions returned
+    # NULL for a view's rowid. Either way it must not fail earlier with
+    # an AttributeError from View lacking Table-only properties
+    from sqlite_utils.utils import sqlite3
+
+    fresh_db["dogs"].insert({"id": 1, "name": "Cleo"}, pk="id")
+    fresh_db.create_view("dog_names", "select name from dogs")
+    try:
+        result = list(fresh_db["dog_names"].pks_and_rows_where())
+    except sqlite3.OperationalError:
+        pass  # SQLite 3.36+: no such column: rowid
+    else:
+        # Older SQLite returns NULL rowids for views
+        assert result == [(None, {"rowid": None, "name": "Cleo"})]
+
+
+def test_pks_and_rows_where_compound_pk_declaration_order(fresh_db):
+    # Compound pks are returned in PRIMARY KEY declaration order
+    fresh_db.execute("create table t (b text, a text, primary key (a, b))")
+    fresh_db["t"].insert({"a": "A", "b": "B"})
+    pks_and_rows = list(fresh_db["t"].pks_and_rows_where())
+    assert pks_and_rows == [(("A", "B"), {"b": "B", "a": "A"})]

tests/test_rows_from_file.py

@@ -48,7 +48,7 @@ def test_rows_from_file_extra_fields_strategies(ignore_extras, extras_key, expec
 
 def test_rows_from_file_error_on_string_io():
     with pytest.raises(TypeError) as ex:
-        rows_from_file(StringIO("id,name\r\n1,Cleo"))
+        rows_from_file(StringIO("id,name\r\n1,Cleo"))  # type: ignore[arg-type]
     assert ex.value.args == (
         "rows_from_file() requires a file-like object that supports peek(), such as io.BytesIO",
     )

tests/test_sniff.py

@@ -6,7 +6,7 @@ import pytest
 sniff_dir = pathlib.Path(__file__).parent / "sniff"
 
 
-@pytest.mark.parametrize("filepath", sniff_dir.glob("example*"))
+@pytest.mark.parametrize("filepath", sorted(sniff_dir.glob("example*")))
 def test_sniff(tmpdir, filepath):
     db_path = str(tmpdir / "test.db")
     runner = CliRunner()

tests/test_tracer.py

@@ -50,7 +50,7 @@ def test_with_tracer():
     with db.tracer(tracer):
         list(dogs.search("Cleopaws"))
 
-    assert len(collected) == 5
+    assert len(collected) == 4
     assert collected == [
         (
             "SELECT name FROM sqlite_master\n"
@@ -64,13 +64,12 @@ def test_with_tracer():
             "        )\n"
             "    )",
             {
-                "like": '%VIRTUAL TABLE%USING FTS%content="dogs"%',
+                "like": "%VIRTUAL TABLE%USING FTS%content=[dogs]%",
                 "like2": '%VIRTUAL TABLE%USING FTS%content="dogs"%',
                 "table": "dogs",
             },
         ),
         ("select name from sqlite_master where type = 'view'", None),
-        ("select name from sqlite_master where type = 'view'", None),
         ("select sql from sqlite_master where name = ?", ("dogs_fts",)),
         (
             'with "original" as (\n'
@@ -94,4 +93,4 @@ def test_with_tracer():
 
     # Outside the with block collected should not be appended to
     dogs.insert({"name": "Cleopaws"})
-    assert len(collected) == 5
+    assert len(collected) == 4

tests/test_transform.py

@@ -1,4 +1,6 @@
-from sqlite_utils.db import ForeignKey, TransformError
+import sqlite3
+
+from sqlite_utils.db import ForeignKey, TransactionError, TransformError
 from sqlite_utils.utils import OperationalError
 import pytest
 
@@ -224,6 +226,40 @@ def test_transform_rename_pk(fresh_db):
     )
 
