[ie/ntv.ru] Swap metadata priority (#14939 )

Fix 5dde0d0c9f Authored by: seproDev
[ie/ntv.ru] Rework extractor (#14934 )
2025-12-16 06:05:41 +07:00 · 2025-11-08 11:41:17 +01:00 · 2025-11-08 10:51:21 +01:00 · 2025-11-08 18:30:43 +13:00
6 changed files with 218 additions and 73 deletions
--- a/test/test_networking.py
+++ b/test/test_networking.py
@@ -3,6 +3,7 @@
 # Allow direct execution
 import os
 import sys
+from unittest.mock import MagicMock

 import pytest

@@ -614,8 +615,11 @@ def test_source_address(self, handler):
    @pytest.mark.skip_handler('CurlCFFI', 'not supported by curl-cffi')
    def test_gzip_trailing_garbage(self, handler):
        with handler() as rh:
-            data = validate_and_send(rh, Request(f'http://localhost:{self.http_port}/trailing_garbage')).read().decode()
+            res = validate_and_send(rh, Request(f'http://localhost:{self.http_port}/trailing_garbage'))
+            data = res.read().decode()
            assert data == '<html><video src="/vid.mp4" /></html>'
+            # Should auto-close and mark the response adaptor as closed
+            assert res.closed

    @pytest.mark.skip_handler('CurlCFFI', 'not applicable to curl-cffi')
    @pytest.mark.skipif(not brotli, reason='brotli support is not installed')
@@ -627,6 +631,8 @@ def test_brotli(self, handler):
                    headers={'ytdl-encoding': 'br'}))
            assert res.headers.get('Content-Encoding') == 'br'
            assert res.read() == b'<html><video src="/vid.mp4" /></html>'
+            # Should auto-close and mark the response adaptor as closed
+            assert res.closed

    def test_deflate(self, handler):
        with handler() as rh:
@@ -636,6 +642,8 @@ def test_deflate(self, handler):
                    headers={'ytdl-encoding': 'deflate'}))
            assert res.headers.get('Content-Encoding') == 'deflate'
            assert res.read() == b'<html><video src="/vid.mp4" /></html>'
+            # Should auto-close and mark the response adaptor as closed
+            assert res.closed

    def test_gzip(self, handler):
        with handler() as rh:
@@ -645,6 +653,8 @@ def test_gzip(self, handler):
                    headers={'ytdl-encoding': 'gzip'}))
            assert res.headers.get('Content-Encoding') == 'gzip'
            assert res.read() == b'<html><video src="/vid.mp4" /></html>'
+            # Should auto-close and mark the response adaptor as closed
+            assert res.closed

    def test_multiple_encodings(self, handler):
        with handler() as rh:
@@ -655,6 +665,8 @@ def test_multiple_encodings(self, handler):
                        headers={'ytdl-encoding': pair}))
                assert res.headers.get('Content-Encoding') == pair
                assert res.read() == b'<html><video src="/vid.mp4" /></html>'
+                # Should auto-close and mark the response adaptor as closed
+                assert res.closed

    @pytest.mark.skip_handler('CurlCFFI', 'not supported by curl-cffi')
    def test_unsupported_encoding(self, handler):
@@ -665,6 +677,8 @@ def test_unsupported_encoding(self, handler):
                    headers={'ytdl-encoding': 'unsupported', 'Accept-Encoding': '*'}))
            assert res.headers.get('Content-Encoding') == 'unsupported'
            assert res.read() == b'raw'
+            # Should auto-close and mark the response adaptor as closed
+            assert res.closed

    def test_read(self, handler):
        with handler() as rh:
@@ -672,9 +686,13 @@ def test_read(self, handler):
                rh, Request(f'http://127.0.0.1:{self.http_port}/headers'))
            assert res.readable()
            assert res.read(1) == b'H'
+            # Ensure we don't close the adaptor yet
+            assert not res.closed
            assert res.read(3) == b'ost'
            assert res.read().decode().endswith('\n\n')
            assert res.read() == b''
+            # Should auto-close and mark the response adaptor as closed
+            assert res.closed

