test/test_YoutubeDL.py

   1 #!/usr/bin/env python3
   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 yt_dlp import YoutubeDL
  16 from yt_dlp.compat import compat_str, compat_urllib_error
  17 from yt_dlp.extractor import YoutubeIE
  18 from yt_dlp.extractor.common import InfoExtractor
  19 from yt_dlp.postprocessor.common import PostProcessor
  20 from yt_dlp.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         info_dict.pop('__original_infodict', None)
  33         self.downloaded_info_dicts.append(info_dict)
  34
  35     def to_screen(self, msg):
  36         self.msgs.append(msg)
  37
  38
  39 def _make_result(formats, **kwargs):
  40     res = {
  41         'formats': formats,
  42         'id': 'testid',
  43         'title': 'testttitle',
  44         'extractor': 'testex',
  45         'extractor_key': 'TestEx',
  46         'webpage_url': 'http://example.com/watch?v=shenanigans',
  47     }
  48     res.update(**kwargs)
  49     return res
  50
  51
  52 class TestFormatSelection(unittest.TestCase):
  53     def test_prefer_free_formats(self):
  54         # Same resolution => download webm
  55         ydl = YDL()
  56         ydl.params['prefer_free_formats'] = True
  57         formats = [
  58             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  59             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  60         ]
  61         info_dict = _make_result(formats)
  62         yie = YoutubeIE(ydl)
  63         yie._sort_formats(info_dict['formats'])
  64         ydl.process_ie_result(info_dict)
  65         downloaded = ydl.downloaded_info_dicts[0]
  66         self.assertEqual(downloaded['ext'], 'webm')
  67
  68         # Different resolution => download best quality (mp4)
  69         ydl = YDL()
  70         ydl.params['prefer_free_formats'] = True
  71         formats = [
  72             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  73             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  74         ]
  75         info_dict['formats'] = formats
  76         yie = YoutubeIE(ydl)
  77         yie._sort_formats(info_dict['formats'])
  78         ydl.process_ie_result(info_dict)
  79         downloaded = ydl.downloaded_info_dicts[0]
  80         self.assertEqual(downloaded['ext'], 'mp4')
  81
  82         # No prefer_free_formats => prefer mp4 and webm
  83         ydl = YDL()
  84         ydl.params['prefer_free_formats'] = False
  85         formats = [
  86             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  87             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  88             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  89         ]
  90         info_dict['formats'] = formats
  91         yie = YoutubeIE(ydl)
  92         yie._sort_formats(info_dict['formats'])
  93         ydl.process_ie_result(info_dict)
  94         downloaded = ydl.downloaded_info_dicts[0]
  95         self.assertEqual(downloaded['ext'], 'mp4')
  96
  97         ydl = YDL()
  98         ydl.params['prefer_free_formats'] = False
  99         formats = [
 100             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
 101             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 102         ]
 103         info_dict['formats'] = formats
 104         yie = YoutubeIE(ydl)
 105         yie._sort_formats(info_dict['formats'])
 106         ydl.process_ie_result(info_dict)
 107         downloaded = ydl.downloaded_info_dicts[0]
 108         self.assertEqual(downloaded['ext'], 'webm')
 109
 110     def test_format_selection(self):
 111         formats = [
 112             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 113             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 114             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 115             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 116             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 117         ]
 118         info_dict = _make_result(formats)
 119
 120         ydl = YDL({'format': '20/47'})
 121         ydl.process_ie_result(info_dict.copy())
 122         downloaded = ydl.downloaded_info_dicts[0]
 123         self.assertEqual(downloaded['format_id'], '47')
 124
 125         ydl = YDL({'format': '20/71/worst'})
 126         ydl.process_ie_result(info_dict.copy())
 127         downloaded = ydl.downloaded_info_dicts[0]
 128         self.assertEqual(downloaded['format_id'], '35')
 129
 130         ydl = YDL()
 131         ydl.process_ie_result(info_dict.copy())
 132         downloaded = ydl.downloaded_info_dicts[0]
 133         self.assertEqual(downloaded['format_id'], '2')
 134
 135         ydl = YDL({'format': 'webm/mp4'})
 136         ydl.process_ie_result(info_dict.copy())
 137         downloaded = ydl.downloaded_info_dicts[0]
 138         self.assertEqual(downloaded['format_id'], '47')
 139
 140         ydl = YDL({'format': '3gp/40/mp4'})
 141         ydl.process_ie_result(info_dict.copy())
 142         downloaded = ydl.downloaded_info_dicts[0]
 143         self.assertEqual(downloaded['format_id'], '35')
 144
 145         ydl = YDL({'format': 'example-with-dashes'})
 146         ydl.process_ie_result(info_dict.copy())
 147         downloaded = ydl.downloaded_info_dicts[0]
 148         self.assertEqual(downloaded['format_id'], 'example-with-dashes')
 149
 150     def test_format_selection_audio(self):
 151         formats = [