 
+def test_transform_preserves_keyword_literal_defaults(fresh_db):
+    # transform() used to requote keyword-literal defaults (DEFAULT TRUE became
+    # DEFAULT 'TRUE'), so a default insert stored the text 'TRUE' instead of the
+    # integer 1 -- silent value corruption on every rebuilt table.
+    fresh_db.execute(
+        "CREATE TABLE t ("
+        " id INTEGER PRIMARY KEY,"
+        " is_active INTEGER DEFAULT TRUE,"
+        " flag INTEGER DEFAULT FALSE,"
+        " note TEXT DEFAULT NULL"
+        ")"
+    )
+    table = fresh_db["t"]
+    table.insert({"id": 1})
+    before = fresh_db.execute("SELECT is_active, flag, note FROM t").fetchone()
+    assert before == (1, 0, None)
+
+    # Rebuild the table via an unrelated change.
+    table.transform(rename={"note": "note2"})
+
+    # The keyword literals stay unquoted in the schema ...
+    assert "DEFAULT TRUE" in table.schema
+    assert "DEFAULT FALSE" in table.schema
+    assert "DEFAULT NULL" in table.schema
+    assert "'TRUE'" not in table.schema
+
+    # ... and a fresh default insert still yields 1 / 0 / NULL, not strings.
+    table.insert({"id": 2})
+    after = fresh_db.execute(
+        "SELECT is_active, flag, note2 FROM t WHERE id = 2"
+    ).fetchone()
+    assert after == (1, 0, None)
+
+
 def test_transform_not_null(fresh_db):
     dogs = fresh_db["dogs"]
     dogs.insert({"id": 1, "name": "Cleo", "age": "5"}, pk="id")
@@ -396,6 +432,165 @@ def test_transform_verify_foreign_keys(fresh_db):
     assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
 
