From 7f3590c43b8ae5cdba3c63e35e786083e3589485 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Fri, 17 Mar 2017 00:00:01 +0700 Subject: [test_InfoExtractor] Add some realworld tests for _extract_jwplayer_data --- test/test_InfoExtractor.py | 93 +++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 92 insertions(+), 1 deletion(-) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 437c7270e..881197afb 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -8,7 +8,7 @@ import sys import unittest sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) -from test.helper import FakeYDL +from test.helper import FakeYDL, expect_dict 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 @@ -84,6 +84,97 @@ 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_extract_jwplayer_data_realworld(self): + # from http://www.suffolk.edu/sjc/ + expect_dict( + self, + self.ie._extract_jwplayer_data(r''' + + ''', None, require_title=False), + { + 'id': 'XEgvuql4', + 'formats': [{ + 'url': 'rtmp://192.138.214.154/live/sjclive', + 'ext': 'flv' + }] + }) + + # from https://www.pornoxo.com/videos/7564/striptease-from-sexy-secretary/ + expect_dict( + self, + self.ie._extract_jwplayer_data(r''' + + ''', 'dummy', require_title=False), + { + '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' + }] + }) + + # from http://www.indiedb.com/games/king-machine/videos + expect_dict( + self, + self.ie._extract_jwplayer_data(r''' + + ''', 'dummy'), + { + 'title': 'king machine trailer 1', + 'thumbnail': 'http://media.indiedb.com/cache/images/games/1/50/49678/thumb_620x2000/king-machine-trailer.mp4.jpg', + 'formats': [{ + 'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode_mp4/king-machine-trailer.mp4', + 'height': 360, + 'ext': 'mp4' + }, { + 'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode720p_mp4/king-machine-trailer.mp4', + 'height': 720, + 'ext': 'mp4' + }] + }) + if __name__ == '__main__': unittest.main() -- cgit 1.4.1 From cb2520802d7b8efdc71d6d97aeca984b5f878716 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Sat, 22 Apr 2017 07:01:00 +0700 Subject: [extractor/common] Improve m3u8 extraction (closes #12211) * Extract m3u8 parsing to separate method * Improve rendition groups extraction * Build stream name according stream GROUP-ID * Ignore reference to AUDIO group without URI when stream has no CODECS + Add test coverage for parsing m3u8 from #11507, #11995, #12211 and twitch vod --- .gitignore | 1 - test/test_InfoExtractor.py | 308 ++++++++++++++++++++++++++- test/testdata/m3u8/pluzz_francetv_11507.m3u8 | 14 ++ test/testdata/m3u8/teamcoco_11995.m3u8 | 16 ++ test/testdata/m3u8/toggle_mobile_12211.m3u8 | 13 ++ test/testdata/m3u8/twitch_vod.m3u8 | 20 ++ youtube_dl/extractor/common.py | 154 +++++++++----- 7 files changed, 466 insertions(+), 60 deletions(-) create mode 100644 test/testdata/m3u8/pluzz_francetv_11507.m3u8 create mode 100644 test/testdata/m3u8/teamcoco_11995.m3u8 create mode 100644 test/testdata/m3u8/toggle_mobile_12211.m3u8 create mode 100644 test/testdata/m3u8/twitch_vod.m3u8 (limited to 'test/test_InfoExtractor.py') diff --git a/.gitignore b/.gitignore index 9ce4b5e2d..bef084baa 100644 --- a/.gitignore +++ b/.gitignore @@ -36,7 +36,6 @@ updates_key.pem *.swf *.part *.swp -test/testdata test/local_parameters.json .tox youtube-dl.zsh diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 881197afb..57205f6ae 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -3,12 +3,13 @@ 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 +from test.helper import FakeYDL, expect_dict, expect_value 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 @@ -175,6 +176,311 @@ 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 + # 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', + [{ + 'url': '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', + 'ext': 'mp4', + 'format_id': 'meta', + 'format_note': 'Quality selection URL', + 'protocol': 'm3u8', + 'preference': -100, + 'resolution': 'multiple' + }, { + 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0', + 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0', + 'ext': 'mp4', + 'format_id': '180', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.66.30', + 'tbr': 180, + 'width': 256, + 'height': 144, + }, { + 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_1_av.m3u8?null=0', + 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_1_av.m3u8?null=0', + 'ext': 'mp4', + 'format_id': '303', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.66.30', + 'tbr': 303, + 'width': 320, + 'height': 180, + }, { + 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_2_av.m3u8?null=0', + 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_2_av.m3u8?null=0', + 'ext': 'mp4', + 'format_id': '575', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.66.30', + 'tbr': 575, + 'width': 512, + 'height': 288, + }, { + 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_3_av.m3u8?null=0', + 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_3_av.m3u8?null=0', + 'ext': 'mp4', + 'format_id': '831', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.77.30', + 'tbr': 831, + 'width': 704, + 'height': 396, + }, { + 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_4_av.m3u8?null=0', + 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_4_av.m3u8?null=0', + 'ext': 'mp4', + 'protocol': 'm3u8', + 'format_id': '1467', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.77.30', + 'tbr': 1467, + 'width': 1024, + 'height': 576, + }] + ), + ( + # https://github.com/rg3/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', + [{ + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', + 'ext': 'mp4', + 'format_id': 'meta', + 'format_note': 'Quality selection URL', + 'protocol': 'm3u8', + 'preference': -100, + 'resolution': 'multiple', + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-160k_v4.m3u8', + 'ext': 'mp4', + 'format_id': 'audio-0-Default', + 'protocol': 'm3u8', + 'vcodec': 'none', + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8', + 'ext': 'mp4', + 'format_id': 'audio-1-Default', + 'protocol': 'm3u8', + 'vcodec': 'none', + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8', + 'ext': 'mp4', + 'format_id': '71', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.5', + 'vcodec': 'none', + 'tbr': 71, + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', + 'ext': 'mp4', + 'format_id': '413', + 'protocol': 'm3u8', + 'acodec': 'none', + 'vcodec': 'avc1.42001e', + 'tbr': 413, + 'width': 400, + 'height': 224, + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', + 'ext': 'mp4', + 'format_id': '522', + 'protocol': 'm3u8', + 'acodec': 'none', + 'vcodec': 'avc1.42001e', + 'tbr': 522, + 'width': 400, + 'height': 224, + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-1m_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-1m_v4.m3u8', + 'ext': 'mp4', + 'format_id': '1205', + 'protocol': 'm3u8', + 'acodec': 'none', + 'vcodec': 'avc1.4d001e', + 'tbr': 1205, + 'width': 640, + 'height': 360, + }, { + 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-2m_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-2m_v4.m3u8', + 'ext': 'mp4', + 'format_id': '2374', + 'protocol': 'm3u8', + 'acodec': 'none', + 'vcodec': 'avc1.4d001f', + 'tbr': 2374, + 'width': 1024, + 'height': 576, + }] + ), + ( + # https://github.com/rg3/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', + [{ + 'url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', + 'ext': 'mp4', + 'format_id': 'meta', + 'format_note': 'Quality