scripts/evidence_audit.py

#!/usr/bin/env python3
"""Offline structure-only audit of claim/source records. Does NOT verify factual accuracy."""
import argparse,csv,json,re,sys
from pathlib import Path
VALID={'verified','source-reported','inference','opinion','contradicted','stale','unverified'}
URL_RE=re.compile(r'^https?://[^\s]+$',re.I)
DOI_RE=re.compile(r'^10\.\d{4,9}/\S+$',re.I)
def read_csv(path):
 with Path(path).open(encoding='utf-8-sig',newline='') as f:return list(csv.DictReader(f))
def audit(claims,sources):
 problems=[]; src={}
 for i,row in enumerate(sources,2):
  sid=(row.get('source_id') or '').strip(); locator=(row.get('url_or_doi') or '').strip()
  if not sid: problems.append(f'sources row {i}: missing source_id');continue
  if sid in src:problems.append(f'sources row {i}: duplicate source_id {sid}')
  src[sid]=row
  if not locator:problems.append(f'source {sid}: missing locator')
  elif not(URL_RE.match(locator) or DOI_RE.match(locator)):problems.append(f'source {sid}: locator format not URL/DOI')
 seen=set(); counts={}
 for i,row in enumerate(claims,2):
  cid=(row.get('claim_id') or '').strip();claim=(row.get('claim') or '').strip(); status=(row.get('status') or '').strip().lower()
  if not cid:problems.append(f'claims row {i}: missing claim_id')
  elif cid in seen:problems.append(f'claims row {i}: duplicate claim_id {cid}')
  seen.add(cid)
  if not claim:problems.append(f'claims row {i}: missing claim text')
  if status not in VALID:problems.append(f'claim {cid or i}: invalid status {status}')
  counts[status]=counts.get(status,0)+1
  sids=[v.strip() for v in (row.get('source_ids') or '').split(';') if v.strip()]
  if status in {'verified','source-reported','contradicted','stale'} and not sids:
   problems.append(f'claim {cid or i}: evidence-required status without source')
  for s in sids:
   if s not in src:problems.append(f'claim {cid or i}: unknown source_id {s}')
 return {'total_claims':len(claims),'total_sources':len(sources),'status_counts':counts,'structural_issues':problems,'passed':not problems,'limitations':'Checks IDs, status and locator format only; DOES NOT fetch sources or verify claims, quotes or link reachability.'}
def main(argv=None):
 p=argparse.ArgumentParser(description=__doc__);p.add_argument('claims');p.add_argument('sources');a=p.parse_args(argv)
 result=audit(read_csv(a.claims),read_csv(a.sources));print(json.dumps(result,indent=2));return 0 if result['passed'] else 1
if __name__=='__main__':sys.exit(main())