 152             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 153             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 154             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 155             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 156         ]
 157         info_dict = _make_result(formats)
 158
 159         ydl = YDL({'format': 'bestaudio'})
 160         ydl.process_ie_result(info_dict.copy())
 161         downloaded = ydl.downloaded_info_dicts[0]
 162         self.assertEqual(downloaded['format_id'], 'audio-high')
 163
 164         ydl = YDL({'format': 'worstaudio'})
 165         ydl.process_ie_result(info_dict.copy())
 166         downloaded = ydl.downloaded_info_dicts[0]
 167         self.assertEqual(downloaded['format_id'], 'audio-low')
 168
 169         formats = [
 170             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 171             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 172         ]
 173         info_dict = _make_result(formats)
 174
 175         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 176         ydl.process_ie_result(info_dict.copy())
 177         downloaded = ydl.downloaded_info_dicts[0]
 178         self.assertEqual(downloaded['format_id'], 'vid-high')
 179
 180     def test_format_selection_audio_exts(self):
 181         formats = [
 182             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 184             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 185             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 186             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 187         ]
 188
 189         info_dict = _make_result(formats)
 190         ydl = YDL({'format': 'best'})
 191         ie = YoutubeIE(ydl)
 192         ie._sort_formats(info_dict['formats'])
 193         ydl.process_ie_result(copy.deepcopy(info_dict))
 194         downloaded = ydl.downloaded_info_dicts[0]
 195         self.assertEqual(downloaded['format_id'], 'aac-64')
 196
 197         ydl = YDL({'format': 'mp3'})
 198         ie = YoutubeIE(ydl)
 199         ie._sort_formats(info_dict['formats'])
 200         ydl.process_ie_result(copy.deepcopy(info_dict))
 201         downloaded = ydl.downloaded_info_dicts[0]
 202         self.assertEqual(downloaded['format_id'], 'mp3-64')
 203
 204         ydl = YDL({'prefer_free_formats': True})
 205         ie = YoutubeIE(ydl)
 206         ie._sort_formats(info_dict['formats'])
 207         ydl.process_ie_result(copy.deepcopy(info_dict))
 208         downloaded = ydl.downloaded_info_dicts[0]
 209         self.assertEqual(downloaded['format_id'], 'ogg-64')
 210
 211     def test_format_selection_video(self):
 212         formats = [
 213             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 214             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 215             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 216         ]
 217         info_dict = _make_result(formats)
 218
 219         ydl = YDL({'format': 'bestvideo'})
 220         ydl.process_ie_result(info_dict.copy())
 221         downloaded = ydl.downloaded_info_dicts[0]
 222         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 223
 224         ydl = YDL({'format': 'worstvideo'})
 225         ydl.process_ie_result(info_dict.copy())
 226         downloaded = ydl.downloaded_info_dicts[0]
 227         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 228
 229         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 230         ydl.process_ie_result(info_dict.copy())
 231         downloaded = ydl.downloaded_info_dicts[0]
 232         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 233
 234         formats = [
 235             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 236         ]
 237         info_dict = _make_result(formats)
 238
 239         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 240         ydl.process_ie_result(info_dict.copy())
 241         downloaded = ydl.downloaded_info_dicts[0]
 242         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 243
 244     def test_format_selection_string_ops(self):
 245         formats = [
 246             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 247             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 248         ]
 249         info_dict = _make_result(formats)
 250
 251         # equals (=)
 252         ydl = YDL({'format': '[format_id=abc-cba]'})
 253         ydl.process_ie_result(info_dict.copy())
 254         downloaded = ydl.downloaded_info_dicts[0]
 255         self.assertEqual(downloaded['format_id'], 'abc-cba')
 256
 257         # does not equal (!=)
 258         ydl = YDL({'format': '[format_id!=abc-cba]'})
 259         ydl.process_ie_result(info_dict.copy())
 260         downloaded = ydl.downloaded_info_dicts[0]
 261         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 262
 263         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 264         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 265
 266         # starts with (^=)
 267         ydl = YDL({'format': '[format_id^=abc]'})
 268         ydl.process_ie_result(info_dict.copy())
 269         downloaded = ydl.downloaded_info_dicts[0]
 270         self.assertEqual(downloaded['format_id'], 'abc-cba')
 271
 272         # does not start with (!^=)
 273         ydl = YDL({'format': '[format_id!^=abc]'})
 274         ydl.process_ie_result(info_dict.copy())
 275         downloaded = ydl.downloaded_info_dicts[0]