    def test_request_disable_proxy(self, handler):
        for proxy_proto in handler._SUPPORTED_PROXY_SCHEMES or ['http']:
@@ -875,11 +893,31 @@ def test_file_urls(self, handler):

        with handler(enable_file_urls=True) as rh:
            res = validate_and_send(rh, req)
-            assert res.read() == b'foobar'
-            res.close()
+            assert res.read(1) == b'f'
+            assert not res.fp.closed
+            assert res.read() == b'oobar'
+            # Should automatically close the underlying file object
+            assert res.fp.closed

        os.unlink(tf.name)

+    def test_data_uri_auto_close(self, handler):
+        with handler() as rh:
+            res = validate_and_send(rh, Request('data:text/plain,hello%20world'))
+            assert res.read() == b'hello world'
+            # Should automatically close the underlying file object
+            assert res.fp.closed
+            assert res.closed
+
+    def test_http_response_auto_close(self, handler):
+        with handler() as rh:
+            res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/gen_200'))
+            assert res.read() == b'<html></html>'
+            # Should automatically close the underlying file object in the HTTP Response
+            assert isinstance(res.fp, http.client.HTTPResponse)
+            assert res.fp.fp is None
+            assert res.closed
+
    def test_http_error_returns_content(self, handler):
        # urllib HTTPError will try close the underlying response if reference to the HTTPError object is lost
        def get_response():
@@ -1012,6 +1050,14 @@ def mock_close(*args, **kwargs):
        rh.close()
        assert called

+    def test_http_response_auto_close(self, handler):
+        with handler() as rh:
+            res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/gen_200'))
+            assert res.read() == b'<html></html>'
+            # Should automatically close the underlying file object in the HTTP Response
+            assert res.fp.closed
+            assert res.closed
+

@pytest.mark.parametrize('handler', ['CurlCFFI'], indirect=True)
 class TestCurlCFFIRequestHandler(TestRequestHandlerBase):
@@ -1177,6 +1223,14 @@ def close(self):
        assert res4.closed
        assert res4._buffer == b''

+    def test_http_response_auto_close(self, handler):
+        with handler() as rh:
+            res = validate_and_send(rh, Request(f'http://127.0.0.1:{self.http_port}/gen_200'))
+            assert res.read() == b'<html></html>'
+            # Should automatically close the underlying file object in the HTTP Response
+            assert res.fp.closed
+            assert res.closed
+

 def run_validation(handler, error, req, **handler_kwargs):
    with handler(**handler_kwargs) as rh:
@@ -2032,6 +2086,30 @@ def test_compat(self):
            assert res.info() is res.headers
            assert res.getheader('test') == res.get_header('test')

+    def test_auto_close(self):
+        # Should mark the response as closed if the underlying file is closed
+        class AutoCloseBytesIO(io.BytesIO):
+            def read(self, size=-1, /):
+                data = super().read(size)
+                self.close()
+                return data
+
+        fp = AutoCloseBytesIO(b'test')
+        res = Response(fp, url='test://', headers={}, status=200)
+        assert not res.closed
+        res.read()
+        assert res.closed
+
+    def test_close(self):
+        # Should not call close() on the underlying file when already closed
+        fp = MagicMock()
+        fp.closed = False
+        res = Response(fp, url='test://', headers={}, status=200)
+        res.close()
+        fp.closed = True
+        res.close()
+        assert fp.close.call_count == 1
+

 class TestImpersonateTarget:
    @pytest.mark.parametrize('target_str,expected', [
--- a/yt_dlp/extractor/ntvru.py
+++ b/yt_dlp/extractor/ntvru.py
@@ -1,17 +1,40 @@
 from .common import InfoExtractor
 from ..utils import (
    int_or_none,
-    strip_or_none,
+    parse_iso8601,
    unescapeHTML,
+    url_or_none,
    xpath_text,
 )
+from ..utils.traversal import traverse_obj


