_ 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_format_selection_string_ops(self):
 243         formats = [
 244             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 245             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 246         ]
 247         info_dict = _make_result(formats)
 248
 249         # equals (=)
 250         ydl = YDL({'format': '[format_id=abc-cba]'})
 251         ydl.process_ie_result(info_dict.copy())
 252         downloaded = ydl.downloaded_info_dicts[0]
 253         self.assertEqual(downloaded['format_id'], 'abc-cba')
 254
 255         # does not equal (!=)
 256         ydl = YDL({'format': '[format_id!=abc-cba]'})
 257         ydl.process_ie_result(info_dict.copy())
 258         downloaded = ydl.downloaded_info_dicts[0]
 259         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 260
 261         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 262         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 263
 264         # starts with (^=)
 265         ydl = YDL({'format': '[format_id^=abc]'})
 266         ydl.process_ie_result(info_dict.copy())
 267         downloaded = ydl.downloaded_info_dicts[0]
 268         self.assertEqual(downloaded['format_id'], 'abc-cba')
 269
 270         # does not start with (!^=)
 271         ydl = YDL({'format': '[format_id!^=abc]'})
 272         ydl.process_ie_result(info_dict.copy())
 273         downloaded = ydl.downloaded_info_dicts[0]
 274         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 275
 276         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 277         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 278
 279         # ends with ($=)
 280         ydl = YDL({'format': '[format_id$=cba]'})
 281         ydl.process_ie_result(info_dict.copy())
 282         downloaded = ydl.downloaded_info_dicts[0]
 283         self.assertEqual(downloaded['format_id'], 'abc-cba')
 284
 285         # does not end with (!$=)
 286         ydl = YDL({'format': '[format_id!$=cba]'})
 287         ydl.process_ie_result(info_dict.copy())
 288         downloaded = ydl.downloaded_info_dicts[0]
 289         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 290
 291         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 292         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 293
 294         # contains (*=)
 295         ydl = YDL({'format': '[format_id*=bc-cb]'})
 296         ydl.process_ie_result(info_dict.copy())
 297         downloaded = ydl.downloaded_info_dicts[0]
 298         self.assertEqual(downloaded['format_id'], 'abc-cba')
 299
 300         # does not contain (!*=)
 301         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 302         ydl.process_ie_result(info_dict.copy())
 303         downloaded = ydl.downloaded_info_dicts[0]
 304         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 305
 306         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 307         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 308
 309         ydl = YDL({'format': '[format_id!*=-]'})
 310         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 311
 312     def test_youtube_format_selection(self):
 313         order = [
 314             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 315             # Apple HTTP Live Streaming
 316             '96', '95', '94', '93', '92', '132', '151',
 317             # 3D
 318             '85', '84', '102', '83', '101', '82', '100',
 319             # Dash video
 320             '137', '248', '136', '247', '135', '246',
 321             '245', '244', '134', '243', '133', '242', '160',
 322             # Dash audio
 323             '141', '172', '140', '171', '139',
 324         ]
 325
 326         def format_info(f_id):
 327             info = YoutubeIE._formats[f_id].copy()
 328
 329             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 330             # and 'vcodec', while in tests such information is incomplete since
 331             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 332             # test_YoutubeDL.test_youtube_format_selection is broken without
 333             # this fix
 334             if 'acodec' in info and 'vcodec' not in info:
 335                 info['vcodec'] = 'none'
 336             elif 'vcodec' in info and 'acodec' not in info:
 337                 info['acodec'] = 'none'
 338
 339             info['format_id'] = f_id
 340             info['url'] = 'url:' + f_id
 341             return info
 342         formats_order = [format_info(f_id) for f_id in order]
 343
 344         info_dict = _make_result(list(formats_order), extractor='youtube')
 345         ydl = YDL({'format': 'bestvideo+bestaudio'})
 346         yie = YoutubeIE(ydl)