 276         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 277
 278         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 279         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 280
 281         # ends with ($=)
 282         ydl = YDL({'format': '[format_id$=cba]'})
 283         ydl.process_ie_result(info_dict.copy())
 284         downloaded = ydl.downloaded_info_dicts[0]
 285         self.assertEqual(downloaded['format_id'], 'abc-cba')
 286
 287         # does not end with (!$=)
 288         ydl = YDL({'format': '[format_id!$=cba]'})
 289         ydl.process_ie_result(info_dict.copy())
 290         downloaded = ydl.downloaded_info_dicts[0]
 291         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 292
 293         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 294         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 295
 296         # contains (*=)
 297         ydl = YDL({'format': '[format_id*=bc-cb]'})
 298         ydl.process_ie_result(info_dict.copy())
 299         downloaded = ydl.downloaded_info_dicts[0]
 300         self.assertEqual(downloaded['format_id'], 'abc-cba')
 301
 302         # does not contain (!*=)
 303         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 304         ydl.process_ie_result(info_dict.copy())
 305         downloaded = ydl.downloaded_info_dicts[0]
 306         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 307
 308         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 309         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 310
 311         ydl = YDL({'format': '[format_id!*=-]'})
 312         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 313
 314     def test_youtube_format_selection(self):
 315         # FIXME: Rewrite in accordance with the new format sorting options
 316         return
 317
 318         order = [
 319             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 320             # Apple HTTP Live Streaming
 321             '96', '95', '94', '93', '92', '132', '151',
 322             # 3D
 323             '85', '84', '102', '83', '101', '82', '100',
 324             # Dash video
 325             '137', '248', '136', '247', '135', '246',
 326             '245', '244', '134', '243', '133', '242', '160',
 327             # Dash audio
 328             '141', '172', '140', '171', '139',
 329         ]
 330
 331         def format_info(f_id):
 332             info = YoutubeIE._formats[f_id].copy()
 333
 334             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 335             # and 'vcodec', while in tests such information is incomplete since
 336             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 337             # test_YoutubeDL.test_youtube_format_selection is broken without
 338             # this fix
 339             if 'acodec' in info and 'vcodec' not in info:
 340                 info['vcodec'] = 'none'
 341             elif 'vcodec' in info and 'acodec' not in info:
 342                 info['acodec'] = 'none'
 343
 344             info['format_id'] = f_id
 345             info['url'] = 'url:' + f_id
 346             return info
 347         formats_order = [format_info(f_id) for f_id in order]
 348
 349         info_dict = _make_result(list(formats_order), extractor='youtube')
 350         ydl = YDL({'format': 'bestvideo+bestaudio'})
 351         yie = YoutubeIE(ydl)
 352         yie._sort_formats(info_dict['formats'])
 353         ydl.process_ie_result(info_dict)
 354         downloaded = ydl.downloaded_info_dicts[0]
 355         self.assertEqual(downloaded['format_id'], '248+172')
 356         self.assertEqual(downloaded['ext'], 'mp4')
 357
 358         info_dict = _make_result(list(formats_order), extractor='youtube')
 359         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 360         yie = YoutubeIE(ydl)
 361         yie._sort_formats(info_dict['formats'])
 362         ydl.process_ie_result(info_dict)
 363         downloaded = ydl.downloaded_info_dicts[0]
 364         self.assertEqual(downloaded['format_id'], '38')
 365
 366         info_dict = _make_result(list(formats_order), extractor='youtube')
 367         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 368         yie = YoutubeIE(ydl)
 369         yie._sort_formats(info_dict['formats'])
 370         ydl.process_ie_result(info_dict)
 371         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 372         self.assertEqual(downloaded_ids, ['137', '141'])
 373
 374         info_dict = _make_result(list(formats_order), extractor='youtube')
 375         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 376         yie = YoutubeIE(ydl)
 377         yie._sort_formats(info_dict['formats'])
 378         ydl.process_ie_result(info_dict)
 379         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 380         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 381
 382         info_dict = _make_result(list(formats_order), extractor='youtube')
 383         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 384         yie = YoutubeIE(ydl)
 385         yie._sort_formats(info_dict['formats'])
 386         ydl.process_ie_result(info_dict)
 387         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 388         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 389
 390         info_dict = _make_result(list(formats_order), extractor='youtube')
 391         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 392         yie = YoutubeIE(ydl)
