#!/usr/bin/env python3
"""OpenAlex 고신뢰 매치율 측정 — References 보유 논문(학술 추정) 표본."""
import asyncio, os, re

def toks(s):
    return set(re.findall(r'[a-z0-9]+', (s or '').lower()))
def sim(a, b):
    ta, tb = toks(a), toks(b)
    if not ta or not tb: return 0.0
    return len(ta & tb) / len(ta | tb)

async def main():
    import asyncpg, httpx
    conn = await asyncpg.connect(os.environ['DATABASE_URL'].replace('+asyncpg', ''))
    rows = await conn.fetch("SELECT id, title FROM documents WHERE material_type='paper' "
                            "AND doc_kind='standard' AND deleted_at IS NULL AND title IS NOT NULL "
                            "AND coalesce(md_content,extracted_text) ~* 'references|참고문헌' "
                            "ORDER BY id LIMIT 40")
    hi = mid = lo = 0; hits = []
    async with httpx.AsyncClient(timeout=20) as client:
        for r in rows:
            title = re.sub(r'\s+', ' ', r['title']).strip()
            try:
                resp = await client.get("https://api.openalex.org/works",
                    params={"search": title[:200], "per_page": 1, "mailto": "hyun49196@gmail.com"})
                res = (resp.json().get("results") or [])
                if not res: lo += 1; continue
                s = sim(title, res[0].get("title"))
                if s >= 0.6: hi += 1; hits.append((s, title[:40], (res[0].get('title') or '')[:40], res[0].get('cited_by_count'), len(res[0].get('referenced_works') or [])))
                elif s >= 0.4: mid += 1
                else: lo += 1
            except Exception: lo += 1
    print(f"표본={len(rows)} 고신뢰(≥0.6)={hi} 중간(0.4~0.6)={mid} 저신뢰/무매치={lo}")
    print("고신뢰 매치 샘플:")
    for s, a, b, cb, rf in hits[:8]:
        print(f"  sim={s:.2f} cited={cb} refs={rf} | {a} ≈ {b}")
    await conn.close()

asyncio.run(main())