test/test_YoutubeDL.py

   1 #!/usr/bin/env python3
   2
   3 # Allow direct execution
   4 import os
   5 import sys
   6 import unittest
   7
   8 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
   9
  10
  11 import copy
  12 import json
  13
  14 from test.helper import FakeYDL, assertRegexpMatches, try_rm
  15 from yt_dlp import YoutubeDL
  16 from yt_dlp.compat import compat_os_name
  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 (
  21     ExtractorError,
  22     LazyList,
  23     OnDemandPagedList,
  24     int_or_none,
  25     match_filter_func,
  26 )
  27 from yt_dlp.utils.traversal import traverse_obj
  28
  29 TEST_URL = 'http://localhost/sample.mp4'
  30
  31
  32 class YDL(FakeYDL):
  33     def __init__(self, *args, **kwargs):
  34         super().__init__(*args, **kwargs)
  35         self.downloaded_info_dicts = []
  36         self.msgs = []
  37
  38     def process_info(self, info_dict):
  39         self.downloaded_info_dicts.append(info_dict.copy())
  40
  41     def to_screen(self, msg, *args, **kwargs):
  42         self.msgs.append(msg)
  43
  44     def dl(self, *args, **kwargs):
  45         assert False, 'Downloader must not be invoked for test_YoutubeDL'
  46
  47
  48 def _make_result(formats, **kwargs):
  49     res = {
  50         'formats': formats,
  51         'id': 'testid',
  52         'title': 'testttitle',
  53         'extractor': 'testex',
  54         'extractor_key': 'TestEx',
  55         'webpage_url': 'http://example.com/watch?v=shenanigans',
  56     }
  57     res.update(**kwargs)
  58     return res
  59
  60
  61 class TestFormatSelection(unittest.TestCase):
  62     def test_prefer_free_formats(self):
  63         # Same resolution => download webm
  64         ydl = YDL()
  65         ydl.params['prefer_free_formats'] = True
  66         formats = [
  67             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  68             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  69         ]
  70         info_dict = _make_result(formats)
  71         ydl.sort_formats(info_dict)
  72         ydl.process_ie_result(info_dict)
  73         downloaded = ydl.downloaded_info_dicts[0]
  74         self.assertEqual(downloaded['ext'], 'webm')
  75
  76         # Different resolution => download best quality (mp4)
  77         ydl = YDL()
  78         ydl.params['prefer_free_formats'] = True
  79         formats = [
  80             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  81             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  82         ]
  83         info_dict['formats'] = formats
  84         ydl.sort_formats(info_dict)
  85         ydl.process_ie_result(info_dict)
  86         downloaded = ydl.downloaded_info_dicts[0]
  87         self.assertEqual(downloaded['ext'], 'mp4')
  88
  89         # No prefer_free_formats => prefer mp4 and webm
  90         ydl = YDL()
  91         ydl.params['prefer_free_formats'] = False
  92         formats = [
  93             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  94             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  95             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  96         ]
  97         info_dict['formats'] = formats
  98         ydl.sort_formats(info_dict)
  99         ydl.process_ie_result(info_dict)
 100         downloaded = ydl.downloaded_info_dicts[0]
 101         self.assertEqual(downloaded['ext'], 'mp4')
 102
 103         ydl = YDL()
 104         ydl.params['prefer_free_formats'] = False
 105         formats = [
 106             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
 107             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 108         ]
 109         info_dict['formats'] = formats
 110         ydl.sort_formats(info_dict)
 111         ydl.process_ie_result(info_dict)
 112         downloaded = ydl.downloaded_info_dicts[0]
 113         self.assertEqual(downloaded['ext'], 'webm')
 114
 115     def test_format_selection(self):
 116         formats = [
 117             {'format_id': '35', 'ext': 'mp4', 'preference': 0, 'url': TEST_URL},
 118             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 119             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 120             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 121             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 122         ]
 123         info_dict = _make_result(formats)
 124
 125         def test(inp, *expected, multi=False):
 126             ydl = YDL({
 127                 'format': inp,
 128                 'allow_multiple_video_streams': multi,
 129                 'allow_multiple_audio_streams': multi,
 130             })
 131             ydl.process_ie_result(info_dict.copy())
 132             downloaded = map(lambda x: x['format_id'], ydl.downloaded_info_dicts)
 133             self.assertEqual(list(downloaded), list(expected))
 134
 135         test('20/47', '47')
 136         test('20/71/worst', '35')
 137         test(None, '2')
 138         test('webm/mp4', '47')
 139         test('3gp/40/mp4', '35')
 140         test('example-with-dashes', 'example-with-dashes')
 141         test('all', '2', '47', '45', 'example-with-dashes', '35')
 142         test('mergeall', '2+47+45+example-with-dashes+35', multi=True)
