Raphaël G. Git Repositories - youtubedl/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.compat import compat_str
  16 from youtube_dl.extractor import YoutubeIE
  17 from youtube_dl.postprocessor.common import PostProcessor
  18 from youtube_dl.utils import match_filter_func
  19
  20 TEST_URL = 'http://localhost/sample.mp4'
  21
  22
  23 class YDL(FakeYDL):
  24     def __init__(self, *args, **kwargs):
  25         super(YDL, self).__init__(*args, **kwargs)
  26         self.downloaded_info_dicts = []
  27         self.msgs = []
  28
  29     def process_info(self, info_dict):
  30         self.downloaded_info_dicts.append(info_dict)
  31
  32     def to_screen(self, msg):
  33         self.msgs.append(msg)
  34
  35
  36 def _make_result(formats, **kwargs):
  37     res = {
  38         'formats': formats,
  39         'id': 'testid',
  40         'title': 'testttitle',
  41         'extractor': 'testex',
  42     }
  43     res.update(**kwargs)
  44     return res
  45
  46
  47 class TestFormatSelection(unittest.TestCase):
  48     def test_prefer_free_formats(self):
  49         # Same resolution => download webm
  50         ydl = YDL()
  51         ydl.params['prefer_free_formats'] = True
  52         formats = [
  53             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  54             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  55         ]
  56         info_dict = _make_result(formats)
  57         yie = YoutubeIE(ydl)
  58         yie._sort_formats(info_dict['formats'])
  59         ydl.process_ie_result(info_dict)
  60         downloaded = ydl.downloaded_info_dicts[0]
  61         self.assertEqual(downloaded['ext'], 'webm')
  62
  63         # Different resolution => download best quality (mp4)
  64         ydl = YDL()
  65         ydl.params['prefer_free_formats'] = True
  66         formats = [
  67             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  68             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  69         ]
  70         info_dict['formats'] = formats
  71         yie = YoutubeIE(ydl)
  72         yie._sort_formats(info_dict['formats'])
  73         ydl.process_ie_result(info_dict)
  74         downloaded = ydl.downloaded_info_dicts[0]
  75         self.assertEqual(downloaded['ext'], 'mp4')
  76
  77         # No prefer_free_formats => prefer mp4 and flv for greater compatibility
  78         ydl = YDL()
  79         ydl.params['prefer_free_formats'] = False
  80         formats = [
  81             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  82             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  83             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  84         ]
  85         info_dict['formats'] = formats
  86         yie = YoutubeIE(ydl)
  87         yie._sort_formats(info_dict['formats'])
  88         ydl.process_ie_result(info_dict)
  89         downloaded = ydl.downloaded_info_dicts[0]
  90         self.assertEqual(downloaded['ext'], 'mp4')
  91
  92         ydl = YDL()
  93         ydl.params['prefer_free_formats'] = False
  94         formats = [
  95             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  96             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  97         ]
  98         info_dict['formats'] = formats
  99         yie = YoutubeIE(ydl)
 100         yie._sort_formats(info_dict['formats'])
 101         ydl.process_ie_result(info_dict)
 102         downloaded = ydl.downloaded_info_dicts[0]
 103         self.assertEqual(downloaded['ext'], 'flv')
 104
 105     def test_format_selection(self):
 106         formats = [
 107             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 108             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 109             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 110             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 111         ]
 112         info_dict = _make_result(formats)
 113
 114         ydl = YDL({'format': '20/47'})
 115         ydl.process_ie_result(info_dict.copy())
 116         downloaded = ydl.downloaded_info_dicts[0]
 117         self.assertEqual(downloaded['format_id'], '47')
 118
 119         ydl = YDL({'format': '20/71/worst'})
 120         ydl.process_ie_result(info_dict.copy())
 121         downloaded = ydl.downloaded_info_dicts[0]
 122         self.assertEqual(downloaded['format_id'], '35')
 123
 124         ydl = YDL()
