Function run_sync

scripts/opml_sync.py:551–662

(
    tiny_path: Path,
    full_path: Path,
    mode: str,
    fallback_category: str,
    timeout: float,
    retries: int,
    workers: int,
    state_file: Path,
    delete_threshold: int,
    max_probe_bytes: int,
    checker: Optional[Callable[[str], Any]] = None,
)

def run_sync(
    tiny_path: Path,
    full_path: Path,
    mode: str,
    fallback_category: str,
    timeout: float,
    retries: int,
    workers: int,
    state_file: Path,
    delete_threshold: int,
    max_probe_bytes: int,
    checker: Optional[Callable[[str], Any]] = None,
) -> Tuple[SyncStats, bool]:
    tiny_path = Path(tiny_path)
    full_path = Path(full_path)
    state_file = Path(state_file)

    if mode not in {"check", "apply"}:
        raise ValueError(f"Unsupported mode: {mode}")

    tiny_original = tiny_path.read_bytes()
    full_original = full_path.read_bytes()
    state_original = state_file.read_bytes() if state_file.exists() else b""

    tiny_tree = ET.parse(tiny_path)
    full_tree = ET.parse(full_path)

    tiny_body = get_body(tiny_tree, tiny_path)
    full_body = get_body(full_tree, full_path)

    stats = SyncStats()
    stats.tiny_links_before = len(collect_rss_urls(tiny_body))
    stats.full_links_before = len(collect_rss_urls(full_body))

    all_urls = set(collect_rss_urls(tiny_body)) | set(collect_rss_urls(full_body))
    check_fn = checker or HttpFeedChecker(
        timeout=timeout,
        retries=retries,
        user_agent=DEFAULT_USER_AGENT,
        max_probe_bytes=max_probe_bytes,
    )
    check_results = check_urls_parallel(all_urls, check_fn, workers=workers)

    stats.checked_urls = len(check_results)
    stats.alive_urls = sum(1 for result in check_results.values() if result.alive)
    stats.dead_urls = stats.checked_urls - stats.alive_urls

    previous_state = load_health_state(state_file)
    next_state_payload, removable_urls, hard_fail_urls, transient_fail_urls = build_next_health_state(
        urls=all_urls,
        check_results=check_results,
        previous_state=previous_state,
        delete_threshold=max(1, delete_threshold),
    )
    stats.hard_fail_urls = hard_fail_urls
    stats.transient_fail_urls = transient_fail_urls

    (