_ Git - youtube-dl/blob - 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         'extractor_key': 'TestEx',
  45     }
  46     res.update(**kwargs)
  47     return res
  48
  49
  50 class TestFormatSelection(unittest.TestCase):
  51     def test_prefer_free_formats(self):
  52         # Same resolution => download webm
  53         ydl = YDL()
  54         ydl.params['prefer_free_formats'] = True
  55         formats = [
  56             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  57             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  58         ]
  59         info_dict = _make_result(formats)
  60         yie = YoutubeIE(ydl)
  61         yie._sort_formats(info_dict['formats'])
  62         ydl.process_ie_result(info_dict)
  63         downloaded = ydl.downloaded_info_dicts[0]
  64         self.assertEqual(downloaded['ext'], 'webm')
  65
  66         # Different resolution => download best quality (mp4)
  67         ydl = YDL()
  68         ydl.params['prefer_free_formats'] = True
  69         formats = [
  70             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  71             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  72         ]
  73         info_dict['formats'] = formats
  74         yie = YoutubeIE(ydl)
  75         yie._sort_formats(info_dict['formats'])
  76         ydl.process_ie_result(info_dict)
  77         downloaded = ydl.downloaded_info_dicts[0]
  78         self.assertEqual(downloaded['ext'], 'mp4')
  79
  80         # No prefer_free_formats => prefer mp4 and flv for greater compatibility
  81         ydl = YDL()
  82         ydl.params['prefer_free_formats'] = False
  83         formats = [
  84             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  85             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  86             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  87         ]
  88         info_dict['formats'] = formats
  89         yie = YoutubeIE(ydl)
  90         yie._sort_formats(info_dict['formats'])
  91         ydl.process_ie_result(info_dict)
  92         downloaded = ydl.downloaded_info_dicts[0]
  93         self.assertEqual(downloaded['ext'], 'mp4')
  94
  95         ydl = YDL()
  96         ydl.params['prefer_free_formats'] = False
  97         formats = [
  98             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  99             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 100         ]
 101         info_dict['formats'] = formats
 102         yie = YoutubeIE(ydl)
 103         yie._sort_formats(info_dict['formats'])
 104         ydl.process_ie_result(info_dict)
 105         downloaded = ydl.downloaded_info_dicts[0]
 106         self.assertEqual(downloaded['ext'], 'flv')
 107
 108     def test_format_selection(self):
 109         formats = [
 110             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 111             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 112             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 113             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 114             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 115         ]
 116         info_dict = _make_result(formats)
 117
 118         ydl = YDL({'format': '20/47'})
 119         ydl.process_ie_result(info_dict.copy())
 120         downloaded = ydl.downloaded_info_dicts[0]
 121         self.assertEqual(downloaded['format_id'], '47')
 122
 123         ydl = YDL({'format': '20/71/worst'})
 124         ydl.process_ie_result(info_dict.copy())
 125         downloaded = ydl.downloaded_info_dicts[0]
 126         self.assertEqual(downloaded['format_id'], '35')
 127
 128         ydl = YDL()
 129         ydl.process_ie_result(info_dict.copy())
 130         downloaded = ydl.downloaded_info_dicts[0]
 131         self.assertEqual(downloaded['format_id'], '2')
 132
 133         ydl = YDL({'format': 'webm/mp4'})
 134         ydl.process_ie_result(info_dict.copy())
 135         downloaded = ydl.downloaded_info_dicts[0]
 136         self.assertEqual(downloaded['format_id'], '47')
 137
 138         ydl = YDL({'format': '3gp/40/mp4'})
 139         ydl.process_ie_result(info_dict.copy())
 140         downloaded = ydl.downloaded_info_dicts[0]
 141         self.assertEqual(downloaded['format_id'], '35')
 142
 143         ydl = YDL({'format': 'example-with-dashes'})
 144         ydl.process_ie_result(info_dict.copy())
 145         downloaded = ydl.downloaded_info_dicts[0]
 146         self.assertEqual(downloaded['format_id'], 'example-with-dashes')
 147
 148     def test_format_selection_audio(self):
 149         formats = [
 150             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 151             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 152             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 153             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 154         ]
 155         info_dict = _make_result(formats)
 156
 157         ydl = YDL({'format': 'bestaudio'})