 347         yie._sort_formats(info_dict['formats'])
 348         ydl.process_ie_result(info_dict)
 349         downloaded = ydl.downloaded_info_dicts[0]
 350         self.assertEqual(downloaded['format_id'], '137+141')
 351         self.assertEqual(downloaded['ext'], 'mp4')
 352
 353         info_dict = _make_result(list(formats_order), extractor='youtube')
 354         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 355         yie = YoutubeIE(ydl)
 356         yie._sort_formats(info_dict['formats'])
 357         ydl.process_ie_result(info_dict)
 358         downloaded = ydl.downloaded_info_dicts[0]
 359         self.assertEqual(downloaded['format_id'], '38')
 360
 361         info_dict = _make_result(list(formats_order), extractor='youtube')
 362         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 363         yie = YoutubeIE(ydl)
 364         yie._sort_formats(info_dict['formats'])
 365         ydl.process_ie_result(info_dict)
 366         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 367         self.assertEqual(downloaded_ids, ['137', '141'])
 368
 369         info_dict = _make_result(list(formats_order), extractor='youtube')
 370         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 371         yie = YoutubeIE(ydl)
 372         yie._sort_formats(info_dict['formats'])
 373         ydl.process_ie_result(info_dict)
 374         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 375         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 376
 377         info_dict = _make_result(list(formats_order), extractor='youtube')
 378         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 379         yie = YoutubeIE(ydl)
 380         yie._sort_formats(info_dict['formats'])
 381         ydl.process_ie_result(info_dict)
 382         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 383         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 384
 385         info_dict = _make_result(list(formats_order), extractor='youtube')
 386         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 387         yie = YoutubeIE(ydl)
 388         yie._sort_formats(info_dict['formats'])
 389         ydl.process_ie_result(info_dict)
 390         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 391         self.assertEqual(downloaded_ids, ['248+141'])
 392
 393         for f1, f2 in zip(formats_order, formats_order[1:]):
 394             info_dict = _make_result([f1, f2], extractor='youtube')
 395             ydl = YDL({'format': 'best/bestvideo'})
 396             yie = YoutubeIE(ydl)
 397             yie._sort_formats(info_dict['formats'])
 398             ydl.process_ie_result(info_dict)
 399             downloaded = ydl.downloaded_info_dicts[0]
 400             self.assertEqual(downloaded['format_id'], f1['format_id'])
 401
 402             info_dict = _make_result([f2, f1], extractor='youtube')
 403             ydl = YDL({'format': 'best/bestvideo'})
 404             yie = YoutubeIE(ydl)
 405             yie._sort_formats(info_dict['formats'])
 406             ydl.process_ie_result(info_dict)
 407             downloaded = ydl.downloaded_info_dicts[0]
 408             self.assertEqual(downloaded['format_id'], f1['format_id'])
 409
 410     def test_audio_only_extractor_format_selection(self):
 411         # For extractors with incomplete formats (all formats are audio-only or
 412         # video-only) best and worst should fallback to corresponding best/worst
 413         # video-only or audio-only formats (as per
 414         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 415         formats = [
 416             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 417             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 418         ]
 419         info_dict = _make_result(formats)
 420
 421         ydl = YDL({'format': 'best'})
 422         ydl.process_ie_result(info_dict.copy())
 423         downloaded = ydl.downloaded_info_dicts[0]
 424         self.assertEqual(downloaded['format_id'], 'high')
 425
 426         ydl = YDL({'format': 'worst'})
 427         ydl.process_ie_result(info_dict.copy())
 428         downloaded = ydl.downloaded_info_dicts[0]
 429         self.assertEqual(downloaded['format_id'], 'low')
 430
 431     def test_format_not_available(self):
 432         formats = [
 433             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 434             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 435         ]
 436         info_dict = _make_result(formats)
 437
 438         # This must fail since complete video-audio format does not match filter
 439         # and extractor does not provide incomplete only formats (i.e. only
 440         # video-only or audio-only).
