#!/usr/bin/env python3
# Agentpedia 电影词条「配图-人名一致性」机械核验（#71 冻结口径 v1 复跑）
# 输入: page*.json (pin/path/list 快照), content/*.json (manapi content/<pin>), cache/*.bin (metafile 字节)
# 输出: out/rows.csv (17字段), out/summary.json, out/fileinfo.json  —— 确定性输出
import json,glob,os,re,csv,collections,hashlib,struct,sys
BOT='idq1p0zzdyaedg8xsuczqe4vkljdmu07rna7tf278z'
POSTER=re.compile(r'海报[：:]\s*!\[[^\]]*\]\(metafile://([0-9a-f]{64}i0)\)')
PHOTO=re.compile(r'^-\s*(?P<role>\S+?)\s+(?P<name>[^：:]+)[：:]\s*!\[(?P<label>[^\]]*)\]\(metafile://(?P<pin>[0-9a-f]{64}i0)\)')
MASTER=re.compile(r'^-\s*(?P<role>\S+?)[：:]\s*(?P<names>.+)$')
def dims(b):
    if b[:8]==b'\x89PNG\r\n\x1a\n': return ('png',)+struct.unpack('>II',b[16:24])
    if b[:3]==b'\xff\xd8\xff':
        i=2
        while i<len(b)-9:
            if b[i]!=0xFF: i+=1; continue
            m=b[i+1]
            if m in (0xC0,0xC1,0xC2,0xC3,0xC5,0xC6,0xC7,0xC9,0xCA,0xCB,0xCD,0xCE,0xCF):
                h,w=struct.unpack('>HH',b[i+5:i+9]); return ('jpeg',w,h)
            if m in (0xD8,0xD9) or 0xD0<=m<=0xD7: i+=2; continue
            if i+3>=len(b): break
            i+=2+struct.unpack('>H',b[i+2:i+4])[0]
        return ('jpeg',None,None)
    if b[:6] in (b'GIF87a',b'GIF89a'): return ('gif',)+struct.unpack('<HH',b[6:10])
    if b[:4]==b'RIFF' and b[8:12]==b'WEBP': return ('webp',None,None)
    return ('unknown',None,None)
# ---- listing (fixtures 指纹: pin+content 规范化) ----
tsm={}; botpins=set()
for f in sorted(glob.glob('page[0-9].json')):
    for r in (json.load(open(f)).get('data') or {}).get('list') or []:
        tsm[r['id']]=r.get('timestamp',0)
        if r.get('globalMetaId')==BOT: botpins.add(r['id'])
# ---- entries ----
byslug={}
for fn in glob.glob('content/*.json'):
    pin=os.path.basename(fn)[:-5]
    if pin not in botpins: continue
    j=json.load(open(fn))
    if j.get('type')!='create' or not j.get('slug'): continue
    j['_pin']=pin; j['_ts']=tsm.get(pin,0)
    s=j['slug']
    if s not in byslug or j['_ts']>byslug[s]['_ts']: byslug[s]=j
recs=[]
for s,j in sorted(byslug.items()):
    c=j.get('content') or ''; pm=POSTER.search(c); photos=[]; masters=[]; inm=False
    for ln in c.split('\n'):
        l=ln.strip('\r')
        if re.match(r'^主创[：:]',l.strip()): inm=True; continue
        if re.match(r'^获奖记录|^海报[：:]|^refs|^相关词条',l.strip()): inm=False
        m=PHOTO.match(l)
        if m: photos.append(m.groupdict()); continue
        if inm:
            mm=MASTER.match(l)
            if mm: masters.append((mm.group('role'),[x for x in re.split(r'[、,，/]',mm.group('names').strip()) if x]))
    recs.append(dict(slug=s,pin=j['_pin'],poster=(pm.group(1) if pm else None),photos=photos,masters=masters))
# ---- fileinfo ----
fi={}
for f in sorted(glob.glob('cache/*.bin')):
    pin=os.path.basename(f)[:-4]; b=open(f,'rb').read(); t,w,h=dims(b)
    fi[pin]=dict(size=len(b),sha256=hashlib.sha256(b).hexdigest(),md5=hashlib.md5(b).hexdigest(),magic=t,w=w,h=h)
pin2names=collections.defaultdict(set); name2pins=collections.defaultdict(set)
for r in recs:
    for p in r['photos']: pin2names[p['pin']].add(p['name']); name2pins[p['name']].add(p['pin'])
rows=[]
for r in recs:
    mn=set(); 
    for role,ns in r['masters']: mn|=set(ns)
    pn=set(p['name'] for p in r['photos']); gaps=sorted(mn-pn)
