#!/usr/bin/env python3
"""Conclude a board job when systemd fails before worker.py can run."""

import json
import state_store
import os
import re
import sys
from datetime import datetime, timezone
from pathlib import Path

SLUG_RE = re.compile(r"^[a-z0-9][a-z0-9-]{0,60}$")
JOB_ID_RE = re.compile(r"^[a-z0-9][a-z0-9-]{0,120}$")
MAX_ERROR_DETAIL = 240


def now_iso():
    return datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")


def clean_detail(value):
    value = re.sub(r"[\x00-\x1f\x7f]+", " ", str(value or ""))
    value = re.sub(r"\s+", " ", value).strip()
    return value[:MAX_ERROR_DETAIL] or "Systemd konnte den Worker nicht starten."


def load_json(path):
    try:
        value = json.loads(path.read_text(encoding="utf-8"))
    except (OSError, ValueError):
        return None
    return value if isinstance(value, dict) else None


def write_json(path, value):
    tmp = path.with_suffix(".tmp")
    tmp.write_text(json.dumps(value, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
    tmp.chmod(0o600)
    tmp.replace(path)


def candidate_jobs():
    candidates = []
    for env_name in ("BOARD_INBOX_DIR", "BOARD_PROCESSING_DIR"):
        directory = Path(os.environ[env_name])
        if not directory.is_dir():
            continue
        for path in sorted(directory.glob("*.json")):
            payload = load_json(path)
            if not payload:
                continue
            job_id = str(payload.get("job_id", ""))
            slug = str(payload.get("slug", ""))
            intent_path = str(payload.get("intent_path", ""))
            source = str(payload.get("source", "intent"))
            if (
                not JOB_ID_RE.fullmatch(job_id)
                or not SLUG_RE.fullmatch(slug)
                or source not in ("intent", "research")
                or not re.fullmatch(r"docs/(?:intent|research-intent)/[a-z0-9][a-z0-9-]{0,60}\.md", intent_path)
                or Path(intent_path).stem != slug
            ):
                continue
            candidates.append((path, payload, job_id, slug, source))
    return candidates


@state_store.atomic
def update_state(slug, job_id, stage, detail):
    state_path = Path(os.environ["BOARD_STATE_FILE"])
    state = state_store.load()
    cards = state.setdefault("cards", {})
    card = cards.setdefault(slug, {})
    if card.get("last_job_id") != job_id or card.get("last_run_status") in ("done", "no_changes"):
        return False
    card.update(
        {
            "last_job_id": job_id,
            "last_run_status": "failed",
            "last_run_at": now_iso(),
            "failure_stage": stage,
            "failure_detail": clean_detail(detail),
        }
    )
    state_store.save(state)
    return True


def main():
    # systemd runs ExecStopPost after successful services as well.  In that
    # case a newly queued job must never be mistaken for the completed run.
    if os.environ.get("SERVICE_RESULT") == "success":
        return 0
    stage = os.environ.get("BOARD_FAILURE_STAGE", "pre_start")
    if stage not in ("pre_start", "sandbox_setup", "worker"):
        stage = "pre_start"
    candidates = candidate_jobs()
    if len(candidates) != 1:
        print(
            f"board failure handler: expected exactly one valid queue candidate, found {len(candidates)}",
            file=sys.stderr,
        )
        return 0
    path, payload, job_id, slug, _source = candidates[0]
    service_result = os.environ.get("SERVICE_RESULT", "unknown")
    exit_code = os.environ.get("EXIT_CODE", "unknown")
    exit_status = os.environ.get("EXIT_STATUS", "unknown")
    detail = os.environ.get(
        "BOARD_FAILURE_DETAIL",
        f"Systemd konnte den Worker nicht starten ({service_result}, {exit_code} {exit_status}).",
    )
    if update_state(slug, job_id, stage, detail):
        failed = Path(os.environ["BOARD_FAILED_DIR"])
        failed.mkdir(parents=True, exist_ok=True, mode=0o700)
        payload.update({"failed_at": now_iso(), "failure_stage": stage, "error": clean_detail(detail)})
        write_json(path, payload)
        path.replace(failed / path.name)
    else:
        print("board failure handler: job is no longer the card's current run", file=sys.stderr)
    return 0


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