 class NTVRuIE(InfoExtractor):
    IE_NAME = 'ntv.ru'
-    _VALID_URL = r'https?://(?:www\.)?ntv\.ru/(?:[^/]+/)*(?P<id>[^/?#&]+)'
+    _VALID_URL = r'https?://(?:www\.)?ntv\.ru/(?:[^/#?]+/)*(?P<id>[^/?#&]+)'

    _TESTS = [{
+        # JSON Api is geo restricted
+        'url': 'https://www.ntv.ru/peredacha/svoya_igra/m58980/o818800',
+        'md5': '818962a1b52747d446db7cd5be43e142',
+        'info_dict': {
+            'id': '2520563',
+            'ext': 'mp4',
+            'title': 'Участники: Ирина Петрова, Сергей Коновалов, Кристина Кораблина',
+            'description': 'md5:fcbd21cd45238a940b95550f9e178e3e',
+            'thumbnail': r're:^http://.*\.jpg',
+            'duration': 2462,
+            'view_count': int,
+            'comment_count': int,
+            'tags': ['игры и игрушки'],
+            'timestamp': 1761821096,
+            'upload_date': '20251030',
+            'release_timestamp': 1761821096,
+            'release_date': '20251030',
+            'modified_timestamp': 1761821096,
+            'modified_date': '20251030',
+        },
+    }, {
        'url': 'http://www.ntv.ru/novosti/863142/',
        'md5': 'ba7ea172a91cb83eb734cad18c10e723',
        'info_dict': {
@@ -22,31 +45,35 @@ class NTVRuIE(InfoExtractor):
            'thumbnail': r're:^http://.*\.jpg',
            'duration': 136,
            'view_count': int,
+            'comment_count': int,
+            'tags': ['ВМС', 'захват', 'митинги', 'Севастополь', 'Украина'],
+            'timestamp': 1395222013,
+            'upload_date': '20140319',
+            'release_timestamp': 1395222013,
+            'release_date': '20140319',
+            'modified_timestamp': 1395222013,
+            'modified_date': '20140319',
        },
    }, {
-        'url': 'http://www.ntv.ru/video/novosti/750370/',
-        'md5': 'adecff79691b4d71e25220a191477124',
-        'info_dict': {
-            'id': '750370',
-            'ext': 'mp4',
-            'title': 'Родные пассажиров пропавшего Boeing не верят в трагический исход',
-            'description': 'Родные пассажиров пропавшего Boeing не верят в трагический исход',
-            'thumbnail': r're:^http://.*\.jpg',
-            'duration': 172,
-            'view_count': int,
-        },
-        'skip': '404 Not Found',
-    }, {
+        # Requires unescapeHTML
        'url': 'http://www.ntv.ru/peredacha/segodnya/m23700/o232416',
        'md5': '82dbd49b38e3af1d00df16acbeab260c',
        'info_dict': {
            'id': '747480',
            'ext': 'mp4',
-            'title': '«Сегодня». 21 марта 2014 года. 16:00',
-            'description': '«Сегодня». 21 марта 2014 года. 16:00',
+            'title': '"Сегодня". 21 марта 2014 года. 16:00 ',
+            'description': 'md5:bed80745ca72af557433195f51a02785',
            'thumbnail': r're:^http://.*\.jpg',
            'duration': 1496,
            'view_count': int,
+            'comment_count': int,
+            'tags': ['Брюссель', 'гражданство', 'ЕС', 'Крым', 'ОСАГО', 'саммит', 'санкции', 'события', 'чиновники', 'рейтинг'],
+            'timestamp': 1395406951,
+            'upload_date': '20140321',
+            'release_timestamp': 1395406951,
+            'release_date': '20140321',
+            'modified_timestamp': 1395406951,
+            'modified_date': '20140321',
        },
    }, {
        'url': 'https://www.ntv.ru/kino/Koma_film/m70281/o336036/video/',
@@ -54,11 +81,19 @@ class NTVRuIE(InfoExtractor):
        'info_dict': {
            'id': '1126480',
            'ext': 'mp4',
-            'title': 'Остросюжетный фильм «Кома»',
-            'description': 'Остросюжетный фильм «Кома»',
+            'title': 'Остросюжетный фильм "Кома"',
+            'description': 'md5:e79ffd0887425a0f05a58885c408d7d8',
            'thumbnail': r're:^http://.*\.jpg',
-            'duration': 5592,
+            'duration': 5608,
            'view_count': int,
+            'comment_count': int,
+            'tags': ['кино'],
+            'timestamp': 1432868572,
+            'upload_date': '20150529',
+            'release_timestamp': 1432868572,
+            'release_date': '20150529',
+            'modified_timestamp': 1432868572,
+            'modified_date': '20150529',
        },
    }, {
        'url': 'http://www.ntv.ru/serial/Delo_vrachey/m31760/o233916/',
@@ -66,11 +101,19 @@ class NTVRuIE(InfoExtractor):
        'info_dict': {
            'id': '751482',
            'ext': 'mp4',
-            'title': '«Дело врачей»: «Деревце жизни»',
-            'description': '«Дело врачей»: «Деревце жизни»',
+            'title': '"Дело врачей": "Деревце жизни"',
+            'description': 'md5:d6fbf9193f880f50d9cbfbcc954161c1',
            'thumbnail': r're:^http://.*\.jpg',
            'duration': 2590,
            'view_count': int,
+            'comment_count': int,
+            'tags': ['врачи', 'больницы'],
+            'timestamp': 1395882300,
+            'upload_date': '20140327',
+            'release_timestamp': 1395882300,
+            'release_date': '20140327',
+            'modified_timestamp': 1395882300,
+            'modified_date': '20140327',
        },
    }, {
        # Schemeless file URL
@@ -78,48 +121,26 @@ class NTVRuIE(InfoExtractor):
        'only_matching': True,
    }]