 125         ydl.process_ie_result(info_dict.copy())
 126         downloaded = ydl.downloaded_info_dicts[0]
 127         self.assertEqual(downloaded['format_id'], '2')
 128
 129         ydl = YDL({'format': 'webm/mp4'})
 130         ydl.process_ie_result(info_dict.copy())
 131         downloaded = ydl.downloaded_info_dicts[0]
 132         self.assertEqual(downloaded['format_id'], '47')
 133
 134         ydl = YDL({'format': '3gp/40/mp4'})
 135         ydl.process_ie_result(info_dict.copy())
 136         downloaded = ydl.downloaded_info_dicts[0]
 137         self.assertEqual(downloaded['format_id'], '35')
 138
 139     def test_format_selection_audio(self):
 140         formats = [
 141             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 142             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 143             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 144             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 145         ]
 146         info_dict = _make_result(formats)
 147
 148         ydl = YDL({'format': 'bestaudio'})
 149         ydl.process_ie_result(info_dict.copy())
 150         downloaded = ydl.downloaded_info_dicts[0]
 151         self.assertEqual(downloaded['format_id'], 'audio-high')
 152
 153         ydl = YDL({'format': 'worstaudio'})
 154         ydl.process_ie_result(info_dict.copy())
 155         downloaded = ydl.downloaded_info_dicts[0]
 156         self.assertEqual(downloaded['format_id'], 'audio-low')
 157
 158         formats = [
 159             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 160             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 161         ]
 162         info_dict = _make_result(formats)
 163
 164         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 165         ydl.process_ie_result(info_dict.copy())
 166         downloaded = ydl.downloaded_info_dicts[0]
 167         self.assertEqual(downloaded['format_id'], 'vid-high')
 168
 169     def test_format_selection_audio_exts(self):
 170         formats = [
 171             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 172             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 173             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 174             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 175             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 176         ]
 177
 178         info_dict = _make_result(formats)
 179         ydl = YDL({'format': 'best'})
 180         ie = YoutubeIE(ydl)
 181         ie._sort_formats(info_dict['formats'])
 182         ydl.process_ie_result(copy.deepcopy(info_dict))
 183         downloaded = ydl.downloaded_info_dicts[0]
 184         self.assertEqual(downloaded['format_id'], 'aac-64')
 185
 186         ydl = YDL({'format': 'mp3'})
 187         ie = YoutubeIE(ydl)
 188         ie._sort_formats(info_dict['formats'])
 189         ydl.process_ie_result(copy.deepcopy(info_dict))
 190         downloaded = ydl.downloaded_info_dicts[0]
 191         self.assertEqual(downloaded['format_id'], 'mp3-64')
 192
 193         ydl = YDL({'prefer_free_formats': True})
 194         ie = YoutubeIE(ydl)
 195         ie._sort_formats(info_dict['formats'])
 196         ydl.process_ie_result(copy.deepcopy(info_dict))
 197         downloaded = ydl.downloaded_info_dicts[0]
 198         self.assertEqual(downloaded['format_id'], 'ogg-64')
 199
 200     def test_format_selection_video(self):
 201         formats = [
 202             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 203             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 204             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 205         ]
 206         info_dict = _make_result(formats)
 207
 208         ydl = YDL({'format': 'bestvideo'})
 209         ydl.process_ie_result(info_dict.copy())
 210         downloaded = ydl.downloaded_info_dicts[0]
 211         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 212
 213         ydl = YDL({'format': 'worstvideo'})
 214         ydl.process_ie_result(info_dict.copy())
 215         downloaded = ydl.downloaded_info_dicts[0]
 216         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 217
 218     def test_youtube_format_selection(self):
 219         order = [
 220             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '36', '17', '13',
 221             # Apple HTTP Live Streaming
 222             '96', '95', '94', '93', '92', '132', '151',
 223             # 3D
