diff options
Diffstat (limited to 'test/test_InfoExtractor.py')
| -rw-r--r-- | test/test_InfoExtractor.py | 356 |
1 files changed, 331 insertions, 25 deletions
diff --git a/test/test_InfoExtractor.py b/test/test_InfoExtractor.py index 8494105..529da52 100644 --- a/test/test_InfoExtractor.py +++ b/test/test_InfoExtractor.py @@ -1,27 +1,32 @@ #!/usr/bin/env python3 -from __future__ import unicode_literals - # Allow direct execution -import io import os import sys import unittest + sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) -from test.helper import FakeYDL, expect_dict, expect_value, http_server_port -from hypervideo_dl.compat import compat_etree_fromstring, compat_http_server -from hypervideo_dl.extractor.common import InfoExtractor -from hypervideo_dl.extractor import YoutubeIE, get_info_extractor -from hypervideo_dl.utils import encode_data_uri, strip_jsonp, ExtractorError, RegexNotFoundError + +import http.server import threading +from test.helper import FakeYDL, expect_dict, expect_value, http_server_port +from hypervideo_dl.compat import compat_etree_fromstring +from hypervideo_dl.extractor import YoutubeIE, get_info_extractor +from hypervideo_dl.extractor.common import InfoExtractor +from hypervideo_dl.utils import ( + ExtractorError, + RegexNotFoundError, + encode_data_uri, + strip_jsonp, +) TEAPOT_RESPONSE_STATUS = 418 TEAPOT_RESPONSE_BODY = "<h1>418 I'm a teapot</h1>" -class InfoExtractorTestRequestHandler(compat_http_server.BaseHTTPRequestHandler): +class InfoExtractorTestRequestHandler(http.server.BaseHTTPRequestHandler): def log_message(self, format, *args): pass @@ -36,7 +41,9 @@ class InfoExtractorTestRequestHandler(compat_http_server.BaseHTTPRequestHandler) class DummyIE(InfoExtractor): - pass + def _sort_formats(self, formats, field_preference=[]): + self._downloader.sort_formats( + {'formats': formats, '_format_sort_fields': field_preference}) class TestInfoExtractor(unittest.TestCase): @@ -500,6 +507,24 @@ class TestInfoExtractor(unittest.TestCase): }], }) + # from https://0000.studio/ + # with type attribute but without extension in URL + expect_dict( + self, + self.ie._parse_html5_media_entries( + 'https://0000.studio', + r''' + <video src="https://d1ggyt9m8pwf3g.cloudfront.net/protected/ap-northeast-1:1864af40-28d5-492b-b739-b32314b1a527/archive/clip/838db6a7-8973-4cd6-840d-8517e4093c92" + controls="controls" type="video/mp4" preload="metadata" autoplay="autoplay" playsinline class="object-contain"> + </video> + ''', None)[0], + { + 'formats': [{ + 'url': 'https://d1ggyt9m8pwf3g.cloudfront.net/protected/ap-northeast-1:1864af40-28d5-492b-b739-b32314b1a527/archive/clip/838db6a7-8973-4cd6-840d-8517e4093c92', + 'ext': 'mp4', + }], + }) + def test_extract_jwplayer_data_realworld(self): # from http://www.suffolk.edu/sjc/ expect_dict( @@ -1011,8 +1036,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ ] for m3u8_file, m3u8_url, expected_formats, expected_subs in _TEST_CASES: - with io.open('./test/testdata/m3u8/%s.m3u8' % m3u8_file, - mode='r', encoding='utf-8') as f: + with open('./test/testdata/m3u8/%s.m3u8' % m3u8_file, encoding='utf-8') as f: formats, subs = self.ie._parse_m3u8_formats_and_subtitles( f.read(), m3u8_url, ext='mp4') self.ie._sort_formats(formats) @@ -1357,10 +1381,9 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ ] for mpd_file, mpd_url, mpd_base_url, expected_formats, expected_subtitles in _TEST_CASES: - with io.open('./test/testdata/mpd/%s.mpd' % mpd_file, - mode='r', encoding='utf-8') as f: + with open('./test/testdata/mpd/%s.mpd' % mpd_file, encoding='utf-8') as f: formats, subtitles = self.ie._parse_mpd_formats_and_subtitles( - compat_etree_fromstring(f.read().encode('utf-8')), + compat_etree_fromstring(f.read().encode()), mpd_base_url=mpd_base_url, mpd_url=mpd_url) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) @@ -1546,13 +1569,298 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ ] }, ), + ( + 'ec-3_test', + 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + [{ + 'format_id': 'audio_deu_1-224', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'isma', + 'tbr': 224, + 'asr': 48000, + 'vcodec': 'none', + 'acodec': 'EC-3', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'audio', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 0, + 'height': 0, + 'fourcc': 'EC-3', + 'language': 'deu', + 'codec_private_data': '00063F000000AF87FBA7022DFB42A4D405CD93843BDD0700200F00', + 'sampling_rate': 48000, + 'channels': 6, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'audio_ext': 'isma', + 'video_ext': 'none', + 'abr': 224, + }, { + 'format_id': 'audio_deu-127', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'isma', + 'tbr': 127, + 'asr': 48000, + 'vcodec': 'none', + 'acodec': 'AACL', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'audio', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 0, + 'height': 0, + 'fourcc': 'AACL', + 'language': 'deu', + 'codec_private_data': '1190', + 'sampling_rate': 48000, + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'audio_ext': 'isma', + 'video_ext': 'none', + 'abr': 127, + }, { + 'format_id': 'video_deu-23', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 384, + 'height': 216, + 'tbr': 23, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 