-    _VIDEO_ID_REGEXES = [
-        r'<meta property="og:url" content="https?://www\.ntv\.ru/video/(\d+)',
-        r'<meta property="og:video:(?:url|iframe)" content="https?://www\.ntv\.ru/embed/(\d+)',
-        r'<video embed=[^>]+><id>(\d+)</id>',
-        r'<video restriction[^>]+><key>(\d+)</key>',
-    ]
-
    def _real_extract(self, url):
-        video_id = self._match_id(url)
+        display_id = self._match_id(url)
+        webpage = self._download_webpage(url, display_id)

-        webpage = self._download_webpage(url, video_id)
-
-        video_url = self._og_search_property(
-            ('video', 'video:iframe'), webpage, default=None)
-        if video_url:
-            video_id = self._search_regex(
-                r'https?://(?:www\.)?ntv\.ru/video/(?:embed/)?(\d+)',
-                video_url, 'video id', default=None)
-
-        if not video_id:
-            video_id = self._html_search_regex(
-                self._VIDEO_ID_REGEXES, webpage, 'video id')
+        video_id = self._html_search_regex(
+            r'<meta property="ya:ovs:feed_url" content="https?://www\.ntv\.ru/(?:exp/)?video/(\d+)', webpage, 'video id')

        player = self._download_xml(
            f'http://www.ntv.ru/vi{video_id}/',
            video_id, 'Downloading video XML')

