push_log.py

"""Push log: record what was pushed, deduped, and filtered for analytics."""

from __future__ import annotations

import json
import logging
from datetime import UTC, datetime
from pathlib import Path

log = logging.getLogger(__name__)


def write_push_log(
    log_dir: str | Path,
    flow: str,
    items_pushed: list[dict],
    items_deduped: int,
    items_filtered: int,
    items_raw: int,
    alerts: list[dict],
) -> Path:
    """Append a push log entry as JSONL (one line per push cycle).

    Returns the log file path.
    """
    log_path = Path(log_dir) / "push_log.jsonl"
    log_path.parent.mkdir(parents=True, exist_ok=True)

    entry = {
        "ts": datetime.now(UTC).isoformat(),
        "flow": flow,
        "raw": items_raw,
        "after_dedup": items_raw - items_deduped,
        "deduped": items_deduped,
        "filtered_seen": items_filtered,
        "pushed": len(items_pushed),
        "items": items_pushed,
        "alerts": [a.get("msg", "") for a in alerts] if alerts else [],
    }

    with open(log_path, "a", encoding="utf-8") as f:
        f.write(json.dumps(entry, ensure_ascii=False) + "\n")

    log.info(
        f"push log: {log_path} (raw={items_raw} dedup=-{items_deduped} seen=-{items_filtered} pushed={len(items_pushed)})"
    )
    return log_path