X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/cc52de43568d8cd58c7e2ef4e5cecf609da28a9c..54a63e80af82791d2f0985bd0176bb182963fd5f:/test/test_InfoExtractor.py diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 9b6672a1d..31e8f8244 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -1,27 +1,32 @@ #!/usr/bin/env python3 -from __future__ import unicode_literals - # Allow direct execution -import io import os import sys import unittest + sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) -from test.helper import FakeYDL, expect_dict, expect_value, http_server_port -from yt_dlp.compat import compat_etree_fromstring, compat_http_server -from yt_dlp.extractor.common import InfoExtractor -from yt_dlp.extractor import YoutubeIE, get_info_extractor -from yt_dlp.utils import encode_data_uri, strip_jsonp, ExtractorError, RegexNotFoundError + +import http.server import threading +from test.helper import FakeYDL, expect_dict, expect_value, http_server_port +from yt_dlp.compat import compat_etree_fromstring +from yt_dlp.extractor import YoutubeIE, get_info_extractor +from yt_dlp.extractor.common import InfoExtractor +from yt_dlp.utils import ( + ExtractorError, + RegexNotFoundError, + encode_data_uri, + strip_jsonp, +) TEAPOT_RESPONSE_STATUS = 418 TEAPOT_RESPONSE_BODY = "

418 I'm a teapot

" -class InfoExtractorTestRequestHandler(compat_http_server.BaseHTTPRequestHandler): +class InfoExtractorTestRequestHandler(http.server.BaseHTTPRequestHandler): def log_message(self, format, *args): pass @@ -35,13 +40,15 @@ def do_GET(self): assert False -class TestIE(InfoExtractor): - pass +class DummyIE(InfoExtractor): + def _sort_formats(self, formats, field_preference=[]): + self._downloader.sort_formats( + {'formats': formats, '_format_sort_fields': field_preference}) class TestInfoExtractor(unittest.TestCase): def setUp(self): - self.ie = TestIE(FakeYDL()) + self.ie = DummyIE(FakeYDL()) def test_ie_key(self): self.assertEqual(get_info_extractor(YoutubeIE.ie_key()), YoutubeIE) @@ -62,6 +69,7 @@ def test_opengraph(self): + ''' self.assertEqual(ie._og_search_title(html), 'Foo') self.assertEqual(ie._og_search_description(html), 'Some video\'s description ') @@ -74,6 +82,7 @@ def test_opengraph(self): self.assertEqual(ie._og_search_property(('test0', 'test1'), html), 'foo > < bar') self.assertRaises(RegexNotFoundError, ie._og_search_property, 'test0', html, None, fatal=True) self.assertRaises(RegexNotFoundError, ie._og_search_property, ('test0', 'test00'), html, None, fatal=True) + self.assertEqual(ie._og_search_property('test4', html), 'unquoted-value') def test_html_search_meta(self): ie = self.ie @@ -99,10 +108,10 @@ def test_html_search_meta(self): self.assertRaises(RegexNotFoundError, ie._html_search_meta, ('z', 'x'), html, None, fatal=True) def test_search_json_ld_realworld(self): - # https://github.com/ytdl-org/youtube-dl/issues/23306 - expect_dict( - self, - self.ie._search_json_ld(r'''''', None), - { - 'title': '1 On 1 With Kleio', - 'description': 'Kleio Valentien', - 'url': 'https://gvideo.eporner.com/xN49A1cT3eB/xN49A1cT3eB.mp4', - 'timestamp': 1449347075, - 'duration': 743.0, - 'view_count': 1120958, - 'width': 1920, - 'height': 1080, - }) + ''', + { + 'title': '1 On 1 With Kleio', + 'description': 'Kleio Valentien', + 'url': 'https://gvideo.eporner.com/xN49A1cT3eB/xN49A1cT3eB.mp4', + 'timestamp': 1449347075, + 'duration': 743.0, + 'view_count': 1120958, + 'width': 1920, + 'height': 1080, + }, + {}, + ), + ( + r'''''', + { + 'timestamp': 1636523400, + 'title': 'md5:91fe569e952e4d146485740ae927662b', + }, + {'expected_type': 