_ 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$=low]'})
 228         ydl.process_ie_result(info_dict.copy())
 229         downloaded = ydl.downloaded_info_dicts[0]
 230         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 231
 232         formats = [
 233             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 234         ]
 235         info_dict = _make_result(formats)
 236
 237         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 238         ydl.process_ie_result(info_dict.copy())
 239         downloaded = ydl.downloaded_info_dicts[0]
 240         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 241
 242     def test_format_selection_string_ops(self):
 243         formats = [
 244             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 245         ]
 246         info_dict = _make_result(formats)
 247
 248         # equals (=)
 249         ydl = YDL({'format': '[format_id=abc-cba]'})
 250         ydl.process_ie_result(info_dict.copy())
 251         downloaded = ydl.downloaded_info_dicts[0]
 252         self.assertEqual(downloaded['format_id'], 'abc-cba')
 253
 254         # does not equal (!=)
 255         ydl = YDL({'format': '[format_id!=abc-cba]'})
 256         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 257
 258         # starts with (^=)
 259         ydl = YDL({'format': '[format_id^=abc]'})
 260         ydl.process_ie_result(info_dict.copy())
 261         downloaded = ydl.downloaded_info_dicts[0]
 262         self.assertEqual(downloaded['format_id'], 'abc-cba')
 263
 264         # does not start with (!^=)
 265         ydl = YDL({'format': '[format_id!^=abc-cba]'})
 266         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 267
 268         # ends with ($=)
 269         ydl = YDL({'format': '[format_id$=cba]'})
 270         ydl.process_ie_result(info_dict.copy())
 271         downloaded = ydl.downloaded_info_dicts[0]
 272         self.assertEqual(downloaded['format_id'], 'abc-cba')
 273
 274         # does not end with (!$=)
 275         ydl = YDL({'format': '[format_id!$=abc-cba]'})
 276         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 277
 278         # contains (*=)
 279         ydl = YDL({'format': '[format_id*=-]'})
 280         ydl.process_ie_result(info_dict.copy())
 281         downloaded = ydl.downloaded_info_dicts[0]
 282         self.assertEqual(downloaded['format_id'], 'abc-cba')
 283
 284         # does not contain (!*=)
 285         ydl = YDL({'format': '[format_id!*=-]'})
 286         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 287
 288     def test_youtube_format_selection(self):
 289         order = [
 290             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 291             # Apple HTTP Live Streaming
 292             '96', '95', '94', '93', '92', '132', '151',
 293             # 3D
 294             '85', '84', '102', '83', '101', '82', '100',
 295             # Dash video
 296             '137', '248', '136', '247', '135', '246',
 297             '245', '244', '134', '243', '133', '242', '160',
 298             # Dash audio
 299             '141', '172', '140', '171', '139',
 300         ]
 301
 302         def format_info(f_id):
 303             info = YoutubeIE._formats[f_id].copy()
 304
 305             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 306             # and 'vcodec', while in tests such information is incomplete since
 307             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 308             # test_YoutubeDL.test_youtube_format_selection is broken without
 309             # this fix
 310             if 'acodec' in info and 'vcodec' not in info:
 311                 info['vcodec'] = 'none'
 312             elif 'vcodec' in info and 'acodec' not in info:
 313                 info['acodec'] = 'none'
 314
 315             info['format_id'] = f_id
 316             info['url'] = 'url:' + f_id
 317             return info
 318         formats_order = [format_info(f_id) for f_id in order]
 319
 320         info_dict = _make_result(list(formats_order), extractor='youtube')
 321         ydl = YDL({'format': 'bestvideo+bestaudio'})
 322         yie = YoutubeIE(ydl)
 323         yie._sort_formats(info_dict['formats'])
 324         ydl.process_ie_result(info_dict)
 325         downloaded = ydl.downloaded_info_dicts[0]
 326         self.assertEqual(downloaded['format_id'], '137+141')
 327         self.assertEqual(downloaded['ext'], 'mp4')
 328
 329         info_dict = _make_result(list(formats_order), extractor='youtube')
 330         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 331         yie = YoutubeIE(ydl)
 332         yie._sort_formats(info_dict['formats'])
 333         ydl.process_ie_result(info_dict)
 334         downloaded = ydl.downloaded_info_dicts[0]
 335         self.assertEqual(downloaded['format_id'], '38')
 336
 337         info_dict = _make_result(list(formats_order), extractor='youtube')
 338         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 339         yie = YoutubeIE(ydl)
 340         yie._sort_formats(info_dict['formats'])
 341         ydl.process_ie_result(info_dict)
 342         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 343         self.assertEqual(downloaded_ids, ['137', '141'])
 344
 345         info_dict = _make_result(list(formats_order), extractor='youtube')
 346         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 347         yie = YoutubeIE(ydl)
 348         yie._sort_formats(info_dict['formats'])
