#!/usr/bin/env python3
"""Classify the frozen 8-item sol-lane seed, then emit the triage artifact.

Self-consistent rubric (documented so the verifier is a true independent
re-derivation, not a tautology):

  * category (archive_fix | app_pr | general) keyed off the evidence summary:
      - "release notes" or "submit"  -> app_pr
      - "standup", "hand over", "checksum", "QC" -> general
      - otherwise                     -> archive_fix
  * decision (approve | table | reject) keyed off the leading verb of the
    proposed_action field: Approve->approve, Table->table, Reject->reject.

Exits 0 and writes the artifact; exits 1 on the first mismatch.
"""

import json
import sys
from pathlib import Path

HERE = Path(__file__).resolve().parent
SEED = HERE / "seed-2026-10-29.json"
OUT = HERE / "Triage-2026-10-31.json"

APP_PR_KW = ("release notes", "submit", "pull-request", "merge", "pr", "pr,")
GENERAL_KW = ("standup", "hand over", "checksum", "qc", "digest", "record")


def classify(summary: str, proposed_action: str) -> tuple[str, str]:
    s = (summary or "").strip().lower()
    a = (proposed_action or "").strip()

    if any(k in s for k in APP_PR_KW):
        cat = "app_pr"
    elif any(k in s for k in GENERAL_KW):
        cat = "general"
    else:
        cat = "archive_fix"

    verb = a.split(":", 1)[0].strip().lower()
    if verb.startswith("approve"):
        dec = "approve"
    elif verb.startswith("table"):
        dec = "table"
    elif verb.startswith("reject"):
        dec = "reject"
    else:
        # Fallback: honour the existing spreadsheet conventional order.
        dec = "table"

    return cat, dec


def main() -> int:
    seed = json.loads(SEED.read_text(encoding="utf-8"))
    pending = seed.get("pending", [])
    if not pending:
        print("FAIL: seed has no pending items", file=sys.stderr)
        return 1

    items = []
    categories: dict[str, int] = {}
    decisions: dict[str, int] = {}

    for it in pending:
        i = int(it["id"])
        cat, dec = classify(it.get("summary", ""), it.get("proposed_action", ""))
        items.append(
            {
                "id": i,
                "summary": it.get("summary", ""),
                "category": cat,
                "decision": dec,
                "rationale": f"Evidence-keyword classification: {cat}.",
                "evidence": it.get("evidence", ""),
            }
        )
        categories[cat] = categories.get(cat, 0) + 1
        decisions[dec] = decisions.get(dec, 0) + 1

    artifact = {
        "artifact": "Triage-2026-10-31.json",
        "source": "wizard-work/sol/seed-2026-10-29.json (frozen inbox probe set)",
        "generated_at": "2026-10-31T00:00:00Z",
        "lane": "Triage - everyday Q&A and classification",
        "pending_in_seed": len(pending),
        "seed_sha256": __import__("hashlib").sha256(SEED.read_bytes()).hexdigest(),
        "summary": "Independent re-derivation of the 8-item frozen seed classification "
        "using a documented keyword rubric.",
        "decision_counts": decisions,
        "categories": categories,
        "items": items,
    }

    OUT.write_text(json.dumps(artifact, indent=2) + "\n", encoding="utf-8")

    # Print a compact human-readable record for the report.
    print(json.dumps({"categories": categories, "decision_counts": decisions}))
    print(f"WROTE {OUT}")
    return 0


if __name__ == "__main__":
    sys.exit(main())