selection URL', + 'protocol': 'm3u8', + 'preference': -100, + 'resolution': 'multiple' + }, { + 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_sa2ntrdg/name/a.mp4/index.m3u8', + 'ext': 'mp4', + 'format_id': 'audio-English', + 'protocol': 'm3u8', + 'language': 'eng', + 'vcodec': 'none', + }, { + 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_r7y0nitg/name/a.mp4/index.m3u8', + 'ext': 'mp4', + 'format_id': 'audio-Undefined', + 'protocol': 'm3u8', + 'language': 'und', + 'vcodec': 'none', + }, { + 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_qlk9hlzr/name/a.mp4/index.m3u8', + 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_qlk9hlzr/name/a.mp4/index.m3u8', + 'ext': 'mp4', + 'format_id': '155', + 'protocol': 'm3u8', + 'tbr': 155, + 'width': 320, + 'height': 180, + }, { + 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_oefackmi/name/a.mp4/index.m3u8', + 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_oefackmi/name/a.mp4/index.m3u8', + 'ext': 'mp4', + 'format_id': '502', + 'protocol': 'm3u8', + 'tbr': 502, + 'width': 480, + 'height': 270, + }, { + 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_vyg9pj7k/name/a.mp4/index.m3u8', + 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_vyg9pj7k/name/a.mp4/index.m3u8', + 'ext': 'mp4', + 'format_id': '827', + 'protocol': 'm3u8', + 'tbr': 827, + 'width': 640, + 'height': 360, + }, { + 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_50n4psvx/name/a.mp4/index.m3u8', + 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_50n4psvx/name/a.mp4/index.m3u8', + 'ext': 'mp4', + 'format_id': '1396', + 'protocol': 'm3u8', + 'tbr': 1396, + 'width': 854, + 'height': 480, + }] + ), + ( + # http://www.twitch.tv/riotgames/v/6528877 + 'twitch_vod', + 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', + [{ + 'url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', + 'ext': 'mp4', + 'format_id': 'meta', + 'format_note': 'Quality selection URL', + 'protocol': 'm3u8', + 'preference': -100, + 'resolution': 'multiple' + }, { + 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8', + 'ext': 'mp4', + 'format_id': 'Audio Only', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'none', + 'tbr': 182, + }, { + 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8', + 'ext': 'mp4', + 'format_id': 'Mobile', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.42C00D', + 'tbr': 280, + 'width': 400, + 'height': 226, + }, { + 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/low/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/low/index-muted-HM49I092CC.m3u8', + 'ext': 'mp4', + 'format_id': 'Low', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.42C01E', + 'tbr': 628, + 'width': 640, + 'height': 360, + }, { + 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/medium/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/medium/index-muted-HM49I092CC.m3u8', + 'ext': 'mp4', + 'format_id': 'Medium', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.42C01E', + 'tbr': 893, + 'width': 852, + 'height': 480, + }, { + 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/high/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/high/index-muted-HM49I092CC.m3u8', + 'ext': 'mp4', + 'format_id': 'High', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.42C01F', + 'tbr': 1603, + 'width': 1280, + 'height': 720, + }, { + 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/chunked/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/chunked/index-muted-HM49I092CC.m3u8', + 'ext': 'mp4', + 'format_id': 'Source', + 'protocol': 'm3u8', + 'acodec': 'mp4a.40.2', + 'vcodec': 'avc1.100.31', + 'tbr': 3214, + 'width': 1280, + 'height': 720, + }] + ) + ] + + for m3u8_file, m3u8_url, expected_formats in _TEST_CASES: + with io.open('./test/testdata/m3u8/%s.m3u8' % m3u8_file, + mode='r', encoding='utf-8') as f: + formats = self.ie._parse_m3u8_formats( + f.read(), m3u8_url, ext='mp4') + self.ie._sort_formats(formats) + expect_value(self, formats, expected_formats, None) + if __name__ == '__main__': unittest.main() diff --git a/test/testdata/m3u8/pluzz_francetv_11507.m3u8 b/test/testdata/m3u8/pluzz_francetv_11507.m3u8 new file mode 100644 index 000000000..0809f5aa0 --- /dev/null +++ b/test/testdata/m3u8/pluzz_francetv_11507.m3u8 @@ -0,0 +1,14 @@ +#EXTM3U + #EXT-X-VERSION:5 + #EXT-X-MEDIA:TYPE=SUBTITLES,GROUP-ID="subs",NAME="Francais",DEFAULT=NO,FORCED=NO,URI="http://replayftv-pmd.francetv.fr/subtitles/2017/16/156589847-1492488987.m3u8",LANGUAGE="fra" + #EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="aac",LANGUAGE="fra",NAME="Francais",DEFAULT=YES, AUTOSELECT=YES +#EXT-X-STREAM-INF:SUBTITLES="subs",AUDIO="aac",PROGRAM-ID=1,BANDWIDTH=180000,RESOLUTION=256x144,CODECS="avc1.66.30, mp4a.40.2" +http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0 +#EXT-X-STREAM-INF:SUBTITLES="subs",AUDIO="aac",PROGRAM-ID=1,BANDWIDTH=303000,RESOLUTION=320x180,CODECS="avc1.66.30, mp4a.40.2" +http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_1_av.m3u8?null=0 +#EXT-X-STREAM-INF:SUBTITLES="subs",AUDIO="aac",PROGRAM-ID=1,BANDWIDTH=575000,RESOLUTION=512x288,CODECS="avc1.66.30, mp4a.40.2" +http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_2_av.m3u8?null=0 +#EXT-X-STREAM-INF:SUBTITLES="subs",AUDIO="aac",PROGRAM-ID=1,BANDWIDTH=831000,RESOLUTION=704x396,CODECS="avc1.77.30, mp4a.40.2" +http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_3_av.m3u8?null=0 +#EXT-X-STREAM-INF:SUBTITLES="subs",AUDIO="aac",PROGRAM-ID=1,BANDWIDTH=1467000,RESOLUTION=1024x576,CODECS="avc1.77.30, mp4a.40.2" +http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_4_av.m3u8?null=0 diff --git a/test/testdata/m3u8/teamcoco_11995.m3u8 b/test/testdata/m3u8/teamcoco_11995.m3u8 new file mode 100644 index 000000000..a6e421697 --- /dev/null +++ b/test/testdata/m3u8/teamcoco_11995.m3u8 @@ -0,0 +1,16 @@ +#EXTM3U +#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio-0",NAME="Default",AUTOSELECT=YES,DEFAULT=YES,URI="hls/CONAN_020217_Highlight_show-audio-160k_v4.m3u8" +#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio-1",NAME="Default",AUTOSELECT=YES,DEFAULT=YES,URI="hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8" +#EXT-X-I-FRAME-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=37862000,CODECS="avc1.4d001f",URI="hls/CONAN_020217_Highlight_show-2m_iframe.m3u8" +#EXT-X-I-FRAME-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=18750000,CODECS="avc1.4d001e",URI="hls/CONAN_020217_Highlight_show-1m_iframe.m3u8" +#EXT-X-I-FRAME-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=6535000,CODECS="avc1.42001e",URI="hls/CONAN_020217_Highlight_show-400k_iframe.m3u8" +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=2374000,RESOLUTION=1024x576,CODECS="avc1.4d001f,mp4a.40.2",AUDIO="audio-0" +hls/CONAN_020217_Highlight_show-2m_v4.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1205000,RESOLUTION=640x360,CODECS="avc1.4d001e,mp4a.40.2",AUDIO="audio-0" +hls/CONAN_020217_Highlight_show-1m_v4.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=522000,RESOLUTION=400x224,CODECS="avc1.42001e,mp4a.40.2",AUDIO="audio-0" +hls/CONAN_020217_Highlight_show-400k_v4.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=413000,RESOLUTION=400x224,CODECS="avc1.42001e,mp4a.40.5",AUDIO="audio-1" +hls/CONAN_020217_Highlight_show-400k_v4.