"""合成离线样本检查。与同名 JSON 及文中列出的附件放在同一目录运行。"""
from pathlib import Path, PurePosixPath
import base64, hashlib, io, inspect, json, math, re, subprocess, sys, tempfile, wave, zipfile
from collections import Counter, defaultdict
from datetime import date, datetime
from decimal import Decimal
sys.dont_write_bytecode=True

def sha(b): return hashlib.sha256(b).hexdigest()
def rejected(fn):
    try: fn()
    except (ValueError,TypeError,KeyError,UnicodeError,OverflowError): return True
    return False

def git(root,*args,input=None,check=True):
    r=subprocess.run(['git','-c','core.autocrlf=false',*args],cwd=root,input=input,capture_output=True,check=check)
    return r

def verify(d,base):
    import unicodedata
    def normalized(t):return ' '.join(unicodedata.normalize('NFC',t).split())
    groups=defaultdict(list)
    for row in d['rows']:
        n=normalized(row['text']);groups[sha(n.encode())].append({**row,'raw_hash':sha(row['text'].encode()),'normalized':n})
    leaks=[v for v in groups.values() if len({x['split'] for x in v})>1]
    assert len(leaks)==d['expected_leak_groups'];assert [x['id'] for x in leaks[0]]==['q1','q2']
    assert normalized(d['rows'][0]['text'])!=normalized(d['rows'][2]['text'])!=normalized(d['rows'][3]['text'])
    fixed={key:sorted({x['split'] for x in values})[0] for key,values in groups.items()}
    sets=defaultdict(set)
    for key,split in fixed.items():sets[split].add(key)
    assert not sets['fewshot']&sets['test'] and not sets['train']&sets['test']
    return {'leak_group_ids':[[x['id'] for x in v] for v in leaks],'fixed_group_count':len(fixed),'semantic_paraphrases_detected':False}

if __name__=='__main__':
    base=Path(__file__).resolve().parent
    data=json.loads((base/'a038.json').read_text('utf-8'))
    result=verify(data,base)
    print(json.dumps({'id':'A038','passed':True,'result':result},ensure_ascii=True))
