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 webm
  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'], 'webm')
 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         return
 315         # disabled for now - this needs some changes
 316
 317         order = [
 318             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 319             # Apple HTTP Live Streaming
 320             '96', '95', '94', '93', '92', '132', '151',
 321             # 3D
 322             '85', '84', '102', '83', '101', '82', '100',
 323             # Dash video
 324             '137', '248', '136', '247', '135', '246',
 325             '245', '244', '134', '243', '133', '242', '160',
 326             # Dash audio
 327             '141', '172', '140', '171', '139',
 328         ]
 329
 330         def format_info(f_id):
 331             info = YoutubeIE._formats[f_id].copy()
 332
 333             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 334             # and 'vcodec', while in tests such information is incomplete since
 335             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 336             # test_YoutubeDL.test_youtube_format_selection is broken without
 337             # this fix
 338             if 'acodec' in info and 'vcodec' not in info:
 339                 info['vcodec'] = 'none'
 340             elif 'vcodec' in info and 'acodec' not in info:
 341                 info['acodec'] = 'none'
 342
 343             info['format_id'] = f_id
 344             info['url'] = 'url:' + f_id
 345             return info
 346         formats_order = [format_info(f_id) for f_id in order]
 347
 348         info_dict = _make_result(list(formats_order), extractor='youtube')
 349         ydl = YDL({'format': 'bestvideo+bestaudio'})
 350         yie = YoutubeIE(ydl)
 351         yie._sort_formats(info_dict['formats'])
 352         ydl.process_ie_result(info_dict)
 353         downloaded = ydl.downloaded_info_dicts[0]
 354         self.assertEqual(downloaded['format_id'], '248+172')
 355         self.assertEqual(downloaded['ext'], 'mp4')
 356
 357         info_dict = _make_result(list(formats_order), extractor='youtube')
 358         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 359         yie = YoutubeIE(ydl)
 360         yie._sort_formats(info_dict['formats'])
 361         ydl.process_ie_result(info_dict)
 362         downloaded = ydl.downloaded_info_dicts[0]
 363         self.assertEqual(downloaded['format_id'], '38')
 364
 365         info_dict = _make_result(list(formats_order), extractor='youtube')
 366         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 367         yie = YoutubeIE(ydl)
 368         yie._sort_formats(info_dict['formats'])
 369         ydl.process_ie_result(info_dict)
 370         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 371         self.assertEqual(downloaded_ids, ['137', '141'])
 372
 373         info_dict = _make_result(list(formats_order), extractor='youtube')
 374         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 375         yie = YoutubeIE(ydl)
 376         yie._sort_formats(info_dict['formats'])
 377         ydl.process_ie_result(info_dict)
 378         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 379         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 380
 381         info_dict = _make_result(list(formats_order), extractor='youtube')
 382         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 383         yie = YoutubeIE(ydl)
 384         yie._sort_formats(info_dict['formats'])
 385         ydl.process_ie_result(info_dict)
 386         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 387         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 388
 389         info_dict = _make_result(list(formats_order), extractor='youtube')
 390         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 391         yie = YoutubeIE(ydl)
 392         yie._sort_formats(info_dict['formats'])
 393         ydl.process_ie_result(info_dict)
 394         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 395         self.assertEqual(downloaded_ids, ['248+141'])
 396
 397         for f1, f2 in zip(formats_order, formats_order[1:]):
 398             info_dict = _make_result([f1, f2], extractor='youtube')
 399             ydl = YDL({'format': 'best/bestvideo'})
 400             yie = YoutubeIE(ydl)
 401             yie._sort_formats(info_dict['formats'])
 402             ydl.process_ie_result(info_dict)
 403             downloaded = ydl.downloaded_info_dicts[0]
 404             self.assertEqual(downloaded['format_id'], f1['format_id'])
 405
 406             info_dict = _make_result([f2, f1], extractor='youtube')
 407             ydl = YDL({'format': 'best/bestvideo'})
 408             yie = YoutubeIE(ydl)
 409             yie._sort_formats(info_dict['formats'])
 410             ydl.process_ie_result(info_dict)