 393         yie._sort_formats(info_dict['formats'])
 394         ydl.process_ie_result(info_dict)
 395         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 396         self.assertEqual(downloaded_ids, ['248+141'])
 397
 398         for f1, f2 in zip(formats_order, formats_order[1:]):
 399             info_dict = _make_result([f1, f2], extractor='youtube')
 400             ydl = YDL({'format': 'best/bestvideo'})
 401             yie = YoutubeIE(ydl)
 402             yie._sort_formats(info_dict['formats'])
 403             ydl.process_ie_result(info_dict)
 404             downloaded = ydl.downloaded_info_dicts[0]
 405             self.assertEqual(downloaded['format_id'], f1['format_id'])
 406
 407             info_dict = _make_result([f2, f1], extractor='youtube')
 408             ydl = YDL({'format': 'best/bestvideo'})
 409             yie = YoutubeIE(ydl)
 410             yie._sort_formats(info_dict['formats'])
 411             ydl.process_ie_result(info_dict)
 412             downloaded = ydl.downloaded_info_dicts[0]
 413             self.assertEqual(downloaded['format_id'], f1['format_id'])
 414
 415     def test_audio_only_extractor_format_selection(self):
 416         # For extractors with incomplete formats (all formats are audio-only or
 417         # video-only) best and worst should fallback to corresponding best/worst
 418         # video-only or audio-only formats (as per
 419         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 420         formats = [
 421             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 422             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 423         ]
 424         info_dict = _make_result(formats)
 425
 426         ydl = YDL({'format': 'best'})
 427         ydl.process_ie_result(info_dict.copy())
 428         downloaded = ydl.downloaded_info_dicts[0]
 429         self.assertEqual(downloaded['format_id'], 'high')
 430
 431         ydl = YDL({'format': 'worst'})
 432         ydl.process_ie_result(info_dict.copy())
 433         downloaded = ydl.downloaded_info_dicts[0]
 434         self.assertEqual(downloaded['format_id'], 'low')
 435
 436     def test_format_not_available(self):
 437         formats = [
 438             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 439             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 440         ]
 441         info_dict = _make_result(formats)
 442
 443         # This must fail since complete video-audio format does not match filter
 444         # and extractor does not provide incomplete only formats (i.e. only
 445         # video-only or audio-only).
 446         ydl = YDL({'format': 'best[height>360]'})
 447         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 448
 449     def test_format_selection_issue_10083(self):
 450         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 451         formats = [
 452             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 453             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 454             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 455         ]
 456         info_dict = _make_result(formats)
 457
 458         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 459         ydl.process_ie_result(info_dict.copy())
 460         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 461
 462     def test_invalid_format_specs(self):
 463         def assert_syntax_error(format_spec):
 464             ydl = YDL({'format': format_spec})
 465             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 466             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 467
 468         assert_syntax_error('bestvideo,,best')
 469         assert_syntax_error('+bestaudio')
 470         assert_syntax_error('bestvideo+')
 471         assert_syntax_error('/')
 472
 473     def test_format_filtering(self):
 474         formats = [
 475             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 476             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 477             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 478             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 479             {'format_id': 'E', 'filesize': 3000},
 480             {'format_id': 'F'},
 481             {'format_id': 'G', 'filesize': 1000000},
 482         ]
 483         for f in formats:
 484             f['url'] = 'http://_/'
 485             f['ext'] = 'unknown'
 486         info_dict = _make_result(formats)
 487
 488         ydl = YDL({'format': 'best[filesize<3000]'})
 489         ydl.process_ie_result(info_dict)
 490         downloaded = ydl.downloaded_info_dicts[0]
 491         self.assertEqual(downloaded['format_id'], 'D')
 492
 493         ydl = YDL({'format': 'best[filesize<=3000]'})
 494         ydl.process_ie_result(info_dict)
 495         downloaded = ydl.downloaded_info_dicts[0]
 496         self.assertEqual(downloaded['format_id'], 'E')
 497
 498         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 499         ydl.process_ie_result(info_dict)
 500         downloaded = ydl.downloaded_info_dicts[0]
 501         self.assertEqual(downloaded['format_id'], 'F')
 502
 503         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 504         ydl.process_ie_result(info_dict)
 505         downloaded = ydl.downloaded_info_dicts[0]
 506         self.assertEqual(downloaded['format_id'], 'B')
 507