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=71000,CODECS="mp4a.40.5",AUDIO="audio-1" +hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8 diff --git a/test/testdata/m3u8/toggle_mobile_12211.m3u8 b/test/testdata/m3u8/toggle_mobile_12211.m3u8 new file mode 100644 index 000000000..69604e683 --- /dev/null +++ b/test/testdata/m3u8/toggle_mobile_12211.m3u8 @@ -0,0 +1,13 @@ +#EXTM3U +#EXT-X-VERSION:4 +#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio",LANGUAGE="eng",NAME="English",URI="http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_sa2ntrdg/name/a.mp4/index.m3u8" +#EXT-X-MEDIA:TYPE=AUDIO,GROUP-ID="audio",LANGUAGE="und",NAME="Undefined",URI="http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_r7y0nitg/name/a.mp4/index.m3u8" + +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=155648,RESOLUTION=320x180,AUDIO="audio" +http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_qlk9hlzr/name/a.mp4/index.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=502784,RESOLUTION=480x270,AUDIO="audio" +http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_oefackmi/name/a.mp4/index.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=827392,RESOLUTION=640x360,AUDIO="audio" +http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_vyg9pj7k/name/a.mp4/index.m3u8 +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1396736,RESOLUTION=854x480,AUDIO="audio" +http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_50n4psvx/name/a.mp4/index.m3u8 diff --git a/test/testdata/m3u8/twitch_vod.m3u8 b/test/testdata/m3u8/twitch_vod.m3u8 new file mode 100644 index 000000000..7617277ca --- /dev/null +++ b/test/testdata/m3u8/twitch_vod.m3u8 @@ -0,0 +1,20 @@ +#EXTM3U +#EXT-X-TWITCH-INFO:ORIGIN="s3",CLUSTER="edgecast_vod",REGION="EU",MANIFEST-CLUSTER="edgecast_vod",USER-IP="109.171.17.81" +#EXT-X-MEDIA:TYPE=VIDEO,GROUP-ID="chunked",NAME="Source",AUTOSELECT=YES,DEFAULT=YES +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=3214134,CODECS="avc1.100.31,mp4a.40.2",RESOLUTION="1280x720",VIDEO="chunked" +https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/chunked/index-muted-HM49I092CC.m3u8 +#EXT-X-MEDIA:TYPE=VIDEO,GROUP-ID="high",NAME="High",AUTOSELECT=YES,DEFAULT=YES +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=1603789,CODECS="avc1.42C01F,mp4a.40.2",RESOLUTION="1280x720",VIDEO="high" +https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/high/index-muted-HM49I092CC.m3u8 +#EXT-X-MEDIA:TYPE=VIDEO,GROUP-ID="medium",NAME="Medium",AUTOSELECT=YES,DEFAULT=YES +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=893387,CODECS="avc1.42C01E,mp4a.40.2",RESOLUTION="852x480",VIDEO="medium" +https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/medium/index-muted-HM49I092CC.m3u8 +#EXT-X-MEDIA:TYPE=VIDEO,GROUP-ID="low",NAME="Low",AUTOSELECT=YES,DEFAULT=YES +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=628347,CODECS="avc1.42C01E,mp4a.40.2",RESOLUTION="640x360",VIDEO="low" +https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/low/index-muted-HM49I092CC.m3u8 +#EXT-X-MEDIA:TYPE=VIDEO,GROUP-ID="mobile",NAME="Mobile",AUTOSELECT=YES,DEFAULT=YES +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=280474,CODECS="avc1.42C00D,mp4a.40.2",RESOLUTION="400x226",VIDEO="mobile" +https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8 +#EXT-X-MEDIA:TYPE=VIDEO,GROUP-ID="audio_only",NAME="Audio Only",AUTOSELECT=NO,DEFAULT=NO +#EXT-X-STREAM-INF:PROGRAM-ID=1,BANDWIDTH=182725,CODECS="mp4a.40.2",VIDEO="audio_only" +https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8 diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 61d97ab72..359c549c5 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -1312,17 +1312,25 @@ class InfoExtractor(object): entry_protocol='m3u8', preference=None, m3u8_id=None, note=None, errnote=None, fatal=True, live=False): - res = self._download_webpage_handle( m3u8_url, video_id, note=note or 'Downloading m3u8 information', errnote=errnote or 'Failed to download m3u8 information', fatal=fatal) + if res is False: return [] + m3u8_doc, urlh = res m3u8_url = urlh.geturl() + return self._parse_m3u8_formats( + m3u8_doc, m3u8_url, ext=ext, entry_protocol=entry_protocol, + preference=preference, m3u8_id=m3u8_id, live=live) + + def _parse_m3u8_formats(self, m3u8_doc, m3u8_url, ext=None, + entry_protocol='m3u8', preference=None, + m3u8_id=None, live=False): if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access return [] @@ -1333,19 +1341,21 @@ class InfoExtractor(object): if re.match(r'^https?://', u) else compat_urlparse.urljoin(m3u8_url, u)) - # We should try extracting formats only from master playlists [1], i.e. - # playlists that describe available qualities. On the other hand media - # playlists [2] should be returned as is since they contain just the media - # without qualities renditions. + # References: + # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-21 + # 2. https://github.com/rg3/youtube-dl/issues/12211 + + # We should try extracting formats only from master playlists [1, 4.3.4], + # i.e. playlists that describe available qualities. On the other hand + # media playlists [1, 4.3.3] should be returned as is since they contain + # just the media without qualities renditions. # Fortunately, master playlist can be easily distinguished from media - # playlist based on particular tags availability. As of [1, 2] master - # playlist tags MUST NOT appear in a media playist and vice versa. - # As of [3] #EXT-X-TARGETDURATION tag is REQUIRED for every media playlist - # and MUST NOT appear in master playlist thus we can clearly detect media - # playlist with this criterion. - # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.4 - # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3 - # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.1 + # playlist based on particular tags availability. As of [1, 4.3.3, 4.3.4] + # master playlist tags MUST NOT appear in a media playist and vice versa. + # As of [1, 4.3.3.1] #EXT-X-TARGETDURATION tag is REQUIRED for every + # media playlist and MUST NOT appear in master playlist thus we can + # clearly detect media playlist with this criterion. + if '#EXT-X-TARGETDURATION' in m3u8_doc: # media playlist, return as is return [{ 'url': m3u8_url, @@ -1354,52 +1364,67 @@ class InfoExtractor(object): 'protocol': entry_protocol, 'preference': preference, }] - audio_in_video_stream = {} - last_info = {} - last_media = {} + + groups = {} + last_stream_inf = {} + + def extract_media(x_media_line): + media = parse_m3u8_attributes(x_media_line) + # As per [1, 4.3.4.1] TYPE, GROUP-ID and NAME are REQUIRED + media_type, group_id, name = media.get('TYPE'), media.get('GROUP-ID'), media.get('NAME') + if not (media_type and group_id and name): + return + groups.setdefault(group_id, []).append(media) + if media_type not in ('VIDEO', 'AUDIO'): + return + media_url = media.get('URI') + if media_url: + format_id = [] + for v in (group_id, name): + if v: + format_id.append(v) + f = { + 'format_id': '-'.join(format_id), + 'url': format_url(media_url), + 'language': media.get('LANGUAGE'), + 'ext': ext, + 'protocol': entry_protocol, + 'preference': preference, + } + if media_type == 'AUDIO': + f['vcodec'] = 'none' + formats.append(f) + + def build_stream_name(): + # Despite specification does not mention NAME attribute for + # EXT-X-STREAM-INF it still sometimes may be present + stream_name = last_stream_inf.get('NAME') + if stream_name: + return stream_name + # If there is no NAME in EXT-X-STREAM-INF it will be obtained + # from corresponding rendition group + stream_group_id = last_stream_inf.get('VIDEO') + if not stream_group_id: + return + stream_group = groups.get(stream_group_id) + if