 143
 144     def test_format_selection_audio(self):
 145         formats = [
 146             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 147             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 148             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 149             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 150         ]
 151         info_dict = _make_result(formats)
 152
 153         ydl = YDL({'format': 'bestaudio'})
 154         ydl.process_ie_result(info_dict.copy())
 155         downloaded = ydl.downloaded_info_dicts[0]
 156         self.assertEqual(downloaded['format_id'], 'audio-high')
 157
 158         ydl = YDL({'format': 'worstaudio'})
 159         ydl.process_ie_result(info_dict.copy())
 160         downloaded = ydl.downloaded_info_dicts[0]
 161         self.assertEqual(downloaded['format_id'], 'audio-low')
 162
 163         formats = [
 164             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 165             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 166         ]
 167         info_dict = _make_result(formats)
 168
 169         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 170         ydl.process_ie_result(info_dict.copy())
 171         downloaded = ydl.downloaded_info_dicts[0]
 172         self.assertEqual(downloaded['format_id'], 'vid-high')
 173
 174     def test_format_selection_audio_exts(self):
 175         formats = [
 176             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 177             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 178             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 179             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 180             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 181         ]
 182
 183         info_dict = _make_result(formats)
 184         ydl = YDL({'format': 'best'})
 185         ydl.sort_formats(info_dict)
 186         ydl.process_ie_result(copy.deepcopy(info_dict))
 187         downloaded = ydl.downloaded_info_dicts[0]
 188         self.assertEqual(downloaded['format_id'], 'aac-64')
 189
 190         ydl = YDL({'format': 'mp3'})
 191         ydl.sort_formats(info_dict)
 192         ydl.process_ie_result(copy.deepcopy(info_dict))
 193         downloaded = ydl.downloaded_info_dicts[0]
 194         self.assertEqual(downloaded['format_id'], 'mp3-64')
 195
 196         ydl = YDL({'prefer_free_formats': True})
 197         ydl.sort_formats(info_dict)
 198         ydl.process_ie_result(copy.deepcopy(info_dict))
 199         downloaded = ydl.downloaded_info_dicts[0]
 200         self.assertEqual(downloaded['format_id'], 'ogg-64')
 201
 202     def test_format_selection_video(self):
 203         formats = [
 204             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 205             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 206             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 207         ]
 208         info_dict = _make_result(formats)
 209
 210         ydl = YDL({'format': 'bestvideo'})
 211         ydl.process_ie_result(info_dict.copy())
 212         downloaded = ydl.downloaded_info_dicts[0]
 213         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 214
 215         ydl = YDL({'format': 'worstvideo'})
 216         ydl.process_ie_result(info_dict.copy())
 217         downloaded = ydl.downloaded_info_dicts[0]
 218         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 219
 220         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 221         ydl.process_ie_result(info_dict.copy())
 222         downloaded = ydl.downloaded_info_dicts[0]
 223         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 224
 225         formats = [
 226             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 227         ]
 228         info_dict = _make_result(formats)
 229
 230         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 231         ydl.process_ie_result(info_dict.copy())
 232         downloaded = ydl.downloaded_info_dicts[0]
 233         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 234
 235     def test_format_selection_string_ops(self):
 236         formats = [
 237             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 238             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 239         ]
 240         info_dict = _make_result(formats)
 241
 242         # equals (=)
 243         ydl = YDL({'format': '[format_id=abc-cba]'})
 244         ydl.process_ie_result(info_dict.copy())
 245         downloaded = ydl.downloaded_info_dicts[0]
 246         self.assertEqual(downloaded['format_id'], 'abc-cba')
 247
 248         # does not equal (!=)
 249         ydl = YDL({'format': '[format_id!=abc-cba]'})
 250         ydl.process_ie_result(info_dict.copy())
 251         downloaded = ydl.downloaded_info_dicts[0]
 252         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 253
 254         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 255         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 256
 257         # starts with (^=)
 258         ydl = YDL({'format': '[format_id^=abc]'})
 259         ydl.process_ie_result(info_dict.copy())
 260         downloaded = ydl.downloaded_info_dicts[0]