 158         ydl.process_ie_result(info_dict.copy())
 159         downloaded = ydl.downloaded_info_dicts[0]
 160         self.assertEqual(downloaded['format_id'], 'audio-high')
 161
 162         ydl = YDL({'format': 'worstaudio'})
 163         ydl.process_ie_result(info_dict.copy())
 164         downloaded = ydl.downloaded_info_dicts[0]
 165         self.assertEqual(downloaded['format_id'], 'audio-low')
 166
 167         formats = [
 168             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 169             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 170         ]
 171         info_dict = _make_result(formats)
 172
 173         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 174         ydl.process_ie_result(info_dict.copy())
 175         downloaded = ydl.downloaded_info_dicts[0]
 176         self.assertEqual(downloaded['format_id'], 'vid-high')
 177
 178     def test_format_selection_audio_exts(self):
 179         formats = [
 180             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 181             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 182             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 184             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 185         ]
 186
 187         info_dict = _make_result(formats)
 188         ydl = YDL({'format': 'best'})
 189         ie = YoutubeIE(ydl)
 190         ie._sort_formats(info_dict['formats'])
 191         ydl.process_ie_result(copy.deepcopy(info_dict))
 192         downloaded = ydl.downloaded_info_dicts[0]
 193         self.assertEqual(downloaded['format_id'], 'aac-64')
 194
 195         ydl = YDL({'format': 'mp3'})
 196         ie = YoutubeIE(ydl)
 197         ie._sort_formats(info_dict['formats'])
 198         ydl.process_ie_result(copy.deepcopy(info_dict))
 199         downloaded = ydl.downloaded_info_dicts[0]
 200         self.assertEqual(downloaded['format_id'], 'mp3-64')
 201
 202         ydl = YDL({'prefer_free_formats': True})
 203         ie = YoutubeIE(ydl)
 204         ie._sort_formats(info_dict['formats'])
 205         ydl.process_ie_result(copy.deepcopy(info_dict))
 206         downloaded = ydl.downloaded_info_dicts[0]
 207         self.assertEqual(downloaded['format_id'], 'ogg-64')
 208
 209     def test_format_selection_video(self):
 210         formats = [
 211             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 212             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 213             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 214         ]
 215         info_dict = _make_result(formats)
 216
 217         ydl = YDL({'format': 'bestvideo'})
 218         ydl.process_ie_result(info_dict.copy())
 219         downloaded = ydl.downloaded_info_dicts[0]
 220         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 221
 222         ydl = YDL({'format': 'worstvideo'})
 223         ydl.process_ie_result(info_dict.copy())
 224         downloaded = ydl.downloaded_info_dicts[0]
 225         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 226
 227         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 228         ydl.process_ie_result(info_dict.copy())
 229         downloaded = ydl.downloaded_info_dicts[0]
 230         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 231
 232         formats = [
 233             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 234         ]
 235         info_dict = _make_result(formats)
 236
 237         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 238         ydl.process_ie_result(info_dict.copy())
 239         downloaded = ydl.downloaded_info_dicts[0]
 240         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 241
 242     def test_youtube_format_selection(self):
 243         order = [
 244             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 245             # Apple HTTP Live Streaming
 246             '96', '95', '94', '93', '92', '132', '151',
 247             # 3D
 248             '85', '84', '102', '83', '101', '82', '100',
 249             # Dash video
 250             '137', '248', '136', '247', '135', '246',
 251             '245', '244', '134', '243', '133', '242', '160',
 252             # Dash audio
 253             '141', '172', '140', '171', '139',
 254         ]
 255
 256         def format_info(f_id):
 257             info = YoutubeIE._formats[f_id].copy()
 258
 259             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 260             # and 'vcodec', while in tests such information is incomplete since
 261             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 262             # test_YoutubeDL.test_youtube_format_selection is broken without
 263             # this fix
 264             if 'acodec' in info and 'vcodec' not in info:
 265                 info['vcodec'] = 'none'
 266             elif 'vcodec' in info and 'acodec' not in info:
 267                 info['acodec'] = 'none'
 268
 269             info['format_id'] = f_id