384, + 'height': 216, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '000000016742C00CDB06077E5C05A808080A00000300020000030009C0C02EE0177CC6300F142AE00000000168CA8DC8', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 23, + }, { + 'format_id': 'video_deu-403', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 400, + 'height': 224, + 'tbr': 403, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 400, + 'height': 224, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4014E98323B602D4040405000003000100000300320F1429380000000168EAECF2', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 403, + }, { + 'format_id': 'video_deu-680', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 640, + 'height': 360, + 'tbr': 680, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 640, + 'height': 360, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D401EE981405FF2E02D4040405000000300100000030320F162D3800000000168EAECF2', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 680, + }, { + 'format_id': 'video_deu-1253', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 640, + 'height': 360, + 'tbr': 1253, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 640, + 'height': 360, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D401EE981405FF2E02D4040405000000300100000030320F162D3800000000168EAECF2', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 1253, + }, { + 'format_id': 'video_deu-2121', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 768, + 'height': 432, + 'tbr': 2121, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 768, + 'height': 432, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D401EECA0601BD80B50101014000003000400000300C83C58B6580000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 2121, + }, { + 'format_id': 'video_deu-3275', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 1280, + 'height': 720, + 'tbr': 3275, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 1280, + 'height': 720, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4020ECA02802DD80B501010140000003004000000C83C60C65800000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 3275, + }, { + 'format_id': 'video_deu-5300', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 1920, + 'height': 1080, + 'tbr': 5300, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 1920, + 'height': 1080, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4028ECA03C0113F2E02D4040405000000300100000030320F18319600000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 5300, + }, { + 'format_id': 'video_deu-8079', + 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest', + 'ext': 'ismv', + 'width': 1920, + 'height': 1080, + 'tbr': 8079, + 'vcodec': 'AVC1', + 'acodec': 'none', + 'protocol': 'ism', + '_download_params': + { + 'stream_type': 'video', + 'duration': 370000000, + 'timescale': 10000000, + 'width': 1920, + 'height': 1080, + 'fourcc': 'AVC1', + 'language': 'deu', + 'codec_private_data': '00000001674D4028ECA03C0113F2E02D4040405000000300100000030320F18319600000000168E93B3C80', + 'channels': 2, + 'bits_per_sample': 16, + 'nal_unit_length_field': 4 + }, + 'video_ext': 'ismv', + 'audio_ext': 'none', + 'vbr': 8079, + }], + {}, + ), ] for ism_file, ism_url, expected_formats, expected_subtitles in _TEST_CASES: - with io.open('./test/testdata/ism/%s.Manifest' % ism_file, - mode='r', encoding='utf-8') as f: + with open('./test/testdata/ism/%s.Manifest' % ism_file, encoding='utf-8') as f: formats, subtitles = self.ie._parse_ism_formats_and_subtitles( - compat_etree_fromstring(f.read().encode('utf-8')), ism_url=ism_url) + compat_etree_fromstring(f.read().encode()), ism_url=ism_url) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) expect_value(self, subtitles, expected_subtitles, None) @@ -1576,10 +1884,9 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ ] for f4m_file, f4m_url, expected_formats in _TEST_CASES: - with io.open('./test/testdata/f4m/%s.f4m' % f4m_file, - mode='r', encoding='utf-8') as f: + with open('./test/testdata/f4m/%s.f4m' % f4m_file, encoding='utf-8') as f: formats = self.ie._parse_f4m_formats( - compat_etree_fromstring(f.read().encode('utf-8')), + compat_etree_fromstring(f.read().encode()), f4m_url, None) self.ie._sort_formats(formats) expect_value(self, formats, expected_formats, None) @@ -1624,10 +1931,9 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ ] for xspf_file, xspf_url, expected_entries in _TEST_CASES: - with io.open('./test/testdata/xspf/%s.xspf' % xspf_file, - mode='r', encoding='utf-8') as f: + with open('./test/testdata/xspf/%s.xspf' % xspf_file, encoding='utf-8') as f: entries = self.ie._parse_xspf( - compat_etree_fromstring(f.read().encode('utf-8')), + compat_etree_fromstring(f.read().encode()), xspf_file, xspf_url=xspf_url, xspf_base_url=xspf_url) expect_value(self, entries, expected_entries, None) for i in range(len(entries)): @@ -1640,7 +1946,7 @@ jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/ # or the underlying `_download_webpage_handle` returning no content # when a response matches `expected_status`. - httpd = compat_http_server.HTTPServer( + httpd = http.server.HTTPServer( ('127.0.0.1', 0), InfoExtractorTestRequestHandler) port = http_server_port(httpd) server_thread = threading.Thread(target=httpd.serve_forever) |