 411             downloaded = ydl.downloaded_info_dicts[0]
 412             self.assertEqual(downloaded['format_id'], f1['format_id'])
 413
 414     def test_audio_only_extractor_format_selection(self):
 415         # For extractors with incomplete formats (all formats are audio-only or
 416         # video-only) best and worst should fallback to corresponding best/worst
 417         # video-only or audio-only formats (as per
 418         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 419         formats = [
 420             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 421             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 422         ]
 423         info_dict = _make_result(formats)
 424
 425         ydl = YDL({'format': 'best'})
 426         ydl.process_ie_result(info_dict.copy())
 427         downloaded = ydl.downloaded_info_dicts[0]
 428         self.assertEqual(downloaded['format_id'], 'high')
 429
 430         ydl = YDL({'format': 'worst'})
 431         ydl.process_ie_result(info_dict.copy())
 432         downloaded = ydl.downloaded_info_dicts[0]
 433         self.assertEqual(downloaded['format_id'], 'low')
 434
 435     def test_format_not_available(self):
 436         formats = [
 437             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 438             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 439         ]
 440         info_dict = _make_result(formats)
 441
 442         # This must fail since complete video-audio format does not match filter
 443         # and extractor does not provide incomplete only formats (i.e. only
 444         # video-only or audio-only).
 445         ydl = YDL({'format': 'best[height>360]'})
 446         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 447
 448     def test_format_selection_issue_10083(self):
 449         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 450         formats = [
 451             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 452             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 453             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 454         ]
 455         info_dict = _make_result(formats)
 456
 457         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 458         ydl.process_ie_result(info_dict.copy())
 459         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 460
 461     def test_invalid_format_specs(self):
 462         def assert_syntax_error(format_spec):
 463             ydl = YDL({'format': format_spec})
 464             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 465             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 466
 467         assert_syntax_error('bestvideo,,best')
 468         assert_syntax_error('+bestaudio')
 469         assert_syntax_error('bestvideo+')
 470         assert_syntax_error('/')
 471
 472     def test_format_filtering(self):
 473         formats = [
 474             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 475             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 476             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 477             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 478             {'format_id': 'E', 'filesize': 3000},
 479             {'format_id': 'F'},
 480             {'format_id': 'G', 'filesize': 1000000},
 481         ]
 482         for f in formats:
 483             f['url'] = 'http://_/'
 484             f['ext'] = 'unknown'
 485         info_dict = _make_result(formats)
 486
 487         ydl = YDL({'format': 'best[filesize<3000]'})
 488         ydl.process_ie_result(info_dict)
 489         downloaded = ydl.downloaded_info_dicts[0]
 490         self.assertEqual(downloaded['format_id'], 'D')
 491
 492         ydl = YDL({'format': 'best[filesize<=3000]'})
 493         ydl.process_ie_result(info_dict)
 494         downloaded = ydl.downloaded_info_dicts[0]
 495         self.assertEqual(downloaded['format_id'], 'E')
 496
 497         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 498         ydl.process_ie_result(info_dict)
 499         downloaded = ydl.downloaded_info_dicts[0]
 500         self.assertEqual(downloaded['format_id'], 'F')
 501
 502         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 503         ydl.process_ie_result(info_dict)
 504         downloaded = ydl.downloaded_info_dicts[0]
 505         self.assertEqual(downloaded['format_id'], 'B')
 506
 507         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 508         ydl.process_ie_result(info_dict)
 509         downloaded = ydl.downloaded_info_dicts[0]
 510         self.assertEqual(downloaded['format_id'], 'C')
 511
 512         ydl = YDL({'format': '[filesize>?1]'})
 513         ydl.process_ie_result(info_dict)
 514         downloaded = ydl.downloaded_info_dicts[0]
 515         self.assertEqual(downloaded['format_id'], 'G')
