import json,sys,platform

from html.parser import HTMLParser
import re
class Main(HTMLParser):
 def __init__(self):super().__init__();self.depth=0;self.skip=0;self.text=[];self.links=[];self.headings=[];self.mainCount=0;self.verbatim=False
 def handle_starttag(self,t,a):
  if t=='main':self.depth+=1;self.mainCount+=1
  if self.depth and t in ['pre','code']:self.verbatim=True
  if self.depth and t in ['script','style','nav','aside']:self.skip+=1
  if self.depth and not self.skip and t=='a':self.links.append(dict(a).get('href',''))
 def handle_endtag(self,t):
  if self.depth and t in ['script','style','nav','aside']:self.skip-=1
  if t=='main':self.depth-=1
 def handle_data(self,v):
  if self.depth and not self.skip:self.text.append(v)
def extract(s):
 p=Main();p.feed(s);text=re.sub(r'\s+',' ',' '.join(p.text)).strip();reasons=[]
 if p.mainCount!=1:reasons.append('main-count-must-be-one')
 if not text:reasons.append('empty-main-text')
 if p.depth!=0 or p.skip!=0:reasons.append('unbalanced-tracked-tags')
 if p.verbatim:reasons.append('verbatim-policy-required')
 return {'text':text,'links':p.links,'mainCount':p.mainCount,'comparisonReady':not reasons,'holdReasons':reasons}
def compare(before,after):
 left=extract(before);right=extract(after);ready=left['comparisonReady'] and right['comparisonReady']
 return {'comparisonStatus':'compared' if ready else 'hold-extraction','textEqual':left['text']==right['text'] if ready else None,'linksEqual':left['links']==right['links'] if ready else None,'before':left,'after':right}
a='<nav>방문 10</nav><main><h2>검사</h2><p>기한은 10월 20일.</p><a href="/v1">자료</a></main>'
b='<nav>방문 11</nav><main><h2>검사</h2> <p>기한은 10월 20일.</p><a href="/v1">자료</a></main>'
c=b.replace('20일','21일');e=b.replace('/v1','/v2')
cases=[('navigation-only',b),('deadline-changed',c),('href-only',e)]
result={'inputs':{'baseline':a,**dict(cases)},'baseline':extract(a),'results':[]}
for name,document in cases:
 comparison=compare(a,document);result['results'].append({'name':name,'rawEqual':a==document,'comparisonStatus':comparison['comparisonStatus'],'textEqual':comparison['textEqual'],'linksEqual':comparison['linksEqual'],'extracted':comparison['after']})
assert [(r['textEqual'],r['linksEqual']) for r in result['results']]==[(True,True),(False,True),(True,False)]
contract_inputs=[('missing-main','<div>로그인 안내</div>','<div>로그인 안내</div>'),('empty-main','<main> </main>','<main> </main>'),('multiple-main','<main>첫째</main><main>둘째</main>','<main>첫째</main><main>둘째</main>'),('pre-whitespace','<main><pre>a  b</pre></main>','<main><pre>a b</pre></main>')]
result['contractCases']=[{'name':name,'inputs':[before,after],'result':compare(before,after)}for name,before,after in contract_inputs]
assert all(x['result']['comparisonStatus']=='hold-extraction' and x['result']['textEqual']is None and x['result']['linksEqual']is None for x in result['contractCases'])
result['comparisonContract']='Exactly one nonempty main; no pre/code without a separate whitespace policy. Failure is hold, never equal. This parser does not reproduce browser DOM correction or prove page identity.'

from datetime import datetime as _runtime_datetime, timezone as _runtime_timezone
result.update({'fixtureCreatedOn':'2026-10-15','executedAt':_runtime_datetime.now(_runtime_timezone.utc).isoformat(),'runtime':platform.python_version(),'synthetic':True})
print(json.dumps(result,ensure_ascii=False,indent=2))