 261         self.assertEqual(downloaded['format_id'], 'abc-cba')
 262
 263         # does not start with (!^=)
 264         ydl = YDL({'format': '[format_id!^=abc]'})
 265         ydl.process_ie_result(info_dict.copy())
 266         downloaded = ydl.downloaded_info_dicts[0]
 267         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 268
 269         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 270         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 271
 272         # ends with ($=)
 273         ydl = YDL({'format': '[format_id$=cba]'})
 274         ydl.process_ie_result(info_dict.copy())
 275         downloaded = ydl.downloaded_info_dicts[0]
 276         self.assertEqual(downloaded['format_id'], 'abc-cba')
 277
 278         # does not end with (!$=)
 279         ydl = YDL({'format': '[format_id!$=cba]'})
 280         ydl.process_ie_result(info_dict.copy())
 281         downloaded = ydl.downloaded_info_dicts[0]
 282         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 283
 284         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 285         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 286
 287         # contains (*=)
 288         ydl = YDL({'format': '[format_id*=bc-cb]'})
 289         ydl.process_ie_result(info_dict.copy())
 290         downloaded = ydl.downloaded_info_dicts[0]
 291         self.assertEqual(downloaded['format_id'], 'abc-cba')
 292
 293         # does not contain (!*=)
 294         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 295         ydl.process_ie_result(info_dict.copy())
 296         downloaded = ydl.downloaded_info_dicts[0]
 297         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 298
 299         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 300         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 301
 302         ydl = YDL({'format': '[format_id!*=-]'})
 303         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 304
 305     def test_youtube_format_selection(self):
 306         # FIXME: Rewrite in accordance with the new format sorting options
 307         return
 308
 309         order = [
 310             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 311             # Apple HTTP Live Streaming
 312             '96', '95', '94', '93', '92', '132', '151',
 313             # 3D
 314             '85', '84', '102', '83', '101', '82', '100',
 315             # Dash video
 316             '137', '248', '136', '247', '135', '246',
 317             '245', '244', '134', '243', '133', '242', '160',
 318             # Dash audio
 319             '141', '172', '140', '171', '139',
 320         ]
 321
 322         def format_info(f_id):
 323             info = YoutubeIE._formats[f_id].copy()
 324
 325             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 326             # and 'vcodec', while in tests such information is incomplete since
 327             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 328             # test_YoutubeDL.test_youtube_format_selection is broken without
 329             # this fix
 330             if 'acodec' in info and 'vcodec' not in info:
 331                 info['vcodec'] = 'none'
 332             elif 'vcodec' in info and 'acodec' not in info:
 333                 info['acodec'] = 'none'
 334
 335             info['format_id'] = f_id
 336             info['url'] = 'url:' + f_id
 337             return info
 338         formats_order = [format_info(f_id) for f_id in order]
 339
 340         info_dict = _make_result(list(formats_order), extractor='youtube')
 341         ydl = YDL({'format': 'bestvideo+bestaudio'})
 342         ydl.sort_formats(info_dict)
 343         ydl.process_ie_result(info_dict)
 344         downloaded = ydl.downloaded_info_dicts[0]
 345         self.assertEqual(downloaded['format_id'], '248+172')
 346         self.assertEqual(downloaded['ext'], 'mp4')
 347
 348         info_dict = _make_result(list(formats_order), extractor='youtube')
 349         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 350         ydl.sort_formats(info_dict)
 351         ydl.process_ie_result(info_dict)
 352         downloaded = ydl.downloaded_info_dicts[0]
 353         self.assertEqual(downloaded['format_id'], '38')
 354
 355         info_dict = _make_result(list(formats_order), extractor='youtube')
 356         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 357         ydl.sort_formats(info_dict)
 358         ydl.process_ie_result(info_dict)
 359         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 360         self.assertEqual(downloaded_ids, ['137', '141'])
 361
 362         info_dict = _make_result(list(formats_order), extractor='youtube')
 363         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 364         ydl.sort_formats(info_dict)
 365         ydl.process_ie_result(info_dict)
 366         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 367         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 368
 369         info_dict = _make_result(list(formats_order), extractor='youtube')
 370         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 371         ydl.sort_formats(info_dict)
 372         ydl.process_ie_result(info_dict)
 373         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 374         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 375
