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_dlc import YoutubeDL
  16 from youtube_dlc.compat import compat_str, compat_urllib_error
  17 from youtube_dlc.extractor import YoutubeIE
  18 from youtube_dlc.extractor.common import InfoExtractor
  19 from youtube_dlc.postprocessor.common import PostProcessor
  20 from youtube_dlc.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         'webpage_url': 'http://example.com/watch?v=shenanigans',
  46     }
  47     res.update(**kwargs)
  48     return res
  49
  50
  51 class TestFormatSelection(unittest.TestCase):
  52     def test_prefer_free_formats(self):
  53         # Same resolution => download webm
  54         ydl = YDL()
  55         ydl.params['prefer_free_formats'] = True
  56         formats = [
  57             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  58             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  59         ]
  60         info_dict = _make_result(formats)
  61         yie = YoutubeIE(ydl)
  62         yie._sort_formats(info_dict['formats'])
  63         ydl.process_ie_result(info_dict)
  64         downloaded = ydl.downloaded_info_dicts[0]
  65         self.assertEqual(downloaded['ext'], 'webm')
  66
  67         # Different resolution => download best quality (mp4)
  68         ydl = YDL()
  69         ydl.params['prefer_free_formats'] = True
  70         formats = [
  71             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  72             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  73         ]
  74         info_dict['formats'] = formats
  75         yie = YoutubeIE(ydl)
  76         yie._sort_formats(info_dict['formats'])
  77         ydl.process_ie_result(info_dict)
  78         downloaded = ydl.downloaded_info_dicts[0]
  79         self.assertEqual(downloaded['ext'], 'mp4')
  80
  81         # No prefer_free_formats => prefer mp4 and flv for greater compatibility
  82         ydl = YDL()
  83         ydl.params['prefer_free_formats'] = False
  84         formats = [
  85             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  86             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  87             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  88         ]
  89         info_dict['formats'] = formats
  90         yie = YoutubeIE(ydl)
  91         yie._sort_formats(info_dict['formats'])
  92         ydl.process_ie_result(info_dict)
  93         downloaded = ydl.downloaded_info_dicts[0]
  94         self.assertEqual(downloaded['ext'], 'mp4')
  95
  96         ydl = YDL()
  97         ydl.params['prefer_free_formats'] = False
  98         formats = [
  99             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
 100             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 101         ]
 102         info_dict['formats'] = formats
 103         yie = YoutubeIE(ydl)
 104         yie._sort_formats(info_dict['formats'])
 105         ydl.process_ie_result(info_dict)
 106         downloaded = ydl.downloaded_info_dicts[0]
 107         self.assertEqual(downloaded['ext'], 'flv')
 108
 109     def test_format_selection(self):
 110         formats = [
 111             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 112             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 113             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 114             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 115             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 116         ]
 117         info_dict = _make_result(formats)
 118
 119         ydl = YDL({'format': '20/47'})
 120         ydl.process_ie_result(info_dict.copy())
 121         downloaded = ydl.downloaded_info_dicts[0]
 122         self.assertEqual(downloaded['format_id'], '47')
 123
 124         ydl = YDL({'format': '20/71/worst'})
 125         ydl.process_ie_result(info_dict.copy())
 126         downloaded = ydl.downloaded_info_dicts[0]
 127         self.assertEqual(downloaded['format_id'], '35')
 128
 129         ydl = YDL()
 130         ydl.process_ie_result(info_dict.copy())
 131         downloaded = ydl.downloaded_info_dicts[0]
 132         self.assertEqual(downloaded['format_id'], '2')
 133
 134         ydl = YDL({'format': 'webm/mp4'})
 135         ydl.process_ie_result(info_dict.copy())
 136         downloaded = ydl.downloaded_info_dicts[0]
 137         self.assertEqual(downloaded['format_id'], '47')
 138
 139         ydl = YDL({'format': '3gp/40/mp4'})
 140         ydl.process_ie_result(info_dict.copy())
 141         downloaded = ydl.downloaded_info_dicts[0]
 142         self.assertEqual(downloaded['format_id'], '35')
 143
 144         ydl = YDL({'format': 'example-with-dashes'})
 145         ydl.process_ie_result(info_dict.copy())
 146         downloaded = ydl.downloaded_info_dicts[0]
 147         self.assertEqual(downloaded['format_id'], 'example-with-dashes')
 148
 149     def test_format_selection_audio(self):
 150         formats = [