 270             info['url'] = 'url:' + f_id
 271             return info
 272         formats_order = [format_info(f_id) for f_id in order]
 273
 274         info_dict = _make_result(list(formats_order), extractor='youtube')
 275         ydl = YDL({'format': 'bestvideo+bestaudio'})
 276         yie = YoutubeIE(ydl)
 277         yie._sort_formats(info_dict['formats'])
 278         ydl.process_ie_result(info_dict)
 279         downloaded = ydl.downloaded_info_dicts[0]
 280         self.assertEqual(downloaded['format_id'], '137+141')
 281         self.assertEqual(downloaded['ext'], 'mp4')
 282
 283         info_dict = _make_result(list(formats_order), extractor='youtube')
 284         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 285         yie = YoutubeIE(ydl)
 286         yie._sort_formats(info_dict['formats'])
 287         ydl.process_ie_result(info_dict)
 288         downloaded = ydl.downloaded_info_dicts[0]
 289         self.assertEqual(downloaded['format_id'], '38')
 290
 291         info_dict = _make_result(list(formats_order), extractor='youtube')
 292         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 293         yie = YoutubeIE(ydl)
 294         yie._sort_formats(info_dict['formats'])
 295         ydl.process_ie_result(info_dict)
 296         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 297         self.assertEqual(downloaded_ids, ['137', '141'])
 298
 299         info_dict = _make_result(list(formats_order), extractor='youtube')
 300         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 301         yie = YoutubeIE(ydl)
 302         yie._sort_formats(info_dict['formats'])
 303         ydl.process_ie_result(info_dict)
 304         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 305         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 306
 307         info_dict = _make_result(list(formats_order), extractor='youtube')
 308         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 309         yie = YoutubeIE(ydl)
 310         yie._sort_formats(info_dict['formats'])
 311         ydl.process_ie_result(info_dict)
 312         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 313         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 314
 315         info_dict = _make_result(list(formats_order), extractor='youtube')
 316         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 317         yie = YoutubeIE(ydl)
 318         yie._sort_formats(info_dict['formats'])
 319         ydl.process_ie_result(info_dict)
 320         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 321         self.assertEqual(downloaded_ids, ['248+141'])
 322
 323         for f1, f2 in zip(formats_order, formats_order[1:]):
 324             info_dict = _make_result([f1, f2], extractor='youtube')
 325             ydl = YDL({'format': 'best/bestvideo'})
 326             yie = YoutubeIE(ydl)
 327             yie._sort_formats(info_dict['formats'])
 328             ydl.process_ie_result(info_dict)
 329             downloaded = ydl.downloaded_info_dicts[0]
 330             self.assertEqual(downloaded['format_id'], f1['format_id'])
 331
 332             info_dict = _make_result([f2, f1], extractor='youtube')
 333             ydl = YDL({'format': 'best/bestvideo'})
 334             yie = YoutubeIE(ydl)
 335             yie._sort_formats(info_dict['formats'])
 336             ydl.process_ie_result(info_dict)
 337             downloaded = ydl.downloaded_info_dicts[0]
 338             self.assertEqual(downloaded['format_id'], f1['format_id'])
 339
 340     def test_audio_only_extractor_format_selection(self):
 341         # For extractors with incomplete formats (all formats are audio-only or
 342         # video-only) best and worst should fallback to corresponding best/worst
 343         # video-only or audio-only formats (as per
 344         # https://github.com/rg3/youtube-dl/pull/5556)
 345         formats = [
 346             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 347             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 348         ]
 349         info_dict = _make_result(formats)
 350
 351         ydl = YDL({'format': 'best'})
 352         ydl.process_ie_result(info_dict.copy())
 353         downloaded = ydl.downloaded_info_dicts[0]
 354         self.assertEqual(downloaded['format_id'], 'high')
 355
 356         ydl = YDL({'format': 'worst'})
 357         ydl.process_ie_result(info_dict.copy())
 358         downloaded = ydl.downloaded_info_dicts[0]
 359         self.assertEqual(downloaded['format_id'], 'low')
 360
 361     def test_format_not_available(self):
 362         formats = [
 363             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 364             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 365         ]
 366         info_dict = _make_result(formats)
 367
 368         # This must fail since complete video-audio format does not match filter
 369         # and extractor does not provide incomplete only formats (i.e. only
 370         # video-only or audio-only).