not stream_group: + return stream_group_id + rendition = stream_group[0] + return rendition.get('NAME') or stream_group_id + for line in m3u8_doc.splitlines(): if line.startswith('#EXT-X-STREAM-INF:'): - last_info = parse_m3u8_attributes(line) + last_stream_inf = parse_m3u8_attributes(line) elif line.startswith('#EXT-X-MEDIA:'): - media = parse_m3u8_attributes(line) - media_type = media.get('TYPE') - if media_type in ('VIDEO', 'AUDIO'): - group_id = media.get('GROUP-ID') - media_url = media.get('URI') - if media_url: - format_id = [] - for v in (group_id, media.get('NAME')): - if v: - format_id.append(v) - f = { - 'format_id': '-'.join(format_id), - 'url': format_url(media_url), - 'language': media.get('LANGUAGE'), - 'ext': ext, - 'protocol': entry_protocol, - 'preference': preference, - } - if media_type == 'AUDIO': - f['vcodec'] = 'none' - if group_id and not audio_in_video_stream.get(group_id): - audio_in_video_stream[group_id] = False - formats.append(f) - else: - # When there is no URI in EXT-X-MEDIA let this tag's - # data be used by regular URI lines below - last_media = media - if media_type == 'AUDIO' and group_id: - audio_in_video_stream[group_id] = True + extract_media(line) elif line.startswith('#') or not line.strip(): continue else: - tbr = int_or_none(last_info.get('AVERAGE-BANDWIDTH') or last_info.get('BANDWIDTH'), scale=1000) + tbr = int_or_none(last_stream_inf.get('AVERAGE-BANDWIDTH') or last_stream_inf.get('BANDWIDTH'), scale=1000) format_id = [] if m3u8_id: format_id.append(m3u8_id) - # Despite specification does not mention NAME attribute for - # EXT-X-STREAM-INF it still sometimes may be present - stream_name = last_info.get('NAME') or last_media.get('NAME') + stream_name = build_stream_name() # Bandwidth of live streams may differ over time thus making # format_id unpredictable. So it's better to keep provided # format_id intact. @@ -1412,11 +1437,11 @@ class InfoExtractor(object): 'manifest_url': manifest_url, 'tbr': tbr, 'ext': ext, - 'fps': float_or_none(last_info.get('FRAME-RATE')), + 'fps': float_or_none(last_stream_inf.get('FRAME-RATE')), 'protocol': entry_protocol, 'preference': preference, } - resolution = last_info.get('RESOLUTION') + resolution = last_stream_inf.get('RESOLUTION') if resolution: mobj = re.search(r'(?P\d+)[xX](?P\d+)', resolution) if mobj: @@ -1432,13 +1457,26 @@ class InfoExtractor(object): 'vbr': vbr, 'abr': abr, }) - f.update(parse_codecs(last_info.get('CODECS'))) - if audio_in_video_stream.get(last_info.get('AUDIO')) is False and f['vcodec'] != 'none': - # TODO: update acodec for audio only formats with the same GROUP-ID - f['acodec'] = 'none' + codecs = parse_codecs(last_stream_inf.get('CODECS')) + f.update(codecs) + audio_group_id = last_stream_inf.get('AUDIO') + # As per [1, 4.3.4.1.1] any EXT-X-STREAM-INF tag which + # references a rendition group MUST have a CODECS attribute. + # However, this is not always respected, for example, [2] + # contains EXT-X-STREAM-INF tag which references AUDIO + # rendition group but does not have CODECS and despite + # referencing audio group an audio group, it represents + # a complete (with audio and video) format. So, for such cases + # we will ignore references to rendition groups and treat them + # as complete formats. + if audio_group_id and codecs and f.get('vcodec') != 'none': + audio_group = groups.get(audio_group_id) + if audio_group and audio_group[0].get('URI'): + # TODO: update acodec for audio only formats with + # the same GROUP-ID + f['acodec'] = 'none' formats.append(f) - last_info = {} - last_media = {} + last_stream_inf = {} return formats @staticmethod -- cgit 1.4.1 From ddd258f92270f48649b57ba4288027a2433af079 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Sun, 23 Apr 2017 11:49:57 +0700 Subject: [test_InfoExtractor] Add m3u8 parsing test for NAME attribute in EXT-X-STREAM-INF tag --- test/test_InfoExtractor.py | 63 +++++++++++++++++++++++++++++++++++------- youtube_dl/extractor/common.py | 4 ++- 2 files changed, 56 insertions(+), 11 deletions(-) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 57205f6ae..2a3c3f7cb 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -361,7 +361,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'ext': 'mp4', 'format_id': '155', 'protocol': 'm3u8', - 'tbr': 155, + 'tbr': 155.648, 'width': 320, 'height': 180, }, { @@ -370,7 +370,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'ext': 'mp4', 'format_id': '502', 'protocol': 'm3u8', - 'tbr': 502, + 'tbr': 502.784, 'width': 480, 'height': 270, }, { @@ -379,7 +379,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'ext': 'mp4', 'format_id': '827', 'protocol': 'm3u8', - 'tbr': 827, + 'tbr': 827.392, 'width': 640, 'height': 360, }, { @@ -388,7 +388,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'ext': 'mp4', 'format_id': '1396', 'protocol': 'm3u8', - 'tbr': 1396, + 'tbr': 1396.736, 'width': 854, 'height': 480, }] @@ -413,7 +413,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'protocol': 'm3u8', 'acodec': 'mp4a.40.2', 'vcodec': 'none', - 'tbr': 182, + 'tbr': 182.725, }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8', 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8', @@ -422,7 +422,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'protocol': 'm3u8', 'acodec': 'mp4a.40.2', 'vcodec': 'avc1.42C00D', - 'tbr': 280, + 'tbr': 280.474, 'width': 400, 'height': 226, }, { @@ -433,7 +433,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'protocol': 'm3u8', 'acodec': 'mp4a.40.2', 'vcodec': 'avc1.42C01E', - 'tbr': 628, + 'tbr': 628.347, 'width': 640, 'height': 360, }, { @@ -444,7 +444,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'protocol': 'm3u8', 'acodec': 'mp4a.40.2', 'vcodec': 'avc1.42C01E', - 'tbr': 893, + 'tbr': 893.387, 'width': 852, 'height': 480, }, { @@ -455,7 +455,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'protocol': 'm3u8', 'acodec': 'mp4a.40.2', 'vcodec': 'avc1.42C01F', - 'tbr': 1603, + 'tbr': 1603.789, 'width': 1280, 'height': 720, }, { @@ -466,7 +466,50 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'protocol': 'm3u8', 'acodec': 'mp4a.40.2', 'vcodec': 'avc1.100.31', - 'tbr': 3214, + 'tbr': 3214.134, + 'width': 1280, + 'height': 720, + }] + ), + ( + # http://www.vidio.com/watch/165683-dj_ambred-booyah-live-2015 + # EXT-X-STREAM-INF tag with NAME attribute that is not defined + # in HLS specification + 'vidio', + 'https://www.vidio.com/videos/165683/playlist.m3u8', + [{ + 'url': 'https://www.vidio.com/videos/165683/playlist.m3u8', + 'ext': 'mp4', + 'format_id': 'meta', + 'format_note': 'Quality selection URL', + 'protocol': 'm3u8', + 'preference': -100, + 'resolution': 'multiple' + }, { + 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b300.mp4.m3u8', + 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b300.mp4.m3u8', + 'ext': 'mp4', + 'format_id': '270p 3G', + 'protocol': 'm3u8', + 'tbr': 300, + 'width': 480, + 'height': 270, + }, { + 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b600.mp4.m3u8', + 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b600.mp4.m3u8', + 'ext': 'mp4', + 'format_id': '360p SD', + 'protocol': 'm3u8', + 'tbr': 600, + 'width': 640, + 'height': 360, + }, { + 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b1200.mp4.m3u8', + 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b1200.mp4.m3u8', + 'ext': 'mp4', + 'format_id': '720p HD', + 'protocol': 'm3u8', + 'tbr': 1200, 'width': 1280, 'height': 720, }] diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 2a099480f..6d023106e 