'NewsArticle'}, + ), + ( + r''' + ''', + { + 'chapters': [ + {'title': 'Explosie Turnhout', 'start_time': 70, 'end_time': 440}, + {'title': 'Jaarwisseling', 'start_time': 440, 'end_time': 1179}, + {'title': 'Natuurbranden Colorado', 'start_time': 1179, 'end_time': 1263}, + {'title': 'Klimaatverandering', 'start_time': 1263, 'end_time': 1367}, + {'title': 'Zacht weer', 'start_time': 1367, 'end_time': 1383}, + {'title': 'Financiële balans', 'start_time': 1383, 'end_time': 1484}, + {'title': 'Club Brugge', 'start_time': 1484, 'end_time': 1575}, + {'title': 'Mentale gezondheid bij topsporters', 'start_time': 1575, 'end_time': 1728}, + {'title': 'Olympische Winterspelen', 'start_time': 1728, 'end_time': 1873}, + {'title': 'Sober oudjaar in Nederland', 'start_time': 1873, 'end_time': 2079.23}, + ], + 'title': 'Het journaal - Aflevering 365 (Seizoen 2021)', + }, {}, + ), + ( + # test multiple thumbnails in a list + r''' +''', + { + 'thumbnails': [{'url': 'https://www.rainews.it/cropgd/640x360/dl/img/2021/12/30/1640886376927_GettyImages.jpg'}], + }, + {}, + ), + ( + # test single thumbnail + r''' +''', + { + 'thumbnails': [{'url': 'https://www.rainews.it/cropgd/640x360/dl/img/2021/12/30/1640886376927_GettyImages.jpg'}], + }, + {}, + ), + ] + for html, expected_dict, search_json_ld_kwargs in _TESTS: + expect_dict( + self, + self.ie._search_json_ld(html, None, **search_json_ld_kwargs), + expected_dict, + ) def test_download_json(self): uri = encode_data_uri(b'{"foo": "blah"}', 'application/json') @@ -203,7 +366,7 @@ def test_parse_html5_media_entries(self): 'height': 740, 'tbr': 1500, }], - 'thumbnail': '//pics.r18.com/digital/amateur/mgmr105/mgmr105jp.jpg' + 'thumbnail': '//pics.r18.com/digital/amateur/mgmr105/mgmr105jp.jpg', }) # from https://www.csfd.cz/ @@ -256,9 +419,9 @@ def test_parse_html5_media_entries(self): 'height': 1080, }], 'subtitles': { - 'cs': [{'url': 'https://video.csfd.cz/files/subtitles/163/344/163344115_4c388b.srt'}] + 'cs': [{'url': 'https://video.csfd.cz/files/subtitles/163/344/163344115_4c388b.srt'}], }, - 'thumbnail': 'https://img.csfd.cz/files/images/film/video/preview/163/344/163344118_748d20.png?h360' + 'thumbnail': 'https://img.csfd.cz/files/images/film/video/preview/163/344/163344118_748d20.png?h360', }) # from https://tamasha.com/v/Kkdjw @@ -289,7 +452,7 @@ def test_parse_html5_media_entries(self): 'ext': 'mp4', 'format_id': '144p', 'height': 144, - }] + }], }) # from https://www.directvnow.com @@ -307,7 +470,7 @@ def test_parse_html5_media_entries(self): 'formats': [{ 'ext': 'mp4', 'url': 'https://cdn.directv.com/content/dam/dtv/prod/website_directvnow-international/videos/DTVN_hdr_HBO_v3.mp4', - }] + }], }) # from https://www.directvnow.com @@ -325,7 +488,7 @@ def test_parse_html5_media_entries(self): 'formats': [{ 'url': 'https://cdn.directv.com/content/dam/dtv/prod/website_directvnow-international/videos/DTVN_hdr_HBO_v3.mp4', 'ext': 'mp4', - }] + }], }) # from https://www.klarna.com/uk/ @@ -346,6 +509,24 @@ def test_parse_html5_media_entries(self): }], }) + # from https://0000.studio/ + # with type attribute but without extension in URL + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://0000.studio', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://d1ggyt9m8pwf3g.cloudfront.net/protected/ap-northeast-1:1864af40-28d5-492b-b739-b32314b1a527/archive/clip/838db6a7-8973-4cd6-840d-8517e4093c92', + 'ext': 'mp4', + }], + }) + def test_extract_jwplayer_data_realworld(self): # from http://www.suffolk.edu/sjc/ expect_dict( @@ -366,8 +547,8 @@ def test_extract_jwplayer_data_realworld(self): 