 371         ydl = YDL({'format': 'best[height>360]'})
 372         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 373
 374     def test_invalid_format_specs(self):
 375         def assert_syntax_error(format_spec):
 376             ydl = YDL({'format': format_spec})
 377             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 378             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 379
 380         assert_syntax_error('bestvideo,,best')
 381         assert_syntax_error('+bestaudio')
 382         assert_syntax_error('bestvideo+')
 383         assert_syntax_error('/')
 384
 385     def test_format_filtering(self):
 386         formats = [
 387             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 388             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 389             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 390             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 391             {'format_id': 'E', 'filesize': 3000},
 392             {'format_id': 'F'},
 393             {'format_id': 'G', 'filesize': 1000000},
 394         ]
 395         for f in formats:
 396             f['url'] = 'http://_/'
 397             f['ext'] = 'unknown'
 398         info_dict = _make_result(formats)
 399
 400         ydl = YDL({'format': 'best[filesize<3000]'})
 401         ydl.process_ie_result(info_dict)
 402         downloaded = ydl.downloaded_info_dicts[0]
 403         self.assertEqual(downloaded['format_id'], 'D')
 404
 405         ydl = YDL({'format': 'best[filesize<=3000]'})
 406         ydl.process_ie_result(info_dict)
 407         downloaded = ydl.downloaded_info_dicts[0]
 408         self.assertEqual(downloaded['format_id'], 'E')
 409
 410         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 411         ydl.process_ie_result(info_dict)
 412         downloaded = ydl.downloaded_info_dicts[0]
 413         self.assertEqual(downloaded['format_id'], 'F')
 414
 415         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 416         ydl.process_ie_result(info_dict)
 417         downloaded = ydl.downloaded_info_dicts[0]
 418         self.assertEqual(downloaded['format_id'], 'B')
 419
 420         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 421         ydl.process_ie_result(info_dict)
 422         downloaded = ydl.downloaded_info_dicts[0]
 423         self.assertEqual(downloaded['format_id'], 'C')
 424
 425         ydl = YDL({'format': '[filesize>?1]'})
 426         ydl.process_ie_result(info_dict)
 427         downloaded = ydl.downloaded_info_dicts[0]
 428         self.assertEqual(downloaded['format_id'], 'G')
 429
 430         ydl = YDL({'format': '[filesize<1M]'})
 431         ydl.process_ie_result(info_dict)
 432         downloaded = ydl.downloaded_info_dicts[0]
 433         self.assertEqual(downloaded['format_id'], 'E')
 434
 435         ydl = YDL({'format': '[filesize<1MiB]'})
 436         ydl.process_ie_result(info_dict)
 437         downloaded = ydl.downloaded_info_dicts[0]
 438         self.assertEqual(downloaded['format_id'], 'G')
 439
 440         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 441         ydl.process_ie_result(info_dict)
 442         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 443         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 444
 445         ydl = YDL({'format': 'best[height<40]'})
 446         try:
 447             ydl.process_ie_result(info_dict)
 448         except ExtractorError:
 449             pass
 450         self.assertEqual(ydl.downloaded_info_dicts, [])
 451
 452
 453 class TestYoutubeDL(unittest.TestCase):
 454     def test_subtitles(self):
 455         def s_formats(lang, autocaption=False):
 456             return [{
 457                 'ext': ext,
 458                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 459                 '_auto': autocaption,
 460             } for ext in ['vtt', 'srt', 'ass']]
 461         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 462         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 463         info_dict = {
 464             'id': 'test',
 465             'title': 'Test',
 466             'url': 'http://localhost/video.mp4',
 467             'subtitles': subtitles,
 468             'automatic_captions': auto_captions,
