#!/usr/bin/env python3
+
# Allow direct execution
import os
import sys
import unittest
+
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
-from test.helper import FakeYDL, expect_dict, expect_value, http_server_port
-from yt_dlp.compat import compat_etree_fromstring, compat_http_server
-from yt_dlp.extractor.common import InfoExtractor
-from yt_dlp.extractor import YoutubeIE, get_info_extractor
-from yt_dlp.utils import encode_data_uri, strip_jsonp, ExtractorError, RegexNotFoundError
+
+import http.server
import threading
+from test.helper import FakeYDL, expect_dict, expect_value, http_server_port
+from yt_dlp.compat import compat_etree_fromstring
+from yt_dlp.extractor import YoutubeIE, get_info_extractor
+from yt_dlp.extractor.common import InfoExtractor
+from yt_dlp.utils import (
+ ExtractorError,
+ RegexNotFoundError,
+ encode_data_uri,
+ strip_jsonp,
+)
TEAPOT_RESPONSE_STATUS = 418
TEAPOT_RESPONSE_BODY = "<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
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):
<meta name="og:test1" content='foo > < bar'/>
<meta name="og:test2" content="foo >//< bar"/>
<meta property=og-test3 content='Ill-formatted opengraph'/>
+ <meta property=og:test4 content=unquoted-value/>
'''
self.assertEqual(ie._og_search_title(html), 'Foo')
self.assertEqual(ie._og_search_description(html), 'Some video\'s description ')
self.assertEqual(ie._og_search_property(('test0', 'test1'), html), 'foo > < bar')
self.assertRaises(RegexNotFoundError, ie._og_search_property, 'test0', html, None, fatal=True)
self.assertRaises(RegexNotFoundError, ie._og_search_property, ('test0', 'test00'), html, None, fatal=True)
+ self.assertEqual(ie._og_search_property('test4', html), 'unquoted-value')
def test_html_search_meta(self):
ie = self.ie
}],
})
+ # 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(
'acodec': 'mp4a.40.2',
'video_ext': 'mp4',
'audio_ext': 'none',
- 'vbr': 263.851,
- 'abr': 0,
}, {
'format_id': '577',
'format_index': None,
'acodec': 'mp4a.40.2',
'video_ext': 'mp4',
'audio_ext': 'none',
- 'vbr': 577.61,
- 'abr': 0,
}, {
'format_id': '915',
'format_index': None,
'acodec': 'mp4a.40.2',
'video_ext': 'mp4',
'audio_ext': 'none',
- 'vbr': 915.905,
- 'abr': 0,
}, {
'format_id': '1030',
'format_index': None,
'acodec': 'mp4a.40.2',
'video_ext': 'mp4',
'audio_ext': 'none',
- 'vbr': 1030.138,
- 'abr': 0,
}, {
'format_id': '1924',
'format_index': None,
'acodec': 'mp4a.40.2',
'video_ext': 'mp4',
'audio_ext': 'none',
- 'vbr': 1924.009,
- 'abr': 0,
}],
{
'en': [{
for mpd_file, mpd_url, mpd_base_url, expected_formats, expected_subtitles in _TEST_CASES:
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)
'vcodec': 'none',
'acodec': 'AACL',
'protocol': 'ism',
+ 'audio_channels': 2,
'_download_params': {
'stream_type': 'audio',
'duration': 8880746666,
'bits_per_sample': 16,
'nal_unit_length_field': 4
},
- 'audio_ext': 'isma',
- 'video_ext': 'none',
- 'abr': 128,
}, {
'format_id': 'video-100',
'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest',
'bits_per_sample': 16,
'nal_unit_length_field': 4
},
- 'video_ext': 'ismv',
- 'audio_ext': 'none',
- 'vbr': 100,
}, {
'format_id': 'video-326',
'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest',
'bits_per_sample': 16,
'nal_unit_length_field': 4
},
- 'video_ext': 'ismv',
- 'audio_ext': 'none',
- 'vbr': 326,
}, {
'format_id': 'video-698',
'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest',
'bits_per_sample': 16,
'nal_unit_length_field': 4
},
- 'video_ext': 'ismv',
- 'audio_ext': 'none',
- 'vbr': 698,
}, {
'format_id': 'video-1493',
'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest',
'bits_per_sample': 16,
'nal_unit_length_field': 4
},
- 'video_ext': 'ismv',
- 'audio_ext': 'none',
- 'vbr': 1493,
}, {
'format_id': 'video-4482',
'url': 'https://sdn-global-streaming-cache-3qsdn.akamaized.net/stream/3144/files/17/07/672975/3144-kZT4LWMQw6Rh7Kpd.ism/Manifest',
'bits_per_sample': 16,
'nal_unit_length_field': 4
},
- 'video_ext': 'ismv',
- 'audio_ext': 'none',
- 'vbr': 4482,
}],
{
'eng': [
]
},
),
+ (
+ 'ec-3_test',
+ 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ [{
+ 'format_id': 'audio_deu-127',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'isma',
+ 'tbr': 127,
+ 'asr': 48000,
+ 'vcodec': 'none',
+ 'acodec': 'AACL',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ 'audio_channels': 2,
+ '_download_params': {
+ 'stream_type': 'audio',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 0,
+ 'height': 0,
+ 'fourcc': 'AACL',
+ 'language': 'deu',
+ 'codec_private_data': '1190',
+ 'sampling_rate': 48000,
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'audio_deu_1-224',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'isma',
+ 'tbr': 224,
+ 'asr': 48000,
+ 'vcodec': 'none',
+ 'acodec': 'EC-3',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ 'audio_channels': 6,