 224             '85', '84', '102', '83', '101', '82', '100',
 225             # Dash video
 226             '137', '248', '136', '247', '135', '246',
 227             '245', '244', '134', '243', '133', '242', '160',
 228             # Dash audio
 229             '141', '172', '140', '171', '139',
 230         ]
 231
 232         for f1id, f2id in zip(order, order[1:]):
 233             f1 = YoutubeIE._formats[f1id].copy()
 234             f1['format_id'] = f1id
 235             f1['url'] = 'url:' + f1id
 236             f2 = YoutubeIE._formats[f2id].copy()
 237             f2['format_id'] = f2id
 238             f2['url'] = 'url:' + f2id
 239
 240             info_dict = _make_result([f1, f2], extractor='youtube')
 241             ydl = YDL({'format': 'best/bestvideo'})
 242             yie = YoutubeIE(ydl)
 243             yie._sort_formats(info_dict['formats'])
 244             ydl.process_ie_result(info_dict)
 245             downloaded = ydl.downloaded_info_dicts[0]
 246             self.assertEqual(downloaded['format_id'], f1id)
 247
 248             info_dict = _make_result([f2, f1], extractor='youtube')
 249             ydl = YDL({'format': 'best/bestvideo'})
 250             yie = YoutubeIE(ydl)
 251             yie._sort_formats(info_dict['formats'])
 252             ydl.process_ie_result(info_dict)
 253             downloaded = ydl.downloaded_info_dicts[0]
 254             self.assertEqual(downloaded['format_id'], f1id)
 255
 256     def test_format_filtering(self):
 257         formats = [
 258             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 259             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 260             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 261             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 262             {'format_id': 'E', 'filesize': 3000},
 263             {'format_id': 'F'},
 264             {'format_id': 'G', 'filesize': 1000000},
 265         ]
 266         for f in formats:
 267             f['url'] = 'http://_/'
 268             f['ext'] = 'unknown'
 269         info_dict = _make_result(formats)
 270
 271         ydl = YDL({'format': 'best[filesize<3000]'})
 272         ydl.process_ie_result(info_dict)
 273         downloaded = ydl.downloaded_info_dicts[0]
 274         self.assertEqual(downloaded['format_id'], 'D')
 275
 276         ydl = YDL({'format': 'best[filesize<=3000]'})
 277         ydl.process_ie_result(info_dict)
 278         downloaded = ydl.downloaded_info_dicts[0]
 279         self.assertEqual(downloaded['format_id'], 'E')
 280
 281         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 282         ydl.process_ie_result(info_dict)
 283         downloaded = ydl.downloaded_info_dicts[0]
 284         self.assertEqual(downloaded['format_id'], 'F')
 285
 286         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 287         ydl.process_ie_result(info_dict)
 288         downloaded = ydl.downloaded_info_dicts[0]
 289         self.assertEqual(downloaded['format_id'], 'B')
 290
 291         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 292         ydl.process_ie_result(info_dict)
 293         downloaded = ydl.downloaded_info_dicts[0]
 294         self.assertEqual(downloaded['format_id'], 'C')
 295
 296         ydl = YDL({'format': '[filesize>?1]'})
 297         ydl.process_ie_result(info_dict)
 298         downloaded = ydl.downloaded_info_dicts[0]
 299         self.assertEqual(downloaded['format_id'], 'G')
 300
 301         ydl = YDL({'format': '[filesize<1M]'})
 302         ydl.process_ie_result(info_dict)
 303         downloaded = ydl.downloaded_info_dicts[0]
 304         self.assertEqual(downloaded['format_id'], 'E')
 305
 306         ydl = YDL({'format': '[filesize<1MiB]'})
 307         ydl.process_ie_result(info_dict)
 308         downloaded = ydl.downloaded_info_dicts[0]
 309         self.assertEqual(downloaded['format_id'], 'G')
 310
 311
 312 class TestYoutubeDL(unittest.TestCase):
 313     def test_subtitles(self):
 314         def s_formats(lang, autocaption=False):
 315             return [{
 316                 'ext': ext,
 317                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 318                 '_auto': autocaption,
 319             } for ext in ['vtt', 'srt', 'ass']]
 320         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 321         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 322         info_dict = {
 323             'id': 'test',