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -1397,7 +1397,9 @@ class InfoExtractor(object): def build_stream_name(): # Despite specification does not mention NAME attribute for - # EXT-X-STREAM-INF it still sometimes may be present + # EXT-X-STREAM-INF tag (see [1] or vidio test in + # test_parse_m3u8_formats) it still sometimes may be present + # 1. http://www.vidio.com/watch/165683-dj_ambred-booyah-live-2015 stream_name = last_stream_inf.get('NAME') if stream_name: return stream_name -- cgit 1.4.1 From ff99fe529e52b2465f1d973e69df01a6391568d6 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Tue, 25 Apr 2017 22:07:10 +0700 Subject: Don't list master m3u8 playlists in format list (closes #12832) --- test/test_InfoExtractor.py | 40 -------------------------------------- youtube_dl/extractor/anvato.py | 7 +------ youtube_dl/extractor/common.py | 4 ++-- youtube_dl/extractor/funnyordie.py | 3 +-- youtube_dl/extractor/gamespot.py | 3 +-- youtube_dl/extractor/lego.py | 2 +- youtube_dl/extractor/pbs.py | 2 +- youtube_dl/extractor/r7.py | 3 +-- youtube_dl/extractor/ted.py | 2 +- youtube_dl/extractor/tvp.py | 3 +-- youtube_dl/extractor/viewster.py | 3 +-- 11 files changed, 11 insertions(+), 61 deletions(-) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 2a3c3f7cb..14abf03c4 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -184,14 +184,6 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ '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', [{ - 'url': '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', - 'ext': 'mp4', - 'format_id': 'meta', - 'format_note': 'Quality selection URL', - 'protocol': 'm3u8', - 'preference': -100, - 'resolution': 'multiple' - }, { 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0', 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0', 'ext': 'mp4', @@ -254,14 +246,6 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'teamcoco_11995', 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', [{ - 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', - 'ext': 'mp4', - 'format_id': 'meta', - 'format_note': 'Quality selection URL', - 'protocol': 'm3u8', - 'preference': -100, - 'resolution': 'multiple', - }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-160k_v4.m3u8', 'ext': 'mp4', 'format_id': 'audio-0-Default', @@ -334,14 +318,6 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ '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', [{ - 'url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', - 'ext': 'mp4', - 'format_id': 'meta', - 'format_note': 'Quality selection URL', - 'protocol': 'm3u8', - 'preference': -100, - 'resolution': 'multiple' - }, { 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_sa2ntrdg/name/a.mp4/index.m3u8', 'ext': 'mp4', 'format_id': 'audio-English', @@ -398,14 +374,6 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'twitch_vod', 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', [{ - 'url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', - 'ext': 'mp4', - 'format_id': 'meta', - 'format_note': 'Quality selection URL', - 'protocol': 'm3u8', - 'preference': -100, - 'resolution': 'multiple' - }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8', 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8', 'ext': 'mp4', @@ -478,14 +446,6 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'vidio', 'https://www.vidio.com/videos/165683/playlist.m3u8', [{ - 'url': 'https://www.vidio.com/videos/165683/playlist.m3u8', - 'ext': 'mp4', - 'format_id': 'meta', - 'format_note': 'Quality selection URL', - 'protocol': 'm3u8', - 'preference': -100, - 'resolution': 'multiple' - }, { 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b300.mp4.m3u8', 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b300.mp4.m3u8', 'ext': 'mp4', diff --git a/youtube_dl/extractor/anvato.py b/youtube_dl/extractor/anvato.py index 623f44dce..9fd91c2f6 100644 --- a/youtube_dl/extractor/anvato.py +++ b/youtube_dl/extractor/anvato.py @@ -178,12 +178,7 @@ class AnvatoIE(InfoExtractor): } if ext == 'm3u8' or media_format in ('m3u8', 'm3u8-variant'): - # Not using _extract_m3u8_formats here as individual media - # playlists are also included in published_urls. - if tbr is None: - formats.append(self._m3u8_meta_format(video_url, ext='mp4', m3u8_id='hls')) - continue - else: + if tbr is not None: a_format.update({ 'format_id': '-'.join(filter(None, ['hls', compat_str(tbr)])), 'ext': 'mp4', diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 8b3f04c61..6d01f0800 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -1334,7 +1334,7 @@ class InfoExtractor(object): if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access return [] - formats = [self._m3u8_meta_format(m3u8_url, ext, preference, m3u8_id)] + formats = [] format_url = lambda u: ( u @@ -1438,7 +1438,7 @@ class InfoExtractor(object): f = { 'format_id': '-'.join(format_id), 'url': manifest_url, - 'manifest_url': manifest_url, + 'manifest_url': m3u8_url, 'tbr': tbr, 'ext': ext, 'fps': float_or_none(last_stream_inf.get('FRAME-RATE')), diff --git a/youtube_dl/extractor/funnyordie.py b/youtube_dl/extractor/funnyordie.py index 81c0ce9a3..49409369c 100644 --- a/youtube_dl/extractor/funnyordie.py +++ b/youtube_dl/extractor/funnyordie.py @@ -58,8 +58,7 @@ class FunnyOrDieIE(InfoExtractor): m3u8_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False) source_formats = list(filter( - lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', - m3u8_formats)) + lambda f: f.get('vcodec') != 'none', m3u8_formats)) bitrates = [int(bitrate) for bitrate in re.findall(r'[,/]v(\d+)(?=[,/])', m3u8_url)] bitrates.sort() diff --git a/youtube_dl/extractor/gamespot.py b/youtube_dl/extractor/gamespot.py index 682c49e79..00d311158 100644 --- a/youtube_dl/extractor/gamespot.py +++ b/youtube_dl/extractor/gamespot.py @@ -78,8 +78,7 @@ class GameSpotIE(OnceIE): if m3u8_formats: self._sort_formats(m3u8_formats) m3u8_formats = list(filter( - lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', - m3u8_formats)) + lambda f: f.get('vcodec') != 'none', m3u8_formats)) if len(qualities) == len(m3u8_formats): for q, m3u8_format in zip(qualities, m3u8_formats): f = m3u8_format.copy() diff --git a/youtube_dl/extractor/lego.py b/youtube_dl/extractor/lego.py index d3bca6435..b312e77f1 100644 --- a/youtube_dl/extractor/lego.py +++ b/youtube_dl/extractor/lego.py @@ -86,7 +86,7 @@ class LEGOIE(InfoExtractor): formats = self._extract_akamai_formats( '%si/s/public/%s_,%s,.mp4.csmil/master.m3u8' % (streaming_base, path, streaming_path), video_id) m3u8_formats = list(filter( - lambda f: f.get('protocol') == 'm3u8_native' and f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', + lambda f: f.get('protocol') == 'm3u8_native' and f.get('vcodec') != 'none', formats)) if len(m3u8_formats) == len(self._BITRATES): self._sort_formats(m3u8_formats) diff --git a/youtube_dl/extractor/pbs.py b/youtube_dl/extractor/pbs.py index 3e51b4dd7..6166dc2ad 100644 --- a/youtube_dl/extractor/pbs.py +++ b/youtube_dl/extractor/pbs.py @@ -515,7 +515,7 @@ class PBSIE(InfoExtractor): http_url = format_url self._remove_duplicate_formats(formats) m3u8_formats = list(filter( - lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', + lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none', formats)) if http_url: for m3u8_format in m3u8_formats: diff --git a/youtube_dl/extractor/r7.py b/youtube_dl/extractor/r7.py index ed38c77eb..e2202d603 100644 --- a/youtube_dl/extractor/r7.py +++ b/youtube_dl/extractor/r7.py @@ -62,8 +62,7 @@ class R7IE(InfoExtractor): # m3u8 format always matches the http format, let's copy metadata from # one to another m3u8_formats = list(filter( - lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', - formats)) + lambda f: f.get('vcodec') != 'none', formats)) if len(m3u8_formats) == 1: f_copy = m3u8_formats[0].copy() f_copy.update(f) diff --git a/youtube_dl/extractor/ted.py b/youtube_dl/extractor/ted.py index 1b1afab32..3f3c681ae 100644 --- a/youtube_dl/extractor/ted.py +++ b/youtube_dl/extractor/ted.py @@ -210,7 +210,7 @@ class TEDIE(InfoExtractor): resources.get('stream'), video_name, 'mp4', m3u8_id=format_id, fatal=False)) m3u8_formats = list(filter( - lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', + lambda f: f.get('protocol') == 'm3u8' and f.get('vcodec') != 'none', formats)) if http_url: for m3u8_format in m3u8_formats: diff --git a/youtube_dl/extractor/tvp.py b/youtube_dl/extractor/tvp.py index 06ea2b40a..c5b3288ad 100644 --- a/youtube_dl/extractor/tvp.py +++ b/youtube_dl/extractor/tvp.py @@ -150,8 +150,7 @@ class TVPEmbedIE(InfoExtractor): 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False) self._sort_formats(m3u8_formats) m3u8_formats = list(filter( - lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', - m3u8_formats)) + lambda f: f.get('vcodec') != 'none', m3u8_formats)) formats.extend(m3u8_formats) for i, m3u8_format in enumerate(m3u8_formats, 2): http_url = '%s-%d.mp4' % (video_url_base, i) diff --git a/youtube_dl/extractor/viewster.py b/youtube_dl/extractor/viewster.py index fcf0cb100..d5d5b4c69 100644 --- a/youtube_dl/extractor/viewster.py +++ b/youtube_dl/extractor/viewster.py @@ -176,8 +176,7 @@ class ViewsterIE(InfoExtractor): if m3u8_formats: self._sort_formats(m3u8_formats) m3u8_formats = list(filter( - lambda f: f.get('vcodec') != 'none' and f.get('resolution') != 'multiple', - m3u8_formats)) + lambda f: f.get('vcodec') != 'none', m3u8_formats)) if len(qualities) == len(m3u8_formats): for q, m3u8_format in zip(qualities, m3u8_formats): f = m3u8_format.copy() -- cgit 1.4.1 From 30bb6ce1a49f37f11afdca96fa7fd62ec56395c3 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Fri, 28 Apr 2017 03:01:43 +0700 Subject: [test_InfoExtractor] Fix test_parse_m3u8_formats --- test/test_InfoExtractor.py | 50 +++++++++++++++++++++++++--------------------- 1 file changed, 27 insertions(+), 23 deletions(-) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 14abf03c4..6f52e11f7 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -185,7 +185,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ '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', [{ 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0', - 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_0_av.m3u8?null=0', + 'manifest_url': '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', 'ext': 'mp4', 'format_id': '180', 'protocol': 'm3u8', @@ -196,7 +196,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 144, }, { 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_1_av.m3u8?null=0', - 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_1_av.m3u8?null=0', + 'manifest_url': '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', 'ext': 'mp4', 'format_id': '303', 'protocol': 'm3u8', @@ -207,7 +207,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 180, }, { 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_2_av.m3u8?null=0', - 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_2_av.m3u8?null=0', + 'manifest_url': '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', 'ext': 'mp4', 'format_id': '575', 'protocol': 'm3u8', @@ -218,7 +218,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 288, }, { 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_3_av.m3u8?null=0', - 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_3_av.m3u8?null=0', + 'manifest_url': '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', 'ext': 'mp4', 'format_id': '831', 'protocol': 'm3u8', @@ -229,7 +229,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 396, }, { 'url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_4_av.m3u8?null=0', - 'manifest_url': 'http://replayftv-vh.akamaihd.net/i/streaming-adaptatif_france-dom-tom/2017/S16/J2/156589847-58f59130c1f52-,standard1,standard2,standard3,standard4,standard5,.mp4.csmil/index_4_av.m3u8?null=0', + 'manifest_url': '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', 'ext': 'mp4', 'protocol': 'm3u8', 'format_id': '1467', @@ -247,19 +247,21 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', [{ 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-160k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': 'audio-0-Default', 'protocol': 'm3u8', 'vcodec': 'none', }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': 'audio-1-Default', 'protocol': 'm3u8', 'vcodec': 'none', }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8', - 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-audio-64k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': '71', 'protocol': 'm3u8', @@ -268,7 +270,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'tbr': 71, }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', - 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': '413', 'protocol': 'm3u8', @@ -279,7 +281,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 224, }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', - 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-400k_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': '522', 'protocol': 'm3u8', @@ -290,7 +292,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 224, }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-1m_v4.m3u8', - 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-1m_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': '1205', 'protocol': 'm3u8', @@ -301,7 +303,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 360, }, { 'url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-2m_v4.m3u8', - 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/hls/CONAN_020217_Highlight_show-2m_v4.m3u8', + 'manifest_url': 'http://ak.storage-w.teamcococdn.com/cdn/2017-02/98599/ed8f/main.m3u8', 'ext': 'mp4', 'format_id': '2374', 'protocol': 'm3u8', @@ -319,6 +321,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', [{ 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_sa2ntrdg/name/a.mp4/index.m3u8', + 'manifest_url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', 'ext': 'mp4', 'format_id': 'audio-English', 'protocol': 'm3u8', @@ -326,6 +329,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'vcodec': 'none', }, { 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_r7y0nitg/name/a.mp4/index.m3u8', + 'manifest_url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', 'ext': 'mp4', 'format_id': 'audio-Undefined', 'protocol': 'm3u8', @@ -333,7 +337,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'vcodec': 'none', }, { 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_qlk9hlzr/name/a.mp4/index.m3u8', - 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_qlk9hlzr/name/a.mp4/index.m3u8', + 'manifest_url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', 'ext': 'mp4', 'format_id': '155', 'protocol': 'm3u8', @@ -342,7 +346,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 180, }, { 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_oefackmi/name/a.mp4/index.m3u8', - 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/2/pv/1/flavorId/0_oefackmi/name/a.mp4/index.m3u8', + 