test/test_YoutubeDL.py

   1 #!/usr/bin/env python
   2 # coding: utf-8
   3
   4 from __future__ import unicode_literals
   5
   6 # Allow direct execution
   7 import os
   8 import sys
   9 import unittest
  10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12 import copy
  13
  14 from test.helper import FakeYDL, assertRegexpMatches
  15 from youtube_dl import YoutubeDL
  16 from youtube_dl.compat import compat_str, compat_urllib_error
  17 from youtube_dl.extractor import YoutubeIE
  18 from youtube_dl.extractor.common import InfoExtractor
  19 from youtube_dl.postprocessor.common import PostProcessor
  20 from youtube_dl.utils import ExtractorError, match_filter_func
  21
  22 TEST_URL = 'http://localhost/sample.mp4'
  23
  24
  25 class YDL(FakeYDL):
  26     def __init__(self, *args, **kwargs):
  27         super(YDL, self).__init__(*args, **kwargs)
  28         self.downloaded_info_dicts = []
  29         self.msgs = []
  30
  31     def process_info(self, info_dict):
  32         self.downloaded_info_dicts.append(info_dict)
  33
  34     def to_screen(self, msg):
  35         self.msgs.append(msg)
  36
  37
  38 def _make_result(formats, **kwargs):
  39     res = {
  40         'formats': formats,
  41         'id': 'testid',
  42         'title': 'testttitle',
  43         'extractor': 'testex',
  44     }
  45     res.update(**kwargs)
  46     return res
  47
  48
  49 class TestFormatSelection(unittest.TestCase):
  50     def test_prefer_free_formats(self):
  51         # Same resolution => download webm
  52         ydl = YDL()
  53         ydl.params['prefer_free_formats'] = True
  54         formats = [
  55             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  56             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  57         ]
  58         info_dict = _make_result(formats)
  59         yie = YoutubeIE(ydl)
  60         yie._sort_formats(info_dict['formats'])
  61         ydl.process_ie_result(info_dict)
  62         downloaded = ydl.downloaded_info_dicts[0]
  63         self.assertEqual(downloaded['ext'], 'webm')
  64
  65         # Different resolution => download best quality (mp4)
  66         ydl = YDL()
  67         ydl.params['prefer_free_formats'] = True
  68         formats = [
  69             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  70             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  71         ]
  72         info_dict['formats'] = formats
  73         yie = YoutubeIE(ydl)
  74         yie._sort_formats(info_dict['formats'])
  75         ydl.process_ie_result(info_dict)
  76         downloaded = ydl.downloaded_info_dicts[0]
  77         self.assertEqual(downloaded['ext'], 'mp4')
  78
  79         # No prefer_free_formats => prefer mp4 and flv for greater compatibility
  80         ydl = YDL()
  81         ydl.params['prefer_free_formats'] = False
  82         formats = [
  83             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  84             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  85             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  86         ]
  87         info_dict['formats'] = formats
  88         yie = YoutubeIE(ydl)
  89         yie._sort_formats(info_dict['formats'])
  90         ydl.process_ie_result(info_dict)
  91         downloaded = ydl.downloaded_info_dicts[0]
  92         self.assertEqual(downloaded['ext'], 'mp4')
  93
  94         ydl = YDL()
  95         ydl.params['prefer_free_formats'] = False
  96         formats = [
  97             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  98             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  99         ]
 100         info_dict['formats'] = formats
 101         yie = YoutubeIE(ydl)
 102         yie._sort_formats(info_dict['formats'])
 103         ydl.process_ie_result(info_dict)
 104         downloaded = ydl.downloaded_info_dicts[0]
 105         self.assertEqual(downloaded['ext'], 'flv')
 106
 107     def test_format_selection(self):
 108         formats = [
 109             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 110             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 111             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 112             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 113             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 114         ]
 115         info_dict = _make_result(formats)
 116
 117         ydl = YDL({'format': '20/47'})
 118         ydl.process_ie_result(info_dict.copy())