 441         ydl = YDL({'format': 'best[height>360]'})
 442         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 443
 444     def test_format_selection_issue_10083(self):
 445         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 446         formats = [
 447             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 448             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 449             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 450         ]
 451         info_dict = _make_result(formats)
 452
 453         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 454         ydl.process_ie_result(info_dict.copy())
 455         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 456
 457     def test_invalid_format_specs(self):
 458         def assert_syntax_error(format_spec):
 459             ydl = YDL({'format': format_spec})
 460             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 461             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 462
 463         assert_syntax_error('bestvideo,,best')
 464         assert_syntax_error('+bestaudio')
 465         assert_syntax_error('bestvideo+')
 466         assert_syntax_error('/')
 467
 468     def test_format_filtering(self):
 469         formats = [
 470             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 471             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 472             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 473             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 474             {'format_id': 'E', 'filesize': 3000},
 475             {'format_id': 'F'},
 476             {'format_id': 'G', 'filesize': 1000000},
 477         ]
 478         for f in formats:
 479             f['url'] = 'http://_/'
 480             f['ext'] = 'unknown'
 481         info_dict = _make_result(formats)
 482
 483         ydl = YDL({'format': 'best[filesize<3000]'})
 484         ydl.process_ie_result(info_dict)
 485         downloaded = ydl.downloaded_info_dicts[0]
 486         self.assertEqual(downloaded['format_id'], 'D')
 487
 488         ydl = YDL({'format': 'best[filesize<=3000]'})
 489         ydl.process_ie_result(info_dict)
 490         downloaded = ydl.downloaded_info_dicts[0]
 491         self.assertEqual(downloaded['format_id'], 'E')
 492
 493         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 494         ydl.process_ie_result(info_dict)
 495         downloaded = ydl.downloaded_info_dicts[0]
 496         self.assertEqual(downloaded['format_id'], 'F')
 497
 498         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 499         ydl.process_ie_result(info_dict)
 500         downloaded = ydl.downloaded_info_dicts[0]
 501         self.assertEqual(downloaded['format_id'], 'B')
 502
 503         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 504         ydl.process_ie_result(info_dict)
 505         downloaded = ydl.downloaded_info_dicts[0]
 506         self.assertEqual(downloaded['format_id'], 'C')
 507
 508         ydl = YDL({'format': '[filesize>?1]'})
 509         ydl.process_ie_result(info_dict)
 510         downloaded = ydl.downloaded_info_dicts[0]
 511         self.assertEqual(downloaded['format_id'], 'G')
 512
 513         ydl = YDL({'format': '[filesize<1M]'})
 514         ydl.process_ie_result(info_dict)
 515         downloaded = ydl.downloaded_info_dicts[0]
 516         self.assertEqual(downloaded['format_id'], 'E')
 517
 518         ydl = YDL({'format': '[filesize<1MiB]'})
 519         ydl.process_ie_result(info_dict)
 520         downloaded = ydl.downloaded_info_dicts[0]
 521         self.assertEqual(downloaded['format_id'], 'G')
 522
 523         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 524         ydl.process_ie_result(info_dict)
 525         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 526         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 527
 528         ydl = YDL({'format': 'best[height<40]'})
 529         try:
 530             ydl.process_ie_result(info_dict)
 531         except ExtractorError:
 532             pass
 533         self.assertEqual(ydl.downloaded_info_dicts, [])
 534
 535     def test_default_format_spec(self):
 536         ydl = YDL({'simulate': True})
 537         self.assertEqual(ydl._default_format_spec({}), 'bestvideo+bestaudio/best')
 538
 539         ydl = YDL({})
 540         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 541
 542         ydl = YDL({'simulate': True})
 543         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo+bestaudio/best')
 544
 545         ydl = YDL({'outtmpl': '-'})
 546         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 547
 548         ydl = YDL({})
 549         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo+bestaudio/best')
 550         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 551
 552
 553 class TestYoutubeDL(unittest.TestCase):
