"""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):
    words=d['words'];segments=[];start=0
    for i,word in enumerate(words):
        assert math.isfinite(word['start']) and math.isfinite(word['end']) and word['start']<=word['end']
        if i:assert words[i-1]['end']<=word['start']
        pause=i+1<len(words) and words[i+1]['start']-word['end']>=d['pause_threshold']
        if word['text'].endswith(('。','！','？','.','!','?')) or pause or i==len(words)-1:
            segments.append({'word_start':start,'word_end':i+1,'start':words[start]['start'],'end':word['end'],'text':''.join(w['text'] for w in words[start:i+1])});start=i+1
    ranges=[[s['word_start'],s['word_end']] for s in segments];assert ranges==d['expected_ranges']
    def valid(rows):
        indexes=[i for s in rows for i in range(s['word_start'],s['word_end'])]
        return indexes==list(range(len(words))) and all(s['start']<=w['start']<=w['end']<=s['end'] for s in rows for w in words[s['word_start']:s['word_end']])
    assert valid(segments);bad=[dict(s) for s in segments];bad[1]['word_start']=1;assert not valid(bad)
    assert segments[-1]['text']=='结束' and not valid(segments[:-1])
    return {'sentences':segments,'all_words_once':True,'last_unpunctuated_sentence_kept':True,'transcription_model_called':False}

if __name__=='__main__':
    base=Path(__file__).resolve().parent
    data=json.loads((base/'a043.json').read_text('utf-8'))
    result=verify(data,base)
    print(json.dumps({'id':'A043','passed':True,'result':result},ensure_ascii=True))