Function parse_feed_metadata_stream

scripts/add_feed_to_tiny.py:193–281

(
    stream: BinaryIO,
    feed_url: str,
    max_bytes: int = MAX_FEED_BYTES,
)

def parse_feed_metadata_stream(
    stream: BinaryIO,
    feed_url: str,
    max_bytes: int = MAX_FEED_BYTES,
) -> FeedMetadata:
    parser = ET.XMLPullParser(events=("start", "end"))
    path: List[str] = []
    root_name = ""
    title = ""
    html_url = ""
    atom_fallback_html_url = ""
    bytes_read = 0

    while True:
        chunk = stream.read(READ_CHUNK_SIZE)
        if not chunk:
            break
        bytes_read += len(chunk)
        if bytes_read > max_bytes:
            raise ValueError(f"Feed payload is too large (> {max_bytes} bytes)")

        try:
            parser.feed(chunk)
        except ET.ParseError as exc:
            raise ValueError(f"RSS/Atom XML parse failed: {exc}") from exc

        for event, elem in parser.read_events():
            local_name = strip_namespace(elem.tag)

            if event == "start":
                path.append(local_name)

                if len(path) == 1:
                    root_name = local_name
                    if root_name not in {"rss", "feed", "rdf"}:
                        raise ValueError(f"Unsupported feed root tag: {root_name}")
                elif root_name == "feed" and len(path) == 2:
                    if local_name == "entry":
                        return _build_feed_metadata(
                            root_name,
                            title,
                            html_url or atom_fallback_html_url,
                            feed_url,
                        )
                    if local_name == "link":
                        href = normalize_url(elem.attrib.get("href", ""))
                        if href:
                            rel = (elem.attrib.get("rel", "alternate") or "alternate").strip().lower()
                            if rel in {"alternate", ""}:
                                html_url = href
                            elif not atom_fallback_html_url:
                                atom_fallback_html_url = href
                            if title and html_url:
                                return _build_feed_metadata(root_name, title, html_url, feed_url)
                continue

            if root_name in {"rss", "rdf"}:
                if len(path) == 3 and path[1] == "channel":