_ Git - youtube-dl/blob - test/test_YoutubeDL.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 import copy
  12
  13 from test.helper import FakeYDL, assertRegexpMatches
  14 from youtube_dl import YoutubeDL
  15 from youtube_dl.extractor import YoutubeIE
  16 from youtube_dl.postprocessor.common import PostProcessor
  17 from youtube_dl.utils import match_filter_func
  18
  19 TEST_URL = 'http://localhost/sample.mp4'
  20
  21
  22 class YDL(FakeYDL):
  23     def __init__(self, *args, **kwargs):
  24         super(YDL, self).__init__(*args, **kwargs)
  25         self.downloaded_info_dicts = []
  26         self.msgs = []
  27
  28     def process_info(self, info_dict):
  29         self.downloaded_info_dicts.append(info_dict)
  30
  31     def to_screen(self, msg):
  32         self.msgs.append(msg)
  33
  34
  35 def _make_result(formats, **kwargs):
  36     res = {
  37         'formats': formats,
  38         'id': 'testid',
  39         'title': 'testttitle',
  40         'extractor': 'testex',
  41     }
  42     res.update(**kwargs)
  43     return res
  44
  45
  46 class TestFormatSelection(unittest.TestCase):
  47     def test_prefer_free_formats(self):
  48         # Same resolution => download webm
  49         ydl = YDL()
  50         ydl.params['prefer_free_formats'] = True
  51         formats = [
  52             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  53             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  54         ]
  55         info_dict = _make_result(formats)
  56         yie = YoutubeIE(ydl)
  57         yie._sort_formats(info_dict['formats'])
  58         ydl.process_ie_result(info_dict)
  59         downloaded = ydl.downloaded_info_dicts[0]
  60         self.assertEqual(downloaded['ext'], 'webm')
  61
  62         # Different resolution => download best quality (mp4)
  63         ydl = YDL()
  64         ydl.params['prefer_free_formats'] = True
  65         formats = [
  66             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  67             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  68         ]
  69         info_dict['formats'] = formats
  70         yie = YoutubeIE(ydl)
  71         yie._sort_formats(info_dict['formats'])
  72         ydl.process_ie_result(info_dict)
  73         downloaded = ydl.downloaded_info_dicts[0]
  74         self.assertEqual(downloaded['ext'], 'mp4')
  75
  76         # No prefer_free_formats => prefer mp4 and flv for greater compatibility
  77         ydl = YDL()
  78         ydl.params['prefer_free_formats'] = False
  79         formats = [
  80             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  81             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  82             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  83         ]
  84         info_dict['formats'] = formats
  85         yie = YoutubeIE(ydl)
  86         yie._sort_formats(info_dict['formats'])
  87         ydl.process_ie_result(info_dict)
  88         downloaded = ydl.downloaded_info_dicts[0]
  89         self.assertEqual(downloaded['ext'], 'mp4')
  90
  91         ydl = YDL()
  92         ydl.params['prefer_free_formats'] = False
  93         formats = [
  94             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  95             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  96         ]
  97         info_dict['formats'] = formats
  98         yie = YoutubeIE(ydl)
  99         yie._sort_formats(info_dict['formats'])
 100         ydl.process_ie_result(info_dict)
 101         downloaded = ydl.downloaded_info_dicts[0]
 102         self.assertEqual(downloaded['ext'], 'flv')
 103
 104     def test_format_limit(self):
 105         formats = [
 106             {'format_id': 'meh', 'url': 'http://example.com/meh', 'preference': 1},
 107             {'format_id': 'good', 'url': 'http://example.com/good', 'preference': 2},
 108             {'format_id': 'great', 'url': 'http://example.com/great', 'preference': 3},
 109             {'format_id': 'excellent', 'url': 'http://example.com/exc', 'preference': 4},
 110         ]
 111         info_dict = _make_result(formats)
 112
 113         ydl = YDL()
 114         ydl.process_ie_result(info_dict)
 115         downloaded = ydl.downloaded_info_dicts[0]
 116         self.assertEqual(downloaded['format_id'], 'excellent')
 117
 118         ydl = YDL({'format_limit': 'good'})