-        title = strip_or_none(unescapeHTML(xpath_text(player, './data/title', 'title', fatal=True)))
-
        video = player.find('./data/video')

        formats = []
        for format_id in ['', 'hi', 'webm']:
-            file_ = xpath_text(video, f'./{format_id}file')
-            if not file_:
+            video_url = url_or_none(xpath_text(video, f'./{format_id}file'))
+            if not video_url:
                continue
-            if file_.startswith('//'):
-                file_ = self._proto_relative_url(file_)
-            elif not file_.startswith('http'):
-                file_ = 'http://media.ntv.ru/vod/' + file_
            formats.append({
-                'url': file_,
+                'url': video_url,
                'filesize': int_or_none(xpath_text(video, f'./{format_id}size')),
            })
        hls_manifest = xpath_text(video, './playback/hls')
@@ -131,12 +152,28 @@ def _real_extract(self, url):
            formats.extend(self._extract_mpd_formats(
                dash_manifest, video_id, mpd_id='dash', fatal=False))

+        metadata = self._download_xml(
+            f'https://www.ntv.ru/exp/video/{video_id}', video_id, 'Downloading XML metadata', fatal=False)
+
        return {
-            'id': xpath_text(video, './id'),
-            'title': title,
-            'description': strip_or_none(unescapeHTML(xpath_text(player, './data/description'))),
-            'thumbnail': xpath_text(video, './splash'),
-            'duration': int_or_none(xpath_text(video, './totaltime')),
-            'view_count': int_or_none(xpath_text(video, './views')),
+            'id': video_id,
            'formats': formats,
+            **traverse_obj(player, {
+                'title': ('data/title/text()', ..., {str}, {unescapeHTML}, any),
+                'description': ('data/description/text()', ..., {str}, {unescapeHTML}, any),
+                'duration': ('data/video/totaltime/text()', ..., {int_or_none}, any),
+                'view_count': ('data/video/views/text()', ..., {int_or_none}, any),
+                'thumbnail': ('data/video/splash/text()', ..., {url_or_none}, any),
+            }),
+            **traverse_obj(metadata, {
+                'title': ('{*}title/text()', ..., {str}, {unescapeHTML}, any),
+                'description': ('{*}description/text()', ..., {str}, {unescapeHTML}, any),
+                'duration': ('{*}duration/text()', ..., {int_or_none}, any),
+                'timestamp': ('{*}create_date/text()', ..., {parse_iso8601}, any),
+                'release_timestamp': ('{*}upload_date/text()', ..., {parse_iso8601}, any),
+                'modified_timestamp': ('{*}modify_date/text()', ..., {parse_iso8601}, any),
+                'tags': ('{*}tag/text()', ..., {str}, {lambda x: x.split(',')}, ..., {str.strip}, filter),
+                'view_count': ('{*}stats/views_total/text()', ..., {int_or_none}, any),
+                'comment_count': ('{*}stats/comments/text()', ..., {int_or_none}, any),
+            }),
        }
--- a/yt_dlp/networking/_curlcffi.py
+++ b/yt_dlp/networking/_curlcffi.py
@@ -96,7 +96,10 @@ def __init__(self, response: curl_cffi.requests.Response):

    def read(self, amt=None):
        try:
-            return self.fp.read(amt)
+            res = self.fp.read(amt)
+            if self.fp.closed:
+                self.close()
+            return res
        except curl_cffi.requests.errors.RequestsError as e:
            if e.code == CurlECode.PARTIAL_FILE:
                content_length = e.response and int_or_none(e.response.headers.get('Content-Length'))
--- a/yt_dlp/networking/_requests.py
+++ b/yt_dlp/networking/_requests.py
@@ -119,17 +119,22 @@ def __init__(self, res: requests.models.Response):