 516
 517         ydl = YDL({'format': '[filesize<1M]'})
 518         ydl.process_ie_result(info_dict)
 519         downloaded = ydl.downloaded_info_dicts[0]
 520         self.assertEqual(downloaded['format_id'], 'E')
 521
 522         ydl = YDL({'format': '[filesize<1MiB]'})
 523         ydl.process_ie_result(info_dict)
 524         downloaded = ydl.downloaded_info_dicts[0]
 525         self.assertEqual(downloaded['format_id'], 'G')
 526
 527         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 528         ydl.process_ie_result(info_dict)
 529         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 530         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 531
 532         ydl = YDL({'format': 'best[height<40]'})
 533         try:
 534             ydl.process_ie_result(info_dict)
 535         except ExtractorError:
 536             pass
 537         self.assertEqual(ydl.downloaded_info_dicts, [])
 538
 539     def test_default_format_spec(self):
 540         ydl = YDL({'simulate': True})
 541         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 542
 543         ydl = YDL({})
 544         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 545
 546         ydl = YDL({'simulate': True})
 547         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo*+bestaudio/best')
 548
 549         ydl = YDL({'outtmpl': '-'})
 550         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 551
 552         ydl = YDL({})
 553         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo*+bestaudio/best')
 554         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 555
 556
 557 class TestYoutubeDL(unittest.TestCase):
 558     def test_subtitles(self):
 559         def s_formats(lang, autocaption=False):
 560             return [{
 561                 'ext': ext,
 562                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 563                 '_auto': autocaption,
 564             } for ext in ['vtt', 'srt', 'ass']]
 565         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 566         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 567         info_dict = {
 568             'id': 'test',
 569             'title': 'Test',
 570             'url': 'http://localhost/video.mp4',
 571             'subtitles': subtitles,
 572             'automatic_captions': auto_captions,
 573             'extractor': 'TEST',
 574             'webpage_url': 'http://example.com/watch?v=shenanigans',
 575         }
 576
 577         def get_info(params={}):
 578             params.setdefault('simulate', True)
 579             ydl = YDL(params)
 580             ydl.report_warning = lambda *args, **kargs: None
 581             return ydl.process_video_result(info_dict, download=False)
 582
 583         result = get_info()
 584         self.assertFalse(result.get('requested_subtitles'))
 585         self.assertEqual(result['subtitles'], subtitles)
 586         self.assertEqual(result['automatic_captions'], auto_captions)
 587
 588         result = get_info({'writesubtitles': True})
 589         subs = result['requested_subtitles']
 590         self.assertTrue(subs)
 591         self.assertEqual(set(subs.keys()), set(['en']))
 592         self.assertTrue(subs['en'].get('data') is None)
 593         self.assertEqual(subs['en']['ext'], 'ass')
 594
 595         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 596         subs = result['requested_subtitles']
 597         self.assertEqual(subs['en']['ext'], 'srt')
 598
 599         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 600         subs = result['requested_subtitles']
 601         self.assertTrue(subs)
 602         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 603
 604         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 605         subs = result['requested_subtitles']
 606         self.assertTrue(subs)
 607         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 608         self.assertFalse(subs['es']['_auto'])
 609         self.assertTrue(subs['pt']['_auto'])
 610
 611         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 612         subs = result['requested_subtitles']
 613         self.assertTrue(subs)
 614         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 615         self.assertTrue(subs['es']['_auto'])
 616         self.assertTrue(subs['pt']['_auto'])
 617
 618     def test_add_extra_info(self):
 619         test_dict = {
 620             'extractor': 'Foo',
 621         }
 622         extra_info = {
 623             'extractor': 'Bar',
 624             'playlist': 'funny videos',
 625         }
 626         YDL.add_extra_info(test_dict, extra_info)
 627         self.assertEqual(test_dict['extractor'], 'Foo')
 628         self.assertEqual(test_dict['playlist'], 'funny videos')
 629
 630     def test_prepare_filename(self):
 631         info = {
 632             'id': '1234',