 376         info_dict = _make_result(list(formats_order), extractor='youtube')
 377         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 378         ydl.sort_formats(info_dict)
 379         ydl.process_ie_result(info_dict)
 380         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 381         self.assertEqual(downloaded_ids, ['248+141'])
 382
 383         for f1, f2 in zip(formats_order, formats_order[1:]):
 384             info_dict = _make_result([f1, f2], extractor='youtube')
 385             ydl = YDL({'format': 'best/bestvideo'})
 386             ydl.sort_formats(info_dict)
 387             ydl.process_ie_result(info_dict)
 388             downloaded = ydl.downloaded_info_dicts[0]
 389             self.assertEqual(downloaded['format_id'], f1['format_id'])
 390
 391             info_dict = _make_result([f2, f1], extractor='youtube')
 392             ydl = YDL({'format': 'best/bestvideo'})
 393             ydl.sort_formats(info_dict)
 394             ydl.process_ie_result(info_dict)
 395             downloaded = ydl.downloaded_info_dicts[0]
 396             self.assertEqual(downloaded['format_id'], f1['format_id'])
 397
 398     def test_audio_only_extractor_format_selection(self):
 399         # For extractors with incomplete formats (all formats are audio-only or
 400         # video-only) best and worst should fallback to corresponding best/worst
 401         # video-only or audio-only formats (as per
 402         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 403         formats = [
 404             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 405             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 406         ]
 407         info_dict = _make_result(formats)
 408
 409         ydl = YDL({'format': 'best'})
 410         ydl.process_ie_result(info_dict.copy())
 411         downloaded = ydl.downloaded_info_dicts[0]
 412         self.assertEqual(downloaded['format_id'], 'high')
 413
 414         ydl = YDL({'format': 'worst'})
 415         ydl.process_ie_result(info_dict.copy())
 416         downloaded = ydl.downloaded_info_dicts[0]
 417         self.assertEqual(downloaded['format_id'], 'low')
 418
 419     def test_format_not_available(self):
 420         formats = [
 421             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 422             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 423         ]
 424         info_dict = _make_result(formats)
 425
 426         # This must fail since complete video-audio format does not match filter
 427         # and extractor does not provide incomplete only formats (i.e. only
 428         # video-only or audio-only).
 429         ydl = YDL({'format': 'best[height>360]'})
 430         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 431
 432     def test_format_selection_issue_10083(self):
 433         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 434         formats = [
 435             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 436             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 437             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 438         ]
 439         info_dict = _make_result(formats)
 440
 441         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 442         ydl.process_ie_result(info_dict.copy())
 443         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 444
 445     def test_invalid_format_specs(self):
 446         def assert_syntax_error(format_spec):
 447             self.assertRaises(SyntaxError, YDL, {'format': format_spec})
 448
 449         assert_syntax_error('bestvideo,,best')
 450         assert_syntax_error('+bestaudio')
 451         assert_syntax_error('bestvideo+')
 452         assert_syntax_error('/')
 453         assert_syntax_error('[720<height]')
 454
 455     def test_format_filtering(self):
 456         formats = [
 457             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 458             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 459             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 460             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 461             {'format_id': 'E', 'filesize': 3000},
 462             {'format_id': 'F'},
 463             {'format_id': 'G', 'filesize': 1000000},
 464         ]
 465         for f in formats:
 466             f['url'] = 'http://_/'
 467             f['ext'] = 'unknown'
 468         info_dict = _make_result(formats, _format_sort_fields=('id', ))
 469
 470         ydl = YDL({'format': 'best[filesize<3000]'})
 471         ydl.process_ie_result(info_dict)
 472         downloaded = ydl.downloaded_info_dicts[0]
 473         self.assertEqual(downloaded['format_id'], 'D')
 474
 475         ydl = YDL({'format': 'best[filesize<=3000]'})
 476         ydl.process_ie_result(info_dict)
 477         downloaded = ydl.downloaded_info_dicts[0]
 478         self.assertEqual(downloaded['format_id'], 'E')
 479
 480         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 481         ydl.process_ie_result(info_dict)
 482         downloaded = ydl.downloaded_info_dicts[0]
 483         self.assertEqual(downloaded['format_id'], 'F')
 484
 485         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 486         ydl.process_ie_result(info_dict)