 469             'extractor': 'TEST',
 470         }
 471
 472         def get_info(params={}):
 473             params.setdefault('simulate', True)
 474             ydl = YDL(params)
 475             ydl.report_warning = lambda *args, **kargs: None
 476             return ydl.process_video_result(info_dict, download=False)
 477
 478         result = get_info()
 479         self.assertFalse(result.get('requested_subtitles'))
 480         self.assertEqual(result['subtitles'], subtitles)
 481         self.assertEqual(result['automatic_captions'], auto_captions)
 482
 483         result = get_info({'writesubtitles': True})
 484         subs = result['requested_subtitles']
 485         self.assertTrue(subs)
 486         self.assertEqual(set(subs.keys()), set(['en']))
 487         self.assertTrue(subs['en'].get('data') is None)
 488         self.assertEqual(subs['en']['ext'], 'ass')
 489
 490         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 491         subs = result['requested_subtitles']
 492         self.assertEqual(subs['en']['ext'], 'srt')
 493
 494         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 495         subs = result['requested_subtitles']
 496         self.assertTrue(subs)
 497         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 498
 499         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 500         subs = result['requested_subtitles']
 501         self.assertTrue(subs)
 502         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 503         self.assertFalse(subs['es']['_auto'])
 504         self.assertTrue(subs['pt']['_auto'])
 505
 506         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 507         subs = result['requested_subtitles']
 508         self.assertTrue(subs)
 509         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 510         self.assertTrue(subs['es']['_auto'])
 511         self.assertTrue(subs['pt']['_auto'])
 512
 513     def test_add_extra_info(self):
 514         test_dict = {
 515             'extractor': 'Foo',
 516         }
 517         extra_info = {
 518             'extractor': 'Bar',
 519             'playlist': 'funny videos',
 520         }
 521         YDL.add_extra_info(test_dict, extra_info)
 522         self.assertEqual(test_dict['extractor'], 'Foo')
 523         self.assertEqual(test_dict['playlist'], 'funny videos')
 524
 525     def test_prepare_filename(self):
 526         info = {
 527             'id': '1234',
 528             'ext': 'mp4',
 529             'width': None,
 530             'height': 1080,
 531             'title1': '$PATH',
 532             'title2': '%PATH%',
 533         }
 534
 535         def fname(templ):
 536             ydl = YoutubeDL({'outtmpl': templ})
 537             return ydl.prepare_filename(info)
 538         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 539         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 540         # Replace missing fields with 'NA'
 541         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 542         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 543         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 544         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 545         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 546         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 547         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 548         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 549         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 550         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 551         self.assertEqual(fname('%%'), '%')
 552         self.assertEqual(fname('%%%%'), '%%')
 553         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 554         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 555         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 556         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 557         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 558         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 559
 560     def test_format_note(self):
 561         ydl = YoutubeDL()
 562         self.assertEqual(ydl._format_note({}), '')
 563         assertRegexpMatches(self, ydl._format_note({