 151             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 152             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 153             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 154             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 155         ]
 156         info_dict = _make_result(formats)
 157
 158         ydl = YDL({'format': 'bestaudio'})
 159         ydl.process_ie_result(info_dict.copy())
 160         downloaded = ydl.downloaded_info_dicts[0]
 161         self.assertEqual(downloaded['format_id'], 'audio-high')
 162
 163         ydl = YDL({'format': 'worstaudio'})
 164         ydl.process_ie_result(info_dict.copy())
 165         downloaded = ydl.downloaded_info_dicts[0]
 166         self.assertEqual(downloaded['format_id'], 'audio-low')
 167
 168         formats = [
 169             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 170             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 171         ]
 172         info_dict = _make_result(formats)
 173
 174         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 175         ydl.process_ie_result(info_dict.copy())
 176         downloaded = ydl.downloaded_info_dicts[0]
 177         self.assertEqual(downloaded['format_id'], 'vid-high')
 178
 179     def test_format_selection_audio_exts(self):
 180         formats = [
 181             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 182             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 184             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 185             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 186         ]
 187
 188         info_dict = _make_result(formats)
 189         ydl = YDL({'format': 'best'})
 190         ie = YoutubeIE(ydl)
 191         ie._sort_formats(info_dict['formats'])
 192         ydl.process_ie_result(copy.deepcopy(info_dict))
 193         downloaded = ydl.downloaded_info_dicts[0]
 194         self.assertEqual(downloaded['format_id'], 'aac-64')
 195
 196         ydl = YDL({'format': 'mp3'})
 197         ie = YoutubeIE(ydl)
 198         ie._sort_formats(info_dict['formats'])
 199         ydl.process_ie_result(copy.deepcopy(info_dict))
 200         downloaded = ydl.downloaded_info_dicts[0]
 201         self.assertEqual(downloaded['format_id'], 'mp3-64')
 202
 203         ydl = YDL({'prefer_free_formats': True})
 204         ie = YoutubeIE(ydl)
 205         ie._sort_formats(info_dict['formats'])
 206         ydl.process_ie_result(copy.deepcopy(info_dict))
 207         downloaded = ydl.downloaded_info_dicts[0]
 208         self.assertEqual(downloaded['format_id'], 'ogg-64')
 209
 210     def test_format_selection_video(self):
 211         formats = [
 212             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 213             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 214             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 215         ]
 216         info_dict = _make_result(formats)
 217
 218         ydl = YDL({'format': 'bestvideo'})
 219         ydl.process_ie_result(info_dict.copy())
 220         downloaded = ydl.downloaded_info_dicts[0]
 221         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 222
 223         ydl = YDL({'format': 'worstvideo'})
 224         ydl.process_ie_result(info_dict.copy())
 225         downloaded = ydl.downloaded_info_dicts[0]
 226         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 227
 228         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 229         ydl.process_ie_result(info_dict.copy())
 230         downloaded = ydl.downloaded_info_dicts[0]
 231         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 232
 233         formats = [
 234             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 235         ]
 236         info_dict = _make_result(formats)
 237
 238         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 239         ydl.process_ie_result(info_dict.copy())
 240         downloaded = ydl.downloaded_info_dicts[0]
 241         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 242
 243     def test_format_selection_string_ops(self):
 244         formats = [
 245             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 246             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 247         ]
 248         info_dict = _make_result(formats)
 249
 250         # equals (=)
 251         ydl = YDL({'format': '[format_id=abc-cba]'})
 252         ydl.process_ie_result(info_dict.copy())
 253         downloaded = ydl.downloaded_info_dicts[0]
 254         self.assertEqual(downloaded['format_id'], 'abc-cba')
 255
 256         # does not equal (!=)
 257         ydl = YDL({'format': '[format_id!=abc-cba]'})
 258         ydl.process_ie_result(info_dict.copy())
 259         downloaded = ydl.downloaded_info_dicts[0]
 260         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 261
 262         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 263         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 264
 265         # starts with (^=)
 266         ydl = YDL({'format': '[format_id^=abc]'})