 
+@pytest.mark.parametrize("use_pragma_foreign_keys", [False, True])
+def test_transform_on_delete_cascade_does_not_delete_records(
+    fresh_db, use_pragma_foreign_keys
+):
+    # Transforming a table drops and recreates it - if another table references
+    # it with ON DELETE CASCADE and PRAGMA foreign_keys is on, that drop must
+    # not cascade and delete the referencing records
+    if use_pragma_foreign_keys:
+        fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db.executescript("""
+        CREATE TABLE authors (id INTEGER PRIMARY KEY, name TEXT);
+        CREATE TABLE books (
+            id INTEGER PRIMARY KEY,
+            title TEXT,
+            author_id INTEGER REFERENCES authors(id) ON DELETE CASCADE
+        );
+        """)
+    fresh_db["authors"].insert({"id": 1, "name": "Ursula K. Le Guin"})
+    fresh_db["books"].insert({"id": 1, "title": "The Dispossessed", "author_id": 1})
+    # Transform the table on the other end of the cascading foreign key
+    fresh_db["authors"].transform(rename={"name": "author_name"})
+    assert list(fresh_db["authors"].rows) == [
+        {"id": 1, "author_name": "Ursula K. Le Guin"}
+    ]
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "title": "The Dispossessed", "author_id": 1}
+    ]
+    # Transforming the table with the cascading foreign key should not
+    # delete its records either
+    fresh_db["books"].transform(rename={"title": "book_title"})
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "book_title": "The Dispossessed", "author_id": 1}
+    ]
+    if use_pragma_foreign_keys:
+        assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
+
+
+@pytest.mark.parametrize("on_delete", ["CASCADE", "SET NULL", "SET DEFAULT", "cascade"])
+def test_transform_in_transaction_refuses_destructive_on_delete(fresh_db, on_delete):
+    # PRAGMA foreign_keys is a no-op inside a transaction, so transforming a
+    # table referenced by ON DELETE CASCADE / SET NULL / SET DEFAULT foreign
+    # keys inside an open transaction would fire those actions when the old
+    # table is dropped - transform() should refuse instead
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db.executescript("""
+        CREATE TABLE authors (id INTEGER PRIMARY KEY, name TEXT);
+        CREATE TABLE books (
+            id INTEGER PRIMARY KEY,
+            title TEXT,
+            author_id INTEGER REFERENCES authors(id) ON DELETE {}
+        );
+        """.format(on_delete))
+    fresh_db["authors"].insert({"id": 1, "name": "Ursula K. Le Guin"})
+    fresh_db["books"].insert({"id": 1, "title": "The Dispossessed", "author_id": 1})
+    previous_schema = fresh_db["authors"].schema
+    with fresh_db.atomic():
+        with pytest.raises(TransactionError) as excinfo:
+            fresh_db["authors"].transform(rename={"name": "author_name"})
+    message = str(excinfo.value)
+    assert "books" in message
+    assert "ON DELETE {}".format(on_delete.upper()) in message
+    # Nothing should have changed
+    assert fresh_db["authors"].schema == previous_schema
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "title": "The Dispossessed", "author_id": 1}
+    ]
+    assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
+
+
+def test_transform_in_transaction_refuses_self_referential_cascade(fresh_db):
+    # The copied table carries a foreign key referencing the original table
+    # name, so a self-referential cascade would wipe the copy too
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db.executescript("""
+        CREATE TABLE categories (
+            id INTEGER PRIMARY KEY,
+            name TEXT,
+            parent_id INTEGER REFERENCES categories(id) ON DELETE CASCADE
+        );
+        """)
+    fresh_db["categories"].insert_all(
+        [
+            {"id": 1, "name": "Fiction", "parent_id": None},
+            {"id": 2, "name": "Science Fiction", "parent_id": 1},
+        ]
+    )
+    with fresh_db.atomic():
+        with pytest.raises(TransactionError) as excinfo:
+            fresh_db["categories"].transform(rename={"name": "title"})
+    assert "categories" in str(excinfo.value)
+    assert fresh_db["categories"].count == 2
+
+
+def test_transform_in_transaction_allowed_with_no_action_foreign_key(fresh_db):
+    # An inbound foreign key without a destructive ON DELETE action is safe
+    # inside a transaction thanks to PRAGMA defer_foreign_keys
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db.executescript("""
+        CREATE TABLE authors (id INTEGER PRIMARY KEY, name TEXT);
+        CREATE TABLE books (
+            id INTEGER PRIMARY KEY,
+            title TEXT,
+            author_id INTEGER REFERENCES authors(id)
+        );
+        """)
+    fresh_db["authors"].insert({"id": 1, "name": "Ursula K. Le Guin"})
+    fresh_db["books"].insert({"id": 1, "title": "The Dispossessed", "author_id": 1})
+    with fresh_db.atomic():
+        fresh_db["authors"].transform(rename={"name": "author_name"})
+    assert list(fresh_db["authors"].rows) == [
+        {"id": 1, "author_name": "Ursula K. Le Guin"}
+    ]
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "title": "The Dispossessed", "author_id": 1}
+    ]
+    assert fresh_db.conn.execute("PRAGMA foreign_keys").fetchone()[0]
+
+
+def test_transform_in_transaction_allowed_for_child_table(fresh_db):
+    # The table being transformed only has an outbound foreign key - dropping
+    # it fires no ON DELETE actions, so this is allowed inside a transaction
+    fresh_db.conn.execute("PRAGMA foreign_keys=ON")
+    fresh_db.executescript("""
+        CREATE TABLE authors (id INTEGER PRIMARY KEY, name TEXT);
+        CREATE TABLE books (
+            id INTEGER PRIMARY KEY,
+            title TEXT,
+            author_id INTEGER REFERENCES authors(id) ON DELETE CASCADE
+        );
+        """)
+    fresh_db["authors"].insert({"id": 1, "name": "Ursula K. Le Guin"})
+    fresh_db["books"].insert({"id": 1, "title": "The Dispossessed", "author_id": 1})
+    with fresh_db.atomic():
+        fresh_db["books"].transform(rename={"title": "book_title"})
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "book_title": "The Dispossessed", "author_id": 1}
+    ]
+
+
+def test_transform_in_transaction_allowed_with_foreign_keys_off(fresh_db):
+    # With PRAGMA foreign_keys off (the default) no cascades can fire, so
+    # transform inside a transaction is safe even with a CASCADE schema
+    fresh_db.executescript("""
+        CREATE TABLE authors (id INTEGER PRIMARY KEY, name TEXT);
+        CREATE TABLE books (
+            id INTEGER PRIMARY KEY,
+            title TEXT,
+            author_id INTEGER REFERENCES authors(id) ON DELETE CASCADE
+        );
+        """)
+    fresh_db["authors"].insert({"id": 1, "name": "Ursula K. Le Guin"})
+    fresh_db["books"].insert({"id": 1, "title": "The Dispossessed", "author_id": 1})
+    with fresh_db.atomic():
+        fresh_db["authors"].transform(rename={"name": "author_name"})
+    assert list(fresh_db["books"].rows) == [
+        {"id": 1, "title": "The Dispossessed", "author_id": 1}
+    ]
+
+
 def test_transform_add_foreign_keys_from_scratch(fresh_db):
     _add_country_city_continent(fresh_db)
     fresh_db["places"].insert(_CAVEAU)
@@ -532,13 +727,63 @@ def test_transform_preserves_rowids(fresh_db, table_type):
     assert previous_rows == next_rows
 