'id': 'XEgvuql4', 'formats': [{ 'url': 'rtmp://192.138.214.154/live/sjclive', - 'ext': 'flv' - }] + 'ext': 'flv', + }], }) # from https://www.pornoxo.com/videos/7564/striptease-from-sexy-secretary/ @@ -407,8 +588,8 @@ def test_extract_jwplayer_data_realworld(self): 'thumbnail': 'https://t03.vipstreamservice.com/thumbs/pxo-full/2009-12/14/a4b2157147afe5efa93ce1978e0265289c193874e02597.flv-full-13.jpg', 'formats': [{ 'url': 'https://cdn.pornoxo.com/key=MF+oEbaxqTKb50P-w9G3nA,end=1489689259,ip=104.199.146.27/ip=104.199.146.27/speed=6573765/buffer=3.0/2009-12/4b2157147afe5efa93ce1978e0265289c193874e02597.flv', - 'ext': 'flv' - }] + 'ext': 'flv', + }], }) # from http://www.indiedb.com/games/king-machine/videos @@ -429,12 +610,12 @@ def test_extract_jwplayer_data_realworld(self): 'formats': [{ 'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode_mp4/king-machine-trailer.mp4', 'height': 360, - 'ext': 'mp4' + 'ext': 'mp4', }, { 'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode720p_mp4/king-machine-trailer.mp4', 'height': 720, - 'ext': 'mp4' - }] + 'ext': 'mp4', + }], }) def test_parse_m3u8_formats(self): @@ -685,7 +866,7 @@ def test_parse_m3u8_formats(self): 'height': 1080, 'vcodec': 'avc1.64002a', }], - {} + {}, ), ( 'bipbop_16x9', @@ -736,8 +917,6 @@ def test_parse_m3u8_formats(self): 'acodec': 'mp4a.40.2', 'video_ext': 'mp4', 'audio_ext': 'none', - 'vbr': 263.851, - 'abr': 0, }, { 'format_id': '577', 'format_index': None, @@ -755,8 +934,6 @@ def test_parse_m3u8_formats(self): 'acodec': 'mp4a.40.2', 'video_ext': 'mp4', 'audio_ext': 'none', - 'vbr': 577.61, - 'abr': 0, }, { 'format_id': '915', 'format_index': None, @@ -774,8 +951,6 @@ def test_parse_m3u8_formats(self): 'acodec': 'mp4a.40.2', 'video_ext': 'mp4', 'audio_ext': 'none', - 'vbr': 915.905, - 'abr': 0, }, { 'format_id': '1030', 'format_index': None, @@ -793,8 +968,6 @@ def test_parse_m3u8_formats(self): 'acodec': 'mp4a.40.2', 'video_ext': 'mp4', 'audio_ext': 'none', - 'vbr': 1030.138, - 'abr': 0, }, { 'format_id': '1924', 'format_index': None, @@ -812,53 +985,50 @@ def test_parse_m3u8_formats(self): 'acodec': 'mp4a.40.2', 'video_ext': 'mp4', 'audio_ext': 'none', - 'vbr': 1924.009, - 'abr': 0, }], { 'en': [{ 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/eng/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }, { 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/eng_forced/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }], 'fr': [{ 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/fra/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }, { 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/fra_forced/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }], 'es': [{ 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/spa/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }, { 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/spa_forced/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }], 'ja': [{ 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/jpn/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }, { 'url': 'https://devstreaming-cdn.apple.com/videos/streaming/examples/bipbop_16x9/subtitles/jpn_forced/prog_index.m3u8', 'ext': 'vtt', - 'protocol': 'm3u8_native' + 'protocol': 'm3u8_native', }], - } + }, ), ] for m3u8_file, m3u8_url, expected_formats, expected_subs in _TEST_CASES: - with io.open('./test/testdata/m3u8/%s.m3u8' % m3u8_file, - mode='r', encoding='utf-8') as f: + with open(f'./test/testdata/m3u8/{m3u8_file}.m3u8', encoding='utf-8') as f: formats, subs = self.ie._parse_m3u8_formats_and_subtitles( f.read(), m3u8_url, ext='mp4') self.ie._sort_formats(formats) @@ -1196,17 +1366,16 @@ def test_parse_mpd_formats(self): 'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/manifest.mpd', 'fragment_base_url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/dash/', 'protocol': 'http_dash_segments', - } - ] + }, + ], }, - ) + ), ] for mpd_file, mpd_url, mpd_base_url, expected_formats, expected_subtitles in _TEST_CASES: - with io.open('./test/testdata/mpd/%s.mpd' % mpd_file, - mode='r', encoding='utf-8') as f: + with open(f'./test/testdata/mpd/{mpd_file}.mpd', encoding='utf-8') as f: formats, subtitles = self.ie._parse_mpd_formats_and_subtitles( - compat_etree_fromstring(f.read().encode('utf-8')), + compat_etree_fromstring(f.read().encode()), mpd_base_url=mpd_base_url, mpd_url=mpd_url) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) @@ -1227,6 +1396,7 @@ def test_parse_ism_formats(self): 'vcodec': 'none', 'acodec': 'AACL', 'protocol': 'ism', + 'audio_channels': 2, '_download_params': { 'stream_type': 'audio', 'duration': 8880746666, @@ -1238,11 +1408,8 @@ def test_parse_ism_formats(self): 'sampling_rate': 48000, 'channels': 2, 'bits_per_sample': 16, - 'nal_unit_length_field': 4 + 'nal_unit_length_field': 4, }, - 'audio_ext': 'isma', - 'video_ext': 'none', - 'abr': 128, }, { 'format_id': 'video-100', 'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest', @@ -1264,11 +1431,8 @@ def test_parse_ism_formats(self): 'codec_private_data': '00000001674D401FDA0544EFFC2D002CBC40000003004000000C03C60CA80000000168EF32C8', 'channels': 2, 'bits_per_sample': 16, - 'nal_unit_length_field': 4 + 'nal_unit_length_field': 4, }, - 'video_ext': 'ismv', - 'audio_ext': 'none', - 'vbr': 100, }, { 'format_id': 'video-326', 'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest', @@ -1290,11 +1454,8 @@ def test_parse_ism_formats(self): 'codec_private_data': '00000001674D401FDA0241FE23FFC3BC83BA44000003000400000300C03C60CA800000000168EF32C8', 'channels': 2, 'bits_per_sample': 16, - 'nal_unit_length_field': 4 + 'nal_unit_length_field': 4, }, - 'video_ext': 'ismv', - 'audio_ext': 'none', - 'vbr': 326, }, { 'format_id': 'video-698', 'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest', @@ -1316,11 +1477,8 @@ def test_parse_ism_formats(self): 'codec_private_data': '00000001674D401FDA0350BFB97FF06AF06AD1000003000100000300300F1832A00000000168EF32C8', 'channels': 2, 'bits_per_sample': 16, - 'nal_unit_length_field': 4 + 'nal_unit_length_field': 4, }, - 'video_ext': 'ismv', - 'audio_ext': 'none', - 'vbr': 698, }, { 'format_id': 'video-1493', 'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest', @@ -1342,11 +1500,8 @@ def test_parse_ism_formats(self): 'codec_private_data': '00000001674D401FDA011C3DE6FFF0D890D871000003000100000300300F1832A00000000168EF32C8', 'channels': 2, 'bits_per_sample': 16, - 'nal_unit_length_field': 4 + 'nal_unit_length_field': 4, }, - 'video_ext': 'ismv', - 'audio_ext': 'none', - 'vbr': 1493, }, { 'format_id': 'video-4482', 'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest', @@ -1368,11 +1523,8 @@ def test_parse_ism_formats(self): 'codec_private_data': '00000001674D401FDA01A816F97FFC1ABC1AB440000003004000000C03C60CA80000000168EF32C8', 'channels': 2, 'bits_per_sample': 16, - 'nal_unit_length_field': 4 + 'nal_unit_length_field': 4, }, - 'video_ext': 'ismv', - 'audio_ext': 'none', - 'vbr': 4482, }], { 'eng': [ @@ -1386,19 +1538,277 @@ def test_parse_ism_formats(self): 'duration': 8880746666, 'timescale': 10000000, 'fourcc': 'TTML', - 'codec_private_data': '' - } - } - ] + 'codec_private_data': '', + }, + }, + ], }, ), + ( + 