'manifest_url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', 'ext': 'mp4', 'format_id': '502', 'protocol': 'm3u8', @@ -351,7 +355,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 270, }, { 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_vyg9pj7k/name/a.mp4/index.m3u8', - 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_vyg9pj7k/name/a.mp4/index.m3u8', + 'manifest_url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', 'ext': 'mp4', 'format_id': '827', 'protocol': 'm3u8', @@ -360,7 +364,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 360, }, { 'url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_50n4psvx/name/a.mp4/index.m3u8', - 'manifest_url': 'http://k.toggle.sg/fhls/p/2082311/sp/208231100/serveFlavor/entryId/0_89q6e8ku/v/12/pv/1/flavorId/0_50n4psvx/name/a.mp4/index.m3u8', + 'manifest_url': 'http://cdnapi.kaltura.com/p/2082311/sp/208231100/playManifest/protocol/http/entryId/0_89q6e8ku/format/applehttp/tags/mobile_sd/f/a.m3u8', 'ext': 'mp4', 'format_id': '1396', 'protocol': 'm3u8', @@ -375,7 +379,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', [{ 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8', - 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/audio_only/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', 'ext': 'mp4', 'format_id': 'Audio Only', 'protocol': 'm3u8', @@ -384,7 +388,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'tbr': 182.725, }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8', - 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/mobile/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', 'ext': 'mp4', 'format_id': 'Mobile', 'protocol': 'm3u8', @@ -395,7 +399,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 226, }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/low/index-muted-HM49I092CC.m3u8', - 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/low/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', 'ext': 'mp4', 'format_id': 'Low', 'protocol': 'm3u8', @@ -406,7 +410,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 360, }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/medium/index-muted-HM49I092CC.m3u8', - 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/medium/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', 'ext': 'mp4', 'format_id': 'Medium', 'protocol': 'm3u8', @@ -417,7 +421,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 480, }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/high/index-muted-HM49I092CC.m3u8', - 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/high/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', 'ext': 'mp4', 'format_id': 'High', 'protocol': 'm3u8', @@ -428,7 +432,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 720, }, { 'url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/chunked/index-muted-HM49I092CC.m3u8', - 'manifest_url': 'https://vod.edgecast.hls.ttvnw.net/e5da31ab49_riotgames_15001215120_261543898/chunked/index-muted-HM49I092CC.m3u8', + 'manifest_url': 'https://usher.ttvnw.net/vod/6528877?allow_source=true&allow_audio_only=true&allow_spectre=true&player=twitchweb&nauth=%7B%22user_id%22%3Anull%2C%22vod_id%22%3A6528877%2C%22expires%22%3A1492887874%2C%22chansub%22%3A%7B%22restricted_bitrates%22%3A%5B%5D%7D%2C%22privileged%22%3Afalse%2C%22https_required%22%3Afalse%7D&nauthsig=3e29296a6824a0f48f9e731383f77a614fc79bee', 'ext': 'mp4', 'format_id': 'Source', 'protocol': 'm3u8', @@ -447,7 +451,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'https://www.vidio.com/videos/165683/playlist.m3u8', [{ 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b300.mp4.m3u8', - 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b300.mp4.m3u8', + 'manifest_url': 'https://www.vidio.com/videos/165683/playlist.m3u8', 'ext': 'mp4', 'format_id': '270p 3G', 'protocol': 'm3u8', @@ -456,7 +460,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 270, }, { 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b600.mp4.m3u8', - 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b600.mp4.m3u8', + 'manifest_url': 'https://www.vidio.com/videos/165683/playlist.m3u8', 'ext': 'mp4', 'format_id': '360p SD', 'protocol': 'm3u8', @@ -465,7 +469,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 360, }, { 'url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b1200.mp4.m3u8', - 'manifest_url': 'https://cdn1-a.production.vidio.static6.com/uploads/165683/dj_ambred-4383-b1200.mp4.m3u8', + 'manifest_url': 'https://www.vidio.com/videos/165683/playlist.m3u8', 'ext': 'mp4', 'format_id': '720p HD', 'protocol': 'm3u8', -- cgit 1.4.1 From 485047854376465f95309daad4966971f56728ef Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Tue, 15 Aug 2017 23:58:00 +0700 Subject: [extractor/common] Add support for float durations in _parse_mpd_formats (closes #13919) --- test/test_InfoExtractor.py | 86 ++++++++++++++++++++++++++++++++++++ test/testdata/mpd/float_duration.mpd | 18 ++++++++ youtube_dl/extractor/common.py | 2 +- 3 files changed, 105 insertions(+), 1 deletion(-) create mode 100644 test/testdata/mpd/float_duration.mpd (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 6f52e11f7..f18a823fc 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -10,6 +10,7 @@ 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 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 @@ -488,6 +489,91 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) + def test_parse_mpd_formats(self): + _TEST_CASES = [ + ( + # https://github.com/rg3/youtube-dl/issues/13919 + 'float_duration', + 'http://unknown/manifest.mpd', + [{ + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': '318597', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'none', + 'vcodec': 'avc1.42001f', + 'tbr': 318.597, + 'width': 340, + 'height': 192, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': '638590', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'none', + 'vcodec': 'avc1.42001f', + 'tbr': 638.59, + 'width': 512, + 'height': 288, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': '1022565', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'none', + 'vcodec': 'avc1.4d001f', + 'tbr': 1022.565, + 'width': 688, + 'height': 384, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': '2046506', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'none', + 'vcodec': 'avc1.4d001f', + 'tbr': 2046.506, + 'width': 1024, + 'height': 576, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': '3998017', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'none', + 'vcodec': 'avc1.640029', + 'tbr': 3998.017, + 'width': 1280, + 'height': 720, + }, { + 'manifest_url': 'http://unknown/manifest.mpd', + 'ext': 'mp4', + 'format_id': '5997485', + 'format_note': 'DASH video', + 'protocol': 'http_dash_segments', + 'acodec': 'none', + 'vcodec': 'avc1.640032', + 'tbr': 5997.485, + 'width': 1920, + 'height': 1080, + }] + ), + ] + + for mpd_file, mpd_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) + self.ie._sort_formats(formats) + expect_value(self, formats, expected_formats, None) + if __name__ == '__main__': unittest.main() diff --git a/test/testdata/mpd/float_duration.mpd b/test/testdata/mpd/float_duration.mpd new file mode 100644 index 000000000..8dc1d2d5e --- /dev/null +++ b/test/testdata/mpd/float_duration.mpd @@ -0,0 +1,18 @@ + + + + + + + + + + + + + + + + + + \ No newline at end of file diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 7fe888462..e747258aa 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -1786,7 +1786,7 @@ class