 
-@pytest.mark.parametrize("strict", (False, True))
-def test_transform_strict(fresh_db, strict):
-    dogs = fresh_db.table("dogs", strict=strict)
+@pytest.mark.parametrize(
+    "initial_strict,transform_strict,expected_strict",
+    (
+        (False, None, False),
+        (True, None, True),
+        (False, True, True),
+        (True, False, False),
+    ),
+)
+def test_transform_strict(fresh_db, initial_strict, transform_strict, expected_strict):
+    if not fresh_db.supports_strict:
+        pytest.skip("SQLite version does not support strict tables")
+    dogs = fresh_db.table("dogs", strict=initial_strict)
     dogs.insert({"id": 1, "name": "Cleo"})
-    assert dogs.strict == strict or not fresh_db.supports_strict
-    dogs.transform(not_null={"name"})
-    assert dogs.strict == strict or not fresh_db.supports_strict
+    assert dogs.strict is initial_strict
+    dogs.transform(strict=transform_strict)
+    assert dogs.strict is expected_strict
+
+
+def test_transform_to_strict_with_invalid_data(fresh_db):
+    if not fresh_db.supports_strict:
+        pytest.skip("SQLite version does not support strict tables")
+    dogs = fresh_db["dogs"]
+    dogs.create({"id": int})
+    dogs.insert({"id": "not-an-integer"})
+
+    with pytest.raises(sqlite3.IntegrityError):
+        dogs.transform(strict=True)
+
+    assert dogs.strict is False
+    assert list(dogs.rows) == [{"id": "not-an-integer"}]
+    assert fresh_db.table_names() == ["dogs"]
+
+
+def test_transform_strict_updates_default(fresh_db):
+    if not fresh_db.supports_strict:
+        pytest.skip("SQLite version does not support strict tables")
+    table = fresh_db.table("items", strict=True)
+    table.create({"id": int})
+
+    table.transform(strict=False)
+    assert table.strict is False
+
+    table.create({"id": int}, replace=True)
+    assert table.strict is False
+
+
+@pytest.mark.parametrize("method_name", ("transform", "transform_sql"))
+def test_transform_to_strict_not_supported(fresh_db, method_name):
+    table = fresh_db["items"]
+    table.create({"id": int})
+    fresh_db._supports_strict = False
+
+    with pytest.raises(TransformError, match="SQLite does not support STRICT tables"):
+        getattr(table, method_name)(strict=True)
+
+    assert table.strict is False
 
 
 @pytest.mark.parametrize(
@@ -638,15 +883,13 @@ def test_transform_with_indexes_errors(fresh_db, transform_params):
 def test_transform_with_unique_constraint_implicit_index(fresh_db):
     dogs = fresh_db["dogs"]
     # Create a table with a UNIQUE constraint on 'name', which creates an implicit index
-    fresh_db.execute(
-        """
+    fresh_db.execute("""
         CREATE TABLE dogs (
             id INTEGER PRIMARY KEY,
             name TEXT UNIQUE,
             age INTEGER
         );
-    """
-    )
+    """)
     dogs.insert({"id": 1, "name": "Cleo", "age": 5})
 
