"""Promote ORIGEM_MERCADORIA/CST_ICMS/CSOSN/CST_PIS/CST_COFINS from the local official MOC (Anexo I NF-e/NFC-e) PDF. No network fetch — this manual has no stable public direct-download URL either (session-gated portal), so the authoritative copy lives in sources/raw/bootstrap_v1/. """ import json from datetime import datetime, timezone from pathlib import Path from central.app.datasets.nfe_moc_tables import parse_moc_tables from central.app.datasets.service import DatasetService from central.app.db import SessionLocal SOURCE = Path("sources/raw/bootstrap_v1/moc7-anexo-i-leiaute-e-rv.pdf") SOURCE_URL = f"file://{SOURCE.name}" def main(): raw_bytes = SOURCE.read_bytes() payload = parse_moc_tables(raw_bytes) with SessionLocal() as session: dataset = DatasetService(session).import_moc_tables(payload, SOURCE_URL, datetime.now(timezone.utc)) from collections import Counter domains = Counter(e.domain for e in payload.entries) print(json.dumps({ "status": dataset.status, "dataset_id": dataset.id, "sha256": dataset.sha256, "record_count": dataset.record_count, "domains": dict(domains), }, indent=2)) if __name__ == "__main__": main()