Raphaël G. Git Repositories - youtubedl/blob - test/test_InfoExtractor.py

   1 #!/usr/bin/env python
   2
   3 from __future__ import unicode_literals
   4
   5 # Allow direct execution
   6 import os
   7 import sys
   8 import unittest
   9 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  10
  11 from test.helper import FakeYDL, expect_dict
  12 from youtube_dl.extractor.common import InfoExtractor
  13 from youtube_dl.extractor import YoutubeIE, get_info_extractor
  14 from youtube_dl.utils import encode_data_uri, strip_jsonp, ExtractorError, RegexNotFoundError
  15
  16
  17 class TestIE(InfoExtractor):
  18     pass
  19
  20
  21 class TestInfoExtractor(unittest.TestCase):
  22     def setUp(self):
  23         self.ie = TestIE(FakeYDL())
  24
  25     def test_ie_key(self):
  26         self.assertEqual(get_info_extractor(YoutubeIE.ie_key()), YoutubeIE)
  27
  28     def test_html_search_regex(self):
  29         html = '<p id="foo">Watch this <a href="http://www.youtube.com/watch?v=BaW_jenozKc">video</a></p>'
  30         search = lambda re, *args: self.ie._html_search_regex(re, html, *args)
  31         self.assertEqual(search(r'<p id="foo">(.+?)</p>', 'foo'), 'Watch this video')
  32
  33     def test_opengraph(self):
  34         ie = self.ie
  35         html = '''
  36             <meta name="og:title" content='Foo'/>
  37             <meta content="Some video's description " name="og:description"/>
  38             <meta property='og:image' content='http://domain.com/pic.jpg?key1=val1&amp;key2=val2'/>
  39             <meta content='application/x-shockwave-flash' property='og:video:type'>
  40             <meta content='Foo' property=og:foobar>
  41             <meta name="og:test1" content='foo > < bar'/>
  42             <meta name="og:test2" content="foo >//< bar"/>
  43             '''
  44         self.assertEqual(ie._og_search_title(html), 'Foo')
  45         self.assertEqual(ie._og_search_description(html), 'Some video\'s description ')
  46         self.assertEqual(ie._og_search_thumbnail(html), 'http://domain.com/pic.jpg?key1=val1&key2=val2')
  47         self.assertEqual(ie._og_search_video_url(html, default=None), None)
  48         self.assertEqual(ie._og_search_property('foobar', html), 'Foo')
  49         self.assertEqual(ie._og_search_property('test1', html), 'foo > < bar')
  50         self.assertEqual(ie._og_search_property('test2', html), 'foo >//< bar')
  51         self.assertEqual(ie._og_search_property(('test0', 'test1'), html), 'foo > < bar')
  52         self.assertRaises(RegexNotFoundError, ie._og_search_property, 'test0', html, None, fatal=True)
  53         self.assertRaises(RegexNotFoundError, ie._og_search_property, ('test0', 'test00'), html, None, fatal=True)
  54
  55     def test_html_search_meta(self):
  56         ie = self.ie
  57         html = '''
  58             <meta name="a" content="1" />
  59             <meta name='b' content='2'>
  60             <meta name="c" content='3'>
  61             <meta name=d content='4'>
  62             <meta property="e" content='5' >
  63             <meta content="6" name="f">
  64         '''
  65
  66         self.assertEqual(ie._html_search_meta('a', html), '1')
  67         self.assertEqual(ie._html_search_meta('b', html), '2')
  68         self.assertEqual(ie._html_search_meta('c', html), '3')
  69         self.assertEqual(ie._html_search_meta('d', html), '4')
  70         self.assertEqual(ie._html_search_meta('e', html), '5')
  71         self.assertEqual(ie._html_search_meta('f', html), '6')