 119         downloaded = ydl.downloaded_info_dicts[0]
 120         self.assertEqual(downloaded['format_id'], '47')
 121
 122         ydl = YDL({'format': '20/71/worst'})
 123         ydl.process_ie_result(info_dict.copy())
 124         downloaded = ydl.downloaded_info_dicts[0]
 125         self.assertEqual(downloaded['format_id'], '35')
 126
 127         ydl = YDL()
 128         ydl.process_ie_result(info_dict.copy())
 129         downloaded = ydl.downloaded_info_dicts[0]
 130         self.assertEqual(downloaded['format_id'], '2')
 131
 132         ydl = YDL({'format': 'webm/mp4'})
 133         ydl.process_ie_result(info_dict.copy())
 134         downloaded = ydl.downloaded_info_dicts[0]
 135         self.assertEqual(downloaded['format_id'], '47')
 136
 137         ydl = YDL({'format': '3gp/40/mp4'})
 138         ydl.process_ie_result(info_dict.copy())
 139         downloaded = ydl.downloaded_info_dicts[0]
 140         self.assertEqual(downloaded['format_id'], '35')
 141
 142         ydl = YDL({'format': 'example-with-dashes'})
 143         ydl.process_ie_result(info_dict.copy())
 144         downloaded = ydl.downloaded_info_dicts[0]
 145         self.assertEqual(downloaded['format_id'], 'example-with-dashes')
 146
 147     def test_format_selection_audio(self):
 148         formats = [
 149             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 150             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 151             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 152             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 153         ]
 154         info_dict = _make_result(formats)
 155
 156         ydl = YDL({'format': 'bestaudio'})
 157         ydl.process_ie_result(info_dict.copy())
 158         downloaded = ydl.downloaded_info_dicts[0]
 159         self.assertEqual(downloaded['format_id'], 'audio-high')
 160
 161         ydl = YDL({'format': 'worstaudio'})
 162         ydl.process_ie_result(info_dict.copy())
 163         downloaded = ydl.downloaded_info_dicts[0]
 164         self.assertEqual(downloaded['format_id'], 'audio-low')
 165
 166         formats = [
 167             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 168             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 169         ]
 170         info_dict = _make_result(formats)
 171
 172         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 173         ydl.process_ie_result(info_dict.copy())
 174         downloaded = ydl.downloaded_info_dicts[0]
 175         self.assertEqual(downloaded['format_id'], 'vid-high')
 176
 177     def test_format_selection_audio_exts(self):
 178         formats = [
 179             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 180             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 181             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 182             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 184         ]
 185
 186         info_dict = _make_result(formats)
 187         ydl = YDL({'format': 'best'})
 188         ie = YoutubeIE(ydl)
 189         ie._sort_formats(info_dict['formats'])
 190         ydl.process_ie_result(copy.deepcopy(info_dict))
 191         downloaded = ydl.downloaded_info_dicts[0]
 192         self.assertEqual(downloaded['format_id'], 'aac-64')
 193
 194         ydl = YDL({'format': 'mp3'})
 195         ie = YoutubeIE(ydl)
 196         ie._sort_formats(info_dict['formats'])
 197         ydl.process_ie_result(copy.deepcopy(info_dict))
 198         downloaded = ydl.downloaded_info_dicts[0]
 199         self.assertEqual(downloaded['format_id'], 'mp3-64')
 200
 201         ydl = YDL({'prefer_free_formats': True})
 202         ie = YoutubeIE(ydl)
 203         ie._sort_formats(info_dict['formats'])
 204         ydl.process_ie_result(copy.deepcopy(info_dict))
 205         downloaded = ydl.downloaded_info_dicts[0]
 206         self.assertEqual(downloaded['format_id'], 'ogg-64')
 207
 208     def test_format_selection_video(self):
 209         formats = [
 210             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 211             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 212             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 213         ]
 214         info_dict = _make_result(formats)
 215
 216         ydl = YDL({'format': 'bestvideo'})
 217         ydl.process_ie_result(info_dict.copy())
 218         downloaded = ydl.downloaded_info_dicts[0]
 219         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 220
 221         ydl = YDL({'format': 'worstvideo'})