 267         ydl.process_ie_result(info_dict.copy())
 268         downloaded = ydl.downloaded_info_dicts[0]
 269         self.assertEqual(downloaded['format_id'], 'abc-cba')
 270
 271         # does not start with (!^=)
 272         ydl = YDL({'format': '[format_id!^=abc]'})
 273         ydl.process_ie_result(info_dict.copy())
 274         downloaded = ydl.downloaded_info_dicts[0]
 275         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 276
 277         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 278         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 279
 280         # ends with ($=)
 281         ydl = YDL({'format': '[format_id$=cba]'})
 282         ydl.process_ie_result(info_dict.copy())
 283         downloaded = ydl.downloaded_info_dicts[0]
 284         self.assertEqual(downloaded['format_id'], 'abc-cba')
 285
 286         # does not end with (!$=)
 287         ydl = YDL({'format': '[format_id!$=cba]'})
 288         ydl.process_ie_result(info_dict.copy())
 289         downloaded = ydl.downloaded_info_dicts[0]
 290         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 291
 292         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 293         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 294
 295         # contains (*=)
 296         ydl = YDL({'format': '[format_id*=bc-cb]'})
 297         ydl.process_ie_result(info_dict.copy())
 298         downloaded = ydl.downloaded_info_dicts[0]
 299         self.assertEqual(downloaded['format_id'], 'abc-cba')
 300
 301         # does not contain (!*=)
 302         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 303         ydl.process_ie_result(info_dict.copy())
 304         downloaded = ydl.downloaded_info_dicts[0]
 305         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 306
 307         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 308         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 309
 310         ydl = YDL({'format': '[format_id!*=-]'})
 311         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 312
 313     def test_youtube_format_selection(self):
 314         order = [
 315             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 316             # Apple HTTP Live Streaming
 317             '96', '95', '94', '93', '92', '132', '151',
 318             # 3D
 319             '85', '84', '102', '83', '101', '82', '100',
 320             # Dash video
 321             '137', '248', '136', '247', '135', '246',
 322             '245', '244', '134', '243', '133', '242', '160',
 323             # Dash audio
 324             '141', '172', '140', '171', '139',
 325         ]
 326
 327         def format_info(f_id):
 328             info = YoutubeIE._formats[f_id].copy()
 329
 330             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 331             # and 'vcodec', while in tests such information is incomplete since
 332             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 333             # test_YoutubeDL.test_youtube_format_selection is broken without
 334             # this fix
 335             if 'acodec' in info and 'vcodec' not in info:
 336                 info['vcodec'] = 'none'
 337             elif 'vcodec' in info and 'acodec' not in info:
 338                 info['acodec'] = 'none'
 339
 340             info['format_id'] = f_id
 341             info['url'] = 'url:' + f_id
 342             return info
 343         formats_order = [format_info(f_id) for f_id in order]
 344
 345         info_dict = _make_result(list(formats_order), extractor='youtube')
 346         ydl = YDL({'format': 'bestvideo+bestaudio'})
 347         yie = YoutubeIE(ydl)
 348         yie._sort_formats(info_dict['formats'])
 349         ydl.process_ie_result(info_dict)
 350         downloaded = ydl.downloaded_info_dicts[0]
 351         self.assertEqual(downloaded['format_id'], '137+141')
 352         self.assertEqual(downloaded['ext'], 'mp4')
 353
 354         info_dict = _make_result(list(formats_order), extractor='youtube')
 355         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 356         yie = YoutubeIE(ydl)
 357         yie._sort_formats(info_dict['formats'])
 358         ydl.process_ie_result(info_dict)
 359         downloaded = ydl.downloaded_info_dicts[0]
 360         self.assertEqual(downloaded['format_id'], '38')
 361
 362         info_dict = _make_result(list(formats_order), extractor='youtube')
 363         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 364         yie = YoutubeIE(ydl)
 365         yie._sort_formats(info_dict['formats'])
 366         ydl.process_ie_result(info_dict)
 367         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 368         self.assertEqual(downloaded_ids, ['137', '141'])
 369
 370         info_dict = _make_result(list(formats_order), extractor='youtube')
 371         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 372         yie = YoutubeIE(ydl)
 373         yie._sort_formats(info_dict['formats'])
 374         ydl.process_ie_result(info_dict)
