Files
kb/engine/tests/test_note_titles.py
T

76 lines
2.7 KiB
Python

"""Tests for automatic and backfilled note titles."""
import sqlite3
from kb.ingest.note import auto_title
from kb.maintenance import backfill_note_titles
from kb.maintenance.backfill_note_titles import find_repairs
def test_auto_title_strips_markdown_and_limits_length():
assert auto_title("## Useful heading\nBody") == "Useful heading"
assert auto_title("x" * 100) == "x" * 80
def test_find_repairs_only_selects_synthetic_note_titles():
conn = sqlite3.connect(":memory:")
conn.row_factory = sqlite3.Row
conn.executescript(
"""
CREATE TABLE documents (
id INTEGER PRIMARY KEY, title TEXT, original_filename TEXT, doc_type TEXT
);
CREATE TABLE chunks (
id INTEGER PRIMARY KEY, document_id INTEGER, chunk_index INTEGER,
text TEXT, metadata TEXT
);
INSERT INTO documents VALUES
(1, '7dec828a-1234-4567-89ab-123456789abc_note.note',
'7dec828a-1234-4567-89ab-123456789abc_note.note', 'note'),
(2, 'A deliberate title', 'named.note', 'note');
INSERT INTO chunks VALUES
(1, 1, 0, '# Derived title\nBody', '{}'),
(2, 2, 0, 'Must not replace', '{}');
"""
)
assert find_repairs(conn) == [{
"document_id": 1,
"old_title": "7dec828a-1234-4567-89ab-123456789abc_note.note",
"new_title": "Derived title",
"original_filename": "7dec828a-1234-4567-89ab-123456789abc_note.note",
}]
def test_apply_repairs_refreshes_title_search_text_and_vector(monkeypatch):
conn = sqlite3.connect(":memory:")
conn.row_factory = sqlite3.Row
conn.executescript(
"""
CREATE TABLE documents (id INTEGER PRIMARY KEY, title TEXT, updated_at TEXT);
CREATE TABLE chunks (
id INTEGER PRIMARY KEY, document_id INTEGER, chunk_index INTEGER,
text TEXT, metadata TEXT, enriched_text TEXT
);
CREATE TABLE chunks_vec (embedding BLOB, chunk_id INTEGER);
INSERT INTO documents VALUES (1, 'old', NULL);
INSERT INTO chunks VALUES (3, 1, 0, 'New title\nBody', '{}', 'old text');
INSERT INTO chunks_vec VALUES (X'00', 3);
"""
)
monkeypatch.setattr(
backfill_note_titles.embeddings,
"embed_texts",
lambda texts: [[1.0, 2.0, 3.0] for _ in texts],
)
backfill_note_titles.apply_repairs(conn, [{
"document_id": 1, "new_title": "New title"
}])
assert conn.execute("SELECT title FROM documents").fetchone()[0] == "New title"
assert conn.execute("SELECT enriched_text FROM chunks").fetchone()[0] == (
"New title\n\nNew title\nBody"
)
assert len(conn.execute("SELECT embedding FROM chunks_vec").fetchone()[0]) == 12