 324             'title': 'Test',
 325             'url': 'http://localhost/video.mp4',
 326             'subtitles': subtitles,
 327             'automatic_captions': auto_captions,
 328             'extractor': 'TEST',
 329         }
 330
 331         def get_info(params={}):
 332             params.setdefault('simulate', True)
 333             ydl = YDL(params)
 334             ydl.report_warning = lambda *args, **kargs: None
 335             return ydl.process_video_result(info_dict, download=False)
 336
 337         result = get_info()
 338         self.assertFalse(result.get('requested_subtitles'))
 339         self.assertEqual(result['subtitles'], subtitles)
 340         self.assertEqual(result['automatic_captions'], auto_captions)
 341
 342         result = get_info({'writesubtitles': True})
 343         subs = result['requested_subtitles']
 344         self.assertTrue(subs)
 345         self.assertEqual(set(subs.keys()), set(['en']))
 346         self.assertTrue(subs['en'].get('data') is None)
 347         self.assertEqual(subs['en']['ext'], 'ass')
 348
 349         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 350         subs = result['requested_subtitles']
 351         self.assertEqual(subs['en']['ext'], 'srt')
 352
 353         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 354         subs = result['requested_subtitles']
 355         self.assertTrue(subs)
 356         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 357
 358         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 359         subs = result['requested_subtitles']
 360         self.assertTrue(subs)
 361         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 362         self.assertFalse(subs['es']['_auto'])
 363         self.assertTrue(subs['pt']['_auto'])
 364
 365         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 366         subs = result['requested_subtitles']
 367         self.assertTrue(subs)
 368         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 369         self.assertTrue(subs['es']['_auto'])
 370         self.assertTrue(subs['pt']['_auto'])
 371
 372     def test_add_extra_info(self):
 373         test_dict = {
 374             'extractor': 'Foo',
 375         }
 376         extra_info = {
 377             'extractor': 'Bar',
 378             'playlist': 'funny videos',
 379         }
 380         YDL.add_extra_info(test_dict, extra_info)
 381         self.assertEqual(test_dict['extractor'], 'Foo')
 382         self.assertEqual(test_dict['playlist'], 'funny videos')
 383
 384     def test_prepare_filename(self):
 385         info = {
 386             'id': '1234',
 387             'ext': 'mp4',
 388             'width': None,
 389         }
 390
 391         def fname(templ):
 392             ydl = YoutubeDL({'outtmpl': templ})
 393             return ydl.prepare_filename(info)
 394         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 395         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 396         # Replace missing fields with 'NA'
 397         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 398
 399     def test_format_note(self):
 400         ydl = YoutubeDL()
 401         self.assertEqual(ydl._format_note({}), '')
 402         assertRegexpMatches(self, ydl._format_note({
 403             'vbr': 10,
 404         }), '^\s*10k$')
 405
 406     def test_postprocessors(self):
 407         filename = 'post-processor-testfile.mp4'
 408         audiofile = filename + '.mp3'
 409
 410         class SimplePP(PostProcessor):
 411             def run(self, info):
 412                 with open(audiofile, 'wt') as f:
 413                     f.write('EXAMPLE')
 414                 return [info['filepath']], info
 415
 416         def run_pp(params, PP):
 417             with open(filename, 'wt') as f:
 418                 f.write('EXAMPLE')
 419             ydl = YoutubeDL(params)
 420             ydl.add_post_processor(PP())
 421             ydl.post_process(filename, {'filepath': filename})
 422
 423         run_pp({'keepvideo': True}, SimplePP)
 424         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 425         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 426         os.unlink(filename)
 427         os.unlink(audiofile)
 428
 429         run_pp({'keepvideo': False}, SimplePP)
 430         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 431         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 432         os.unlink(audiofile)