+ '_download_params': {
+ 'stream_type': 'audio',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 0,
+ 'height': 0,
+ 'fourcc': 'EC-3',
+ 'language': 'deu',
+ 'codec_private_data': '00063F000000AF87FBA7022DFB42A4D405CD93843BDD0700200F00',
+ 'sampling_rate': 48000,
+ 'channels': 6,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-23',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 384,
+ 'height': 216,
+ 'tbr': 23,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 384,
+ 'height': 216,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '000000016742C00CDB06077E5C05A808080A00000300020000030009C0C02EE0177CC6300F142AE00000000168CA8DC8',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-403',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 400,
+ 'height': 224,
+ 'tbr': 403,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 400,
+ 'height': 224,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D4014E98323B602D4040405000003000100000300320F1429380000000168EAECF2',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-680',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 640,
+ 'height': 360,
+ 'tbr': 680,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 640,
+ 'height': 360,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D401EE981405FF2E02D4040405000000300100000030320F162D3800000000168EAECF2',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-1253',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 640,
+ 'height': 360,
+ 'tbr': 1253,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'vbr': 1253,
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 640,
+ 'height': 360,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D401EE981405FF2E02D4040405000000300100000030320F162D3800000000168EAECF2',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-2121',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 768,
+ 'height': 432,
+ 'tbr': 2121,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 768,
+ 'height': 432,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D401EECA0601BD80B50101014000003000400000300C83C58B6580000000168E93B3C80',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-3275',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 1280,
+ 'height': 720,
+ 'tbr': 3275,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 1280,
+ 'height': 720,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D4020ECA02802DD80B501010140000003004000000C83C60C65800000000168E93B3C80',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-5300',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 1920,
+ 'height': 1080,
+ 'tbr': 5300,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 1920,
+ 'height': 1080,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D4028ECA03C0113F2E02D4040405000000300100000030320F18319600000000168E93B3C80',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }, {
+ 'format_id': 'video_deu-8079',
+ 'url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'manifest_url': 'https://smstr01.dmm.t-online.de/smooth24/smoothstream_m1/streaming/sony/9221438342941275747/636887760842957027/25_km_h-Trailer-9221571562372022953_deu_20_1300k_HD_H_264_ISMV.ism/Manifest',
+ 'ext': 'ismv',
+ 'width': 1920,
+ 'height': 1080,
+ 'tbr': 8079,
+ 'vcodec': 'AVC1',
+ 'acodec': 'none',
+ 'protocol': 'ism',
+ 'language': 'deu',
+ '_download_params': {
+ 'stream_type': 'video',
+ 'duration': 370000000,
+ 'timescale': 10000000,
+ 'width': 1920,
+ 'height': 1080,
+ 'fourcc': 'AVC1',
+ 'language': 'deu',
+ 'codec_private_data': '00000001674D4028ECA03C0113F2E02D4040405000000300100000030320F18319600000000168E93B3C80',
+ 'channels': 2,
+ 'bits_per_sample': 16,
+ 'nal_unit_length_field': 4
+ },
+ }],
+ {},
+ ),
]
for ism_file, ism_url, expected_formats, expected_subtitles in _TEST_CASES:
with 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)
for f4m_file, f4m_url, expected_formats in _TEST_CASES:
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)
for xspf_file, xspf_url, expected_entries in _TEST_CASES:
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)):
# 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)
expected_status=TEAPOT_RESPONSE_STATUS)
self.assertEqual(content, TEAPOT_RESPONSE_BODY)
+ def test_search_nextjs_data(self):
+ data = '<script id="__NEXT_DATA__" type="application/json">{"props":{}}</script>'
+ self.assertEqual(self.ie._search_nextjs_data(data, None), {'props': {}})
+ self.assertEqual(self.ie._search_nextjs_data('', None, fatal=False), {})
+ self.assertEqual(self.ie._search_nextjs_data('', None, default=None), None)
+ self.assertEqual(self.ie._search_nextjs_data('', None, default={}), {})
+ with self.assertRaises(DeprecationWarning):
+ self.assertEqual(self.ie._search_nextjs_data('', None, default='{}'), {})
+
if __name__ == '__main__':
unittest.main()