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