 119         assert ydl.params['format_limit'] == 'good'
 120         ydl.process_ie_result(info_dict.copy())
 121         downloaded = ydl.downloaded_info_dicts[0]
 122         self.assertEqual(downloaded['format_id'], 'good')
 123
 124         ydl = YDL({'format_limit': 'great', 'format': 'all'})
 125         ydl.process_ie_result(info_dict.copy())
 126         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'meh')
 127         self.assertEqual(ydl.downloaded_info_dicts[1]['format_id'], 'good')
 128         self.assertEqual(ydl.downloaded_info_dicts[2]['format_id'], 'great')
 129         self.assertTrue('3' in ydl.msgs[0])
 130
 131         ydl = YDL()
 132         ydl.params['format_limit'] = 'excellent'
 133         ydl.process_ie_result(info_dict.copy())
 134         downloaded = ydl.downloaded_info_dicts[0]
 135         self.assertEqual(downloaded['format_id'], 'excellent')
 136
 137     def test_format_selection(self):
 138         formats = [
 139             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 140             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 141             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 142             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 143         ]
 144         info_dict = _make_result(formats)
 145
 146         ydl = YDL({'format': '20/47'})
 147         ydl.process_ie_result(info_dict.copy())
 148         downloaded = ydl.downloaded_info_dicts[0]
 149         self.assertEqual(downloaded['format_id'], '47')
 150
 151         ydl = YDL({'format': '20/71/worst'})
 152         ydl.process_ie_result(info_dict.copy())
 153         downloaded = ydl.downloaded_info_dicts[0]
 154         self.assertEqual(downloaded['format_id'], '35')
 155
 156         ydl = YDL()
 157         ydl.process_ie_result(info_dict.copy())
 158         downloaded = ydl.downloaded_info_dicts[0]
 159         self.assertEqual(downloaded['format_id'], '2')
 160
 161         ydl = YDL({'format': 'webm/mp4'})
 162         ydl.process_ie_result(info_dict.copy())
 163         downloaded = ydl.downloaded_info_dicts[0]
 164         self.assertEqual(downloaded['format_id'], '47')
 165
 166         ydl = YDL({'format': '3gp/40/mp4'})
 167         ydl.process_ie_result(info_dict.copy())
 168         downloaded = ydl.downloaded_info_dicts[0]
 169         self.assertEqual(downloaded['format_id'], '35')
 170
 171     def test_format_selection_audio(self):
 172         formats = [
 173             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 174             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 175             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 176             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 177         ]
 178         info_dict = _make_result(formats)
 179
 180         ydl = YDL({'format': 'bestaudio'})
 181         ydl.process_ie_result(info_dict.copy())
 182         downloaded = ydl.downloaded_info_dicts[0]
 183         self.assertEqual(downloaded['format_id'], 'audio-high')
 184
 185         ydl = YDL({'format': 'worstaudio'})
 186         ydl.process_ie_result(info_dict.copy())
 187         downloaded = ydl.downloaded_info_dicts[0]
 188         self.assertEqual(downloaded['format_id'], 'audio-low')
 189
 190         formats = [
 191             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 192             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 193         ]
 194         info_dict = _make_result(formats)
 195
 196         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 197         ydl.process_ie_result(info_dict.copy())
 198         downloaded = ydl.downloaded_info_dicts[0]
 199         self.assertEqual(downloaded['format_id'], 'vid-high')
 200
 201     def test_format_selection_audio_exts(self):
 202         formats = [
 203             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 204             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 205             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 206             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 207             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 208         ]
 209
 210         info_dict = _make_result(formats)
 211         ydl = YDL({'format': 'best'})
 212         ie = YoutubeIE(ydl)
 213         ie._sort_formats(info_dict['formats'])
 214         ydl.process_ie_result(copy.deepcopy(info_dict))