     # Attempt to transform the table without modifying 'name'

tests/test_upsert.py

@@ -49,6 +49,89 @@ def test_upsert_error_if_no_pk(fresh_db):
         table.upsert({"id": 1, "name": "Cleo"})
 
 
+@pytest.mark.parametrize("use_old_upsert", (False, True))
+def test_upsert_empty_record_errors(use_old_upsert):
+    db = Database(memory=True, use_old_upsert=use_old_upsert)
+    table = db["table"]
+    table.insert({"id": 1, "name": "Cleo"}, pk="id")
+    with pytest.raises(PrimaryKeyRequired):
+        table.upsert({}, pk="id")
+    with pytest.raises(PrimaryKeyRequired):
+        table.upsert_all([{}, {}], pk="id")
+    # No rows can have been inserted
+    assert table.count == 1
+
+
+@pytest.mark.parametrize("use_old_upsert", (False, True))
+def test_upsert_missing_pk_value_errors(use_old_upsert):
+    db = Database(memory=True, use_old_upsert=use_old_upsert)
+    table = db["table"]
+    table.insert({"id": 1, "name": "Cleo"}, pk="id")
+    # Records that omit the pk column entirely
+    with pytest.raises(PrimaryKeyRequired):
+        table.upsert_all([{"name": "Pancakes"}, {"name": "Marnie"}], pk="id")
+    # A record with an explicit None pk value can never conflict
+    with pytest.raises(PrimaryKeyRequired):
+        table.upsert({"id": None, "name": "Pancakes"}, pk="id")
+    assert list(table.rows) == [{"id": 1, "name": "Cleo"}]
+
+
+def test_upsert_missing_compound_pk_value_errors(fresh_db):
+    table = fresh_db["table"]
+    table.insert({"a": "x", "b": "y", "v": 1}, pk=("a", "b"))
+    # Missing one component of the detected compound primary key
+    with pytest.raises(PrimaryKeyRequired):
+        table.upsert({"a": "x", "v": 2})
+    assert list(table.rows) == [{"a": "x", "b": "y", "v": 1}]
+
+
+def test_upsert_error_if_existing_table_has_no_pk(fresh_db):
+    table = fresh_db.create_table("table", {"id": int, "name": str})
+    with pytest.raises(PrimaryKeyRequired):
+        table.upsert({"id": 1, "name": "Cleo"})
+
+
+@pytest.mark.parametrize("use_old_upsert", (False, True))
+def test_upsert_uses_compound_pk_from_existing_table(use_old_upsert):
+    # https://github.com/simonw/sqlite-utils/issues/629
+    db = Database(memory=True, use_old_upsert=use_old_upsert)
+    db.execute("""
+        create table summary (
+            Source text,
+            Object text,
+            Category text,
+            Count integer,
+            primary key (Source, Object, Category)
+        )
+        """)
+    table = db["summary"]
+    table.upsert(
+        {
+            "Source": "Client A",
+            "Object": "Accounts",
+            "Category": "All",
+            "Count": 3,
+        }
+    )
+    assert table.last_pk == ("Client A", "Accounts", "All")
+    table.upsert(
+        {
+            "Source": "Client A",
+            "Object": "Accounts",
+            "Category": "All",
+            "Count": 4,
+        }
+    )
+    assert list(table.rows) == [
+        {
+            "Source": "Client A",
+            "Object": "Accounts",
+            "Category": "All",
+            "Count": 4,
+        }
+    ]
+
+
 def test_upsert_with_hash_id(fresh_db):
     table = fresh_db["table"]
     table.upsert({"foo": "bar"}, hash_id="pk")