 508         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 509         ydl.process_ie_result(info_dict)
 510         downloaded = ydl.downloaded_info_dicts[0]
 511         self.assertEqual(downloaded['format_id'], 'C')
 512
 513         ydl = YDL({'format': '[filesize>?1]'})
 514         ydl.process_ie_result(info_dict)
 515         downloaded = ydl.downloaded_info_dicts[0]
 516         self.assertEqual(downloaded['format_id'], 'G')
 517
 518         ydl = YDL({'format': '[filesize<1M]'})
 519         ydl.process_ie_result(info_dict)
 520         downloaded = ydl.downloaded_info_dicts[0]
 521         self.assertEqual(downloaded['format_id'], 'E')
 522
 523         ydl = YDL({'format': '[filesize<1MiB]'})
 524         ydl.process_ie_result(info_dict)
 525         downloaded = ydl.downloaded_info_dicts[0]
 526         self.assertEqual(downloaded['format_id'], 'G')
 527
 528         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 529         ydl.process_ie_result(info_dict)
 530         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 531         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 532
 533         ydl = YDL({'format': 'best[height<40]'})
 534         try:
 535             ydl.process_ie_result(info_dict)
 536         except ExtractorError:
 537             pass
 538         self.assertEqual(ydl.downloaded_info_dicts, [])
 539
 540     def test_default_format_spec(self):
 541         ydl = YDL({'simulate': True})
 542         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 543
 544         ydl = YDL({})
 545         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 546
 547         ydl = YDL({'simulate': True})
 548         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo*+bestaudio/best')
 549
 550         ydl = YDL({'outtmpl': '-'})
 551         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 552
 553         ydl = YDL({})
 554         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo*+bestaudio/best')
 555         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 556
 557
 558 class TestYoutubeDL(unittest.TestCase):
 559     def test_subtitles(self):
 560         def s_formats(lang, autocaption=False):
 561             return [{
 562                 'ext': ext,
 563                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 564                 '_auto': autocaption,
 565             } for ext in ['vtt', 'srt', 'ass']]
 566         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 567         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 568         info_dict = {
 569             'id': 'test',
 570             'title': 'Test',
 571             'url': 'http://localhost/video.mp4',
 572             'subtitles': subtitles,
 573             'automatic_captions': auto_captions,
 574             'extractor': 'TEST',
 575             'webpage_url': 'http://example.com/watch?v=shenanigans',
 576         }
 577
 578         def get_info(params={}):
 579             params.setdefault('simulate', True)
 580             ydl = YDL(params)
 581             ydl.report_warning = lambda *args, **kargs: None
 582             return ydl.process_video_result(info_dict, download=False)
 583
 584         result = get_info()
 585         self.assertFalse(result.get('requested_subtitles'))
 586         self.assertEqual(result['subtitles'], subtitles)
 587         self.assertEqual(result['automatic_captions'], auto_captions)
 588
 589         result = get_info({'writesubtitles': True})
 590         subs = result['requested_subtitles']
 591         self.assertTrue(subs)
 592         self.assertEqual(set(subs.keys()), set(['en']))
 593         self.assertTrue(subs['en'].get('data') is None)
 594         self.assertEqual(subs['en']['ext'], 'ass')
 595
 596         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 597         subs = result['requested_subtitles']
 598         self.assertEqual(subs['en']['ext'], 'srt')
 599
 600         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 601         subs = result['requested_subtitles']
 602         self.assertTrue(subs)
 603         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 604
 605         result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']})
 606         subs = result['requested_subtitles']
 607         self.assertTrue(subs)
 608         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 609
 610         result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']})
 611         subs = result['requested_subtitles']
 612         self.assertTrue(subs)
 613         self.assertEqual(set(subs.keys()), set(['fr']))
 614
 615         result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']})
 616         subs = result['requested_subtitles']
 617         self.assertTrue(subs)
 618         self.assertEqual(set(subs.keys()), set(['en']))
 619
 620         result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']})
 621         subs = result['requested_subtitles']
 622         self.assertTrue(subs)
 623         self.assertEqual(set(subs.keys()), set(['es', 'en']))
 624
 625         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 626         subs = result['requested_subtitles']
 627         self.assertTrue(subs)
 628         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 629         self.assertFalse(subs['es']['_auto'])
 630         self.assertTrue(subs['pt']['_auto'])
 631
 632         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 633         subs = result['requested_subtitles']