 215         downloaded = ydl.downloaded_info_dicts[0]
 216         self.assertEqual(downloaded['format_id'], 'aac-64')
 217
 218         ydl = YDL({'format': 'mp3'})
 219         ie = YoutubeIE(ydl)
 220         ie._sort_formats(info_dict['formats'])
 221         ydl.process_ie_result(copy.deepcopy(info_dict))
 222         downloaded = ydl.downloaded_info_dicts[0]
 223         self.assertEqual(downloaded['format_id'], 'mp3-64')
 224
 225         ydl = YDL({'prefer_free_formats': True})
 226         ie = YoutubeIE(ydl)
 227         ie._sort_formats(info_dict['formats'])
 228         ydl.process_ie_result(copy.deepcopy(info_dict))
 229         downloaded = ydl.downloaded_info_dicts[0]
 230         self.assertEqual(downloaded['format_id'], 'ogg-64')
 231
 232     def test_format_selection_video(self):
 233         formats = [
 234             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 235             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 236             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 237         ]
 238         info_dict = _make_result(formats)
 239
 240         ydl = YDL({'format': 'bestvideo'})
 241         ydl.process_ie_result(info_dict.copy())
 242         downloaded = ydl.downloaded_info_dicts[0]
 243         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 244
 245         ydl = YDL({'format': 'worstvideo'})
 246         ydl.process_ie_result(info_dict.copy())
 247         downloaded = ydl.downloaded_info_dicts[0]
 248         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 249
 250     def test_youtube_format_selection(self):
 251         order = [
 252             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '36', '17', '13',
 253             # Apple HTTP Live Streaming
 254             '96', '95', '94', '93', '92', '132', '151',
 255             # 3D
 256             '85', '84', '102', '83', '101', '82', '100',
 257             # Dash video
 258             '137', '248', '136', '247', '135', '246',
 259             '245', '244', '134', '243', '133', '242', '160',
 260             # Dash audio
 261             '141', '172', '140', '171', '139',
 262         ]
 263
 264         for f1id, f2id in zip(order, order[1:]):
 265             f1 = YoutubeIE._formats[f1id].copy()
 266             f1['format_id'] = f1id
 267             f1['url'] = 'url:' + f1id
 268             f2 = YoutubeIE._formats[f2id].copy()
 269             f2['format_id'] = f2id
 270             f2['url'] = 'url:' + f2id
 271
 272             info_dict = _make_result([f1, f2], extractor='youtube')
 273             ydl = YDL()
 274             yie = YoutubeIE(ydl)
 275             yie._sort_formats(info_dict['formats'])
 276             ydl.process_ie_result(info_dict)
 277             downloaded = ydl.downloaded_info_dicts[0]
 278             self.assertEqual(downloaded['format_id'], f1id)
 279
 280             info_dict = _make_result([f2, f1], extractor='youtube')
 281             ydl = YDL()
 282             yie = YoutubeIE(ydl)
 283             yie._sort_formats(info_dict['formats'])
 284             ydl.process_ie_result(info_dict)
 285             downloaded = ydl.downloaded_info_dicts[0]
 286             self.assertEqual(downloaded['format_id'], f1id)
 287
 288     def test_format_filtering(self):
 289         formats = [
 290             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 291             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 292             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 293             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 294             {'format_id': 'E', 'filesize': 3000},
 295             {'format_id': 'F'},
 296             {'format_id': 'G', 'filesize': 1000000},
 297         ]
 298         for f in formats:
 299             f['url'] = 'http://_/'
 300             f['ext'] = 'unknown'
 301         info_dict = _make_result(formats)
 302
 303         ydl = YDL({'format': 'best[filesize<3000]'})
 304         ydl.process_ie_result(info_dict)
 305         downloaded = ydl.downloaded_info_dicts[0]
 306         self.assertEqual(downloaded['format_id'], 'D')
 307
 308         ydl = YDL({'format': 'best[filesize<=3000]'})
 309         ydl.process_ie_result(info_dict)
 310         downloaded = ydl.downloaded_info_dicts[0]
 311         self.assertEqual(downloaded['format_id'], 'E')
 312