 222         ydl.process_ie_result(info_dict.copy())
 223         downloaded = ydl.downloaded_info_dicts[0]
 224         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 225
 226         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 227         ydl.process_ie_result(info_dict.copy())
 228         downloaded = ydl.downloaded_info_dicts[0]
 229         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 230
 231         formats = [
 232             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 233         ]
 234         info_dict = _make_result(formats)
 235
 236         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 237         ydl.process_ie_result(info_dict.copy())
 238         downloaded = ydl.downloaded_info_dicts[0]
 239         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 240
 241     def test_youtube_format_selection(self):
 242         order = [
 243             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 244             # Apple HTTP Live Streaming
 245             '96', '95', '94', '93', '92', '132', '151',
 246             # 3D
 247             '85', '84', '102', '83', '101', '82', '100',
 248             # Dash video
 249             '137', '248', '136', '247', '135', '246',
 250             '245', '244', '134', '243', '133', '242', '160',
 251             # Dash audio
 252             '141', '172', '140', '171', '139',
 253         ]
 254
 255         def format_info(f_id):
 256             info = YoutubeIE._formats[f_id].copy()
 257
 258             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 259             # and 'vcodec', while in tests such information is incomplete since
 260             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 261             # test_YoutubeDL.test_youtube_format_selection is broken without
 262             # this fix
 263             if 'acodec' in info and 'vcodec' not in info:
 264                 info['vcodec'] = 'none'
 265             elif 'vcodec' in info and 'acodec' not in info:
 266                 info['acodec'] = 'none'
 267
 268             info['format_id'] = f_id
 269             info['url'] = 'url:' + f_id
 270             return info
 271         formats_order = [format_info(f_id) for f_id in order]
 272
 273         info_dict = _make_result(list(formats_order), extractor='youtube')
 274         ydl = YDL({'format': 'bestvideo+bestaudio'})
 275         yie = YoutubeIE(ydl)
 276         yie._sort_formats(info_dict['formats'])
 277         ydl.process_ie_result(info_dict)
 278         downloaded = ydl.downloaded_info_dicts[0]
 279         self.assertEqual(downloaded['format_id'], '137+141')
 280         self.assertEqual(downloaded['ext'], 'mp4')
 281
 282         info_dict = _make_result(list(formats_order), extractor='youtube')
 283         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 284         yie = YoutubeIE(ydl)
 285         yie._sort_formats(info_dict['formats'])
 286         ydl.process_ie_result(info_dict)
 287         downloaded = ydl.downloaded_info_dicts[0]
 288         self.assertEqual(downloaded['format_id'], '38')
 289
 290         info_dict = _make_result(list(formats_order), extractor='youtube')
 291         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 292         yie = YoutubeIE(ydl)
 293         yie._sort_formats(info_dict['formats'])
 294         ydl.process_ie_result(info_dict)
 295         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 296         self.assertEqual(downloaded_ids, ['137', '141'])
 297
 298         info_dict = _make_result(list(formats_order), extractor='youtube')
 299         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 300         yie = YoutubeIE(ydl)
 301         yie._sort_formats(info_dict['formats'])
 302         ydl.process_ie_result(info_dict)
 303         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 304         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 305
 306         info_dict = _make_result(list(formats_order), extractor='youtube')
 307         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 308         yie = YoutubeIE(ydl)
 309         yie._sort_formats(info_dict['formats'])
 310         ydl.process_ie_result(info_dict)
 311         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 312         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 313
 314         info_dict = _make_result(list(formats_order), extractor='youtube')
 315         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 316         yie = YoutubeIE(ydl)
 317         yie._sort_formats(info_dict['formats'])