'ec-3_test', + 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + [{ + 'format_id': 'audio_deu-127', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'isma', + 'tbr': 127, + 'asr': 48000, + 'vcodec': 'none', + 'acodec': 'AACL', + 'protocol': 'ism', + 'language': 'deu', + 'audio_channels': 2, + '_download_params': { + 'stream_type': 'audio', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 0, + 'height': 0, + 'fourcc': 'AACL', + 'language': 'deu', + 'codec_private_data': '1190', + 'sampling_rate': 48000, + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'audio_deu_1-224', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'isma', + 'tbr': 224, + 'asr': 48000, + 'vcodec': 'none', + 'acodec': 'EC-3', + 'protocol': 'ism', + 'language': 'deu', + 'audio_channels': 6, + '_download_params': { + 'stream_type': 'audio', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 0, + 'height': 0, + 'fourcc': 'EC-3', + 'language': 'deu', + 'codec_private_data': '00063F000000AF87FBA7022DFB42A4D405CD93843BDD0700200F00', + 'sampling_rate': 48000, + 'channels': 6, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-23', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 384, + 'height': 216, + 'tbr': 23, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 384, + 'height': 216, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '000000016742C00CDB06077E5C05A808080A00000300020000030009C0C02EE0177CC6300F142AE00000000168CA8DC8', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-403', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 400, + 'height': 224, + 'tbr': 403, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 400, + 'height': 224, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4014E98323B602D4040405000003000100000300320F1429380000000168EAECF2', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-680', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 640, + 'height': 360, + 'tbr': 680, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 640, + 'height': 360, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D401EE981405FF2E02D4040405000000300100000030320F162D3800000000168EAECF2', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-1253', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 640, + 'height': 360, + 'tbr': 1253, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'vbr': 1253, + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 640, + 'height': 360, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D401EE981405FF2E02D4040405000000300100000030320F162D3800000000168EAECF2', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-2121', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 768, + 'height': 432, + 'tbr': 2121, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 768, + 'height': 432, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D401EECA0601BD80B50101014000003000400000300C83C58B6580000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-3275', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 1280, + 'height': 720, + 'tbr': 3275, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 1280, + 'height': 720, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4020ECA02802DD80B501010140000003004000000C83C60C65800000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-5300', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 1920, + 'height': 1080, + 'tbr': 5300, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 1920, + 'height': 1080, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4028ECA03C0113F2E02D4040405000000300100000030320F18319600000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }, { + 