 313         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 314         ydl.process_ie_result(info_dict)
 315         downloaded = ydl.downloaded_info_dicts[0]
 316         self.assertEqual(downloaded['format_id'], 'F')
 317
 318         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 319         ydl.process_ie_result(info_dict)
 320         downloaded = ydl.downloaded_info_dicts[0]
 321         self.assertEqual(downloaded['format_id'], 'B')
 322
 323         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 324         ydl.process_ie_result(info_dict)
 325         downloaded = ydl.downloaded_info_dicts[0]
 326         self.assertEqual(downloaded['format_id'], 'C')
 327
 328         ydl = YDL({'format': '[filesize>?1]'})
 329         ydl.process_ie_result(info_dict)
 330         downloaded = ydl.downloaded_info_dicts[0]
 331         self.assertEqual(downloaded['format_id'], 'G')
 332
 333         ydl = YDL({'format': '[filesize<1M]'})
 334         ydl.process_ie_result(info_dict)
 335         downloaded = ydl.downloaded_info_dicts[0]
 336         self.assertEqual(downloaded['format_id'], 'E')
 337
 338         ydl = YDL({'format': '[filesize<1MiB]'})
 339         ydl.process_ie_result(info_dict)
 340         downloaded = ydl.downloaded_info_dicts[0]
 341         self.assertEqual(downloaded['format_id'], 'G')
 342
 343
 344 class TestYoutubeDL(unittest.TestCase):
 345     def test_subtitles(self):
 346         def s_formats(lang, autocaption=False):
 347             return [{
 348                 'ext': ext,
 349                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 350                 '_auto': autocaption,
 351             } for ext in ['vtt', 'srt', 'ass']]
 352         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 353         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 354         info_dict = {
 355             'id': 'test',
 356             'title': 'Test',
 357             'url': 'http://localhost/video.mp4',
 358             'subtitles': subtitles,
 359             'automatic_captions': auto_captions,
 360             'extractor': 'TEST',
 361         }
 362
 363         def get_info(params={}):
 364             params.setdefault('simulate', True)
 365             ydl = YDL(params)
 366             ydl.report_warning = lambda *args, **kargs: None
 367             return ydl.process_video_result(info_dict, download=False)
 368
 369         result = get_info()
 370         self.assertFalse(result.get('requested_subtitles'))
 371         self.assertEqual(result['subtitles'], subtitles)
 372         self.assertEqual(result['automatic_captions'], auto_captions)
 373
 374         result = get_info({'writesubtitles': True})
 375         subs = result['requested_subtitles']
 376         self.assertTrue(subs)
 377         self.assertEqual(set(subs.keys()), set(['en']))
 378         self.assertTrue(subs['en'].get('data') is None)
 379         self.assertEqual(subs['en']['ext'], 'ass')
 380
 381         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 382         subs = result['requested_subtitles']
 383         self.assertEqual(subs['en']['ext'], 'srt')
 384
 385         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 386         subs = result['requested_subtitles']
 387         self.assertTrue(subs)
 388         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 389
 390         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 391         subs = result['requested_subtitles']
 392         self.assertTrue(subs)
 393         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 394         self.assertFalse(subs['es']['_auto'])
 395         self.assertTrue(subs['pt']['_auto'])
 396
 397         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 398         subs = result['requested_subtitles']
 399         self.assertTrue(subs)
 400         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 401         self.assertTrue(subs['es']['_auto'])
 402         self.assertTrue(subs['pt']['_auto'])
 403
 404     def test_add_extra_info(self):
 405         test_dict = {
 406             'extractor': 'Foo',
 407         }
 408         extra_info = {
 409             'extractor': 'Bar',
 410             'playlist': 'funny videos',
 411         }
 412         YDL.add_extra_info(test_dict, extra_info)
 413         self.assertEqual(test_dict['extractor'], 'Foo')
 414         self.assertEqual(test_dict['playlist'], 'funny videos')