 375         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 376         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 377
 378         info_dict = _make_result(list(formats_order), extractor='youtube')
 379         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 380         yie = YoutubeIE(ydl)
 381         yie._sort_formats(info_dict['formats'])
 382         ydl.process_ie_result(info_dict)
 383         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 384         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 385
 386         info_dict = _make_result(list(formats_order), extractor='youtube')
 387         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 388         yie = YoutubeIE(ydl)
 389         yie._sort_formats(info_dict['formats'])
 390         ydl.process_ie_result(info_dict)
 391         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 392         self.assertEqual(downloaded_ids, ['248+141'])
 393
 394         for f1, f2 in zip(formats_order, formats_order[1:]):
 395             info_dict = _make_result([f1, f2], extractor='youtube')
 396             ydl = YDL({'format': 'best/bestvideo'})
 397             yie = YoutubeIE(ydl)
 398             yie._sort_formats(info_dict['formats'])
 399             ydl.process_ie_result(info_dict)
 400             downloaded = ydl.downloaded_info_dicts[0]
 401             self.assertEqual(downloaded['format_id'], f1['format_id'])
 402
 403             info_dict = _make_result([f2, f1], extractor='youtube')
 404             ydl = YDL({'format': 'best/bestvideo'})
 405             yie = YoutubeIE(ydl)
 406             yie._sort_formats(info_dict['formats'])
 407             ydl.process_ie_result(info_dict)
 408             downloaded = ydl.downloaded_info_dicts[0]
 409             self.assertEqual(downloaded['format_id'], f1['format_id'])
 410
 411     def test_audio_only_extractor_format_selection(self):
 412         # For extractors with incomplete formats (all formats are audio-only or
 413         # video-only) best and worst should fallback to corresponding best/worst
 414         # video-only or audio-only formats (as per
 415         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 416         formats = [
 417             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 418             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 419         ]
 420         info_dict = _make_result(formats)
 421
 422         ydl = YDL({'format': 'best'})
 423         ydl.process_ie_result(info_dict.copy())
 424         downloaded = ydl.downloaded_info_dicts[0]
 425         self.assertEqual(downloaded['format_id'], 'high')
 426
 427         ydl = YDL({'format': 'worst'})
 428         ydl.process_ie_result(info_dict.copy())
 429         downloaded = ydl.downloaded_info_dicts[0]
 430         self.assertEqual(downloaded['format_id'], 'low')
 431
 432     def test_format_not_available(self):
 433         formats = [
 434             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 435             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 436         ]
 437         info_dict = _make_result(formats)
 438
 439         # This must fail since complete video-audio format does not match filter
 440         # and extractor does not provide incomplete only formats (i.e. only
 441         # video-only or audio-only).
 442         ydl = YDL({'format': 'best[height>360]'})
 443         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 444
 445     def test_format_selection_issue_10083(self):
 446         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 447         formats = [
 448             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 449             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 450             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 451         ]
 452         info_dict = _make_result(formats)
 453
 454         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 455         ydl.process_ie_result(info_dict.copy())
 456         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 457
 458     def test_invalid_format_specs(self):
 459         def assert_syntax_error(format_spec):
 460             ydl = YDL({'format': format_spec})
 461             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 462             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 463
 464         assert_syntax_error('bestvideo,,best')
 465         assert_syntax_error('+bestaudio')
 466         assert_syntax_error('bestvideo+')
 467         assert_syntax_error('/')
 468
 469     def test_format_filtering(self):
 470         formats = [
 471             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 472             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 473             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 474             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 475             {'format_id': 'E', 'filesize': 3000},
 476             {'format_id': 'F'},
 477             {'format_id': 'G', 'filesize': 1000000},
 478         ]
 479         for f in formats:
 480             f['url'] = 'http://_/'
 481             f['ext'] = 'unknown'
 482         info_dict = _make_result(formats)
 483
 484         ydl = YDL({'format': 'best[filesize<3000]'})
 485         ydl.process_ie_result(info_dict)
 486         downloaded = ydl.downloaded_info_dicts[0]
 487         self.assertEqual(downloaded['format_id'], 'D')
 488
 489         ydl = YDL({'format': 'best[filesize<=3000]'})
 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': 'best[filesize <= ? 3000]'})
 495         ydl.process_ie_result(info_dict)
 496         downloaded = ydl.downloaded_info_dicts[0]
 497         self.assertEqual(downloaded['format_id'], 'F')
 498
 499         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 500         ydl.process_ie_result(info_dict)
 501         downloaded = ydl.downloaded_info_dicts[0]
 502         self.assertEqual(downloaded['format_id'], 'B')
 503
 504         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 505         ydl.process_ie_result(info_dict)
 506         downloaded = ydl.downloaded_info_dicts[0]
 507         self.assertEqual(downloaded['format_id'], 'C')
 508
 509         ydl = YDL({'format': '[filesize>?1]'})
 510         ydl.process_ie_result(info_dict)
 511         downloaded = ydl.downloaded_info_dicts[0]
 512         self.assertEqual(downloaded['format_id'], 'G')
 513
 514         ydl = YDL({'format': '[filesize<1M]'})
 515         ydl.process_ie_result(info_dict)
 516         downloaded = ydl.downloaded_info_dicts[0]
 517         self.assertEqual(downloaded['format_id'], 'E')
 518
 519         ydl = YDL({'format': '[filesize<1MiB]'})
 520         ydl.process_ie_result(info_dict)
 521         downloaded = ydl.downloaded_info_dicts[0]
 522         self.assertEqual(downloaded['format_id'], 'G')
 523
 524         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 525         ydl.process_ie_result(info_dict)
 526         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 527         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 528
 529         ydl = YDL({'format': 'best[height<40]'})
 530         try:
 531             ydl.process_ie_result(info_dict)
 532         except ExtractorError:
 533             pass
 534         self.assertEqual(ydl.downloaded_info_dicts, [])
 535
 536     def test_default_format_spec(self):
 537         ydl = YDL({'simulate': True})
 538         self.assertEqual(ydl._default_format_spec({}), 'bestvideo+bestaudio/best')
 539
 540         ydl = YDL({})
 541         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 542
 543         ydl = YDL({'simulate': True})
 544         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo+bestaudio/best')
 545
 546         ydl = YDL({'outtmpl': '-'})
 547         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 548
 549         ydl = YDL({})
 550         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo+bestaudio/best')
 551         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 552
 553
 554 class TestYoutubeDL(unittest.TestCase):
 555     def test_subtitles(self):
 556         def s_formats(lang, autocaption=False):
 557             return [{
 558                 'ext': ext,
 559                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 560                 '_auto': autocaption,
 561             } for ext in ['vtt', 'srt', 'ass']]
 562         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 563         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 564         info_dict = {
 565             'id': 'test',
 566             'title': 'Test',
 567             'url': 'http://localhost/video.mp4',
 568             'subtitles': subtitles,
 569             'automatic_captions': auto_captions,
 570             'extractor': 'TEST',
 571             'webpage_url': 'http://example.com/watch?v=shenanigans',
 572         }
 573
 574         def get_info(params={}):
 575             params.setdefault('simulate', True)
 576             ydl = YDL(params)
 577             ydl.report_warning = lambda *args, **kargs: None
 578             return ydl.process_video_result(info_dict, download=False)
 579
 580         result = get_info()
 581         self.assertFalse(result.get('requested_subtitles'))
 582         self.assertEqual(result['subtitles'], subtitles)
 583         self.assertEqual(result['automatic_captions'], auto_captions)
 584
 585         result = get_info({'writesubtitles': True})
 586         subs = result['requested_subtitles']
 587         self.assertTrue(subs)
 588         self.assertEqual(set(subs.keys()), set(['en']))
 589         self.assertTrue(subs['en'].get('data') is None)
 590         self.assertEqual(subs['en']['ext'], 'ass')
 591
 592         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 593         subs = result['requested_subtitles']
 594         self.assertEqual(subs['en']['ext'], 'srt')
 595
 596         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 597         subs = result['requested_subtitles']
 598         self.assertTrue(subs)
 599         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 600
