"""Controlled local extraction demonstration, NOT an AI citation experiment."""
import datetime
import hashlib
import json
import pathlib
import platform
import re
import sys
from html.parser import HTMLParser

ROOT=pathlib.Path(__file__).parent
SENTENCES=[
    '이 안내문은 신청 전에 준비할 자료 세 가지를 설명합니다.',
    '신청자는 연락처, 요청 범위, 희망 일정을 제출합니다.'
]
EXPECTED=' '.join(SENTENCES)
FRAGMENTS=['이 안내문은 신청 전에','준비할 자료 세 가지를 설명합니다.','신청자는 연락처, 요청 범위,','희망 일정을 제출합니다.']
FIXTURES={
    'complete-paragraphs':'\n'.join('<p>'+s+'</p>' for s in SENTENCES),
    'complete-sentences-br':'<p>'+'<br>'.join(SENTENCES)+'</p>',
    'fragment-br':'<p>'+'<br>'.join(FRAGMENTS)+'</p>',
    'fragment-paragraphs':''.join('<p>'+s+'</p>' for s in FRAGMENTS),
}

class Extractor(HTMLParser):
    def __init__(self, boundary_aware):
        super().__init__(convert_charrefs=True)
        self.boundary_aware=boundary_aware
        self.parts=[]
    def handle_data(self,data):
        self.parts.append(data)
    def handle_starttag(self,tag,attrs):
        if self.boundary_aware and tag=='br':
            self.parts.append('\n')
    def handle_endtag(self,tag):
        if self.boundary_aware and tag=='p':
            self.parts.append('\n')
    def output(self):
        return ''.join(self.parts)

def normalize(text):return re.sub(r'\s+',' ',text).strip()
def glyphs(text):return re.sub(r'\s+','',text)
def main():
    rows=[]
    for fixture_id,markup in FIXTURES.items():
        for extractor_id,aware in [('data-concat',False),('boundary-aware',True)]:
            parser=Extractor(aware);parser.feed(markup);parser.close()
            output=parser.output();normalized=normalize(output)
            rows.append({'fixtureId':fixture_id,'extractorId':extractor_id,'htmlSha256':hashlib.sha256(markup.encode()).hexdigest(),'extractedText':output,'normalizedText':normalized,'normalizedEqualsExpected':normalized==EXPECTED,'nonWhitespaceCharactersPreserved':glyphs(output)==glyphs(EXPECTED),'exactSentenceMatches':sum(s in normalized for s in SENTENCES)})
    result={'status':'observed-local-extraction-only','executedAt':datetime.datetime.now(datetime.timezone.utc).isoformat(),'pythonVersion':sys.version,'platform':platform.platform(),'scriptSha256':hashlib.sha256(pathlib.Path(__file__).read_bytes()).hexdigest(),'fixtureCount':len(FIXTURES),'extractorCount':2,'observationCount':len(rows),'expectedNormalizedText':EXPECTED,'rows':rows,'scope':'No network requests, no AI calls, no actual CMS sample, no crawling/indexing/citation measurement. Exact string matching is not semantic understanding.','checks':{'sameNonWhitespaceCharactersAllConditions':all(r['nonWhitespaceCharactersPreserved'] for r in rows),'boundaryAwareMatchesAllFixtures':all(r['normalizedEqualsExpected'] for r in rows if r['extractorId']=='boundary-aware')}}
    for name,data in [('parsing-fixtures.json',{'status':'synthetic-neutral-fixtures','sentences':SENTENCES,'fixtures':FIXTURES}),('parsing-observed-results.json',result)]:
        (ROOT/name).write_text(json.dumps(data,ensure_ascii=False,indent=2)+'\n',encoding='utf-8')
    assert all(result['checks'].values())
    print(json.dumps({'observations':len(rows),'checks':result['checks']},ensure_ascii=True))
if __name__=='__main__':main()