 633             'ext': 'mp4',
 634             'width': None,
 635             'height': 1080,
 636             'title1': '$PATH',
 637             'title2': '%PATH%',
 638         }
 639
 640         def fname(templ):
 641             ydl = YoutubeDL({'outtmpl': templ})
 642             return ydl.prepare_filename(info)
 643         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 644         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 645         # Replace missing fields with 'NA'
 646         self.assertEqual(fname('%(uploader_date)s-%(id)s.%(ext)s'), 'NA-1234.mp4')
 647         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 648         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 649         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 650         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 651         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 652         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 653         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 654         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 655         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 656         self.assertEqual(fname('%%'), '%')
 657         self.assertEqual(fname('%%%%'), '%%')
 658         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 659         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 660         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 661         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 662         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 663         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 664
 665     def test_format_note(self):
 666         ydl = YoutubeDL()
 667         self.assertEqual(ydl._format_note({}), '')
 668         assertRegexpMatches(self, ydl._format_note({
 669             'vbr': 10,
 670         }), r'^\s*10k$')
 671         assertRegexpMatches(self, ydl._format_note({
 672             'fps': 30,
 673         }), r'^30fps$')
 674
 675     def test_postprocessors(self):
 676         filename = 'post-processor-testfile.mp4'
 677         audiofile = filename + '.mp3'
 678
 679         class SimplePP(PostProcessor):
 680             def run(self, info):
 681                 with open(audiofile, 'wt') as f:
 682                     f.write('EXAMPLE')
 683                 return [info['filepath']], info
 684
 685         def run_pp(params, PP):
 686             with open(filename, 'wt') as f:
 687                 f.write('EXAMPLE')
 688             ydl = YoutubeDL(params)
 689             ydl.add_post_processor(PP())
 690             ydl.post_process(filename, {'filepath': filename})
 691
 692         run_pp({'keepvideo': True}, SimplePP)
 693         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 694         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 695         os.unlink(filename)
 696         os.unlink(audiofile)
 697
 698         run_pp({'keepvideo': False}, SimplePP)
 699         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 700         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 701         os.unlink(audiofile)
 702
 703         class ModifierPP(PostProcessor):
 704             def run(self, info):
 705                 with open(info['filepath'], 'wt') as f:
 706                     f.write('MODIFIED')
 707                 return [], info
 708
 709         run_pp({'keepvideo': False}, ModifierPP)
 710         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 711         os.unlink(filename)
 712
 713     def test_match_filter(self):
 714         class FilterYDL(YDL):
 715             def __init__(self, *args, **kwargs):
 716                 super(FilterYDL, self).__init__(*args, **kwargs)
 717                 self.params['simulate'] = True
 718
 719             def process_info(self, info_dict):
 720                 super(YDL, self).process_info(info_dict)
 721
 722             def _match_entry(self, info_dict, incomplete):
 723                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 724                 if res is None:
 725                     self.downloaded_info_dicts.append(info_dict)
 726                 return res
 727
 728         first = {
 729             'id': '1',
 730             'url': TEST_URL,
 731             'title': 'one',
 732             'extractor': 'TEST',
 733             'duration': 30,
 734             'filesize': 10 * 1024,
 735             'playlist_id': '42',
 736             'uploader': "變態妍字幕版 太妍 тест",
 737             'creator': "тест ' 123 ' тест--",
 738             'webpage_url': 'http://example.com/watch?v=shenanigans',
 739         }
 740         second = {
 741             'id': '2',
 742             'url': TEST_URL,
 743             'title': 'two',
 744             'extractor': 'TEST',
 745             'duration': 10,