 318         ydl.process_ie_result(info_dict)
 319         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 320         self.assertEqual(downloaded_ids, ['248+141'])
 321
 322         for f1, f2 in zip(formats_order, formats_order[1:]):
 323             info_dict = _make_result([f1, f2], extractor='youtube')
 324             ydl = YDL({'format': 'best/bestvideo'})
 325             yie = YoutubeIE(ydl)
 326             yie._sort_formats(info_dict['formats'])
 327             ydl.process_ie_result(info_dict)
 328             downloaded = ydl.downloaded_info_dicts[0]
 329             self.assertEqual(downloaded['format_id'], f1['format_id'])
 330
 331             info_dict = _make_result([f2, f1], extractor='youtube')
 332             ydl = YDL({'format': 'best/bestvideo'})
 333             yie = YoutubeIE(ydl)
 334             yie._sort_formats(info_dict['formats'])
 335             ydl.process_ie_result(info_dict)
 336             downloaded = ydl.downloaded_info_dicts[0]
 337             self.assertEqual(downloaded['format_id'], f1['format_id'])
 338
 339     def test_audio_only_extractor_format_selection(self):
 340         # For extractors with incomplete formats (all formats are audio-only or
 341         # video-only) best and worst should fallback to corresponding best/worst
 342         # video-only or audio-only formats (as per
 343         # https://github.com/rg3/youtube-dl/pull/5556)
 344         formats = [
 345             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 346             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 347         ]
 348         info_dict = _make_result(formats)
 349
 350         ydl = YDL({'format': 'best'})
 351         ydl.process_ie_result(info_dict.copy())
 352         downloaded = ydl.downloaded_info_dicts[0]
 353         self.assertEqual(downloaded['format_id'], 'high')
 354
 355         ydl = YDL({'format': 'worst'})
 356         ydl.process_ie_result(info_dict.copy())
 357         downloaded = ydl.downloaded_info_dicts[0]
 358         self.assertEqual(downloaded['format_id'], 'low')
 359
 360     def test_format_not_available(self):
 361         formats = [
 362             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 363             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 364         ]
 365         info_dict = _make_result(formats)
 366
 367         # This must fail since complete video-audio format does not match filter
 368         # and extractor does not provide incomplete only formats (i.e. only
 369         # video-only or audio-only).
 370         ydl = YDL({'format': 'best[height>360]'})
 371         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 372
 373     def test_invalid_format_specs(self):
 374         def assert_syntax_error(format_spec):
 375             ydl = YDL({'format': format_spec})
 376             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 377             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 378
 379         assert_syntax_error('bestvideo,,best')
 380         assert_syntax_error('+bestaudio')
 381         assert_syntax_error('bestvideo+')
 382         assert_syntax_error('/')
 383
 384     def test_format_filtering(self):
 385         formats = [
 386             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 387             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 388             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 389             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 390             {'format_id': 'E', 'filesize': 3000},
 391             {'format_id': 'F'},
 392             {'format_id': 'G', 'filesize': 1000000},
 393         ]
 394         for f in formats:
 395             f['url'] = 'http://_/'
 396             f['ext'] = 'unknown'
 397         info_dict = _make_result(formats)
 398
 399         ydl = YDL({'format': 'best[filesize<3000]'})
 400         ydl.process_ie_result(info_dict)
 401         downloaded = ydl.downloaded_info_dicts[0]
 402         self.assertEqual(downloaded['format_id'], 'D')
 403
 404         ydl = YDL({'format': 'best[filesize<=3000]'})
 405         ydl.process_ie_result(info_dict)
 406         downloaded = ydl.downloaded_info_dicts[0]
 407         self.assertEqual(downloaded['format_id'], 'E')
 408
 409         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 410         ydl.process_ie_result(info_dict)
 411         downloaded = ydl.downloaded_info_dicts[0]
 412         self.assertEqual(downloaded['format_id'], 'F')
 413
