GTI_GERENCIADOR_FISCAL/fiscal/scripts/generate_dry_run_plan.py

"""Generate the dry-run ERP write plan artifact for the 20-product sample.

Calls the running fiscal-hub /v1/classify/batch (read-only classification,
no Firebird access, no dataset creation) and turns each decision into
field-level write proposals via central.app.dry_run_planner. Never issues
an ERP write.
"""
import argparse
import json
import sys
import urllib.request
from pathlib import Path

ROOT = Path(__file__).resolve().parents[1]
sys.path.insert(0, str(ROOT / "central"))

from app.dry_run_planner import build_field_proposals, mva_evidence, split_by_eligibility  # noqa: E402


def classify_batch(hub_url, token, products):
    payload = {"products": [{"current_ncm": p["CLASSIFICACAO_FISCAL"], "description": p["PRODUTO"]} for p in products]}
    request = urllib.request.Request(
        f"{hub_url}/v1/classify/batch",
        data=json.dumps(payload, ensure_ascii=True).encode(),
        headers={"Content-Type": "application/json", "Authorization": f"Bearer {token}"},
        method="POST",
    )
    with urllib.request.urlopen(request) as response:
        return json.load(response)["results"]


def main():
    parser = argparse.ArgumentParser()
    parser.add_argument("--sample", default=str(ROOT / "docs" / "SAMPLE_PRODUCTS_V1.json"))
    parser.add_argument("--hub-url", default="http://localhost:8000")
    parser.add_argument("--token", required=True)
    parser.add_argument("--output", default=str(ROOT / "docs" / "DRY_RUN_PLAN_V1.json"))
    parser.add_argument("--artifact-name", default="DRY_RUN_PLAN_V1")
    parser.add_argument("--firebird-current", default=None, help="JSON file with fresh CODIGO/IPI/CEST read from Firebird (read-only). Omit to keep current values unknown.")
    args = parser.parse_args()

    sample = json.loads(Path(args.sample).read_text(encoding="utf-8"))
    products = sample["sample_without_ncm"]
    decisions = classify_batch(args.hub_url, args.token, products)

    fresh_by_code = {}
    if args.firebird_current:
        for row in json.loads(Path(args.firebird_current).read_text(encoding="utf-8")):
            fresh_by_code[row["CODIGO"]] = row

    field_proposals, mva_records = [], []
    for product, decision in zip(products, decisions):
        code = product["CODIGO"]
        fresh = fresh_by_code.get(code)
        if fresh is not None:
            current = {"ipi": fresh.get("IPI"), "cest": fresh.get("CEST"), "ex_tipi": fresh.get("EX_TIPI")}
        else:
            current = {}
            if "CEST" in product:
                current["cest"] = product["CEST"]
        field_proposals.extend(build_field_proposals(code, decision, current))
        mva_records.append(mva_evidence(code, decision))

    split = split_by_eligibility(field_proposals)
    ipi_rows = [p for p in field_proposals if p["field"] == "IPI"]
    ex_rows = [p for p in field_proposals if p["field"] == "EX_TIPI"]
    cest_rows = [p for p in field_proposals if p["field"] == "CEST"]
    ipi_metrics = {
        "ipi_numeric_proposals": sum(1 for p in ipi_rows if p["status"] in ("SAME_VALUE", "CHANGE_REQUIRED", "CURRENT_UNKNOWN")),
        "ipi_non_numeric_status": sum(1 for p in ipi_rows if p["status"] == "NON_NUMERIC_IPI_STATUS"),
        "ipi_source_blank": sum(1 for p in ipi_rows if p["status"] == "SOURCE_BLANK"),
        "ipi_same_value": sum(1 for p in ipi_rows if p["status"] == "SAME_VALUE"),
        "ipi_change_required": sum(1 for p in ipi_rows if p["status"] == "CHANGE_REQUIRED"),
        "ipi_write_eligible": sum(1 for p in ipi_rows if p["write_eligible"]),
        # every emitted EX_TIPI row implies TIPI had at least one EX-specific entry for that NCM.
        "ex_tipi_db_match_products": len(ex_rows),
        "ex_tipi_concrete_proposals": sum(1 for p in ex_rows if p["status"] in ("SAME_VALUE", "CHANGE_REQUIRED", "CURRENT_UNKNOWN")),
        "ex_tipi_candidates": sum(1 for p in ex_rows if p["status"] == "CANDIDATE"),
        "ex_tipi_write_eligible": sum(1 for p in ex_rows if p["write_eligible"]),
    }
    cest_metrics = {
        "cest_db_match_products": sum(1 for p in cest_rows if p["status"] != "NO_CEST_MAPPING_FOUND" and p["reason_code"] != "NO_CEST_MAPPING_FOUND_FOR_NCM"),
        "cest_confirmed": sum(1 for p in cest_rows if p["status"] in ("SAME_VALUE", "CHANGE_REQUIRED", "CURRENT_UNKNOWN")),
        "cest_candidates": sum(1 for p in cest_rows if p["status"] == "CANDIDATE"),
        "cest_unresolved": sum(1 for p in cest_rows if p["status"] == "UNRESOLVED"),
        "cest_write_eligible": sum(1 for p in cest_rows if p["write_eligible"]),
    }

    artifact = {
        "artifact": args.artifact_name,
        "firebird_read": "OK" if fresh_by_code else "BLOCKED_EXTERNAL_MISSING_CREDENTIALS",
        "firebird_writes": 0,
        "products_planned": len(products),
        "current_value_source": (
            f"live read-only SELECT from C000025 via {args.firebird_current}"
            if fresh_by_code else
            "docs/SAMPLE_PRODUCTS_V1.json (CEST) — C000025.IPI has no authorized offline source; FB_USER/FB_PASSWORD unavailable in this environment"
        ),
        "field_proposals": field_proposals,
        "mva_evidence": mva_records,
        "ipi_metrics": ipi_metrics,
        "cest_metrics": cest_metrics,
        "write_eligible_count": len(split["eligible"]),
        "write_blocked_count": len(split["blocked"]),
    }
    Path(args.output).write_text(json.dumps(artifact, indent=2, ensure_ascii=False), encoding="utf-8")
    print(json.dumps({
        "products": len(products),
        "field_proposals": len(field_proposals),
        "write_eligible": len(split["eligible"]),
        "write_blocked": len(split["blocked"]),
        "output": args.output,
    }, indent=2))


if __name__ == "__main__":
    main()