X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/af014acd27e0b471d5903630847eabb26437b46c..9865587cf7b579793ba274067c3e44065d5ff77d:/test/test_InfoExtractor.py diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index f18a823..71f6608 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -9,11 +9,30 @@ 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 -from youtube_dl.compat import compat_etree_fromstring +from test.helper import FakeYDL, expect_dict, expect_value, http_server_port +from youtube_dl.compat import compat_etree_fromstring, compat_http_server from youtube_dl.extractor.common import InfoExtractor from youtube_dl.extractor import YoutubeIE, get_info_extractor from youtube_dl.utils import encode_data_uri, strip_jsonp, ExtractorError, RegexNotFoundError +import threading + + +TEAPOT_RESPONSE_STATUS = 418 +TEAPOT_RESPONSE_BODY = "

418 I'm a teapot

" + + +class InfoExtractorTestRequestHandler(compat_http_server.BaseHTTPRequestHandler): + def log_message(self, format, *args): + pass + + def do_GET(self): + if self.path == '/teapot': + self.send_response(TEAPOT_RESPONSE_STATUS) + self.send_header('Content-Type', 'text/html; charset=utf-8') + self.end_headers() + self.wfile.write(TEAPOT_RESPONSE_BODY.encode()) + else: + assert False class TestIE(InfoExtractor): @@ -42,6 +61,7 @@ class TestInfoExtractor(unittest.TestCase): + ''' self.assertEqual(ie._og_search_title(html), 'Foo') self.assertEqual(ie._og_search_description(html), 'Some video\'s description ') @@ -50,6 +70,7 @@ class TestInfoExtractor(unittest.TestCase): self.assertEqual(ie._og_search_property('foobar', html), 'Foo') self.assertEqual(ie._og_search_property('test1', html), 'foo > < bar') self.assertEqual(ie._og_search_property('test2', html), 'foo >//< bar') + self.assertEqual(ie._og_search_property('test3', html), 'Ill-formatted opengraph') 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) @@ -86,6 +107,184 @@ class TestInfoExtractor(unittest.TestCase): self.assertRaises(ExtractorError, self.ie._download_json, uri, None) self.assertEqual(self.ie._download_json(uri, None, fatal=False), None) + def test_parse_html5_media_entries(self): + # from https://www.r18.com/ + # with kpbs in label + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.r18.com/', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://awscc3001.r18.com/litevideo/freepv/m/mgm/mgmr105/mgmr105_sm_w.mp4', + 'ext': 'mp4', + 'format_id': '300kbps', + 'height': 240, + 'tbr': 300, + }, { + 'url': 'https://awscc3001.r18.com/litevideo/freepv/m/mgm/mgmr105/mgmr105_dm_w.mp4', + 'ext': 'mp4', + 'format_id': '1000kbps', + 'height': 480, + 'tbr': 1000, + }, { + 'url': 'https://awscc3001.r18.com/litevideo/freepv/m/mgm/mgmr105/mgmr105_dmb_w.mp4', + 'ext': 'mp4', + 'format_id': '1500kbps', + 'height': 740, + 'tbr': 1500, + }], + 'thumbnail': '//pics.r18.com/digital/amateur/mgmr105/mgmr105jp.jpg' + }) + + # from https://www.csfd.cz/ + # with width and height + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.csfd.cz/', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327358_eac647.mp4', + 'ext': 'mp4', + 'width': 640, + 'height': 360, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327360_3d2646.mp4', + 'ext': 'mp4', + 'width': 1280, + 'height': 720, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327356_91f258.mp4', + 'ext': 'mp4', + 'width': 1920, + 'height': 1080, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327359_962b4a.webm', + 'ext': 'webm', + 'width': 640, + 'height': 360, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327361_6feee0.webm', + 'ext': 'webm', + 'width': 1280, + 'height': 720, + }, { + 'url': 'https://video.csfd.cz/files/videos/157/750/157750813/163327357_8ab472.webm', + 'ext': 'webm', + 'width': 1920, + 'height': 1080, + }], + 'subtitles': { + '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' + }) + + # from https://tamasha.com/v/Kkdjw + # with height in label + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://tamasha.com/v/Kkdjw', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://s-v2.tamasha.com/statics/videos_file/19/8f/Kkdjw_198feff8577d0057536e905cce1fb61438dd64e0_n_240.mp4', + }, { + 'url': 'https://s-v2.tamasha.com/statics/videos_file/19/8f/Kkdjw_198feff8577d0057536e905cce1fb61438dd64e0_n_240.mp4', + 'ext': 'mp4', + 'format_id': '240p', + 'height': 240, + }, { + 'url': 'https://s-v2.tamasha.com/statics/videos_file/20/00/Kkdjw_200041c66f657fc967db464d156eafbc1ed9fe6f_n_144.mp4', + 'ext': 'mp4', + 'format_id': '144p', + 'height': 144, + }] + }) + + # from https://www.directvnow.com + # with data-src + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.directvnow.com', + r''' + + ''', None)[0], + { + '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 + # with data-src + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.directvnow.com', + r''' + + ''', None)[0], + { + '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/ + # with data-video-src + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://www.directvnow.com', + r''' + + ''', None)[0], + { + 'formats': [{ + 'url': 'https://www.klarna.com/uk/wp-content/uploads/sites/11/2019/01/KL062_Smooth3_0_DogWalking_5s_920x080_.mp4', + 'ext': 'mp4', + }], + }) + def test_extract_jwplayer_data_realworld(self): # from http://www.suffolk.edu/sjc/ expect_dict( @@ -180,7 +379,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ def test_parse_m3u8_formats(self): _TEST_CASES = [ ( - # https://github.com/rg3/youtube-dl/issues/11507 + # https://github.com/ytdl-org/youtube-dl/issues/11507 # http://pluzz.francetv.fr/videos/le_ministere.html 'pluzz_francetv_11507', 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/master.m3u8?caption=2017%2F16%2F156589847-1492488987.m3u8%3Afra%3AFrancais&audiotrack=0%3Afra%3AFrancais', @@ -242,7 +441,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ }] ), ( - # https://github.com/rg3/youtube-dl/issues/11995 + # https://github.com/ytdl-org/youtube-dl/issues/11995 # http://teamcoco.com/video/clueless-gamer-super-bowl-for-honor 'teamcoco_11995', 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', @@ -316,7 +515,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ }] ), ( - # https://github.com/rg3/youtube-dl/issues/12211 + # https://github.com/ytdl-org/youtube-dl/issues/12211 # http://video.toggle.sg/en/series/whoopie-s-world/ep3/478601 'toggle_mobile_12211', 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', @@ -478,7 +677,64 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'width': 1280, 'height': 720, }] - ) + ), + ( + # https://github.com/ytdl-org/youtube-dl/issues/18923 + # https://www.ted.com/talks/boris_hesser_a_grassroots_healthcare_revolution_in_africa + 'ted_18923', + 'http://hls.ted.com/talks/31241.m3u8', + [{ + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/audio/600k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '600k-Audio', + 'vcodec': 'none', + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/audio/600k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '68', + 'vcodec': 'none', + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/64k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '163', + 'acodec': 'none', + 'width': 320, + 'height': 180, + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/180k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '481', + 'acodec': 'none', + 'width': 512, + 'height': 288, + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/320k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '769', + 'acodec': 'none', + 'width': 512, + 'height': 288, + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/450k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '984', + 'acodec': 'none', + 'width': 512, + 'height': 288, + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/600k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '1255', + 'acodec': 'none', + 'width': 640, + 'height': 360, + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/950k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '1693', + 'acodec': 'none', + 'width': 853, + 'height': 480, + }, { + 'url': 'http://hls.ted.com/videos/BorisHesser_2018S/video/1500k.m3u8?nobumpers=true&uniqueId=76011e2b', + 'format_id': '2462', + 'acodec': 'none', + 'width': 1280, + 'height': 720, + }] + ), ] for m3u8_file, m3u8_url, expected_formats in _TEST_CASES: @@ -492,10 +748,22 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ def test_parse_mpd_formats(self): _TEST_CASES = [ ( - # https://github.com/rg3/youtube-dl/issues/13919 + # https://github.com/ytdl-org/youtube-dl/issues/13919 + # Also tests duplicate representation ids, see + # https://github.com/ytdl-org/youtube-dl/issues/15111 'float_duration', - 'http://unknown/manifest.mpd', + 'http://unknown/manifest.mpd', # mpd_url + None, # mpd_base_url [{ + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'm4a', + 'format_id': '318597', + 'format_note': 'DASH audio', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'none', + 'tbr': 61.587, + }, { 'manifest_url': 'http://unknown/manifest.mpd', 'ext': 'mp4', 'format_id': '318597', @@ -562,18 +830,242 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'width': 1920, 'height': 1080, }] - ), + ), ( + # https://github.com/ytdl-org/youtube-dl/pull/14844 + 'urls_only', + 'http://unknown/manifest.mpd', # mpd_url + None, # mpd_base_url + [{ + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_144p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 200, + 'width': 256, + 'height': 144, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_240p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 400, + 'width': 424, + 'height': 240, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_360p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 800, + 'width': 640, + 'height': 360, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_480p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 1200, + 'width': 856, + 'height': 480, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_576p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 1600, + 'width': 1024, + 'height': 576, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_720p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 2400, + 'width': 1280, + 'height': 720, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': 'h264_aac_1080p_m4s', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc3.42c01e', + 'tbr': 4400, + 'width': 1920, + 'height': 1080, + }] + ), ( + # https://github.com/ytdl-org/youtube-dl/issues/20346 + # Media considered unfragmented even though it contains + # Initialization tag + 'unfragmented', + 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', # mpd_url + 'https://v.redd.it/hw1x7rcg7zl21', # mpd_base_url + [{ + 'url': 'https://v.redd.it/hw1x7rcg7zl21/audio', + 'manifest_url': 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', + 'ext': 'm4a', + 'format_id': 'AUDIO-1', + 'format_note': 'DASH audio', + 'container': 'm4a_dash', + 'acodec': 'mp4a.40.2', + 'vcodec': 'none', + 'tbr': 129.87, + 'asr': 48000, + + }, { + 'url': 'https://v.redd.it/hw1x7rcg7zl21/DASH_240', + 'manifest_url': 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', + 'ext': 'mp4', + 'format_id': 'VIDEO-2', + 'format_note': 'DASH video', + 'container': 'mp4_dash', + 'acodec': 'none', + 'vcodec': 'avc1.4d401e', + 'tbr': 608.0, + 'width': 240, + 'height': 240, + 'fps': 30, + }, { + 'url': 'https://v.redd.it/hw1x7rcg7zl21/DASH_360', + 'manifest_url': 'https://v.redd.it/hw1x7rcg7zl21/DASHPlaylist.mpd', + 'ext': 'mp4', + 'format_id': 'VIDEO-1', + 'format_note': 'DASH video', + 'container': 'mp4_dash', + 'acodec': 'none', + 'vcodec': 'avc1.4d401e', + 'tbr': 804.261, + 'width': 360, + 'height': 360, + 'fps': 30, + }] + ) ] - for mpd_file, mpd_url, expected_formats in _TEST_CASES: + for mpd_file, mpd_url, mpd_base_url, expected_formats in _TEST_CASES: with io.open('./test/testdata/mpd/%s.mpd' % mpd_file, mode='r', encoding='utf-8') as f: formats = self.ie._parse_mpd_formats( compat_etree_fromstring(f.read().encode('utf-8')), - mpd_url=mpd_url) + mpd_base_url=mpd_base_url, mpd_url=mpd_url) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) + def test_parse_f4m_formats(self): + _TEST_CASES = [ + ( + # https://github.com/ytdl-org/youtube-dl/issues/14660 + 'custom_base_url', + 'http://api.new.livestream.com/accounts/6115179/events/6764928/videos/144884262.f4m', + [{ + 'manifest_url': 'http://api.new.livestream.com/accounts/6115179/events/6764928/videos/144884262.f4m', + 'ext': 'flv', + 'format_id': '2148', + 'protocol': 'f4m', + '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: + formats = self.ie._parse_f4m_formats( + compat_etree_fromstring(f.read().encode('utf-8')), + f4m_url, None) + self.ie._sort_formats(formats) + expect_value(self, formats, expected_formats, None) + + def test_parse_xspf(self): + _TEST_CASES = [ + ( + 'foo_xspf', + 'https://example.org/src/foo_xspf.xspf', + [{ + 'id': 'foo_xspf', + 'title': 'Pandemonium', + 'description': 'Visit http://bigbrother404.bandcamp.com', + 'duration': 202.416, + 'formats': [{ + 'manifest_url': 'https://example.org/src/foo_xspf.xspf', + 'url': 'https://example.org/src/cd1/track%201.mp3', + }], + }, { + 'id': 'foo_xspf', + 'title': 'Final Cartridge (Nichico Twelve Remix)', + 'description': 'Visit http://bigbrother404.bandcamp.com', + 'duration': 255.857, + 'formats': [{ + 'manifest_url': 'https://example.org/src/foo_xspf.xspf', + 'url': 'https://example.org/%E3%83%88%E3%83%A9%E3%83%83%E3%82%AF%E3%80%80%EF%BC%92.mp3', + }], + }, { + 'id': 'foo_xspf', + 'title': 'Rebuilding Nightingale', + 'description': 'Visit http://bigbrother404.bandcamp.com', + 'duration': 287.915, + 'formats': [{ + 'manifest_url': 'https://example.org/src/foo_xspf.xspf', + 'url': 'https://example.org/src/track3.mp3', + }, { + '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: + entries = self.ie._parse_xspf( + compat_etree_fromstring(f.read().encode('utf-8')), + xspf_file, xspf_url=xspf_url, xspf_base_url=xspf_url) + expect_value(self, entries, expected_entries, None) + for i in range(len(entries)): + expect_dict(self, entries[i], expected_entries[i]) + + def test_response_with_expected_status_returns_content(self): + # Checks for mitigations against the effects of + # that affect Python 3.4.1+, which + # manifest as `_download_webpage`, `_download_xml`, `_download_json`, + # or the underlying `_download_webpage_handle` returning no content + # when a response matches `expected_status`. + + httpd = compat_http_server.HTTPServer( + ('127.0.0.1', 0), InfoExtractorTestRequestHandler) + port = http_server_port(httpd) + server_thread = threading.Thread(target=httpd.serve_forever) + server_thread.daemon = True + server_thread.start() + + (content, urlh) = self.ie._download_webpage_handle( + 'http://127.0.0.1:%d/teapot' % port, None, + expected_status=TEAPOT_RESPONSE_STATUS) + self.assertEqual(content, TEAPOT_RESPONSE_BODY) + if __name__ == '__main__': unittest.main()