 746             'description': 'foo',
 747             'filesize': 5 * 1024,
 748             'playlist_id': '43',
 749             'uploader': "тест 123",
 750             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 751         }
 752         videos = [first, second]
 753
 754         def get_videos(filter_=None):
 755             ydl = FilterYDL({'match_filter': filter_})
 756             for v in videos:
 757                 ydl.process_ie_result(v, download=True)
 758             return [v['id'] for v in ydl.downloaded_info_dicts]
 759
 760         res = get_videos()
 761         self.assertEqual(res, ['1', '2'])
 762
 763         def f(v):
 764             if v['id'] == '1':
 765                 return None
 766             else:
 767                 return 'Video id is not 1'
 768         res = get_videos(f)
 769         self.assertEqual(res, ['1'])
 770
 771         f = match_filter_func('duration < 30')
 772         res = get_videos(f)
 773         self.assertEqual(res, ['2'])
 774
 775         f = match_filter_func('description = foo')
 776         res = get_videos(f)
 777         self.assertEqual(res, ['2'])
 778
 779         f = match_filter_func('description =? foo')
 780         res = get_videos(f)
 781         self.assertEqual(res, ['1', '2'])
 782
 783         f = match_filter_func('filesize > 5KiB')
 784         res = get_videos(f)
 785         self.assertEqual(res, ['1'])
 786
 787         f = match_filter_func('playlist_id = 42')
 788         res = get_videos(f)
 789         self.assertEqual(res, ['1'])
 790
 791         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 792         res = get_videos(f)
 793         self.assertEqual(res, ['1'])
 794
 795         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 796         res = get_videos(f)
 797         self.assertEqual(res, ['2'])
 798
 799         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 800         res = get_videos(f)
 801         self.assertEqual(res, ['1'])
 802
 803         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 804         res = get_videos(f)
 805         self.assertEqual(res, ['1'])
 806
 807         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 808         res = get_videos(f)
 809         self.assertEqual(res, [])
 810
 811     def test_playlist_items_selection(self):
 812         entries = [{
 813             'id': compat_str(i),
 814             'title': compat_str(i),
 815             'url': TEST_URL,
 816         } for i in range(1, 5)]
 817         playlist = {
 818             '_type': 'playlist',
 819             'id': 'test',
 820             'entries': entries,
 821             'extractor': 'test:playlist',
 822             'extractor_key': 'test:playlist',
 823             'webpage_url': 'http://example.com',
 824         }
 825
 826         def get_downloaded_info_dicts(params):
 827             ydl = YDL(params)
 828             # make a deep copy because the dictionary and nested entries
 829             # can be modified
 830             ydl.process_ie_result(copy.deepcopy(playlist))
 831             return ydl.downloaded_info_dicts
 832
 833         def get_ids(params):
 834             return [int(v['id']) for v in get_downloaded_info_dicts(params)]
 835
 836         result = get_ids({})
 837         self.assertEqual(result, [1, 2, 3, 4])
 838
 839         result = get_ids({'playlistend': 10})
 840         self.assertEqual(result, [1, 2, 3, 4])
 841
 842         result = get_ids({'playlistend': 2})
 843         self.assertEqual(result, [1, 2])
 844
 845         result = get_ids({'playliststart': 10})
 846         self.assertEqual(result, [])
 847
 848         result = get_ids({'playliststart': 2})
 849         self.assertEqual(result, [2, 3, 4])
 850
 851         result = get_ids({'playlist_items': '2-4'})
 852         self.assertEqual(result, [2, 3, 4])
 853
 854         result = get_ids({'playlist_items': '2,4'})
 855         self.assertEqual(result, [2, 4])
 856
 857         result = get_ids({'playlist_items': '10'})
 858         self.assertEqual(result, [])
 859
 860         result = get_ids({'playlist_items': '3-10'})
 861         self.assertEqual(result, [3, 4])
 862
 863         result = get_ids({'playlist_items': '2-4,3-4,3'})
 864         self.assertEqual(result, [2, 3, 4])
 865
 866         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
 867         # @{
 868         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 869         self.assertEqual(result[0]['playlist_index'], 2)
 870         self.assertEqual(result[1]['playlist_index'], 3)
 871
 872         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 873         self.assertEqual(result[0]['playlist_index'], 2)
 874         self.assertEqual(result[1]['playlist_index'], 3)
 875         self.assertEqual(result[2]['playlist_index'], 4)
 876
 877         result = get_downloaded_info_dicts({'playlist_items': '4,2'})