 349         ydl.process_ie_result(info_dict)
 350         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 351         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 352
 353         info_dict = _make_result(list(formats_order), extractor='youtube')
 354         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 355         yie = YoutubeIE(ydl)
 356         yie._sort_formats(info_dict['formats'])
 357         ydl.process_ie_result(info_dict)
 358         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 359         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 360
 361         info_dict = _make_result(list(formats_order), extractor='youtube')
 362         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+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, ['248+141'])
 368
 369         for f1, f2 in zip(formats_order, formats_order[1:]):
 370             info_dict = _make_result([f1, f2], extractor='youtube')
 371             ydl = YDL({'format': 'best/bestvideo'})
 372             yie = YoutubeIE(ydl)
 373             yie._sort_formats(info_dict['formats'])
 374             ydl.process_ie_result(info_dict)
 375             downloaded = ydl.downloaded_info_dicts[0]
 376             self.assertEqual(downloaded['format_id'], f1['format_id'])
 377
 378             info_dict = _make_result([f2, f1], extractor='youtube')
 379             ydl = YDL({'format': 'best/bestvideo'})
 380             yie = YoutubeIE(ydl)
 381             yie._sort_formats(info_dict['formats'])
 382             ydl.process_ie_result(info_dict)
 383             downloaded = ydl.downloaded_info_dicts[0]
 384             self.assertEqual(downloaded['format_id'], f1['format_id'])
 385
 386     def test_audio_only_extractor_format_selection(self):
 387         # For extractors with incomplete formats (all formats are audio-only or
 388         # video-only) best and worst should fallback to corresponding best/worst
 389         # video-only or audio-only formats (as per
 390         # https://github.com/rg3/youtube-dl/pull/5556)
 391         formats = [
 392             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 393             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 394         ]
 395         info_dict = _make_result(formats)
 396
 397         ydl = YDL({'format': 'best'})
 398         ydl.process_ie_result(info_dict.copy())
 399         downloaded = ydl.downloaded_info_dicts[0]
 400         self.assertEqual(downloaded['format_id'], 'high')
 401
 402         ydl = YDL({'format': 'worst'})
 403         ydl.process_ie_result(info_dict.copy())
 404         downloaded = ydl.downloaded_info_dicts[0]
 405         self.assertEqual(downloaded['format_id'], 'low')
 406
 407     def test_format_not_available(self):
 408         formats = [
 409             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 410             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 411         ]
 412         info_dict = _make_result(formats)
 413
 414         # This must fail since complete video-audio format does not match filter
 415         # and extractor does not provide incomplete only formats (i.e. only
 416         # video-only or audio-only).
 417         ydl = YDL({'format': 'best[height>360]'})
 418         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 419
 420     def test_format_selection_issue_10083(self):
 421         # See https://github.com/rg3/youtube-dl/issues/10083
 422         formats = [
 423             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 424             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 425             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 426         ]
 427         info_dict = _make_result(formats)
 428
 429         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 430         ydl.process_ie_result(info_dict.copy())
 431         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 432
 433     def test_invalid_format_specs(self):
 434         def assert_syntax_error(format_spec):
 435             ydl = YDL({'format': format_spec})
 436             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 437             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 438
 439         assert_syntax_error('bestvideo,,best')
 440         assert_syntax_error('+bestaudio')
 441         assert_syntax_error('bestvideo+')
 442         assert_syntax_error('/')
 443
 444     def test_format_filtering(self):
 445         formats = [
 446             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 447             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 448             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 449             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 450             {'format_id': 'E', 'filesize': 3000},
 451             {'format_id': 'F'},
 452             {'format_id': 'G', 'filesize': 1000000},
 453         ]
 454         for f in formats:
 455             f['url'] = 'http://_/'
 456             f['ext'] = 'unknown'
 457         info_dict = _make_result(formats)
 458
 459         ydl = YDL({'format': 'best[filesize<3000]'})
 460         ydl.process_ie_result(info_dict)
 461         downloaded = ydl.downloaded_info_dicts[0]
 462         self.assertEqual(downloaded['format_id'], 'D')
 463
 464         ydl = YDL({'format': 'best[filesize<=3000]'})
 465         ydl.process_ie_result(info_dict)
 466         downloaded = ydl.downloaded_info_dicts[0]