 634         self.assertTrue(subs)
 635         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 636         self.assertTrue(subs['es']['_auto'])
 637         self.assertTrue(subs['pt']['_auto'])
 638
 639     def test_add_extra_info(self):
 640         test_dict = {
 641             'extractor': 'Foo',
 642         }
 643         extra_info = {
 644             'extractor': 'Bar',
 645             'playlist': 'funny videos',
 646         }
 647         YDL.add_extra_info(test_dict, extra_info)
 648         self.assertEqual(test_dict['extractor'], 'Foo')
 649         self.assertEqual(test_dict['playlist'], 'funny videos')
 650
 651     def test_prepare_filename(self):
 652         info = {
 653             'id': '1234',
 654             'ext': 'mp4',
 655             'width': None,
 656             'height': 1080,
 657             'title1': '$PATH',
 658             'title2': '%PATH%',
 659             'timestamp': 1618488000,
 660             'formats': [{'id': 'id1'}, {'id': 'id2'}]
 661         }
 662
 663         def fname(templ, na_placeholder='NA'):
 664             params = {'outtmpl': templ}
 665             if na_placeholder != 'NA':
 666                 params['outtmpl_na_placeholder'] = na_placeholder
 667             ydl = YoutubeDL(params)
 668             return ydl.prepare_filename(info)
 669         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 670         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 671         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(id)s.%(ext)s'
 672         # Replace missing fields with 'NA' by default
 673         self.assertEqual(fname(NA_TEST_OUTTMPL), 'NA-NA-1234.mp4')
 674         # Or by provided placeholder
 675         self.assertEqual(fname(NA_TEST_OUTTMPL, na_placeholder='none'), 'none-none-1234.mp4')
 676         self.assertEqual(fname(NA_TEST_OUTTMPL, na_placeholder=''), '--1234.mp4')
 677         self.assertEqual(fname('%(height)s.%(ext)s'), '1080.mp4')
 678         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 679         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 680         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 681         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 682         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 683         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 684         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 685         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 686         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 687         self.assertEqual(fname('%%'), '%')
 688         self.assertEqual(fname('%%%%'), '%%')
 689         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 690         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 691         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 692         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 693         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 694         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 695         self.assertEqual(fname('%(timestamp+-1000>%H-%M-%S)s'), '11-43-20')
 696         self.assertEqual(fname('%(id+1)05d'), '01235')
 697         self.assertEqual(fname('%(width+100)05d'), 'NA')
 698         self.assertEqual(fname('%(formats.0)s').replace("u", ""), "{'id' - 'id1'}")
 699         self.assertEqual(fname('%(formats.-1.id)s'), 'id2')
 700         self.assertEqual(fname('%(formats.2)s'), 'NA')
 701
 702     def test_format_note(self):
 703         ydl = YoutubeDL()
 704         self.assertEqual(ydl._format_note({}), '')
 705         assertRegexpMatches(self, ydl._format_note({
 706             'vbr': 10,
 707         }), r'^\s*10k$')
 708         assertRegexpMatches(self, ydl._format_note({
 709             'fps': 30,
 710         }), r'^30fps$')
 711
 712     def test_postprocessors(self):
 713         filename = 'post-processor-testfile.mp4'
 714         audiofile = filename + '.mp3'
 715
 716         class SimplePP(PostProcessor):
 717             def run(self, info):
 718                 with open(audiofile, 'wt') as f:
 719                     f.write('EXAMPLE')
 720                 return [info['filepath']], info
 721
 722         def run_pp(params, PP):
 723             with open(filename, 'wt') as f:
 724                 f.write('EXAMPLE')
 725             ydl = YoutubeDL(params)
 726             ydl.add_post_processor(PP())
 727             ydl.post_process(filename, {'filepath': filename})
 728
 729         run_pp({'keepvideo': True}, SimplePP)
 730         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 731         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 732         os.unlink(filename)
 733         os.unlink(audiofile)
 734
 735         run_pp({'keepvideo': False}, SimplePP)
 736         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 737         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 738         os.unlink(audiofile)
 739
 740         class ModifierPP(PostProcessor):
 741             def run(self, info):
 742                 with open(info['filepath'], 'wt') as f:
 743                     f.write('MODIFIED')
 744                 return [], info
 745
 746         run_pp({'keepvideo': False}, ModifierPP)
 747         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 748         os.unlink(filename)
 749
 750     def test_match_filter(self):
 751         class FilterYDL(YDL):
 752             def __init__(self, *args, **kwargs):