 414         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 415         ydl.process_ie_result(info_dict)
 416         downloaded = ydl.downloaded_info_dicts[0]
 417         self.assertEqual(downloaded['format_id'], 'B')
 418
 419         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 420         ydl.process_ie_result(info_dict)
 421         downloaded = ydl.downloaded_info_dicts[0]
 422         self.assertEqual(downloaded['format_id'], 'C')
 423
 424         ydl = YDL({'format': '[filesize>?1]'})
 425         ydl.process_ie_result(info_dict)
 426         downloaded = ydl.downloaded_info_dicts[0]
 427         self.assertEqual(downloaded['format_id'], 'G')
 428
 429         ydl = YDL({'format': '[filesize<1M]'})
 430         ydl.process_ie_result(info_dict)
 431         downloaded = ydl.downloaded_info_dicts[0]
 432         self.assertEqual(downloaded['format_id'], 'E')
 433
 434         ydl = YDL({'format': '[filesize<1MiB]'})
 435         ydl.process_ie_result(info_dict)
 436         downloaded = ydl.downloaded_info_dicts[0]
 437         self.assertEqual(downloaded['format_id'], 'G')
 438
 439         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 440         ydl.process_ie_result(info_dict)
 441         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 442         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 443
 444         ydl = YDL({'format': 'best[height<40]'})
 445         try:
 446             ydl.process_ie_result(info_dict)
 447         except ExtractorError:
 448             pass
 449         self.assertEqual(ydl.downloaded_info_dicts, [])
 450
 451
 452 class TestYoutubeDL(unittest.TestCase):
 453     def test_subtitles(self):
 454         def s_formats(lang, autocaption=False):
 455             return [{
 456                 'ext': ext,
 457                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 458                 '_auto': autocaption,
 459             } for ext in ['vtt', 'srt', 'ass']]
 460         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 461         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 462         info_dict = {
 463             'id': 'test',
 464             'title': 'Test',
 465             'url': 'http://localhost/video.mp4',
 466             'subtitles': subtitles,
 467             'automatic_captions': auto_captions,
 468             'extractor': 'TEST',
 469         }
 470
 471         def get_info(params={}):
 472             params.setdefault('simulate', True)
 473             ydl = YDL(params)
 474             ydl.report_warning = lambda *args, **kargs: None
 475             return ydl.process_video_result(info_dict, download=False)
 476
 477         result = get_info()
 478         self.assertFalse(result.get('requested_subtitles'))
 479         self.assertEqual(result['subtitles'], subtitles)
 480         self.assertEqual(result['automatic_captions'], auto_captions)
 481
 482         result = get_info({'writesubtitles': True})
 483         subs = result['requested_subtitles']
 484         self.assertTrue(subs)
 485         self.assertEqual(set(subs.keys()), set(['en']))
 486         self.assertTrue(subs['en'].get('data') is None)
 487         self.assertEqual(subs['en']['ext'], 'ass')
 488
 489         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 490         subs = result['requested_subtitles']
 491         self.assertEqual(subs['en']['ext'], 'srt')
 492
 493         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 494         subs = result['requested_subtitles']
 495         self.assertTrue(subs)
 496         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 497
 498         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 499         subs = result['requested_subtitles']
 500         self.assertTrue(subs)
 501         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 502         self.assertFalse(subs['es']['_auto'])
 503         self.assertTrue(subs['pt']['_auto'])
 504
 505         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 506         subs = result['requested_subtitles']
 507         self.assertTrue(subs)
 508         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 509         self.assertTrue(subs['es']['_auto'])
 510         self.assertTrue(subs['pt']['_auto'])
 511
 512     def test_add_extra_info(self):
 513         test_dict = {
 514             'extractor': 'Foo',
 515         }
 516         extra_info = {
 517             'extractor': 'Bar',
 518             'playlist': 'funny videos',
 519         }
 520         YDL.add_extra_info(test_dict, extra_info)
