textless_report.py

"""Human-review report for TMDB posters that appear to be mislabelled textless."""

from __future__ import annotations

from datetime import datetime, timezone
import fcntl
import logging
from pathlib import Path

import config as _cfg


logger = logging.getLogger(__name__)

_HEADER = (
    "# Posters Plus suspected fake textless posters\n"
    "# Generated by PP-OCR detection. Review manually before editing TMDB.\n"
    "# detected_at_utc\tmedia_type\ttmdb_id\tvote_count\ttmdb_page\timage_url\timage_path\n"
)


def report_fake_textless_poster(
    *,
    media_type: str,
    tmdb_id: str,
    image_path: str,
    vote_count: int | None,
) -> None:
    """Append one deduplicated poster to the human-review report."""
    if not _cfg.TEXTLESS_FAKE_REPORT:
        return

    normalized_type = "tv" if str(media_type).lower() == "tv" else "movie"
    normalized_id = str(tmdb_id).strip()
    normalized_path = "/" + str(image_path).strip().lstrip("/")
    if not normalized_id or normalized_path == "/":
        return

    report_path = Path(_cfg.TEXTLESS_FAKE_REPORT_PATH)
    try:
        report_path.parent.mkdir(parents=True, exist_ok=True)
        with report_path.open("a+", encoding="utf-8") as report:
            fcntl.flock(report.fileno(), fcntl.LOCK_EX)
            report.seek(0)
            contents = report.read()
            marker = f"\t{normalized_path}\n"
            if marker in contents:
                return

            if not contents:
                report.write(_HEADER)
            elif not contents.endswith("\n"):
                report.write("\n")

            detected_at = datetime.now(timezone.utc).isoformat(timespec="seconds")
            tmdb_page = (
                f"https://www.themoviedb.org/{normalized_type}/{normalized_id}"
                "/images/posters?image_language=xx"
            )
            image_url = f"https://image.tmdb.org/t/p/original{normalized_path}"
            votes = "" if vote_count is None else str(vote_count)
            report.write(
                f"{detected_at}\t{normalized_type}\t{normalized_id}\t{votes}\t"
                f"{tmdb_page}\t{image_url}\t{normalized_path}\n"
            )
            report.flush()
    except Exception as exc:
        logger.warning("Could not update fake-textless report: %s", exc)