 467         self.assertEqual(downloaded['format_id'], 'E')
 468
 469         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 470         ydl.process_ie_result(info_dict)
 471         downloaded = ydl.downloaded_info_dicts[0]
 472         self.assertEqual(downloaded['format_id'], 'F')
 473
 474         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 475         ydl.process_ie_result(info_dict)
 476         downloaded = ydl.downloaded_info_dicts[0]
 477         self.assertEqual(downloaded['format_id'], 'B')
 478
 479         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 480         ydl.process_ie_result(info_dict)
 481         downloaded = ydl.downloaded_info_dicts[0]
 482         self.assertEqual(downloaded['format_id'], 'C')
 483
 484         ydl = YDL({'format': '[filesize>?1]'})
 485         ydl.process_ie_result(info_dict)
 486         downloaded = ydl.downloaded_info_dicts[0]
 487         self.assertEqual(downloaded['format_id'], 'G')
 488
 489         ydl = YDL({'format': '[filesize<1M]'})
 490         ydl.process_ie_result(info_dict)
 491         downloaded = ydl.downloaded_info_dicts[0]
 492         self.assertEqual(downloaded['format_id'], 'E')
 493
 494         ydl = YDL({'format': '[filesize<1MiB]'})
 495         ydl.process_ie_result(info_dict)
 496         downloaded = ydl.downloaded_info_dicts[0]
 497         self.assertEqual(downloaded['format_id'], 'G')
 498
 499         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 500         ydl.process_ie_result(info_dict)
 501         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 502         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 503
 504         ydl = YDL({'format': 'best[height<40]'})
 505         try:
 506             ydl.process_ie_result(info_dict)
 507         except ExtractorError:
 508             pass
 509         self.assertEqual(ydl.downloaded_info_dicts, [])
 510
 511     def test_default_format_spec(self):
 512         ydl = YDL({'simulate': True})
 513         self.assertEqual(ydl._default_format_spec({}), 'bestvideo+bestaudio/best')
 514
 515         ydl = YDL({})
 516         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 517
 518         ydl = YDL({'simulate': True})
 519         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo+bestaudio/best')
 520
 521         ydl = YDL({'outtmpl': '-'})
 522         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 523
 524         ydl = YDL({})
 525         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo+bestaudio/best')
 526         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 527
 528
 529 class TestYoutubeDL(unittest.TestCase):
 530     def test_subtitles(self):
 531         def s_formats(lang, autocaption=False):
 532             return [{
 533                 'ext': ext,
 534                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 535                 '_auto': autocaption,
 536             } for ext in ['vtt', 'srt', 'ass']]
 537         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 538         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 539         info_dict = {
 540             'id': 'test',
 541             'title': 'Test',
 542             'url': 'http://localhost/video.mp4',
 543             'subtitles': subtitles,
 544             'automatic_captions': auto_captions,
 545             'extractor': 'TEST',
 546         }
 547
 548         def get_info(params={}):
 549             params.setdefault('simulate', True)
 550             ydl = YDL(params)
 551             ydl.report_warning = lambda *args, **kargs: None
 552             return ydl.process_video_result(info_dict, download=False)
 553
 554         result = get_info()
 555         self.assertFalse(result.get('requested_subtitles'))
 556         self.assertEqual(result['subtitles'], subtitles)
 557         self.assertEqual(result['automatic_captions'], auto_captions)
 558
 559         result = get_info({'writesubtitles': True})
 560         subs = result['requested_subtitles']
 561         self.assertTrue(subs)
 562         self.assertEqual(set(subs.keys()), set(['en']))
 563         self.assertTrue(subs['en'].get('data') is None)
 564         self.assertEqual(subs['en']['ext'], 'ass')
 565
 566         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 567         subs = result['requested_subtitles']
 568         self.assertEqual(subs['en']['ext'], 'srt')
 569
 570         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 571         subs = result['requested_subtitles']
 572         self.assertTrue(subs)
 573         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 574
 575         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 576         subs = result['requested_subtitles']
 577         self.assertTrue(subs)
 578         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 579         self.assertFalse(subs['es']['_auto'])
 580         self.assertTrue(subs['pt']['_auto'])
 581
 582         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 583         subs = result['requested_subtitles']
 584         self.assertTrue(subs)
 585         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 586         self.assertTrue(subs['es']['_auto'])
 587         self.assertTrue(subs['pt']['_auto'])
 588
 589     def test_add_extra_info(self):
 590         test_dict = {
 591             'extractor': 'Foo',
 592         }
 593         extra_info = {
 594             'extractor': 'Bar',
 595             'playlist': 'funny videos',