 601         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 602         subs = result['requested_subtitles']
 603         self.assertTrue(subs)
 604         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 605         self.assertFalse(subs['es']['_auto'])
 606         self.assertTrue(subs['pt']['_auto'])
 607
 608         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 609         subs = result['requested_subtitles']
 610         self.assertTrue(subs)
 611         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 612         self.assertTrue(subs['es']['_auto'])
 613         self.assertTrue(subs['pt']['_auto'])
 614
 615     def test_add_extra_info(self):
 616         test_dict = {
 617             'extractor': 'Foo',
 618         }
 619         extra_info = {
 620             'extractor': 'Bar',
 621             'playlist': 'funny videos',
 622         }
 623         YDL.add_extra_info(test_dict, extra_info)
 624         self.assertEqual(test_dict['extractor'], 'Foo')
 625         self.assertEqual(test_dict['playlist'], 'funny videos')
 626
 627     def test_prepare_filename(self):
 628         info = {
 629             'id': '1234',
 630             'ext': 'mp4',
 631             'width': None,
 632             'height': 1080,
 633             'title1': '$PATH',
 634             'title2': '%PATH%',
 635         }
 636
 637         def fname(templ):
 638             ydl = YoutubeDL({'outtmpl': templ})
 639             return ydl.prepare_filename(info)
 640         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 641         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 642         # Replace missing fields with 'NA'
 643         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 644         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 645         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 646         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 647         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 648         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 649         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 650         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 651         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 652         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 653         self.assertEqual(fname('%%'), '%')
 654         self.assertEqual(fname('%%%%'), '%%')
 655         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 656         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 657         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 658         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 659         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 660         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 661
 662     def test_format_note(self):
 663         ydl = YoutubeDL()
 664         self.assertEqual(ydl._format_note({}), '')
 665         assertRegexpMatches(self, ydl._format_note({
 666             'vbr': 10,
 667         }), r'^\s*10k$')
 668         assertRegexpMatches(self, ydl._format_note({
 669             'fps': 30,
 670         }), r'^30fps$')
 671
 672     def test_postprocessors(self):
 673         filename = 'post-processor-testfile.mp4'
 674         audiofile = filename + '.mp3'
 675
 676         class SimplePP(PostProcessor):
 677             def run(self, info):
 678                 with open(audiofile, 'wt') as f:
 679                     f.write('EXAMPLE')
 680                 return [info['filepath']], info
 681
 682         def run_pp(params, PP):
 683             with open(filename, 'wt') as f:
 684                 f.write('EXAMPLE')
 685             ydl = YoutubeDL(params)
 686             ydl.add_post_processor(PP())
 687             ydl.post_process(filename, {'filepath': filename})
 688
 689         run_pp({'keepvideo': True}, SimplePP)
 690         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 691         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 692         os.unlink(filename)
 693         os.unlink(audiofile)
 694
 695         run_pp({'keepvideo': False}, SimplePP)
 696         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 697         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 698         os.unlink(audiofile)
 699
 700         class ModifierPP(PostProcessor):
 701             def run(self, info):
 702                 with open(info['filepath'], 'wt') as f:
 703                     f.write('MODIFIED')
 704                 return [], info
 705
 706         run_pp({'keepvideo': False}, ModifierPP)
 707         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 708         os.unlink(filename)
 709
 710     def test_match_filter(self):
 711         class FilterYDL(YDL):
 712             def __init__(self, *args, **kwargs):
 713                 super(FilterYDL, self).__init__(*args, **kwargs)
 714                 self.params['simulate'] = True
 715
 716             def process_info(self, info_dict):
 717                 super(YDL, self).process_info(info_dict)
