Dictionary Schema V2

migrate_dictionary_v2.py — 55 lines

Extends dictionary_terms with significance, source tracking, provenance, and confidence columns.

1"""Extend dictionary_terms schema with new columns for enrichment pipeline.
2
3Adds columns for:
4- significance fields (HP-specific and scholarship-wide)
5- source document tracking
6- source method and notes
7- confidence level
8
9Idempotent: safe to run multiple times.
10"""
11
12import sqlite3
13from pathlib import Path
14
15BASE_DIR = Path(__file__).resolve().parent.parent
16DB_PATH = BASE_DIR / "db" / "hp.db"
17
18NEW_COLUMNS = [
19    ("significance_to_hp", "TEXT"),
20    ("significance_to_scholarship", "TEXT"),
21    ("related_scholars", "TEXT"),
22    ("see_also_text", "TEXT"),
23    ("source_documents", "TEXT"),
24    ("source_page_refs", "TEXT"),
25    ("source_quotes_short", "TEXT"),
26    ("source_method", "TEXT"),
27    ("confidence", "TEXT DEFAULT 'MEDIUM'"),
28    ("notes", "TEXT"),
29]
30
31
32def main():
33    conn = sqlite3.connect(DB_PATH)
34    cur = conn.cursor()
35
36    # Get existing columns
37    cur.execute("PRAGMA table_info(dictionary_terms)")
38    existing = {row[1] for row in cur.fetchall()}
39
40    added = 0
41    for col_name, col_type in NEW_COLUMNS:
42        if col_name not in existing:
43            cur.execute(f"ALTER TABLE dictionary_terms ADD COLUMN {col_name} {col_type}")
44            print(f"  Added column: {col_name} ({col_type})")
45            added += 1
46        else:
47            print(f"  Column exists: {col_name}")
48
49    conn.commit()
50    conn.close()
51    print(f"\nDone. Added {added} new columns.")
52
53
54if __name__ == "__main__":
55    main()