 878         self.assertEqual(result[0]['playlist_index'], 4)
 879         self.assertEqual(result[1]['playlist_index'], 2)
 880         # @}
 881
 882     def test_urlopen_no_file_protocol(self):
 883         # see https://github.com/ytdl-org/youtube-dl/issues/8227
 884         ydl = YDL()
 885         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 886
 887     def test_do_not_override_ie_key_in_url_transparent(self):
 888         ydl = YDL()
 889
 890         class Foo1IE(InfoExtractor):
 891             _VALID_URL = r'foo1:'
 892
 893             def _real_extract(self, url):
 894                 return {
 895                     '_type': 'url_transparent',
 896                     'url': 'foo2:',
 897                     'ie_key': 'Foo2',
 898                     'title': 'foo1 title',
 899                     'id': 'foo1_id',
 900                 }
 901
 902         class Foo2IE(InfoExtractor):
 903             _VALID_URL = r'foo2:'
 904
 905             def _real_extract(self, url):
 906                 return {
 907                     '_type': 'url',
 908                     'url': 'foo3:',
 909                     'ie_key': 'Foo3',
 910                 }
 911
 912         class Foo3IE(InfoExtractor):
 913             _VALID_URL = r'foo3:'
 914
 915             def _real_extract(self, url):
 916                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 917
 918         ydl.add_info_extractor(Foo1IE(ydl))
 919         ydl.add_info_extractor(Foo2IE(ydl))
 920         ydl.add_info_extractor(Foo3IE(ydl))
 921         ydl.extract_info('foo1:')
 922         downloaded = ydl.downloaded_info_dicts[0]
 923         self.assertEqual(downloaded['url'], TEST_URL)
 924         self.assertEqual(downloaded['title'], 'foo1 title')
 925         self.assertEqual(downloaded['id'], 'testid')
 926         self.assertEqual(downloaded['extractor'], 'testex')
 927         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 928
 929     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
 930     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
 931
 932         class _YDL(YDL):
 933             def __init__(self, *args, **kwargs):
 934                 super(_YDL, self).__init__(*args, **kwargs)
 935
 936             def trouble(self, s, tb=None):
 937                 pass
 938
 939         ydl = _YDL({
 940             'format': 'extra',
 941             'ignoreerrors': True,
 942         })
 943
 944         class VideoIE(InfoExtractor):
 945             _VALID_URL = r'video:(?P<id>\d+)'
 946
 947             def _real_extract(self, url):
 948                 video_id = self._match_id(url)
 949                 formats = [{
 950                     'format_id': 'default',
 951                     'url': 'url:',
 952                 }]
 953                 if video_id == '0':
 954                     raise ExtractorError('foo')
 955                 if video_id == '2':
 956                     formats.append({
 957                         'format_id': 'extra',
 958                         'url': TEST_URL,
 959                     })
 960                 return {
 961                     'id': video_id,
 962                     'title': 'Video %s' % video_id,
 963                     'formats': formats,
 964                 }
 965
 966         class PlaylistIE(InfoExtractor):
 967             _VALID_URL = r'playlist:'
 968
 969             def _entries(self):
 970                 for n in range(3):
 971                     video_id = compat_str(n)
 972                     yield {
 973                         '_type': 'url_transparent',
 974                         'ie_key': VideoIE.ie_key(),
 975                         'id': video_id,
 976                         'url': 'video:%s' % video_id,
 977                         'title': 'Video Transparent %s' % video_id,
 978                     }
 979
 980             def _real_extract(self, url):
 981                 return self.playlist_result(self._entries())
 982
 983         ydl.add_info_extractor(VideoIE(ydl))
 984         ydl.add_info_extractor(PlaylistIE(ydl))
 985         info = ydl.extract_info('playlist:')
 986         entries = info['entries']
 987         self.assertEqual(len(entries), 3)
 988         self.assertTrue(entries[0] is None)
 989         self.assertTrue(entries[1] is None)
 990         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
 991         downloaded = ydl.downloaded_info_dicts[0]
 992         self.assertEqual(entries[2], downloaded)
 993         self.assertEqual(downloaded['url'], TEST_URL)
 994         self.assertEqual(downloaded['title'], 'Video Transparent 2')
 995         self.assertEqual(downloaded['id'], '2')
 996         self.assertEqual(downloaded['extractor'], 'Video')
 997         self.assertEqual(downloaded['extractor_key'], 'Video')
 998
 999
1000 if __name__ == '__main__':
1001     unittest.main()