 564             'vbr': 10,
 565         }), r'^\s*10k$')
 566         assertRegexpMatches(self, ydl._format_note({
 567             'fps': 30,
 568         }), r'^30fps$')
 569
 570     def test_postprocessors(self):
 571         filename = 'post-processor-testfile.mp4'
 572         audiofile = filename + '.mp3'
 573
 574         class SimplePP(PostProcessor):
 575             def run(self, info):
 576                 with open(audiofile, 'wt') as f:
 577                     f.write('EXAMPLE')
 578                 return [info['filepath']], info
 579
 580         def run_pp(params, PP):
 581             with open(filename, 'wt') as f:
 582                 f.write('EXAMPLE')
 583             ydl = YoutubeDL(params)
 584             ydl.add_post_processor(PP())
 585             ydl.post_process(filename, {'filepath': filename})
 586
 587         run_pp({'keepvideo': True}, SimplePP)
 588         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 589         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 590         os.unlink(filename)
 591         os.unlink(audiofile)
 592
 593         run_pp({'keepvideo': False}, SimplePP)
 594         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 595         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 596         os.unlink(audiofile)
 597
 598         class ModifierPP(PostProcessor):
 599             def run(self, info):
 600                 with open(info['filepath'], 'wt') as f:
 601                     f.write('MODIFIED')
 602                 return [], info
 603
 604         run_pp({'keepvideo': False}, ModifierPP)
 605         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 606         os.unlink(filename)
 607
 608     def test_match_filter(self):
 609         class FilterYDL(YDL):
 610             def __init__(self, *args, **kwargs):
 611                 super(FilterYDL, self).__init__(*args, **kwargs)
 612                 self.params['simulate'] = True
 613
 614             def process_info(self, info_dict):
 615                 super(YDL, self).process_info(info_dict)
 616
 617             def _match_entry(self, info_dict, incomplete):
 618                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 619                 if res is None:
 620                     self.downloaded_info_dicts.append(info_dict)
 621                 return res
 622
 623         first = {
 624             'id': '1',
 625             'url': TEST_URL,
 626             'title': 'one',
 627             'extractor': 'TEST',
 628             'duration': 30,
 629             'filesize': 10 * 1024,
 630             'playlist_id': '42',
 631             'uploader': "變態妍字幕版 太妍 тест",
 632             'creator': "тест ' 123 ' тест--",
 633         }
 634         second = {
 635             'id': '2',
 636             'url': TEST_URL,
 637             'title': 'two',
 638             'extractor': 'TEST',
 639             'duration': 10,
 640             'description': 'foo',
 641             'filesize': 5 * 1024,
 642             'playlist_id': '43',
 643             'uploader': "тест 123",
 644         }
 645         videos = [first, second]
 646
 647         def get_videos(filter_=None):
 648             ydl = FilterYDL({'match_filter': filter_})
 649             for v in videos:
 650                 ydl.process_ie_result(v, download=True)
 651             return [v['id'] for v in ydl.downloaded_info_dicts]
 652
 653         res = get_videos()
 654         self.assertEqual(res, ['1', '2'])
 655
 656         def f(v):
 657             if v['id'] == '1':
 658                 return None
 659             else:
 660                 return 'Video id is not 1'
 661         res = get_videos(f)
 662         self.assertEqual(res, ['1'])
 663
 664         f = match_filter_func('duration < 30')
 665         res = get_videos(f)
 666         self.assertEqual(res, ['2'])
 667
 668         f = match_filter_func('description = foo')
 669         res = get_videos(f)
 670         self.assertEqual(res, ['2'])
 671
 672         f = match_filter_func('description =? foo')
 673         res = get_videos(f)
 674         self.assertEqual(res, ['1', '2'])
 675
 676         f = match_filter_func('filesize > 5KiB')
 677         res = get_videos(f)
 678         self.assertEqual(res, ['1'])
 679
 680         f = match_filter_func('playlist_id = 42')