 753                 super(FilterYDL, self).__init__(*args, **kwargs)
 754                 self.params['simulate'] = True
 755
 756             def process_info(self, info_dict):
 757                 super(YDL, self).process_info(info_dict)
 758
 759             def _match_entry(self, info_dict, incomplete=False):
 760                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 761                 if res is None:
 762                     self.downloaded_info_dicts.append(info_dict)
 763                 return res
 764
 765         first = {
 766             'id': '1',
 767             'url': TEST_URL,
 768             'title': 'one',
 769             'extractor': 'TEST',
 770             'duration': 30,
 771             'filesize': 10 * 1024,
 772             'playlist_id': '42',
 773             'uploader': "變態妍字幕版 太妍 тест",
 774             'creator': "тест ' 123 ' тест--",
 775             'webpage_url': 'http://example.com/watch?v=shenanigans',
 776         }
 777         second = {
 778             'id': '2',
 779             'url': TEST_URL,
 780             'title': 'two',
 781             'extractor': 'TEST',
 782             'duration': 10,
 783             'description': 'foo',
 784             'filesize': 5 * 1024,
 785             'playlist_id': '43',
 786             'uploader': "тест 123",
 787             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 788         }
 789         videos = [first, second]
 790
 791         def get_videos(filter_=None):
 792             ydl = FilterYDL({'match_filter': filter_})
 793             for v in videos:
 794                 ydl.process_ie_result(v, download=True)
 795             return [v['id'] for v in ydl.downloaded_info_dicts]
 796
 797         res = get_videos()
 798         self.assertEqual(res, ['1', '2'])
 799
 800         def f(v):
 801             if v['id'] == '1':
 802                 return None
 803             else:
 804                 return 'Video id is not 1'
 805         res = get_videos(f)
 806         self.assertEqual(res, ['1'])
 807
 808         f = match_filter_func('duration < 30')
 809         res = get_videos(f)
 810         self.assertEqual(res, ['2'])
 811
 812         f = match_filter_func('description = foo')
 813         res = get_videos(f)
 814         self.assertEqual(res, ['2'])
 815
 816         f = match_filter_func('description =? foo')
 817         res = get_videos(f)
 818         self.assertEqual(res, ['1', '2'])
 819
 820         f = match_filter_func('filesize > 5KiB')
 821         res = get_videos(f)
 822         self.assertEqual(res, ['1'])
 823
 824         f = match_filter_func('playlist_id = 42')
 825         res = get_videos(f)
 826         self.assertEqual(res, ['1'])
 827
 828         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 829         res = get_videos(f)
 830         self.assertEqual(res, ['1'])
 831
 832         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 833         res = get_videos(f)
 834         self.assertEqual(res, ['2'])
 835
 836         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 837         res = get_videos(f)
 838         self.assertEqual(res, ['1'])
 839
 840         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 841         res = get_videos(f)
 842         self.assertEqual(res, ['1'])
 843
 844         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 845         res = get_videos(f)
 846         self.assertEqual(res, [])
 847
 848     def test_playlist_items_selection(self):
 849         entries = [{
 850             'id': compat_str(i),
 851             'title': compat_str(i),
 852             'url': TEST_URL,
 853         } for i in range(1, 5)]
 854         playlist = {
 855             '_type': 'playlist',
 856             'id': 'test',
 857             'entries': entries,
 858             'extractor': 'test:playlist',
 859             'extractor_key': 'test:playlist',
 860             'webpage_url': 'http://example.com',
 861         }
 862
 863         def get_downloaded_info_dicts(params):
 864             ydl = YDL(params)
 865             # make a deep copy because the dictionary and nested entries
 866             # can be modified
 867             ydl.process_ie_result(copy.deepcopy(playlist))
 868             return ydl.downloaded_info_dicts
 869
 870         def get_ids(params):
 871             return [int(v['id']) for v in get_downloaded_info_dicts(params)]
 872
 873         result = get_ids({})
 874         self.assertEqual(result, [1, 2, 3, 4])
 875
 876         result = get_ids({'playlistend': 10})
 877         self.assertEqual(result, [1, 2, 3, 4])
 878
 879         result = get_ids({'playlistend': 2})
 880         self.assertEqual(result, [1, 2])
 881
 882         result = get_ids({'playliststart': 10})
 883         self.assertEqual(result, [])
 884
 885         result = get_ids({'playliststart': 2})
 886         self.assertEqual(result, [2, 3, 4])
 887
 888         result = get_ids({'playlist_items': '2-4'})
 889         self.assertEqual(result, [2, 3, 4])
 890
 891         result = get_ids({'playlist_items': '2,4'})
 892         self.assertEqual(result, [2, 4])
 893
 894         result = get_ids({'playlist_items': '10'})
 895         self.assertEqual(result, [])
 896
 897         result = get_ids({'playlist_items': '3-10'})
 898         self.assertEqual(result, [3, 4])
 899
 900         result = get_ids({'playlist_items': '2-4,3-4,3'})
 901         self.assertEqual(result, [2, 3, 4])