 554     def test_subtitles(self):
 555         def s_formats(lang, autocaption=False):
 556             return [{
 557                 'ext': ext,
 558                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 559                 '_auto': autocaption,
 560             } for ext in ['vtt', 'srt', 'ass']]
 561         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 562         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 563         info_dict = {
 564             'id': 'test',
 565             'title': 'Test',
 566             'url': 'http://localhost/video.mp4',
 567             'subtitles': subtitles,
 568             'automatic_captions': auto_captions,
 569             'extractor': 'TEST',
 570         }
 571
 572         def get_info(params={}):
 573             params.setdefault('simulate', True)
 574             ydl = YDL(params)
 575             ydl.report_warning = lambda *args, **kargs: None
 576             return ydl.process_video_result(info_dict, download=False)
 577
 578         result = get_info()
 579         self.assertFalse(result.get('requested_subtitles'))
 580         self.assertEqual(result['subtitles'], subtitles)
 581         self.assertEqual(result['automatic_captions'], auto_captions)
 582
 583         result = get_info({'writesubtitles': True})
 584         subs = result['requested_subtitles']
 585         self.assertTrue(subs)
 586         self.assertEqual(set(subs.keys()), set(['en']))
 587         self.assertTrue(subs['en'].get('data') is None)
 588         self.assertEqual(subs['en']['ext'], 'ass')
 589
 590         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 591         subs = result['requested_subtitles']
 592         self.assertEqual(subs['en']['ext'], 'srt')
 593
 594         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 595         subs = result['requested_subtitles']
 596         self.assertTrue(subs)
 597         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 598
 599         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 600         subs = result['requested_subtitles']
 601         self.assertTrue(subs)
 602         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 603         self.assertFalse(subs['es']['_auto'])
 604         self.assertTrue(subs['pt']['_auto'])
 605
 606         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 607         subs = result['requested_subtitles']
 608         self.assertTrue(subs)
 609         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 610         self.assertTrue(subs['es']['_auto'])
 611         self.assertTrue(subs['pt']['_auto'])
 612
 613     def test_add_extra_info(self):
 614         test_dict = {
 615             'extractor': 'Foo',
 616         }
 617         extra_info = {
 618             'extractor': 'Bar',
 619             'playlist': 'funny videos',
 620         }
 621         YDL.add_extra_info(test_dict, extra_info)
 622         self.assertEqual(test_dict['extractor'], 'Foo')
 623         self.assertEqual(test_dict['playlist'], 'funny videos')
 624
 625     def test_prepare_filename(self):
 626         info = {
 627             'id': '1234',
 628             'ext': 'mp4',
 629             'width': None,
 630             'height': 1080,
 631             'title1': '$PATH',
 632             'title2': '%PATH%',
 633         }
 634
 635         def fname(templ):
 636             ydl = YoutubeDL({'outtmpl': templ})
 637             return ydl.prepare_filename(info)
 638         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 639         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 640         # Replace missing fields with 'NA'
 641         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 642         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 643         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 644         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 645         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 646         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 647         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 648         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 649         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 650         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 651         self.assertEqual(fname('%%'), '%')
 652         self.assertEqual(fname('%%%%'), '%%')
 653         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 654         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 655         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 656         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 657         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 658         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 659
 660     def test_format_note(self):
 661         ydl = YoutubeDL()
 662         self.assertEqual(ydl._format_note({}), '')