 521         self.assertEqual(test_dict['extractor'], 'Foo')
 522         self.assertEqual(test_dict['playlist'], 'funny videos')
 523
 524     def test_prepare_filename(self):
 525         info = {
 526             'id': '1234',
 527             'ext': 'mp4',
 528             'width': None,
 529         }
 530
 531         def fname(templ):
 532             ydl = YoutubeDL({'outtmpl': templ})
 533             return ydl.prepare_filename(info)
 534         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 535         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 536         # Replace missing fields with 'NA'
 537         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 538
 539     def test_format_note(self):
 540         ydl = YoutubeDL()
 541         self.assertEqual(ydl._format_note({}), '')
 542         assertRegexpMatches(self, ydl._format_note({
 543             'vbr': 10,
 544         }), '^\s*10k$')
 545         assertRegexpMatches(self, ydl._format_note({
 546             'fps': 30,
 547         }), '^30fps$')
 548
 549     def test_postprocessors(self):
 550         filename = 'post-processor-testfile.mp4'
 551         audiofile = filename + '.mp3'
 552
 553         class SimplePP(PostProcessor):
 554             def run(self, info):
 555                 with open(audiofile, 'wt') as f:
 556                     f.write('EXAMPLE')
 557                 return [info['filepath']], info
 558
 559         def run_pp(params, PP):
 560             with open(filename, 'wt') as f:
 561                 f.write('EXAMPLE')
 562             ydl = YoutubeDL(params)
 563             ydl.add_post_processor(PP())
 564             ydl.post_process(filename, {'filepath': filename})
 565
 566         run_pp({'keepvideo': True}, SimplePP)
 567         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 568         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 569         os.unlink(filename)
 570         os.unlink(audiofile)
 571
 572         run_pp({'keepvideo': False}, SimplePP)
 573         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 574         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 575         os.unlink(audiofile)
 576
 577         class ModifierPP(PostProcessor):
 578             def run(self, info):
 579                 with open(info['filepath'], 'wt') as f:
 580                     f.write('MODIFIED')
 581                 return [], info
 582
 583         run_pp({'keepvideo': False}, ModifierPP)
 584         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 585         os.unlink(filename)
 586
 587     def test_match_filter(self):
 588         class FilterYDL(YDL):
 589             def __init__(self, *args, **kwargs):
 590                 super(FilterYDL, self).__init__(*args, **kwargs)
 591                 self.params['simulate'] = True
 592
 593             def process_info(self, info_dict):
 594                 super(YDL, self).process_info(info_dict)
 595
 596             def _match_entry(self, info_dict, incomplete):
 597                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 598                 if res is None:
 599                     self.downloaded_info_dicts.append(info_dict)
 600                 return res
 601
 602         first = {
 603             'id': '1',
 604             'url': TEST_URL,
 605             'title': 'one',
 606             'extractor': 'TEST',
 607             'duration': 30,
 608             'filesize': 10 * 1024,
 609             'playlist_id': '42',
 610             'uploader': "變態妍字幕版 太妍 тест",
 611             'creator': "тест ' 123 ' тест--",
 612         }
 613         second = {
 614             'id': '2',
 615             'url': TEST_URL,
 616             'title': 'two',
 617             'extractor': 'TEST',
 618             'duration': 10,
 619             'description': 'foo',
 620             'filesize': 5 * 1024,
 621             'playlist_id': '43',
 622             'uploader': "тест 123",
 623         }
 624         videos = [first, second]
 625
 626         def get_videos(filter_=None):
 627             ydl = FilterYDL({'match_filter': filter_})
 628             for v in videos:
 629                 ydl.process_ie_result(v, download=True)
 630             return [v['id'] for v in ydl.downloaded_info_dicts]
 631
 632         res = get_videos()
 633         self.assertEqual(res, ['1', '2'])
 634
 635         def f(v):
 636             if v['id'] == '1':
 637                 return None
 638             else:
 639                 return 'Video id is not 1'
 640         res = get_videos(f)