 596         }
 597         YDL.add_extra_info(test_dict, extra_info)
 598         self.assertEqual(test_dict['extractor'], 'Foo')
 599         self.assertEqual(test_dict['playlist'], 'funny videos')
 600
 601     def test_prepare_filename(self):
 602         info = {
 603             'id': '1234',
 604             'ext': 'mp4',
 605             'width': None,
 606             'height': 1080,
 607             'title1': '$PATH',
 608             'title2': '%PATH%',
 609         }
 610
 611         def fname(templ):
 612             ydl = YoutubeDL({'outtmpl': templ})
 613             return ydl.prepare_filename(info)
 614         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 615         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 616         # Replace missing fields with 'NA'
 617         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 618         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 619         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 620         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 621         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 622         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 623         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 624         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 625         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 626         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 627         self.assertEqual(fname('%%'), '%')
 628         self.assertEqual(fname('%%%%'), '%%')
 629         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 630         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 631         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 632         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 633         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 634         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 635
 636     def test_format_note(self):
 637         ydl = YoutubeDL()
 638         self.assertEqual(ydl._format_note({}), '')
 639         assertRegexpMatches(self, ydl._format_note({
 640             'vbr': 10,
 641         }), r'^\s*10k$')
 642         assertRegexpMatches(self, ydl._format_note({
 643             'fps': 30,
 644         }), r'^30fps$')
 645
 646     def test_postprocessors(self):
 647         filename = 'post-processor-testfile.mp4'
 648         audiofile = filename + '.mp3'
 649
 650         class SimplePP(PostProcessor):
 651             def run(self, info):
 652                 with open(audiofile, 'wt') as f:
 653                     f.write('EXAMPLE')
 654                 return [info['filepath']], info
 655
 656         def run_pp(params, PP):
 657             with open(filename, 'wt') as f:
 658                 f.write('EXAMPLE')
 659             ydl = YoutubeDL(params)
 660             ydl.add_post_processor(PP())
 661             ydl.post_process(filename, {'filepath': filename})
 662
 663         run_pp({'keepvideo': True}, SimplePP)
 664         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 665         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 666         os.unlink(filename)
 667         os.unlink(audiofile)
 668
 669         run_pp({'keepvideo': False}, SimplePP)
 670         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 671         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 672         os.unlink(audiofile)
 673
 674         class ModifierPP(PostProcessor):
 675             def run(self, info):
 676                 with open(info['filepath'], 'wt') as f:
 677                     f.write('MODIFIED')
 678                 return [], info
 679
 680         run_pp({'keepvideo': False}, ModifierPP)
 681         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 682         os.unlink(filename)
 683
 684     def test_match_filter(self):
 685         class FilterYDL(YDL):
 686             def __init__(self, *args, **kwargs):
 687                 super(FilterYDL, self).__init__(*args, **kwargs)
 688                 self.params['simulate'] = True
 689
 690             def process_info(self, info_dict):
 691                 super(YDL, self).process_info(info_dict)
 692
 693             def _match_entry(self, info_dict, incomplete):
 694                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 695                 if res is None:
 696                     self.downloaded_info_dicts.append(info_dict)
 697                 return res
 698
 699         first = {
 700             'id': '1',
 701             'url': TEST_URL,
 702             'title': 'one',
 703             'extractor': 'TEST',
 704             'duration': 30,
 705             'filesize': 10 * 1024,
 706             'playlist_id': '42',
 707             'uploader': "變態妍字幕版 太妍 тест",
 708             'creator': "тест ' 123 ' тест--",
 709         }
 710         second = {
 711             'id': '2',
 712             'url': TEST_URL,
 713             'title': 'two',
 714             'extractor': 'TEST',
 715             'duration': 10,
 716             'description': 'foo',
 717             'filesize': 5 * 1024,
 718             'playlist_id': '43',
 719             'uploader': "тест 123",
 720         }
 721         videos = [first, second]
 722
 723         def get_videos(filter_=None):
 724             ydl = FilterYDL({'match_filter': filter_})
 725             for v in videos:
 726                 ydl.process_ie_result(v, download=True)