 433
 434         class ModifierPP(PostProcessor):
 435             def run(self, info):
 436                 with open(info['filepath'], 'wt') as f:
 437                     f.write('MODIFIED')
 438                 return [], info
 439
 440         run_pp({'keepvideo': False}, ModifierPP)
 441         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 442         os.unlink(filename)
 443
 444     def test_match_filter(self):
 445         class FilterYDL(YDL):
 446             def __init__(self, *args, **kwargs):
 447                 super(FilterYDL, self).__init__(*args, **kwargs)
 448                 self.params['simulate'] = True
 449
 450             def process_info(self, info_dict):
 451                 super(YDL, self).process_info(info_dict)
 452
 453             def _match_entry(self, info_dict, incomplete):
 454                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 455                 if res is None:
 456                     self.downloaded_info_dicts.append(info_dict)
 457                 return res
 458
 459         first = {
 460             'id': '1',
 461             'url': TEST_URL,
 462             'title': 'one',
 463             'extractor': 'TEST',
 464             'duration': 30,
 465             'filesize': 10 * 1024,
 466         }
 467         second = {
 468             'id': '2',
 469             'url': TEST_URL,
 470             'title': 'two',
 471             'extractor': 'TEST',
 472             'duration': 10,
 473             'description': 'foo',
 474             'filesize': 5 * 1024,
 475         }
 476         videos = [first, second]
 477
 478         def get_videos(filter_=None):
 479             ydl = FilterYDL({'match_filter': filter_})
 480             for v in videos:
 481                 ydl.process_ie_result(v, download=True)
 482             return [v['id'] for v in ydl.downloaded_info_dicts]
 483
 484         res = get_videos()
 485         self.assertEqual(res, ['1', '2'])
 486
 487         def f(v):
 488             if v['id'] == '1':
 489                 return None
 490             else:
 491                 return 'Video id is not 1'
 492         res = get_videos(f)
 493         self.assertEqual(res, ['1'])
 494
 495         f = match_filter_func('duration < 30')
 496         res = get_videos(f)
 497         self.assertEqual(res, ['2'])
 498
 499         f = match_filter_func('description = foo')
 500         res = get_videos(f)
 501         self.assertEqual(res, ['2'])
 502
 503         f = match_filter_func('description =? foo')
 504         res = get_videos(f)
 505         self.assertEqual(res, ['1', '2'])
 506
 507         f = match_filter_func('filesize > 5KiB')
 508         res = get_videos(f)
 509         self.assertEqual(res, ['1'])
 510
 511     def test_playlist_items_selection(self):
 512         entries = [{
 513             'id': compat_str(i),
 514             'title': compat_str(i),
 515             'url': TEST_URL,
 516         } for i in range(1, 5)]
 517         playlist = {
 518             '_type': 'playlist',
 519             'id': 'test',
 520             'entries': entries,
 521             'extractor': 'test:playlist',
 522             'extractor_key': 'test:playlist',
 523             'webpage_url': 'http://example.com',
 524         }
 525
 526         def get_ids(params):
 527             ydl = YDL(params)
 528             # make a copy because the dictionary can be modified
 529             ydl.process_ie_result(playlist.copy())
 530             return [int(v['id']) for v in ydl.downloaded_info_dicts]
 531
 532         result = get_ids({})
 533         self.assertEqual(result, [1, 2, 3, 4])
 534
 535         result = get_ids({'playlistend': 10})
 536         self.assertEqual(result, [1, 2, 3, 4])
 537
 538         result = get_ids({'playlistend': 2})
 539         self.assertEqual(result, [1, 2])
 540
 541         result = get_ids({'playliststart': 10})
 542         self.assertEqual(result, [])
 543
 544         result = get_ids({'playliststart': 2})
 545         self.assertEqual(result, [2, 3, 4])
 546
 547         result = get_ids({'playlist_items': '2-4'})
 548         self.assertEqual(result, [2, 3, 4])
 549
 550         result = get_ids({'playlist_items': '2,4'})
 551         self.assertEqual(result, [2, 4])
 552
 553         result = get_ids({'playlist_items': '10'})
 554         self.assertEqual(result, [])
 555
 556
 557 if __name__ == '__main__':
 558     unittest.main()