Datasets:
File size: 4,348 Bytes
7a088d4 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 | #!/usr/bin/env python3
"""List NDL documents by original missing-character ratio with local links.
Reads a completed unified audit catalog and writes a Markdown table plus a CSV
for an arbitrary [min, max) missing-ratio band. Links point at the pre-clean
text (corpus/text/<PID>.txt) and at the post-clean line in build documents.
No corpus text is modified.
"""
import argparse
import collections
import csv
import json
from pathlib import Path
def ratio_band(value, bands):
for label, low, high in bands:
if low <= value < high:
return label
return 'other'
def main():
p = argparse.ArgumentParser(description=__doc__)
p.add_argument('--audit', type=Path, required=True)
p.add_argument('--min', type=float, required=True, dest='low')
p.add_argument('--max', type=float, required=True, dest='high')
p.add_argument('--output', type=Path, required=True, help='Markdown output path')
args = p.parse_args()
audit = args.audit.resolve()
out = args.output.resolve()
root = Path(__file__).resolve().parents[1]
rows = []
with (audit / 'catalog.jsonl').open() as f:
for line in f:
r = json.loads(line)
if r['source'] != 'ndl':
continue
q = r['quality']['original_quality']
ratio = float(q['missing_ratio'])
if not (args.low <= ratio < args.high):
continue
rows.append((r, q))
rows.sort(key=lambda x: (-x[1]['missing_ratio'], x[0]['id']))
if not rows:
raise SystemExit('no documents in range')
width = (args.high - args.low) / 5
bands = []
for i in range(5):
low = args.low + i * width
high = args.low + (i + 1) * width
label = f'{low*100:.0f}-{high*100:.0f}%'
bands.append((label, low, high))
counts = collections.Counter(ratio_band(q['missing_ratio'], bands) for _, q in rows)
out.parent.mkdir(parents=True, exist_ok=True)
csv_path = out.with_suffix('.csv')
with csv_path.open('w', newline='') as f:
w = csv.writer(f)
w.writerow(['id', 'title', 'missing_ratio', 'missing_count', 'nonspace_chars',
'text_path', 'text_line', 'document_path', 'document_line'])
for r, q in rows:
pid = r['id'].split('/')[1]
w.writerow([r['id'], r['title'], f"{q['missing_ratio']:.6f}", q['missing_count'],
q['nonspace_chars'], f'corpus/text/{pid}.txt', 1,
r['text_ref']['path'], r['text_ref']['line']])
total_chars = sum(q['nonspace_chars'] for _, q in rows)
lines = [
f"# NDL 欠字率 {args.low*100:.0f}–{args.high*100:.0f}% の資料一覧",
'',
f'- 対象: {len(rows)}資料（`{audit.name}/catalog.jsonl` の original missing_ratio 基準）',
f'- 削除前の非空白文字合計: {total_chars:,}字',
f'- 書名の種類: {len({r["title"] for r, _ in rows})}',
f'- CSV: `{csv_path.name}`',
'',
'| 欠字率 | 件数 |',
'|---|---:|',
]
for label, low, high in reversed(bands):
lines.append(f'| {label} | {counts[label]} |')
lines.extend([
f'| 合計 | {len(rows)} |',
'',
'リンク: `txt` は削除前テキスト、`Lxxxx` は削除後本文（documents.jsonl の行）。',
'',
'| # | ID | 書名 | 欠字率 | 欠字/字数 | 本文 | 後処理 |',
'|---|---|---|---:|---:|---|---|',
])
for n, (r, q) in enumerate(rows, 1):
pid = r['id'].split('/')[1]
title = r['title'].replace('|', '\\|')
txt = f'[txt](vscode://file{root}/corpus/text/{pid}.txt:1)'
doc = (f'[L{r["text_ref"]["line"]}](vscode://file{root}/'
f'{r["text_ref"]["path"]}:{r["text_ref"]["line"]})')
lines.append(f'| {n} | {r["id"]} | {title} | {q["missing_ratio"]*100:.1f}% | '
f'{q["missing_count"]}/{q["nonspace_chars"]} | {txt} | {doc} |')
out.write_text('\n'.join(lines) + '\n')
print(json.dumps(dict(documents=len(rows), chars=total_chars,
markdown=str(out), csv=str(csv_path),
bands={label: counts[label] for label, _, _ in bands}),
ensure_ascii=False))
if __name__ == '__main__':
main()
|