 727             return [v['id'] for v in ydl.downloaded_info_dicts]
 728
 729         res = get_videos()
 730         self.assertEqual(res, ['1', '2'])
 731
 732         def f(v):
 733             if v['id'] == '1':
 734                 return None
 735             else:
 736                 return 'Video id is not 1'
 737         res = get_videos(f)
 738         self.assertEqual(res, ['1'])
 739
 740         f = match_filter_func('duration < 30')
 741         res = get_videos(f)
 742         self.assertEqual(res, ['2'])
 743
 744         f = match_filter_func('description = foo')
 745         res = get_videos(f)
 746         self.assertEqual(res, ['2'])
 747
 748         f = match_filter_func('description =? foo')
 749         res = get_videos(f)
 750         self.assertEqual(res, ['1', '2'])
 751
 752         f = match_filter_func('filesize > 5KiB')
 753         res = get_videos(f)
 754         self.assertEqual(res, ['1'])
 755
 756         f = match_filter_func('playlist_id = 42')
 757         res = get_videos(f)
 758         self.assertEqual(res, ['1'])
 759
 760         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 761         res = get_videos(f)
 762         self.assertEqual(res, ['1'])
 763
 764         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 765         res = get_videos(f)
 766         self.assertEqual(res, ['2'])
 767
 768         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 769         res = get_videos(f)
 770         self.assertEqual(res, ['1'])
 771
 772         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 773         res = get_videos(f)
 774         self.assertEqual(res, ['1'])
 775
 776         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 777         res = get_videos(f)
 778         self.assertEqual(res, [])
 779
 780     def test_playlist_items_selection(self):
 781         entries = [{
 782             'id': compat_str(i),
 783             'title': compat_str(i),
 784             'url': TEST_URL,
 785         } for i in range(1, 5)]
 786         playlist = {
 787             '_type': 'playlist',
 788             'id': 'test',
 789             'entries': entries,
 790             'extractor': 'test:playlist',
 791             'extractor_key': 'test:playlist',
 792             'webpage_url': 'http://example.com',
 793         }
 794
 795         def get_ids(params):
 796             ydl = YDL(params)
 797             # make a copy because the dictionary can be modified
 798             ydl.process_ie_result(playlist.copy())
 799             return [int(v['id']) for v in ydl.downloaded_info_dicts]
 800
 801         result = get_ids({})
 802         self.assertEqual(result, [1, 2, 3, 4])
 803
 804         result = get_ids({'playlistend': 10})
 805         self.assertEqual(result, [1, 2, 3, 4])
 806
 807         result = get_ids({'playlistend': 2})
 808         self.assertEqual(result, [1, 2])
 809
 810         result = get_ids({'playliststart': 10})
 811         self.assertEqual(result, [])
 812
 813         result = get_ids({'playliststart': 2})
 814         self.assertEqual(result, [2, 3, 4])
 815
 816         result = get_ids({'playlist_items': '2-4'})
 817         self.assertEqual(result, [2, 3, 4])
 818
 819         result = get_ids({'playlist_items': '2,4'})
 820         self.assertEqual(result, [2, 4])
 821
 822         result = get_ids({'playlist_items': '10'})
 823         self.assertEqual(result, [])
 824
 825         result = get_ids({'playlist_items': '3-10'})
 826         self.assertEqual(result, [3, 4])
 827
 828         result = get_ids({'playlist_items': '2-4,3-4,3'})
 829         self.assertEqual(result, [2, 3, 4])
 830
 831     def test_urlopen_no_file_protocol(self):
 832         # see https://github.com/rg3/youtube-dl/issues/8227
 833         ydl = YDL()
 834         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 835
 836     def test_do_not_override_ie_key_in_url_transparent(self):
 837         ydl = YDL()
 838
 839         class Foo1IE(InfoExtractor):
 840             _VALID_URL = r'foo1:'
 841
 842             def _real_extract(self, url):
 843                 return {
 844                     '_type': 'url_transparent',
 845                     'url': 'foo2:',
 846                     'ie_key': 'Foo2',
 847                     'title': 'foo1 title',
 848                     'id': 'foo1_id',
 849                 }
 850
 851         class Foo2IE(InfoExtractor):
 852             _VALID_URL = r'foo2:'
 853
 854             def _real_extract(self, url):
 855                 return {
 856                     '_type': 'url',
 857                     'url': 'foo3:',
 858                     'ie_key': 'Foo3',
 859                 }
 860
 861         class Foo3IE(InfoExtractor):
 862             _VALID_URL = r'foo3:'
 863
 864             def _real_extract(self, url):
 865                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 866
 867         ydl.add_info_extractor(Foo1IE(ydl))
 868         ydl.add_info_extractor(Foo2IE(ydl))
 869         ydl.add_info_extractor(Foo3IE(ydl))
 870         ydl.extract_info('foo1:')
 871         downloaded = ydl.downloaded_info_dicts[0]
 872         self.assertEqual(downloaded['url'], TEST_URL)
 873         self.assertEqual(downloaded['title'], 'foo1 title')
 874         self.assertEqual(downloaded['id'], 'testid')
 875         self.assertEqual(downloaded['extractor'], 'testex')
 876         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 877
 878
 879 if __name__ == '__main__':
 880     unittest.main()