Export Showcase Data (Legacy)

export_showcase_data.py — 116 lines
Original data.json exporter for the gallery. Superseded by build_site.py.
1"""Export matched references as JSON for the static showcase page."""
2
3import sqlite3
4import json
5from pathlib import Path
6
7BASE_DIR = Path(__file__).resolve().parent.parent
8DB_PATH = BASE_DIR / "db" / "hp.db"
9OUTPUT_PATH = BASE_DIR / "site" / "data.json"
10
11
12def main():
13    conn = sqlite3.connect(DB_PATH)
14    conn.row_factory = sqlite3.Row
15    cur = conn.cursor()
16
17    # Get all HIGH confidence matches (Ch6 BL + Ch9 Siena) plus all matches for showcase
18    cur.execute("""
19        SELECT
20            dr.id as ref_id,
21            dr.thesis_page,
22            dr.signature_ref,
23            dr.manuscript_shelfmark,
24            dr.context_text,
25            dr.marginal_text,
26            dr.chapter_num,
27            i.filename as image_filename,
28            i.folio_number,
29            i.side,
30            i.page_type,
31            COALESCE(i.web_path, i.relative_path) as relative_path,
32            m.confidence,
33            m.match_method,
34            ms.shelfmark,
35            ms.institution,
36            ms.city,
37            sm.quire,
38            sm.leaf_in_quire
39        FROM matches m
40        JOIN dissertation_refs dr ON m.ref_id = dr.id
41        JOIN images i ON m.image_id = i.id
42        JOIN manuscripts ms ON i.manuscript_id = ms.id
43        LEFT JOIN signature_map sm ON sm.signature = dr.signature_ref
44        WHERE m.confidence = 'HIGH'
45        AND i.page_type = 'PAGE'
46        ORDER BY sm.folio_number, i.side, dr.thesis_page
47    """)
48
49    entries = []
50    seen = set()  # Deduplicate by (signature, image)
51
52    for row in cur.fetchall():
53        key = (row['signature_ref'], row['image_filename'])
54        if key in seen:
55            continue
56        seen.add(key)
57
58        entries.append({
59            'ref_id': row['ref_id'],
60            'thesis_page': row['thesis_page'],
61            'signature': row['signature_ref'],
62            'manuscript': row['shelfmark'],
63            'institution': row['institution'],
64            'city': row['city'],
65            'context': row['context_text'],
66            'marginal_text': row['marginal_text'],
67            'chapter': row['chapter_num'],
68            'image_file': row['image_filename'],
69            'image_path': row['relative_path'],
70            'folio_number': row['folio_number'],
71            'side': row['side'],
72            'confidence': row['confidence'],
73            'quire': row['quire'],
74            'leaf_in_quire': row['leaf_in_quire'],
75        })
76
77    # Also get manuscript info for the page
78    cur.execute("SELECT shelfmark, institution, city, description, image_count FROM manuscripts")
79    manuscripts = [dict(row) for row in cur.fetchall()]
80
81    # Summary stats
82    cur.execute("SELECT COUNT(DISTINCT signature_ref) FROM dissertation_refs")
83    unique_sigs = cur.fetchone()[0]
84    cur.execute("SELECT COUNT(*) FROM dissertation_refs")
85    total_refs = cur.fetchone()[0]
86
87    data = {
88        'entries': entries,
89        'manuscripts': manuscripts,
90        'stats': {
91            'total_references': total_refs,
92            'unique_signatures': unique_sigs,
93            'high_confidence_matches': len(entries),
94        }
95    }
96
97    OUTPUT_PATH.parent.mkdir(exist_ok=True)
98    with open(OUTPUT_PATH, 'w', encoding='utf-8') as f:
99        json.dump(data, f, indent=2, ensure_ascii=False)
100
101    print(f"Exported {len(entries)} showcase entries to {OUTPUT_PATH}")
102    print(f"  BL C.60.o.12: {sum(1 for e in entries if e['manuscript'] == 'C.60.o.12')}")
103    print(f"  Siena O.III.38: {sum(1 for e in entries if e['manuscript'] == 'O.III.38')}")
104
105    # Show a few entries
106    print("\nSample entries:")
107    for e in entries[:5]:
108        print(f"  {e['signature']} [{e['manuscript']}] -> {e['image_file']}")
109        if e['marginal_text']:
110            print(f"    Marginal: '{e['marginal_text'][:60]}'")
111
112    conn.close()
113
114
115if __name__ == "__main__":
116    main()