 718
 719             def _match_entry(self, info_dict, incomplete):
 720                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 721                 if res is None:
 722                     self.downloaded_info_dicts.append(info_dict)
 723                 return res
 724
 725         first = {
 726             'id': '1',
 727             'url': TEST_URL,
 728             'title': 'one',
 729             'extractor': 'TEST',
 730             'duration': 30,
 731             'filesize': 10 * 1024,
 732             'playlist_id': '42',
 733             'uploader': "變態妍字幕版 太妍 тест",
 734             'creator': "тест ' 123 ' тест--",
 735             'webpage_url': 'http://example.com/watch?v=shenanigans',
 736         }
 737         second = {
 738             'id': '2',
 739             'url': TEST_URL,
 740             'title': 'two',
 741             'extractor': 'TEST',
 742             'duration': 10,
 743             'description': 'foo',
 744             'filesize': 5 * 1024,
 745             'playlist_id': '43',
 746             'uploader': "тест 123",
 747             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 748         }
 749         videos = [first, second]
 750
 751         def get_videos(filter_=None):
 752             ydl = FilterYDL({'match_filter': filter_})
 753             for v in videos:
 754                 ydl.process_ie_result(v, download=True)
 755             return [v['id'] for v in ydl.downloaded_info_dicts]
 756
 757         res = get_videos()
 758         self.assertEqual(res, ['1', '2'])
 759
 760         def f(v):
 761             if v['id'] == '1':
 762                 return None
 763             else:
 764                 return 'Video id is not 1'
 765         res = get_videos(f)
 766         self.assertEqual(res, ['1'])
 767
 768         f = match_filter_func('duration < 30')
 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, ['2'])
 775
 776         f = match_filter_func('description =? foo')
 777         res = get_videos(f)
 778         self.assertEqual(res, ['1', '2'])
 779
 780         f = match_filter_func('filesize > 5KiB')
 781         res = get_videos(f)
 782         self.assertEqual(res, ['1'])
 783
 784         f = match_filter_func('playlist_id = 42')
 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, ['1'])
 791
 792         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 793         res = get_videos(f)
 794         self.assertEqual(res, ['2'])
 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("creator = 'тест \\' 123 \\' тест--'")
 801         res = get_videos(f)
 802         self.assertEqual(res, ['1'])
 803
 804         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 805         res = get_videos(f)
 806         self.assertEqual(res, [])
 807
 808     def test_playlist_items_selection(self):
 809         entries = [{
 810             'id': compat_str(i),
 811             'title': compat_str(i),
 812             'url': TEST_URL,
 813         } for i in range(1, 5)]
 814         playlist = {
 815             '_type': 'playlist',
 816             'id': 'test',
 817             'entries': entries,
 818             'extractor': 'test:playlist',
 819             'extractor_key': 'test:playlist',
 820             'webpage_url': 'http://example.com',
 821         }
 822
 823         def get_downloaded_info_dicts(params):
 824             ydl = YDL(params)
 825             # make a deep copy because the dictionary and nested entries
 826             # can be modified
 827             ydl.process_ie_result(copy.deepcopy(playlist))
 828             return ydl.downloaded_info_dicts
 829
 830         def get_ids(params):
 831             return [int(v['id']) for v in get_downloaded_info_dicts(params)]
 832
 833         result = get_ids({})
 834         self.assertEqual(result, [1, 2, 3, 4])
 835
 836         result = get_ids({'playlistend': 10})
 837         self.assertEqual(result, [1, 2, 3, 4])
 838
 839         result = get_ids({'playlistend': 2})
 840         self.assertEqual(result, [1, 2])
 841
 842         result = get_ids({'playliststart': 10})
 843         self.assertEqual(result, [])
 844
 845         result = get_ids({'playliststart': 2})
 846         self.assertEqual(result, [2, 3, 4])
 847
 848         result = get_ids({'playlist_items': '2-4'})
 849         self.assertEqual(result, [2, 3, 4])
 850
 851         result = get_ids({'playlist_items': '2,4'})
 852         self.assertEqual(result, [2, 4])
 853
 854         result = get_ids({'playlist_items': '10'})
 855         self.assertEqual(result, [])
 856
 857         result = get_ids({'playlist_items': '3-10'})
 858         self.assertEqual(result, [3, 4])
 859
 860         result = get_ids({'playlist_items': '2-4,3-4,3'})
 861         self.assertEqual(result, [2, 3, 4])
 862
 863         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
 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
 869         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 870         self.assertEqual(result[0]['playlist_index'], 2)