 902
 903         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
 904         # @{
 905         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 906         self.assertEqual(result[0]['playlist_index'], 2)
 907         self.assertEqual(result[1]['playlist_index'], 3)
 908
 909         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 910         self.assertEqual(result[0]['playlist_index'], 2)
 911         self.assertEqual(result[1]['playlist_index'], 3)
 912         self.assertEqual(result[2]['playlist_index'], 4)
 913
 914         result = get_downloaded_info_dicts({'playlist_items': '4,2'})
 915         self.assertEqual(result[0]['playlist_index'], 4)
 916         self.assertEqual(result[1]['playlist_index'], 2)
 917         # @}
 918
 919     def test_urlopen_no_file_protocol(self):
 920         # see https://github.com/ytdl-org/youtube-dl/issues/8227
 921         ydl = YDL()
 922         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 923
 924     def test_do_not_override_ie_key_in_url_transparent(self):
 925         ydl = YDL()
 926
 927         class Foo1IE(InfoExtractor):
 928             _VALID_URL = r'foo1:'
 929
 930             def _real_extract(self, url):
 931                 return {
 932                     '_type': 'url_transparent',
 933                     'url': 'foo2:',
 934                     'ie_key': 'Foo2',
 935                     'title': 'foo1 title',
 936                     'id': 'foo1_id',
 937                 }
 938
 939         class Foo2IE(InfoExtractor):
 940             _VALID_URL = r'foo2:'
 941
 942             def _real_extract(self, url):
 943                 return {
 944                     '_type': 'url',
 945                     'url': 'foo3:',
 946                     'ie_key': 'Foo3',
 947                 }
 948
 949         class Foo3IE(InfoExtractor):
 950             _VALID_URL = r'foo3:'
 951
 952             def _real_extract(self, url):
 953                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 954
 955         ydl.add_info_extractor(Foo1IE(ydl))
 956         ydl.add_info_extractor(Foo2IE(ydl))
 957         ydl.add_info_extractor(Foo3IE(ydl))
 958         ydl.extract_info('foo1:')
 959         downloaded = ydl.downloaded_info_dicts[0]
 960         self.assertEqual(downloaded['url'], TEST_URL)
 961         self.assertEqual(downloaded['title'], 'foo1 title')
 962         self.assertEqual(downloaded['id'], 'testid')
 963         self.assertEqual(downloaded['extractor'], 'testex')
 964         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 965
 966     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
 967     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
 968
 969         class _YDL(YDL):
 970             def __init__(self, *args, **kwargs):
 971                 super(_YDL, self).__init__(*args, **kwargs)
 972
 973             def trouble(self, s, tb=None):
 974                 pass
 975
 976         ydl = _YDL({
 977             'format': 'extra',
 978             'ignoreerrors': True,
 979         })
 980
 981         class VideoIE(InfoExtractor):
 982             _VALID_URL = r'video:(?P<id>\d+)'
 983
 984             def _real_extract(self, url):
 985                 video_id = self._match_id(url)
 986                 formats = [{
 987                     'format_id': 'default',
 988                     'url': 'url:',
 989                 }]
 990                 if video_id == '0':
 991                     raise ExtractorError('foo')
 992                 if video_id == '2':
 993                     formats.append({
 994                         'format_id': 'extra',
 995                         'url': TEST_URL,
 996                     })
 997                 return {
 998                     'id': video_id,
 999                     'title': 'Video %s' % video_id,
1000                     'formats': formats,
1001                 }
1002
1003         class PlaylistIE(InfoExtractor):
1004             _VALID_URL = r'playlist:'
1005
1006             def _entries(self):
1007                 for n in range(3):
1008                     video_id = compat_str(n)
1009                     yield {
1010                         '_type': 'url_transparent',
1011                         'ie_key': VideoIE.ie_key(),
1012                         'id': video_id,
1013                         'url': 'video:%s' % video_id,
1014                         'title': 'Video Transparent %s' % video_id,
1015                     }
1016
1017             def _real_extract(self, url):
1018                 return self.playlist_result(self._entries())
1019
1020         ydl.add_info_extractor(VideoIE(ydl))
1021         ydl.add_info_extractor(PlaylistIE(ydl))
1022         info = ydl.extract_info('playlist:')
1023         entries = info['entries']
1024         self.assertEqual(len(entries), 3)
1025         self.assertTrue(entries[0] is None)
1026         self.assertTrue(entries[1] is None)
1027         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1028         downloaded = ydl.downloaded_info_dicts[0]
1029         self.assertEqual(entries[2], downloaded)
1030         self.assertEqual(downloaded['url'], TEST_URL)
1031         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1032         self.assertEqual(downloaded['id'], '2')
1033         self.assertEqual(downloaded['extractor'], 'Video')
1034         self.assertEqual(downloaded['extractor_key'], 'Video')
1035
1036
1037 if __name__ == '__main__':
1038     unittest.main()