 663         assertRegexpMatches(self, ydl._format_note({
 664             'vbr': 10,
 665         }), r'^\s*10k$')
 666         assertRegexpMatches(self, ydl._format_note({
 667             'fps': 30,
 668         }), r'^30fps$')
 669
 670     def test_postprocessors(self):
 671         filename = 'post-processor-testfile.mp4'
 672         audiofile = filename + '.mp3'
 673
 674         class SimplePP(PostProcessor):
 675             def run(self, info):
 676                 with open(audiofile, 'wt') as f:
 677                     f.write('EXAMPLE')
 678                 return [info['filepath']], info
 679
 680         def run_pp(params, PP):
 681             with open(filename, 'wt') as f:
 682                 f.write('EXAMPLE')
 683             ydl = YoutubeDL(params)
 684             ydl.add_post_processor(PP())
 685             ydl.post_process(filename, {'filepath': filename})
 686
 687         run_pp({'keepvideo': True}, SimplePP)
 688         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 689         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 690         os.unlink(filename)
 691         os.unlink(audiofile)
 692
 693         run_pp({'keepvideo': False}, SimplePP)
 694         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 695         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 696         os.unlink(audiofile)
 697
 698         class ModifierPP(PostProcessor):
 699             def run(self, info):
 700                 with open(info['filepath'], 'wt') as f:
 701                     f.write('MODIFIED')
 702                 return [], info
 703
 704         run_pp({'keepvideo': False}, ModifierPP)
 705         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 706         os.unlink(filename)
 707
 708     def test_match_filter(self):
 709         class FilterYDL(YDL):
 710             def __init__(self, *args, **kwargs):
 711                 super(FilterYDL, self).__init__(*args, **kwargs)
 712                 self.params['simulate'] = True
 713
 714             def process_info(self, info_dict):
 715                 super(YDL, self).process_info(info_dict)
 716
 717             def _match_entry(self, info_dict, incomplete):
 718                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 719                 if res is None:
 720                     self.downloaded_info_dicts.append(info_dict)
 721                 return res
 722
 723         first = {
 724             'id': '1',
 725             'url': TEST_URL,
 726             'title': 'one',
 727             'extractor': 'TEST',
 728             'duration': 30,
 729             'filesize': 10 * 1024,
 730             'playlist_id': '42',
 731             'uploader': "變態妍字幕版 太妍 тест",
 732             'creator': "тест ' 123 ' тест--",
 733         }
 734         second = {
 735             'id': '2',
 736             'url': TEST_URL,
 737             'title': 'two',
 738             'extractor': 'TEST',
 739             'duration': 10,
 740             'description': 'foo',
 741             'filesize': 5 * 1024,
 742             'playlist_id': '43',
 743             'uploader': "тест 123",
 744         }
 745         videos = [first, second]
 746
 747         def get_videos(filter_=None):
 748             ydl = FilterYDL({'match_filter': filter_})
 749             for v in videos:
 750                 ydl.process_ie_result(v, download=True)
 751             return [v['id'] for v in ydl.downloaded_info_dicts]
 752
 753         res = get_videos()
 754         self.assertEqual(res, ['1', '2'])
 755
 756         def f(v):
 757             if v['id'] == '1':
 758                 return None
 759             else:
 760                 return 'Video id is not 1'
 761         res = get_videos(f)
 762         self.assertEqual(res, ['1'])
 763
 764         f = match_filter_func('duration < 30')
 765         res = get_videos(f)
 766         self.assertEqual(res, ['2'])
 767
 768         f = match_filter_func('description = foo')
 769         res = get_videos(f)
 770         self.assertEqual(res, ['2'])
 771
 772         f = match_filter_func('description =? foo')
 773         res = get_videos(f)
 774         self.assertEqual(res, ['1', '2'])
 775
 776         f = match_filter_func('filesize > 5KiB')
 777         res = get_videos(f)
 778         self.assertEqual(res, ['1'])
 779
 780         f = match_filter_func('playlist_id = 42')
 781         res = get_videos(f)
 782         self.assertEqual(res, ['1'])
 783
 784         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 785         res = get_videos(f)
 786         self.assertEqual(res, ['1'])
 787
 788         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 789         res = get_videos(f)
 790         self.assertEqual(res, ['2'])
 791
 792         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 793         res = get_videos(f)
 794         self.assertEqual(res, ['1'])
 795