'format_id': 'video_deu-8079', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 1920, + 'height': 1080, + 'tbr': 8079, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + 'language': 'deu', + '_download_params': { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 1920, + 'height': 1080, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4028ECA03C0113F2E02D4040405000000300100000030320F18319600000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4, + }, + }], + {}, + ), ] for ism_file, ism_url, expected_formats, expected_subtitles in _TEST_CASES: - with io.open('./test/testdata/ism/%s.Manifest' % ism_file, - mode='r', encoding='utf-8') as f: + with open(f'./test/testdata/ism/{ism_file}.Manifest', encoding='utf-8') as f: formats, subtitles = self.ie._parse_ism_formats_and_subtitles( - compat_etree_fromstring(f.read().encode('utf-8')), ism_url=ism_url) + compat_etree_fromstring(f.read().encode()), ism_url=ism_url) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) expect_value(self, subtitles, expected_subtitles, None) @@ -1417,15 +1827,14 @@ def test_parse_f4m_formats(self): 'tbr': 2148, 'width': 1280, 'height': 720, - }] + }], ), ] for f4m_file, f4m_url, expected_formats in _TEST_CASES: - with io.open('./test/testdata/f4m/%s.f4m' % f4m_file, - mode='r', encoding='utf-8') as f: + with open(f'./test/testdata/f4m/{f4m_file}.f4m', encoding='utf-8') as f: formats = self.ie._parse_f4m_formats( - compat_etree_fromstring(f.read().encode('utf-8')), + compat_etree_fromstring(f.read().encode()), f4m_url, None) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) @@ -1464,16 +1873,15 @@ def test_parse_xspf(self): }, { 'manifest_url': 'https://example.org/src/foo_xspf.xspf', 'url': 'https://example.com/track3.mp3', - }] - }] + }], + }], ), ] for xspf_file, xspf_url, expected_entries in _TEST_CASES: - with io.open('./test/testdata/xspf/%s.xspf' % xspf_file, - mode='r', encoding='utf-8') as f: + with open(f'./test/testdata/xspf/{xspf_file}.xspf', encoding='utf-8') as f: entries = self.ie._parse_xspf( - compat_etree_fromstring(f.read().encode('utf-8')), + compat_etree_fromstring(f.read().encode()), xspf_file, xspf_url=xspf_url, xspf_base_url=xspf_url) expect_value(self, entries, expected_entries, None) for i in range(len(entries)): @@ -1486,7 +1894,7 @@ def test_response_with_expected_status_returns_content(self): # or the underlying `_download_webpage_handle` returning no content # when a response matches `expected_status`. - httpd = compat_http_server.HTTPServer( + httpd = http.server.HTTPServer( ('127.0.0.1', 0), InfoExtractorTestRequestHandler) port = http_server_port(httpd) server_thread = threading.Thread(target=httpd.serve_forever) @@ -1494,10 +1902,19 @@ def test_response_with_expected_status_returns_content(self): server_thread.start() (content, urlh) = self.ie._download_webpage_handle( - 'http://127.0.0.1:%d/teapot' % port, None, + f'http://127.0.0.1:{port}/teapot', None, expected_status=TEAPOT_RESPONSE_STATUS) self.assertEqual(content, TEAPOT_RESPONSE_BODY) + def test_search_nextjs_data(self): + data = '' + self.assertEqual(self.ie._search_nextjs_data(data, None), {'props': {}}) + self.assertEqual(self.ie._search_nextjs_data('', None, fatal=False), {}) + self.assertEqual(self.ie._search_nextjs_data('', None, default=None), None) + self.assertEqual(self.ie._search_nextjs_data('', None, default={}), {}) + with self.assertWarns(DeprecationWarning): + self.assertEqual(self.ie._search_nextjs_data('', None, default='{}'), {}) + if __name__ == '__main__': unittest.main()