  72         self.assertEqual(ie._html_search_meta(('a', 'b', 'c'), html), '1')
  73         self.assertEqual(ie._html_search_meta(('c', 'b', 'a'), html), '3')
  74         self.assertEqual(ie._html_search_meta(('z', 'x', 'c'), html), '3')
  75         self.assertRaises(RegexNotFoundError, ie._html_search_meta, 'z', html, None, fatal=True)
  76         self.assertRaises(RegexNotFoundError, ie._html_search_meta, ('z', 'x'), html, None, fatal=True)
  77
  78     def test_download_json(self):
  79         uri = encode_data_uri(b'{"foo": "blah"}', 'application/json')
  80         self.assertEqual(self.ie._download_json(uri, None), {'foo': 'blah'})
  81         uri = encode_data_uri(b'callback({"foo": "blah"})', 'application/javascript')
  82         self.assertEqual(self.ie._download_json(uri, None, transform_source=strip_jsonp), {'foo': 'blah'})
  83         uri = encode_data_uri(b'{"foo": invalid}', 'application/json')
  84         self.assertRaises(ExtractorError, self.ie._download_json, uri, None)
  85         self.assertEqual(self.ie._download_json(uri, None, fatal=False), None)
  86
  87     def test_extract_jwplayer_data_realworld(self):
  88         # from http://www.suffolk.edu/sjc/
  89         expect_dict(
  90             self,
  91             self.ie._extract_jwplayer_data(r'''
  92                 <script type='text/javascript'>
  93                     jwplayer('my-video').setup({
  94                         file: 'rtmp://192.138.214.154/live/sjclive',
  95                         fallback: 'true',
  96                         width: '95%',
  97                       aspectratio: '16:9',
  98                       primary: 'flash',
  99                       mediaid:'XEgvuql4'
 100                     });
 101                 </script>
 102                 ''', None, require_title=False),
 103             {
 104                 'id': 'XEgvuql4',
 105                 'formats': [{
 106                     'url': 'rtmp://192.138.214.154/live/sjclive',
 107                     'ext': 'flv'
 108                 }]
 109             })
 110
 111         # from https://www.pornoxo.com/videos/7564/striptease-from-sexy-secretary/
 112         expect_dict(
 113             self,
 114             self.ie._extract_jwplayer_data(r'''
 115 <script type="text/javascript">
 116     jwplayer("mediaplayer").setup({
 117         'videoid': "7564",
 118         'width': "100%",
 119         'aspectratio': "16:9",
 120         'stretching': "exactfit",
 121         'autostart': 'false',
 122         'flashplayer': "https://t04.vipstreamservice.com/jwplayer/v5.10/player.swf",
 123         'file': "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",
 124         'image': "https://t03.vipstreamservice.com/thumbs/pxo-full/2009-12/14/a4b2157147afe5efa93ce1978e0265289c193874e02597.flv-full-13.jpg",
 125         'filefallback': "https://cdn.pornoxo.com/key=9ZPsTR5EvPLQrBaak2MUGA,end=1489689259,ip=104.199.146.27/ip=104.199.146.27/speed=6573765/buffer=3.0/2009-12/m_4b2157147afe5efa93ce1978e0265289c193874e02597.mp4",
 126         'logo.hide': true,
 127         'skin': "https://t04.vipstreamservice.com/jwplayer/skin/modieus-blk.zip",
 128         'plugins': "https://t04.vipstreamservice.com/jwplayer/dock/dockableskinnableplugin.swf",
 129         'dockableskinnableplugin.piclink': "/index.php?key=ajax-videothumbsn&vid=7564&data=2009-12--14--4b2157147afe5efa93ce1978e0265289c193874e02597.flv--17370",
 130         'controlbar': 'bottom',
 131         'modes': [
 132             {type: 'flash', src: 'https://t04.vipstreamservice.com/jwplayer/v5.10/player.swf'}