 681         res = get_videos(f)
 682         self.assertEqual(res, ['1'])
 683
 684         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 685         res = get_videos(f)
 686         self.assertEqual(res, ['1'])
 687
 688         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 689         res = get_videos(f)
 690         self.assertEqual(res, ['2'])
 691
 692         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 693         res = get_videos(f)
 694         self.assertEqual(res, ['1'])
 695
 696         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 697         res = get_videos(f)
 698         self.assertEqual(res, ['1'])
 699
 700         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 701         res = get_videos(f)
 702         self.assertEqual(res, [])
 703
 704     def test_playlist_items_selection(self):
 705         entries = [{
 706             'id': compat_str(i),
 707             'title': compat_str(i),
 708             'url': TEST_URL,
 709         } for i in range(1, 5)]
 710         playlist = {
 711             '_type': 'playlist',
 712             'id': 'test',
 713             'entries': entries,
 714             'extractor': 'test:playlist',
 715             'extractor_key': 'test:playlist',
 716             'webpage_url': 'http://example.com',
 717         }
 718
 719         def get_ids(params):
 720             ydl = YDL(params)
 721             # make a copy because the dictionary can be modified
 722             ydl.process_ie_result(playlist.copy())
 723             return [int(v['id']) for v in ydl.downloaded_info_dicts]
 724
 725         result = get_ids({})
 726         self.assertEqual(result, [1, 2, 3, 4])
 727
 728         result = get_ids({'playlistend': 10})
 729         self.assertEqual(result, [1, 2, 3, 4])
 730
 731         result = get_ids({'playlistend': 2})
 732         self.assertEqual(result, [1, 2])
 733
 734         result = get_ids({'playliststart': 10})
 735         self.assertEqual(result, [])
 736
 737         result = get_ids({'playliststart': 2})
 738         self.assertEqual(result, [2, 3, 4])
 739
 740         result = get_ids({'playlist_items': '2-4'})
 741         self.assertEqual(result, [2, 3, 4])
 742
 743         result = get_ids({'playlist_items': '2,4'})
 744         self.assertEqual(result, [2, 4])
 745
 746         result = get_ids({'playlist_items': '10'})
 747         self.assertEqual(result, [])
 748
 749     def test_urlopen_no_file_protocol(self):
 750         # see https://github.com/rg3/youtube-dl/issues/8227
 751         ydl = YDL()
 752         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 753
 754     def test_do_not_override_ie_key_in_url_transparent(self):
 755         ydl = YDL()
 756
 757         class Foo1IE(InfoExtractor):
 758             _VALID_URL = r'foo1:'
 759
 760             def _real_extract(self, url):
 761                 return {
 762                     '_type': 'url_transparent',
 763                     'url': 'foo2:',
 764                     'ie_key': 'Foo2',
 765                     'title': 'foo1 title',
 766                     'id': 'foo1_id',
 767                 }
 768
 769         class Foo2IE(InfoExtractor):
 770             _VALID_URL = r'foo2:'
 771
 772             def _real_extract(self, url):
 773                 return {
 774                     '_type': 'url',
 775                     'url': 'foo3:',
 776                     'ie_key': 'Foo3',
 777                 }
 778
 779         class Foo3IE(InfoExtractor):
 780             _VALID_URL = r'foo3:'
 781
 782             def _real_extract(self, url):
 783                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 784
 785         ydl.add_info_extractor(Foo1IE(ydl))
 786         ydl.add_info_extractor(Foo2IE(ydl))
 787         ydl.add_info_extractor(Foo3IE(ydl))
 788         ydl.extract_info('foo1:')
 789         downloaded = ydl.downloaded_info_dicts[0]
 790         self.assertEqual(downloaded['url'], TEST_URL)
 791         self.assertEqual(downloaded['title'], 'foo1 title')
 792         self.assertEqual(downloaded['id'], 'testid')
 793         self.assertEqual(downloaded['extractor'], 'testex')
 794         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 795
 796
 797 if __name__ == '__main__':
 798     unittest.main()