 641         self.assertEqual(res, ['1'])
 642
 643         f = match_filter_func('duration < 30')
 644         res = get_videos(f)
 645         self.assertEqual(res, ['2'])
 646
 647         f = match_filter_func('description = foo')
 648         res = get_videos(f)
 649         self.assertEqual(res, ['2'])
 650
 651         f = match_filter_func('description =? foo')
 652         res = get_videos(f)
 653         self.assertEqual(res, ['1', '2'])
 654
 655         f = match_filter_func('filesize > 5KiB')
 656         res = get_videos(f)
 657         self.assertEqual(res, ['1'])
 658
 659         f = match_filter_func('playlist_id = 42')
 660         res = get_videos(f)
 661         self.assertEqual(res, ['1'])
 662
 663         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 664         res = get_videos(f)
 665         self.assertEqual(res, ['1'])
 666
 667         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 668         res = get_videos(f)
 669         self.assertEqual(res, ['2'])
 670
 671         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 672         res = get_videos(f)
 673         self.assertEqual(res, ['1'])
 674
 675         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 676         res = get_videos(f)
 677         self.assertEqual(res, ['1'])
 678
 679         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 680         res = get_videos(f)
 681         self.assertEqual(res, [])
 682
 683     def test_playlist_items_selection(self):
 684         entries = [{
 685             'id': compat_str(i),
 686             'title': compat_str(i),
 687             'url': TEST_URL,
 688         } for i in range(1, 5)]
 689         playlist = {
 690             '_type': 'playlist',
 691             'id': 'test',
 692             'entries': entries,
 693             'extractor': 'test:playlist',
 694             'extractor_key': 'test:playlist',
 695             'webpage_url': 'http://example.com',
 696         }
 697
 698         def get_ids(params):
 699             ydl = YDL(params)
 700             # make a copy because the dictionary can be modified
 701             ydl.process_ie_result(playlist.copy())
 702             return [int(v['id']) for v in ydl.downloaded_info_dicts]
 703
 704         result = get_ids({})
 705         self.assertEqual(result, [1, 2, 3, 4])
 706
 707         result = get_ids({'playlistend': 10})
 708         self.assertEqual(result, [1, 2, 3, 4])
 709
 710         result = get_ids({'playlistend': 2})
 711         self.assertEqual(result, [1, 2])
 712
 713         result = get_ids({'playliststart': 10})
 714         self.assertEqual(result, [])
 715
 716         result = get_ids({'playliststart': 2})
 717         self.assertEqual(result, [2, 3, 4])
 718
 719         result = get_ids({'playlist_items': '2-4'})
 720         self.assertEqual(result, [2, 3, 4])
 721
 722         result = get_ids({'playlist_items': '2,4'})
 723         self.assertEqual(result, [2, 4])
 724
 725         result = get_ids({'playlist_items': '10'})
 726         self.assertEqual(result, [])
 727
 728     def test_urlopen_no_file_protocol(self):
 729         # see https://github.com/rg3/youtube-dl/issues/8227
 730         ydl = YDL()
 731         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 732
 733     def test_do_not_override_ie_key_in_url_transparent(self):
 734         ydl = YDL()
 735
 736         class Foo1IE(InfoExtractor):
 737             _VALID_URL = r'foo1:'
 738
 739             def _real_extract(self, url):
 740                 return {
 741                     '_type': 'url_transparent',
 742                     'url': 'foo2:',
 743                     'ie_key': 'Foo2',
 744                 }
 745
 746         class Foo2IE(InfoExtractor):
 747             _VALID_URL = r'foo2:'
 748
 749             def _real_extract(self, url):
 750                 return {
 751                     '_type': 'url',
 752                     'url': 'foo3:',
 753                     'ie_key': 'Foo3',
 754                 }
 755
 756         class Foo3IE(InfoExtractor):
 757             _VALID_URL = r'foo3:'
 758
 759             def _real_extract(self, url):
 760                 return _make_result([{'url': TEST_URL}])
 761
 762         ydl.add_info_extractor(Foo1IE(ydl))
 763         ydl.add_info_extractor(Foo2IE(ydl))
 764         ydl.add_info_extractor(Foo3IE(ydl))
 765         ydl.extract_info('foo1:')
 766         downloaded = ydl.downloaded_info_dicts[0]
 767         self.assertEqual(downloaded['url'], TEST_URL)
 768
 769
 770 if __name__ == '__main__':
 771     unittest.main()