        self._requests_response = res

+    def _real_read(self, amt: int | None = None) -> bytes:
+        # Work around issue with `.read(amt)` then `.read()`
+        # See: https://github.com/urllib3/urllib3/issues/3636
+        if amt is None:
+            # compat: py3.9: Python 3.9 preallocates the whole read buffer, read in chunks
+            read_chunk = functools.partial(self.fp.read, 1 << 20, decode_content=True)
+            return b''.join(iter(read_chunk, b''))
+        # Interact with urllib3 response directly.
+        return self.fp.read(amt, decode_content=True)
+
    def read(self, amt: int | None = None):
        try:
-            # Work around issue with `.read(amt)` then `.read()`
-            # See: https://github.com/urllib3/urllib3/issues/3636
-            if amt is None:
-                # compat: py3.9: Python 3.9 preallocates the whole read buffer, read in chunks
-                read_chunk = functools.partial(self.fp.read, 1 << 20, decode_content=True)
-                return b''.join(iter(read_chunk, b''))
-            # Interact with urllib3 response directly.
-            return self.fp.read(amt, decode_content=True)
-
+            data = self._real_read(amt)
+            if self.fp.closed:
+                self.close()
+            return data
        # See urllib3.response.HTTPResponse.read() for exceptions raised on read
        except urllib3.exceptions.SSLError as e:
            raise SSLError(cause=e) from e
--- a/yt_dlp/networking/_urllib.py
+++ b/yt_dlp/networking/_urllib.py
@@ -306,7 +306,25 @@ def __init__(self, res: http.client.HTTPResponse | urllib.response.addinfourl):

    def read(self, amt=None):
        try:
-            return self.fp.read(amt)
+            data = self.fp.read(amt)
+            underlying = getattr(self.fp, 'fp', None)
+            if isinstance(self.fp, http.client.HTTPResponse) and underlying is None:
+                # http.client.HTTPResponse automatically closes itself when fully read
+                self.close()
+            elif isinstance(self.fp, urllib.response.addinfourl) and underlying is not None:
+                # urllib's addinfourl does not close the underlying fp automatically when fully read
+                if isinstance(underlying, io.BytesIO):
+                    # data URLs or in-memory responses (e.g. gzip/deflate/brotli decoded)
+                    if underlying.tell() >= len(underlying.getbuffer()):
+                        self.close()
+                elif isinstance(underlying, io.BufferedReader) and amt is None:
+                    # file URLs.
+                    # XXX: this will not mark the response as closed if it was fully read with amt.
+                    self.close()
+            elif underlying is not None and underlying.closed:
+                # Catch-all for any cases where underlying file is closed
+                self.close()
+            return data
        except Exception as e:
            handle_response_read_exceptions(e)
            raise e
--- a/yt_dlp/networking/common.py
+++ b/yt_dlp/networking/common.py
@@ -554,12 +554,16 @@ def read(self, amt: int | None = None) -> bytes:
        # Expected errors raised here should be of type RequestError or subclasses.
        # Subclasses should redefine this method with more precise error handling.
        try:
-            return self.fp.read(amt)
+            res = self.fp.read(amt)
+            if self.fp.closed:
+                self.close()
+            return res
        except Exception as e:
            raise TransportError(cause=e) from e

    def close(self):
-        self.fp.close()
+        if not self.fp.closed:
+            self.fp.close()
        return super().close()

    def get_header(self, name, default=None):
Author	SHA1	Message	Date
sepro	a86eeaadf2	[ie/ntv.ru] Swap metadata priority (#14939 ) Fix `5dde0d0c9f` Authored by: seproDev	2025-11-08 11:41:17 +01:00
sepro	5dde0d0c9f	[ie/ntv.ru] Rework extractor (#14934 ) Closes #14929, Closes #14761 Authored by: seproDev, anlar Co-authored-by: Anton Larionov <11796525+anlar@users.noreply.github.com>	2025-11-08 10:51:21 +01:00
coletdjnz	5767fb4ab1	[networking] Ensure underlying file object is closed when fully read (#14935 ) Fixes https://github.com/yt-dlp/yt-dlp/issues/14891 Authored by: coletdjnz	2025-11-08 18:30:43 +13:00