#!/usr/bin/env python3
"""Reproducible offline query of one downloaded Magalia corpus SQLite snapshot.

gzip -dk texts.sqlite.gz
python3 query.py texts.sqlite '"δημου"' --layer conservative --format jsonl
No SQL fragments, regex, stemming or hidden cross-corpus deduplication.
"""
import argparse,csv,json,sqlite3,sys,unicodedata,re
from pathlib import Path

def normalize(s):
 return ' '.join(re.findall(r'[^\W_]+',''.join(c for c in unicodedata.normalize('NFD',s.replace('\ufffc',' zzmagaliaboundaryzz ').lower()) if unicodedata.category(c)!='Mn').replace('ς','σ')))

def main():
 p=argparse.ArgumentParser(description=__doc__)
 p.add_argument('database',type=Path);p.add_argument('query');p.add_argument('--layer',choices=['reading','conservative'],default='reading');p.add_argument('--format',choices=['jsonl','csv'],default='jsonl');p.add_argument('--limit',type=int,default=1000)
 a=p.parse_args()
 if 'zzmagaliaboundaryzz' in a.query.lower() or '\ufffc' in a.query or a.query.count('"')%2:p.error('Close quoted phrases and do not query reserved boundary markers.')
 if not a.database.is_file() or not 1<=a.limit<=100000: p.error('Use an existing database and a limit between 1 and 100000.')
 clauses=[]
 for match in re.finditer(r'(-?)(?:"([^"]+)"|([^\s"]+))',a.query):
  phrase=normalize(match[2] or match[3])
  if phrase:clauses.append((bool(match[1]),'"'+phrase.replace('"','""')+'"'))
 positive=[v for neg,v in clauses if not neg]
 if not positive:p.error('At least one positive word or quoted phrase is required.')
 expression=' AND '.join(positive)
 for neg,v in clauses:
  if neg:expression+=' NOT '+v
 expression=a.layer+' : ('+expression+')'
 con=sqlite3.connect(a.database.resolve().as_uri()+'?mode=ro',uri=True)
 rows=con.execute('SELECT r.record_json FROM text_fts f JOIN records r ON r.record_id=f.record_id WHERE text_fts MATCH ? ORDER BY r.record_id LIMIT ?',(expression,a.limit+1)).fetchall()
 if len(rows)>a.limit:print('Result truncated at requested limit.',file=sys.stderr)
 records=[json.loads(r[0]) for r in rows[:a.limit]]
 if a.format=='jsonl':
  for rec in records:print(json.dumps(rec,ensure_ascii=False))
 else:
  fields=['record_id','corpus','source_id','edition','title','language','date','source_sha256','source_url','license',a.layer]
  w=csv.DictWriter(sys.stdout,fieldnames=fields,extrasaction='ignore');w.writeheader()
  for rec in records:
   row={k:rec.get(k) for k in fields}
   for k,v in row.items():
    if isinstance(v,str) and v.startswith(('=','+','-','@','\t','\r')):row[k]="'"+v
   w.writerow(row)
 return 0

if __name__=='__main__':raise SystemExit(main())