 133         ],
 134         'provider': 'http'
 135     });
 136     //noinspection JSAnnotator
 137     invideo.setup({
 138         adsUrl: "/banner-iframe/?zoneId=32",
 139         adsUrl2: "",
 140         autostart: false
 141     });
 142 </script>
 143             ''', 'dummy', require_title=False),
 144             {
 145                 'thumbnail': 'https://t03.vipstreamservice.com/thumbs/pxo-full/2009-12/14/a4b2157147afe5efa93ce1978e0265289c193874e02597.flv-full-13.jpg',
 146                 'formats': [{
 147                     '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',
 148                     'ext': 'flv'
 149                 }]
 150             })
 151
 152         # from http://www.indiedb.com/games/king-machine/videos
 153         expect_dict(
 154             self,
 155             self.ie._extract_jwplayer_data(r'''
 156 <script>
 157 jwplayer("mediaplayer").setup({"abouttext":"Visit Indie DB","aboutlink":"http:\/\/www.indiedb.com\/","displaytitle":false,"autostart":false,"repeat":false,"title":"king machine trailer 1","sharing":{"link":"http:\/\/www.indiedb.com\/games\/king-machine\/videos\/king-machine-trailer-1","code":"<iframe width=\"560\" height=\"315\" src=\"http:\/\/www.indiedb.com\/media\/iframe\/1522983\" frameborder=\"0\" allowfullscreen><\/iframe><br><a href=\"http:\/\/www.indiedb.com\/games\/king-machine\/videos\/king-machine-trailer-1\">king machine trailer 1 - Indie DB<\/a>"},"related":{"file":"http:\/\/rss.indiedb.com\/media\/recommended\/1522983\/feed\/rss.xml","dimensions":"160x120","onclick":"link"},"sources":[{"file":"http:\/\/cdn.dbolical.com\/cache\/videos\/games\/1\/50\/49678\/encode_mp4\/king-machine-trailer.mp4","label":"360p SD","default":"true"},{"file":"http:\/\/cdn.dbolical.com\/cache\/videos\/games\/1\/50\/49678\/encode720p_mp4\/king-machine-trailer.mp4","label":"720p HD"}],"image":"http:\/\/media.indiedb.com\/cache\/images\/games\/1\/50\/49678\/thumb_620x2000\/king-machine-trailer.mp4.jpg","advertising":{"client":"vast","tag":"http:\/\/ads.intergi.com\/adrawdata\/3.0\/5205\/4251742\/0\/1013\/ADTECH;cors=yes;width=560;height=315;referring_url=http:\/\/www.indiedb.com\/games\/king-machine\/videos\/king-machine-trailer-1;content_url=http:\/\/www.indiedb.com\/games\/king-machine\/videos\/king-machine-trailer-1;media_id=1522983;title=king+machine+trailer+1;device=__DEVICE__;model=__MODEL__;os=Windows+OS;osversion=__OSVERSION__;ua=__UA__;ip=109.171.17.81;uniqueid=1522983;tags=__TAGS__;number=58cac25928151;time=1489683033"},"width":620,"height":349}).once("play", function(event) {
 158             videoAnalytics("play");
 159 }).once("complete", function(event) {
 160     videoAnalytics("completed");
 161 });
 162 </script>
 163                 ''', 'dummy'),
 164             {
 165                 'title': 'king machine trailer 1',
 166                 'thumbnail': 'http://media.indiedb.com/cache/images/games/1/50/49678/thumb_620x2000/king-machine-trailer.mp4.jpg',
 167                 'formats': [{
 168                     'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode_mp4/king-machine-trailer.mp4',
 169                     'height': 360,
 170                     'ext': 'mp4'
 171                 }, {
 172                     'url': 'http://cdn.dbolical.com/cache/videos/games/1/50/49678/encode720p_mp4/king-machine-trailer.mp4',
 173                     'height': 720,
 174                     'ext': 'mp4'
 175                 }]
 176             })
 177
 178
 179 if __name__ == '__main__':
 180     unittest.main()