 871         self.assertEqual(result[1]['playlist_index'], 3)
 872         self.assertEqual(result[2]['playlist_index'], 4)
 873
 874         result = get_downloaded_info_dicts({'playlist_items': '4,2'})
 875         self.assertEqual(result[0]['playlist_index'], 4)
 876         self.assertEqual(result[1]['playlist_index'], 2)
 877         # @}
 878
 879     def test_urlopen_no_file_protocol(self):
 880         # see https://github.com/ytdl-org/youtube-dl/issues/8227
 881         ydl = YDL()
 882         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 883
 884     def test_do_not_override_ie_key_in_url_transparent(self):
 885         ydl = YDL()
 886
 887         class Foo1IE(InfoExtractor):
 888             _VALID_URL = r'foo1:'
 889
 890             def _real_extract(self, url):
 891                 return {
 892                     '_type': 'url_transparent',
 893                     'url': 'foo2:',
 894                     'ie_key': 'Foo2',
 895                     'title': 'foo1 title',
 896                     'id': 'foo1_id',
 897                 }
 898
 899         class Foo2IE(InfoExtractor):
 900             _VALID_URL = r'foo2:'
 901
 902             def _real_extract(self, url):
 903                 return {
 904                     '_type': 'url',
 905                     'url': 'foo3:',
 906                     'ie_key': 'Foo3',
 907                 }
 908
 909         class Foo3IE(InfoExtractor):
 910             _VALID_URL = r'foo3:'
 911
 912             def _real_extract(self, url):
 913                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 914
 915         ydl.add_info_extractor(Foo1IE(ydl))
 916         ydl.add_info_extractor(Foo2IE(ydl))
 917         ydl.add_info_extractor(Foo3IE(ydl))
 918         ydl.extract_info('foo1:')
 919         downloaded = ydl.downloaded_info_dicts[0]
 920         self.assertEqual(downloaded['url'], TEST_URL)
 921         self.assertEqual(downloaded['title'], 'foo1 title')
 922         self.assertEqual(downloaded['id'], 'testid')
 923         self.assertEqual(downloaded['extractor'], 'testex')
 924         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 925
 926     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
 927     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
 928
 929         class _YDL(YDL):
 930             def __init__(self, *args, **kwargs):
 931                 super(_YDL, self).__init__(*args, **kwargs)
 932
 933             def trouble(self, s, tb=None):
 934                 pass
 935
 936         ydl = _YDL({
 937             'format': 'extra',
 938             'ignoreerrors': True,
 939         })
 940
 941         class VideoIE(InfoExtractor):
 942             _VALID_URL = r'video:(?P<id>\d+)'
 943
 944             def _real_extract(self, url):
 945                 video_id = self._match_id(url)
 946                 formats = [{
 947                     'format_id': 'default',
 948                     'url': 'url:',
 949                 }]
 950                 if video_id == '0':
 951                     raise ExtractorError('foo')
 952                 if video_id == '2':
 953                     formats.append({
 954                         'format_id': 'extra',
 955                         'url': TEST_URL,
 956                     })
 957                 return {
 958                     'id': video_id,
 959                     'title': 'Video %s' % video_id,
 960                     'formats': formats,
 961                 }
 962
 963         class PlaylistIE(InfoExtractor):
 964             _VALID_URL = r'playlist:'
 965
 966             def _entries(self):
 967                 for n in range(3):
 968                     video_id = compat_str(n)
 969                     yield {
 970                         '_type': 'url_transparent',
 971                         'ie_key': VideoIE.ie_key(),
 972                         'id': video_id,
 973                         'url': 'video:%s' % video_id,
 974                         'title': 'Video Transparent %s' % video_id,
 975                     }
 976
 977             def _real_extract(self, url):
 978                 return self.playlist_result(self._entries())
 979
 980         ydl.add_info_extractor(VideoIE(ydl))
 981         ydl.add_info_extractor(PlaylistIE(ydl))
 982         info = ydl.extract_info('playlist:')
 983         entries = info['entries']
 984         self.assertEqual(len(entries), 3)
 985         self.assertTrue(entries[0] is None)
 986         self.assertTrue(entries[1] is None)
 987         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
 988         downloaded = ydl.downloaded_info_dicts[0]
 989         self.assertEqual(entries[2], downloaded)
 990         self.assertEqual(downloaded['url'], TEST_URL)
 991         self.assertEqual(downloaded['title'], 'Video Transparent 2')
 992         self.assertEqual(downloaded['id'], '2')
 993         self.assertEqual(downloaded['extractor'], 'Video')
 994         self.assertEqual(downloaded['extractor_key'], 'Video')
 995
 996
 997 if __name__ == '__main__':
 998     unittest.main()