Datasette core is gaining OpenTelemetry spans alongside the existing
hand-rolled tracer. This commit only lays the groundwork - no span is
emitted yet.
Core takes a runtime dependency on opentelemetry-api and nothing more.
It deliberately never creates a TracerProvider, configures an exporter,
or touches sampling: that belongs to whoever runs Datasette, normally
via an opentelemetry-instrument agent. Owning a provider in core was
tried in an earlier design and produced a cross-request span leak, a
process-global provider that tests could not tear down, and a sampling
env var that silently blanked output. With no provider installed every
span is a NonRecordingSpan and costs approximately nothing.
datasette/telemetry.py exposes the module-level tracer plus
sql_attribute(), which truncates SQL to 2048 characters. On a public
instance the SQL is attacker-controlled and unbounded - someone can
paste a 10MB query into ?sql= - so it must never reach a telemetry
pipeline verbatim.
opentelemetry-sdk goes in the dev dependency group only, because the
test suite needs it to assert on spans while the package itself must
not import it. tests/test_telemetry.py enforces that by importing
datasette in a fresh interpreter and inspecting sys.modules, which
catches a lazy import inside a function body that a grep would miss.
conftest.py gains a session-scoped autouse fixture installing an SDK
provider with an InMemorySpanExporter. It has to be session-scoped
because set_tracer_provider() is effectively once-per-process - a
second call logs a warning and is ignored. SimpleSpanProcessor rather
than BatchSpanProcessor, so assertions made right after a request never
race a background export thread. The otel_spans fixture that later
tickets assert against is added here too.
test_datasette_package_never_imports_the_sdk is moved to the front of
the run. Late in a serial run the pytest process holds enough threads
that the fork half of subprocess' fork+exec segfaults the interpreter
on macOS/CPython 3.13. That reproduces with any subprocess call in that
position on an unmodified tree, so it is a pre-existing hazard rather
than something this commit introduces; the repo already moves its other
subprocess-spawning tests to the front for related reasons.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
* Run datasette serve startup and uvicorn on a single event loop
* Move the serve-subprocess test plumbing into a conftest fixture
* Fix datasette-litestream URL and trim marker-task test comments
* Explain why serve_with_plugins needs a subprocess and plugin files
* Apply ruff 0.16 and black fixes
* Tweaked some comments
- Use a per-process socket path for the UDS test fixture.
- Clean up stale socket files before and after the fixture runs.
- Close the HTTP client and wait for the Datasette subprocess to exit.
Session-scoped fixtures are cached per worker by pytest itself, so the
manual _ds_client module global is no longer needed.
Refs #2692
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ds_client already caches a single Datasette for the whole session via a
module-level _ds_client global, so the declared fixture scope should
match. With function scope the auto-close plugin correctly closes it
after the first test that uses it, which then breaks every subsequent
test that reuses the cached (now-closed) instance — as seen in the CI
coverage job, which runs serially rather than under pytest-xdist.
Refs #2692
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
- New CSRF protection middleware inspired by Go 1.25 and research by Filippo Valsorda - https://words.filippo.io/csrf/ - this replaces the old CSRF token based protection.
- Removes all instances of `<input type="hidden" name="csrftoken" value="{{ csrftoken() }}">` in the templates - they are no longer needed.
- Removes the `def skip_csrf(datasette, scope):` plugin hook defined in `datasette/hookspecs.py` and its documentation and tests.
- Updated CSRF protection documentation to describe the new approach.
- Upgrade guide now describes the CSRF change.
Simplified Action by moving takes_child/takes_parent logic to Resource
- Removed InstanceResource - global actions are now simply those with resource_class=None
- Resource.parent_class - Replaced parent_name: str with parent_class: type[Resource] | None for direct class references
- Simplified Action dataclass - No more redundant fields, everything is derived from the Resource class structure
- Validation - The __init_subclass__ method now checks parent_class.parent_class to enforce the 2-level hierarchy
Closes#2563
* Ported setup.py to pyproject.toml, refs #2553
* Make fixtures tests less flaky
The in-memory fixtures table was being shared between different
instances of the test client, leading to occasional errors when
running the full test suite.
This fixes issues introduced by the ruff commit e57f391a which converted
Optional[x] to x | None:
- Fixed datasette/app.py line 1024: Dict[id | str, Dict] -> Dict[int | str, Dict]
(was using id built-in function instead of int type)
- Fixed datasette/app.py line 1074: Optional["Resource"] -> "Resource" | None
- Added 'from __future__ import annotations' for Python 3.10 compatibility
- Added TYPE_CHECKING blocks to avoid circular imports
- Removed dead code (unused variable assignments) from cli.py and views
- Removed unused imports flagged by ruff across multiple files
- Fixed test fixtures: moved app_client fixture imports to conftest.py
(fixed 71 test errors caused by fixtures not being registered)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
The permission_allowed hook has been fully replaced by permission_resources_sql.
This commit removes:
- hookspec definition from hookspecs.py
- 4 implementations from default_permissions.py
- implementations from test plugins (my_plugin.py, my_plugin_2.py)
- hook monitoring infrastructure from conftest.py
- references from fixtures.py
- Also fixes test_get_permission to use ds.get_action() instead of ds.get_permission()
- Removes 5th column (source_plugin) from PermissionSQL queries
This completes the migration to the SQL-based permission system.
- Consolidated register_permissions and register_actions hooks in my_plugin.py
- Added permission_resources_sql hook to provide SQL-based permission rules
- Updated conftest.py to reference datasette.actions instead of datasette.permissions
- Updated fixtures.py to include permission_resources_sql hook and remove register_permissions
- Added backwards compatibility support for old datasette-register-permissions config
- Converted test actions (this_is_allowed, this_is_denied, etc.) to use permission_resources_sql
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
* Checkpoint, moving top-level plugin config to datasette.json
* Support database-level and table-level plugin configuration in datasette.yaml
Refs #2093
* Docs for permissions: in metadata, refs #1636
* Refactor default_permissions.py to help with implementation of #1636
* register_permissions() plugin hook, closes#1939 - also refs #1938
* Tests for register_permissions() hook, refs #1939
* Documentation for datasette.permissions, refs #1939
* permission_allowed() falls back on Permission.default, refs #1939
* Raise StartupError on duplicate permissions
* Allow dupe permisisons if exact matches
Context manager with open closes the files after usage.
When the object is already a pathlib.Path i used read_text
write_text functions
In some cases pathlib.Path.open were used in context manager,
it is basically the same as builtin open.
Thanks, Konstantin Baikov!
* Support for generated columns, closes#1116
* Show SQLite version in pytest report header
* Use table_info() if SQLite < 3.26.0
* Cache sqlite_version() rather than re-calculate every time
* Adjust test_database_page for SQLite 3.26.0 or higher