InfoExtractor(object): ms_info['timescale'] = int(timescale) segment_duration = source.get('duration') if segment_duration: - ms_info['segment_duration'] = int(segment_duration) + ms_info['segment_duration'] = float(segment_duration) def extract_Initialization(source): initialization = source.find(_add_ns('Initialization')) -- cgit 1.4.1 From 181e381fda4ddb9083f3834a8bd1bab72c937545 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Sat, 4 Nov 2017 22:12:20 +0700 Subject: [test_InfoExtractor] Add test for #14660 --- test/test_InfoExtractor.py | 26 ++++++++++++++++++++++++++ test/testdata/f4m/custom_base_url.f4m | 10 ++++++++++ 2 files changed, 36 insertions(+) create mode 100644 test/testdata/f4m/custom_base_url.f4m (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index f18a823fc..686c63efa 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -574,6 +574,32 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) + def test_parse_f4m_formats(self): + _TEST_CASES = [ + ( + # https://github.com/rg3/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) if __name__ == '__main__': unittest.main() diff --git a/test/testdata/f4m/custom_base_url.f4m b/test/testdata/f4m/custom_base_url.f4m new file mode 100644 index 000000000..74e1539e8 --- /dev/null +++ b/test/testdata/f4m/custom_base_url.f4m @@ -0,0 +1,10 @@ + + + recorded + http://vod.livestream.com/events/0000000000673980/ + 269.293 + AAAAm2Fic3QAAAAAAAAAAQAAAAPoAAAAAAAEG+0AAAAAAAAAAAAAAAAAAQAAABlhc3J0AAAAAAAAAAABAAAAAQAAAC4BAAAAVmFmcnQAAAAAAAAD6AAAAAAEAAAAAQAAAAAAAAAAAAAXcAAAAC0AAAAAAAQHQAAAE5UAAAAuAAAAAAAEGtUAAAEYAAAAAAAAAAAAAAAAAAAAAAA= + + AgAKb25NZXRhRGF0YQgAAAAIAAhkdXJhdGlvbgBAcNSwIMSbpgAFd2lkdGgAQJQAAAAAAAAABmhlaWdodABAhoAAAAAAAAAJZnJhbWVyYXRlAEA4/7DoLwW3AA12aWRlb2RhdGFyYXRlAECe1DLgjcobAAx2aWRlb2NvZGVjaWQAQBwAAAAAAAAADWF1ZGlvZGF0YXJhdGUAQGSimlvaPKQADGF1ZGlvY29kZWNpZABAJAAAAAAAAAAACQ== + + -- cgit 1.4.1 From f58a5060443229e017a04e79cb37d499b6347fe3 Mon Sep 17 00:00:00 2001 From: Yen Chi Hsuan Date: Mon, 27 Nov 2017 21:30:47 +0800 Subject: [test_InfoExtractor] Fix flake8 --- test/test_InfoExtractor.py | 1 + 1 file changed, 1 insertion(+) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 686c63efa..7b95f883f 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -601,5 +601,6 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) + if __name__ == '__main__': unittest.main() -- cgit 1.4.1 From 41bf647e895faca42cdc2565ea034ed341593f8e Mon Sep 17 00:00:00 2001 From: Petr Novak Date: Sat, 25 Nov 2017 02:13:23 +0100 Subject: [extractor/common] Add support for DASH manifests with SegmentLists with bare SegmentURLs --- test/test_InfoExtractor.py | 83 ++++++++++++++- test/testdata/mpd/urls_only.mpd | 218 ++++++++++++++++++++++++++++++++++++++++ youtube_dl/extractor/common.py | 9 ++ 3 files changed, 309 insertions(+), 1 deletion(-) create mode 100644 test/testdata/mpd/urls_only.mpd (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 7b95f883f..e58452ab5 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -562,7 +562,88 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'width': 1920, 'height': 1080, }] - ), + ), ( + 'urls_only', + 'http://unknown/manifest.mpd', + [{ + '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, + }] + ) ] for mpd_file, mpd_url, expected_formats in _TEST_CASES: diff --git a/test/testdata/mpd/urls_only.mpd b/test/testdata/mpd/urls_only.mpd new file mode 100644 index 000000000..2b9d595d3 --- /dev/null +++ b/test/testdata/mpd/urls_only.mpd @@ -0,0 +1,218 @@ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 8e4ee0deb..15999411b 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -1975,6 +1975,15 @@ class InfoExtractor(object): }) segment_index += 1 representation_ms_info['fragments'] = fragments + elif 'segment_urls' in representation_ms_info: + # Segment URLs with no SegmentTimeline + # Example: https://www.seznam.cz/zpravy/clanek/cesko-zasahne-vitr-o-sile-vichrice-muze-byt-i-zivotu-nebezpecny-39091 + fragments = [] + for segment_url in representation_ms_info['segment_urls']: + fragments.append({ + location_key(segment_url): segment_url, + }) + representation_ms_info['fragments'] = fragments # NB: MPD manifest may contain direct URLs to unfragmented media. # No fragments key is present in this case. if 'fragments' in representation_ms_info: -- cgit 1.4.1 From 78593e294cc4623108d17b1b0b2d26d507953006 Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Sat, 2 Dec 2017 21:22:43 +0700 Subject: Add references for #14844 --- test/test_InfoExtractor.py | 1 + youtube_dl/extractor/common.py | 1 + 2 files changed, 2 insertions(+) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index e58452ab5..8a372d2c9 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -563,6 +563,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ 'height': 1080, }] ), ( + # https://github.com/rg3/youtube-dl/pull/14844 'urls_only', 'http://unknown/manifest.mpd', [{ diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 3baf683d8..80a9c982f 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -1978,6 +1978,7 @@ class InfoExtractor(object): elif 'segment_urls' in representation_ms_info: # Segment URLs with no SegmentTimeline # Example: https://www.seznam.cz/zpravy/clanek/cesko-zasahne-vitr-o-sile-vichrice-muze-byt-i-zivotu-nebezpecny-39091 + # https://github.com/rg3/youtube-dl/pull/14844 fragments = [] segment_duration = float_or_none( representation_ms_info['segment_duration'], -- cgit 1.4.1 From 9d6ac71c27b1dfb662c795ef598dbfd0286682da Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Fri, 29 Dec 2017 23:14:15 +0700 Subject: [extractor/common] Fix extraction of DASH formats with the same representation id (closes #15111) --- test/test_InfoExtractor.py | 11 +++++++++++ youtube_dl/extractor/common.py | 18 ++++++++---------- 2 files changed, 19 insertions(+), 10 deletions(-) (limited to 'test/test_InfoExtractor.py') diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 8a372d2c9..7b31d5198 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -493,9 +493,20 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ _TEST_CASES = [ ( # https://github.com/rg3/youtube-dl/issues/13919 + # Also tests duplicate representation ids, see + # https://github.com/rg3/youtube-dl/issues/15111 'float_duration', 'http://unknown/manifest.mpd', [{ + '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', diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 3b79b8cb4..35d427eec 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -2007,16 +2007,14 @@ class InfoExtractor(object): f['url'] = initialization_url f['fragments'].append({location_key(initialization_url): initialization_url}) f['fragments'].extend(representation_ms_info['fragments']) - try: - existing_format = next( - fo for fo in formats - if fo['format_id'] == representation_id) - except StopIteration: - full_info = formats_dict.get(representation_id, {}).copy() - full_info.update(f) - formats.append(full_info) - else: - existing_format.update(f) + # According to [1, 5.3.5.2, Table 7, page 35] @id of Representation + # is not necessarily unique within a Period thus formats with + # the same `format_id` are quite possible. There are numerous examples + # of such manifests (see https://github.com/rg3/youtube-dl/issues/15111, + # https://github.com/rg3/youtube-dl/issues/13919) + full_info = formats_dict.get(representation_id, {}).copy() + full_info.update(f) + formats.append(full_info) else: self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type) return formats -- cgit 1.4.1