 415
 416     def test_prepare_filename(self):
 417         info = {
 418             'id': '1234',
 419             'ext': 'mp4',
 420             'width': None,
 421         }
 422
 423         def fname(templ):
 424             ydl = YoutubeDL({'outtmpl': templ})
 425             return ydl.prepare_filename(info)
 426         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 427         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 428         # Replace missing fields with 'NA'
 429         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 430
 431     def test_format_note(self):
 432         ydl = YoutubeDL()
 433         self.assertEqual(ydl._format_note({}), '')
 434         assertRegexpMatches(self, ydl._format_note({
 435             'vbr': 10,
 436         }), '^\s*10k$')
 437
 438     def test_postprocessors(self):
 439         filename = 'post-processor-testfile.mp4'
 440         audiofile = filename + '.mp3'
 441
 442         class SimplePP(PostProcessor):
 443             def run(self, info):
 444                 with open(audiofile, 'wt') as f:
 445                     f.write('EXAMPLE')
 446                 info['filepath']
 447                 return False, info
 448
 449         def run_pp(params):
 450             with open(filename, 'wt') as f:
 451                 f.write('EXAMPLE')
 452             ydl = YoutubeDL(params)
 453             ydl.add_post_processor(SimplePP())
 454             ydl.post_process(filename, {'filepath': filename})
 455
 456         run_pp({'keepvideo': True})
 457         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 458         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 459         os.unlink(filename)
 460         os.unlink(audiofile)
 461
 462         run_pp({'keepvideo': False})
 463         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 464         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 465         os.unlink(audiofile)
 466
 467     def test_match_filter(self):
 468         class FilterYDL(YDL):
 469             def __init__(self, *args, **kwargs):
 470                 super(FilterYDL, self).__init__(*args, **kwargs)
 471                 self.params['simulate'] = True
 472
 473             def process_info(self, info_dict):
 474                 super(YDL, self).process_info(info_dict)
 475
 476             def _match_entry(self, info_dict, incomplete):
 477                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 478                 if res is None:
 479                     self.downloaded_info_dicts.append(info_dict)
 480                 return res
 481
 482         first = {
 483             'id': '1',
 484             'url': TEST_URL,
 485             'title': 'one',
 486             'extractor': 'TEST',
 487             'duration': 30,
 488             'filesize': 10 * 1024,
 489         }
 490         second = {
 491             'id': '2',
 492             'url': TEST_URL,
 493             'title': 'two',
 494             'extractor': 'TEST',
 495             'duration': 10,
 496             'description': 'foo',
 497             'filesize': 5 * 1024,
 498         }
 499         videos = [first, second]
 500
 501         def get_videos(filter_=None):
 502             ydl = FilterYDL({'match_filter': filter_})
 503             for v in videos:
 504                 ydl.process_ie_result(v, download=True)
 505             return [v['id'] for v in ydl.downloaded_info_dicts]
 506
 507         res = get_videos()
 508         self.assertEqual(res, ['1', '2'])
 509
 510         def f(v):
 511             if v['id'] == '1':
 512                 return None
 513             else:
 514                 return 'Video id is not 1'
 515         res = get_videos(f)
 516         self.assertEqual(res, ['1'])
 517
 518         f = match_filter_func('duration < 30')
 519         res = get_videos(f)
 520         self.assertEqual(res, ['2'])
 521
 522         f = match_filter_func('description = foo')
 523         res = get_videos(f)
 524         self.assertEqual(res, ['2'])
 525
 526         f = match_filter_func('description =? foo')
 527         res = get_videos(f)
 528         self.assertEqual(res, ['1', '2'])
 529
 530         f = match_filter_func('filesize > 5KiB')
 531         res = get_videos(f)
 532         self.assertEqual(res, ['1'])
 533
 534
 535 if __name__ == '__main__':
 536     unittest.main()