"""Synthetic CSV regression and fixed saved-CSV replay, no network or production changes. Python3 stdlib. Writes only its own TemporaryDirectory. """ from pathlib import Path from datetime import datetime,timezone import json,subprocess,sys,tempfile,hashlib,argparse HEADER='해제구분,해제일\n' CASES=[ ('blank',HEADER+',\n',(1,1,0,0,0,0),'',''), ('marker_only',HEADER+'Y,\n',(1,0,1,0,1,0),'Y',''), ('date_only',HEADER+',2026-01-02\n',(1,0,1,0,0,1),'','2026-01-02'), ('both',HEADER+'Y,2026-01-02\n',(1,0,1,1,0,0),'Y','2026-01-02'), ('quoted_comma',HEADER+'"Y,추가",2026-01-02\n',(1,0,1,1,0,0),'Y,추가','2026-01-02'), ('escaped_quote',HEADER+'"Y""설명",2026-01-02\n',(1,0,1,1,0,0),'Y"설명','2026-01-02'), ('bom_multiline','\ufeff해제구분,해제일\r\n"Y\r\n설명",2026-01-02\r\n',(1,0,1,1,0,0),'Y\r\n설명','2026-01-02'), ('unclosed_empty_quote',HEADER+',"',None,None,None), ('unclosed_marker_quote',HEADER+',"Y',None,None,None), ('quote_tail',HEADER+',"Y"x\n',None,None,None), ('missing_header','다른열,해제일\n,\n',None,None,None), ('duplicate_header','해제구분,해제일,해제구분\n,,\n',None,None,None), ('short_row',HEADER+'Y\n',None,None,None), ('extra_row',HEADER+'Y,2026-01-02,extra\n',None,None,None)] COUNTS=['total_rows','blank_both_fields','rows_with_any_marker','marker_and_date','marker_only','date_only'] def invoke(script,input_file,output,flags): return subprocess.run([sys.executable,*flags,str(script),'--input',str(input_file),'--output',str(output)],capture_output=True,text=True,timeout=10) def run(script,source,original,before=None): script=Path(script).resolve();source=Path(source).resolve();original=Path(original).resolve();baseline=json.loads(original.read_text());checks=[];before_rows=[] with tempfile.TemporaryDirectory(prefix='reader-csv-syntax-')as folder: folder=Path(folder) for flags in [[],['-O']]: label='optimized'if flags else'normal' for name,text,expected,marker,date in CASES: input_file=folder/(name+'.csv');input_file.write_bytes(text.encode('utf-8'));output=folder/(label+'-'+name+'.json') p=invoke(script,input_file,output,flags);accepted=p.returncode==0 if accepted!=(expected is not None):raise ValueError(name+' wrong acceptance: '+p.stderr[-300:]) value=json.loads(output.read_text())if accepted else None if accepted: if tuple(value[k]for k in COUNTS)!=expected:raise ValueError(name+' wrong field count') if value['rows'][0]['original_marker']!=marker or value['rows'][0]['original_date']!=date:raise ValueError(name+' original field changed') if value['source_csv_sha256']!=hashlib.sha256(input_file.read_bytes()).hexdigest():raise ValueError('Input hash mismatch') elif output.exists():raise ValueError('Rejected input created a result') checks.append({'case':name,'python_flags':flags,'input_csv':text,'accepted':accepted,'summary':{k:value[k]for k in COUNTS}if value else None,'no_output_on_rejection':not output.exists()if not accepted else None}) output=folder/(label+'-original.json');p=invoke(script,source,output,flags) if p.returncode:raise ValueError(p.stderr[-500:]) actual=json.loads(output.read_text());without_name=lambda x:{k:v for k,v in x.items()if k!='input_file'} if without_name(actual)!=without_name(baseline):raise ValueError('Original 36-row evidence changed') checks.append({'case':'original_saved_csv','python_flags':flags,'total_rows':actual['total_rows'],'blank_both_fields':actual['blank_both_fields'],'matches_existing_json':True}) valid=folder/'valid.csv';valid.write_bytes((HEADER+',\n').encode());existing=folder/(label+'-existing.json');existing.write_bytes(b'KEEP-EXISTING') p=invoke(script,valid,existing,flags) if p.returncode==0 or existing.read_bytes()!=b'KEEP-EXISTING':raise ValueError('Existing output was replaced') checks.append({'case':'existing_output_preserved','python_flags':flags,'passed':True}) data=valid.read_bytes();p=invoke(script,valid,valid,flags) if p.returncode==0 or valid.read_bytes()!=data:raise ValueError('Source input was changed') checks.append({'case':'same_input_output_rejected','python_flags':flags,'passed':True}) if before: for name,text,expected,marker,date in CASES[7:10]: input_file=folder/('before-'+name+'.csv');input_file.write_bytes(text.encode());output=folder/('before-'+name+'.json');p=invoke(Path(before).resolve(),input_file,output,[]) value=json.loads(output.read_text())if p.returncode==0 else None before_rows.append({'case':name,'accepted':p.returncode==0,'summary':{k:value[k]for k in COUNTS}if value else None}) return {'passed':True,'completed_at_utc':datetime.now(timezone.utc).isoformat(),'kind':'synthetic CSV syntax and saved-field reproduction;not a current cancellation lookup','script_sha256':hashlib.sha256(script.read_bytes()).hexdigest(),'source_sha256':hashlib.sha256(source.read_bytes()).hexdigest(),'synthetic_valid_inputs':7,'synthetic_invalid_inputs':7,'corrected_executions':len(checks),'before_executions':len(before_rows),'checks':checks,'before':before_rows,'limits':'These CSV cases and one stored36-row file only. Not every CSV dialect/file size, current contract validity, upstream API validation or legal advice.'} if __name__=='__main__': p=argparse.ArgumentParser();p.add_argument('script');p.add_argument('csv');p.add_argument('original_json');p.add_argument('--before');args=p.parse_args() print(json.dumps(run(args.script,args.csv,args.original_json,args.before),ensure_ascii=False,indent=2))