    for p in r['photos']:
        f=fi.get(p['pin'],{}); i1b=1 if len(name2pins[p['name']])>1 else 0; i1a=1 if len(pin2names[p['pin']])>1 else 0
        i1c=1 if (not f or f['magic']=='unknown' or f['size']==0) else 0; i1d=1 if p['label']!=p['name'] else 0
        hint=1 if (f.get('w') and f.get('h') and f['w']/f['h']>1.2) else 0
        tier='V3' if (i1a or i1b or i1c or i1d) else 'null'
        rows.append(dict(entry_pin=r['pin'],slug=r['slug'],kind='photo',label=p['label'],name=p['name'],metafile_pin=p['pin'],
            sha256=f.get('sha256',''),w=f.get('w',''),h=f.get('h',''),i1a_dup=i1a,i1b_multi=i1b,i1c_invalid=i1c,
            i1d_label_mismatch=i1d,coverage_gap=0,vision_verdict='null',hint_aspect=hint,final_tier=tier,
            blank_reason='' if tier=='null' else 'I-1 机械命中'))
    for n in gaps:
        rows.append(dict(entry_pin=r['pin'],slug=r['slug'],kind='gap',label='',name=n,metafile_pin='',sha256='',w='',h='',
            i1a_dup=0,i1b_multi=0,i1c_invalid=0,i1d_label_mismatch=0,coverage_gap=1,vision_verdict='null',hint_aspect=0,
            final_tier='V3',blank_reason='I-2 覆盖缺口：无配图｜机械命中→V3 候选'))
    if r['poster']:
        f=fi.get(r['poster'],{}); i1c=1 if (not f or f['magic']=='unknown' or f['size']==0) else 0
        rows.append(dict(entry_pin=r['pin'],slug=r['slug'],kind='poster',label='海报',name='',metafile_pin=r['poster'],
            sha256=f.get('sha256',''),w=f.get('w',''),h=f.get('h',''),i1a_dup=0,i1b_multi=0,i1c_invalid=i1c,
            i1d_label_mismatch=0,coverage_gap=0,vision_verdict='null',hint_aspect=0,final_tier=('V3' if i1c else 'null'),
            blank_reason=('I-1c 无效图' if i1c else '')))
cols=['entry_pin','slug','kind','label','name','metafile_pin','sha256','w','h','i1a_dup','i1b_multi','i1c_invalid','i1d_label_mismatch','coverage_gap','vision_verdict','hint_aspect','final_tier','blank_reason']
os.makedirs('out',exist_ok=True)
with open('out/rows.csv','w',newline='') as fh:
    w=csv.DictWriter(fh,fieldnames=cols); w.writeheader(); [w.writerow(r) for r in rows]
summary=dict(entries=len(recs),movie_entries=sum(1 for r in recs if r['poster']),photo_rows=sum(1 for r in rows if r['kind']=='photo'),
    gap_rows=sum(1 for r in rows if r['kind']=='gap'),poster_rows=sum(1 for r in rows if r['kind']=='poster'),rows_total=len(rows),
    i1a=sum(r['i1a_dup'] for r in rows),i1b=sum(r['i1b_multi'] for r in rows),i1c=sum(r['i1c_invalid'] for r in rows),
    i1d=sum(r['i1d_label_mismatch'] for r in rows),i2=sum(r['coverage_gap'] for r in rows),hint_aspect=sum(r['hint_aspect'] for r in rows),
    tier_V3=sum(1 for r in rows if r['final_tier']=='V3'),tier_null=sum(1 for r in rows if r['final_tier']=='null'),
    unique_metafiles=len(fi),references=sum(1 for r in rows if r['metafile_pin']))
json.dump(summary,open('out/summary.json','w'),ensure_ascii=False,indent=1,sort_keys=True)
json.dump(fi,open('out/fileinfo.json','w'),ensure_ascii=False,indent=1,sort_keys=True)
# fixtures 指纹
h=hashlib.sha256()
for s,j in sorted(byslug.items()):
    h.update((j['_pin']+'|'+s+'|'+hashlib.sha256((j.get('content') or '').encode()).hexdigest()+'\n').encode())
print('fixtures_fingerprint', h.hexdigest())
print('summary', json.dumps(summary,ensure_ascii=False,sort_keys=True))