tests/test_utils.py

@@ -83,3 +83,20 @@ def test_maximize_csv_field_size_limit():
 )
 def test_flatten(input, expected):
     assert utils.flatten(input) == expected
+
+
+@pytest.mark.parametrize(
+    "input,expected",
+    (
+        ([], []),
+        (["id", "name"], ["id", "name"]),
+        (["id", "id"], ["id", "id_2"]),
+        (["id", "id", "id"], ["id", "id_2", "id_3"]),
+        # A renamed duplicate must not clobber a real column called id_2
+        (["id", "id", "id_2"], ["id", "id_3", "id_2"]),
+        (["id_2", "id", "id"], ["id_2", "id", "id_3"]),
+        (["id", "id", "id_2", "id_2"], ["id", "id_3", "id_2", "id_2_2"]),
+    ),
+)
+def test_dedupe_keys(input, expected):
+    assert utils.dedupe_keys(input) == expected

tests/test_wal.py

@@ -1,5 +1,6 @@
 import pytest
 from sqlite_utils import Database
+from sqlite_utils.db import TransactionError
 
 
 @pytest.fixture
@@ -21,3 +22,67 @@ def test_enable_disable_wal(db_path_tmpdir):
     db.disable_wal()
     assert "delete" == db.journal_mode
     assert "test.db-wal" not in [f.basename for f in tmpdir.listdir()]
+
+
+def test_enable_wal_inside_transaction_raises(db_path_tmpdir):
+    db, path, tmpdir = db_path_tmpdir
+    db["test"].insert({"id": 1}, pk="id")
+    with pytest.raises(TransactionError):
+        with db.atomic():
+            db["test"].insert({"id": 2}, pk="id")
+            db.enable_wal()
+    # The atomic() block must have rolled back cleanly and the
+    # journal mode must be unchanged
+    assert db.journal_mode == "delete"
+    assert [r["id"] for r in db["test"].rows] == [1]
+
+
+def test_disable_wal_inside_transaction_raises(db_path_tmpdir):
+    db, path, tmpdir = db_path_tmpdir
+    db.enable_wal()
+    db["test"].insert({"id": 1}, pk="id")
+    with pytest.raises(TransactionError):
+        with db.atomic():
+            db["test"].insert({"id": 2}, pk="id")
+            db.disable_wal()
+    assert db.journal_mode == "wal"
+    assert [r["id"] for r in db["test"].rows] == [1]
+
+
+def test_ensure_autocommit_on(db_path_tmpdir):
+    db, path, tmpdir = db_path_tmpdir
+    previous_isolation_level = db.conn.isolation_level
+    assert previous_isolation_level is not None
+    with db.ensure_autocommit_on():
+        # isolation_level of None means driver-level autocommit mode
+        assert db.conn.isolation_level is None
+    # Restored afterwards
+    assert db.conn.isolation_level == previous_isolation_level
+
+
+def test_enable_wal_noop_inside_transaction_is_allowed(db_path_tmpdir):
+    # Calling enable_wal() when WAL is already enabled is a no-op,
+    # so it is fine inside a transaction
+    db, path, tmpdir = db_path_tmpdir
+    db.enable_wal()
+    with db.atomic():
+        db["test"].insert({"id": 1}, pk="id")
+        db.enable_wal()
+    assert [r["id"] for r in db["test"].rows] == [1]
+
+
+def test_ensure_autocommit_on_inside_transaction_raises(db_path_tmpdir):
+    # Setting isolation_level commits any pending transaction as a side
+    # effect, silently breaking the caller's rollback guarantee - so
+    # entering autocommit mode with a transaction open is an error
+    db, path, tmpdir = db_path_tmpdir
+    db["test"].insert({"id": 1}, pk="id")
+    db.begin()
+    db.execute("insert into test (id) values (2)")
+    with pytest.raises(TransactionError):
+        with db.ensure_autocommit_on():
+            pass
+    # The transaction is still open and can still be rolled back
+    assert db.conn.in_transaction
+    db.rollback()
+    assert [r["id"] for r in db["test"].rows] == [1]