 796         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 797         res = get_videos(f)
 798         self.assertEqual(res, ['1'])
 799
 800         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 801         res = get_videos(f)
 802         self.assertEqual(res, [])
 803
 804     def test_playlist_items_selection(self):
 805         entries = [{
 806             'id': compat_str(i),
 807             'title': compat_str(i),
 808             'url': TEST_URL,
 809         } for i in range(1, 5)]
 810         playlist = {
 811             '_type': 'playlist',
 812             'id': 'test',
 813             'entries': entries,
 814             'extractor': 'test:playlist',
 815             'extractor_key': 'test:playlist',
 816             'webpage_url': 'http://example.com',
 817         }
 818
 819         def get_downloaded_info_dicts(params):
 820             ydl = YDL(params)
 821             # make a deep copy because the dictionary and nested entries
 822             # can be modified
 823             ydl.process_ie_result(copy.deepcopy(playlist))
 824             return ydl.downloaded_info_dicts
 825
 826         def get_ids(params):
 827             return [int(v['id']) for v in get_downloaded_info_dicts(params)]
 828
 829         result = get_ids({})
 830         self.assertEqual(result, [1, 2, 3, 4])
 831
 832         result = get_ids({'playlistend': 10})
 833         self.assertEqual(result, [1, 2, 3, 4])
 834
 835         result = get_ids({'playlistend': 2})
 836         self.assertEqual(result, [1, 2])
 837
 838         result = get_ids({'playliststart': 10})
 839         self.assertEqual(result, [])
 840
 841         result = get_ids({'playliststart': 2})
 842         self.assertEqual(result, [2, 3, 4])
 843
 844         result = get_ids({'playlist_items': '2-4'})
 845         self.assertEqual(result, [2, 3, 4])
 846
 847         result = get_ids({'playlist_items': '2,4'})
 848         self.assertEqual(result, [2, 4])
 849
 850         result = get_ids({'playlist_items': '10'})
 851         self.assertEqual(result, [])
 852
 853         result = get_ids({'playlist_items': '3-10'})
 854         self.assertEqual(result, [3, 4])
 855
 856         result = get_ids({'playlist_items': '2-4,3-4,3'})
 857         self.assertEqual(result, [2, 3, 4])
 858
 859         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
 860         # @{
 861         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 862         self.assertEqual(result[0]['playlist_index'], 2)
 863         self.assertEqual(result[1]['playlist_index'], 3)
 864
 865         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 866         self.assertEqual(result[0]['playlist_index'], 2)
 867         self.assertEqual(result[1]['playlist_index'], 3)
 868         self.assertEqual(result[2]['playlist_index'], 4)
 869
 870         result = get_downloaded_info_dicts({'playlist_items': '4,2'})
 871         self.assertEqual(result[0]['playlist_index'], 4)
 872         self.assertEqual(result[1]['playlist_index'], 2)
 873         # @}
 874
 875     def test_urlopen_no_file_protocol(self):
 876         # see https://github.com/ytdl-org/youtube-dl/issues/8227
 877         ydl = YDL()
 878         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 879
 880     def test_do_not_override_ie_key_in_url_transparent(self):
 881         ydl = YDL()
 882
 883         class Foo1IE(InfoExtractor):
 884             _VALID_URL = r'foo1:'
 885
 886             def _real_extract(self, url):
 887                 return {
 888                     '_type': 'url_transparent',
 889                     'url': 'foo2:',
 890                     'ie_key': 'Foo2',
 891                     'title': 'foo1 title',
 892                     'id': 'foo1_id',
 893                 }
 894
 895         class Foo2IE(InfoExtractor):
 896             _VALID_URL = r'foo2:'
 897
 898             def _real_extract(self, url):
 899                 return {
 900                     '_type': 'url',
 901                     'url': 'foo3:',
 902                     'ie_key': 'Foo3',
 903                 }
 904
 905         class Foo3IE(InfoExtractor):
 906             _VALID_URL = r'foo3:'
 907
 908             def _real_extract(self, url):
 909                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 910
 911         ydl.add_info_extractor(Foo1IE(ydl))
 912         ydl.add_info_extractor(Foo2IE(ydl))
 913         ydl.add_info_extractor(Foo3IE(ydl))
 914         ydl.extract_info('foo1:')
 915         downloaded = ydl.downloaded_info_dicts[0]
 916         self.assertEqual(downloaded['url'], TEST_URL)
 917         self.assertEqual(downloaded['title'], 'foo1 title')
 918         self.assertEqual(downloaded['id'], 'testid')
 919         self.assertEqual(downloaded['extractor'], 'testex')
 920         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 921
 922
 923 if __name__ == '__main__':
 924     unittest.main()