From c6c2a238c3e890384eef6bf9bca062fd784d9157 Mon Sep 17 00:00:00 2001 From: Simon Willison Date: Tue, 2 Dec 2025 16:22:42 -0800 Subject: [PATCH] Fix for stale internal database bug, closes #2605 --- datasette/utils/internal_db.py | 3 +++ tests/test_internal_db.py | 48 ++++++++++++++++++++++++++++++++++ 2 files changed, 51 insertions(+) diff --git a/datasette/utils/internal_db.py b/datasette/utils/internal_db.py index a3afbab2..587ea7b1 100644 --- a/datasette/utils/internal_db.py +++ b/datasette/utils/internal_db.py @@ -116,6 +116,9 @@ async def populate_schema_tables(internal_db, db): database_name = db.name def delete_everything(conn): + conn.execute( + "DELETE FROM catalog_databases WHERE database_name = ?", [database_name] + ) conn.execute( "DELETE FROM catalog_tables WHERE database_name = ?", [database_name] ) diff --git a/tests/test_internal_db.py b/tests/test_internal_db.py index 59516225..7a0d1630 100644 --- a/tests/test_internal_db.py +++ b/tests/test_internal_db.py @@ -91,3 +91,51 @@ async def test_internal_foreign_key_references(ds_client): ) await internal_db.execute_fn(inner) + + +@pytest.mark.asyncio +async def test_stale_catalog_entry_database_fix(tmp_path): + """ + Test for https://github.com/simonw/datasette/issues/2605 + + When the internal database persists across restarts and has entries in + catalog_databases for databases that no longer exist, accessing the + index page should not cause a 500 error (KeyError). + """ + from datasette.app import Datasette + + internal_db_path = str(tmp_path / "internal.db") + data_db_path = str(tmp_path / "data.db") + + # Create a data database file + import sqlite3 + + conn = sqlite3.connect(data_db_path) + conn.execute("CREATE TABLE test_table (id INTEGER PRIMARY KEY)") + conn.close() + + # First Datasette instance: with the data database and persistent internal db + ds1 = Datasette(files=[data_db_path], internal=internal_db_path) + await ds1.invoke_startup() + + # Access the index page to populate the internal catalog + response = await ds1.client.get("/") + assert "data" in ds1.databases + assert response.status_code == 200 + + # Second Datasette instance: reusing internal.db but WITHOUT the data database + # This simulates restarting Datasette after removing a database + ds2 = Datasette(internal=internal_db_path) + await ds2.invoke_startup() + + # The database is not in ds2.databases + assert "data" not in ds2.databases + + # Accessing the index page should NOT cause a 500 error + # This is the bug: it currently raises KeyError when trying to + # access ds.databases["data"] for the stale catalog entry + response = await ds2.client.get("/") + assert response.status_code == 200, ( + f"Index page should return 200, not {response.status_code}. " + "This fails due to stale catalog entries causing KeyError." + )