 487         downloaded = ydl.downloaded_info_dicts[0]
 488         self.assertEqual(downloaded['format_id'], 'B')
 489
 490         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 491         ydl.process_ie_result(info_dict)
 492         downloaded = ydl.downloaded_info_dicts[0]
 493         self.assertEqual(downloaded['format_id'], 'C')
 494
 495         ydl = YDL({'format': '[filesize>?1]'})
 496         ydl.process_ie_result(info_dict)
 497         downloaded = ydl.downloaded_info_dicts[0]
 498         self.assertEqual(downloaded['format_id'], 'G')
 499
 500         ydl = YDL({'format': '[filesize<1M]'})
 501         ydl.process_ie_result(info_dict)
 502         downloaded = ydl.downloaded_info_dicts[0]
 503         self.assertEqual(downloaded['format_id'], 'E')
 504
 505         ydl = YDL({'format': '[filesize<1MiB]'})
 506         ydl.process_ie_result(info_dict)
 507         downloaded = ydl.downloaded_info_dicts[0]
 508         self.assertEqual(downloaded['format_id'], 'G')
 509
 510         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 511         ydl.process_ie_result(info_dict)
 512         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 513         self.assertEqual(downloaded_ids, ['D', 'C', 'B'])
 514
 515         ydl = YDL({'format': 'best[height<40]'})
 516         try:
 517             ydl.process_ie_result(info_dict)
 518         except ExtractorError:
 519             pass
 520         self.assertEqual(ydl.downloaded_info_dicts, [])
 521
 522     def test_default_format_spec(self):
 523         ydl = YDL({'simulate': True})
 524         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 525
 526         ydl = YDL({})
 527         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 528
 529         ydl = YDL({'simulate': True})
 530         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo*+bestaudio/best')
 531
 532         ydl = YDL({'outtmpl': '-'})
 533         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 534
 535         ydl = YDL({})
 536         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo*+bestaudio/best')
 537         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 538
 539
 540 class TestYoutubeDL(unittest.TestCase):
 541     def test_subtitles(self):
 542         def s_formats(lang, autocaption=False):
 543             return [{
 544                 'ext': ext,
 545                 'url': f'http://localhost/video.{lang}.{ext}',
 546                 '_auto': autocaption,
 547             } for ext in ['vtt', 'srt', 'ass']]
 548         subtitles = {l: s_formats(l) for l in ['en', 'fr', 'es']}
 549         auto_captions = {l: s_formats(l, True) for l in ['it', 'pt', 'es']}
 550         info_dict = {
 551             'id': 'test',
 552             'title': 'Test',
 553             'url': 'http://localhost/video.mp4',
 554             'subtitles': subtitles,
 555             'automatic_captions': auto_captions,
 556             'extractor': 'TEST',
 557             'webpage_url': 'http://example.com/watch?v=shenanigans',
 558         }
 559
 560         def get_info(params={}):
 561             params.setdefault('simulate', True)
 562             ydl = YDL(params)
 563             ydl.report_warning = lambda *args, **kargs: None
 564             return ydl.process_video_result(info_dict, download=False)
 565
 566         result = get_info()
 567         self.assertFalse(result.get('requested_subtitles'))
 568         self.assertEqual(result['subtitles'], subtitles)
 569         self.assertEqual(result['automatic_captions'], auto_captions)
 570
 571         result = get_info({'writesubtitles': True})
 572         subs = result['requested_subtitles']
 573         self.assertTrue(subs)
 574         self.assertEqual(set(subs.keys()), {'en'})
 575         self.assertTrue(subs['en'].get('data') is None)
 576         self.assertEqual(subs['en']['ext'], 'ass')
 577
 578         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 579         subs = result['requested_subtitles']
 580         self.assertEqual(subs['en']['ext'], 'srt')
 581
 582         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 583         subs = result['requested_subtitles']
 584         self.assertTrue(subs)
 585         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 586
 587         result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']})
 588         subs = result['requested_subtitles']
 589         self.assertTrue(subs)
 590         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 591
 592         result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']})
 593         subs = result['requested_subtitles']
 594         self.assertTrue(subs)
 595         self.assertEqual(set(subs.keys()), {'fr'})
 596
 597         result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']})
 598         subs = result['requested_subtitles']
 599         self.assertTrue(subs)
 600         self.assertEqual(set(subs.keys()), {'en'})
 601
 602         result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']})
 603         subs = result['requested_subtitles']
 604         self.assertTrue(subs)
 605         self.assertEqual(set(subs.keys()), {'es', 'en'})
 606
 607         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 608         subs = result['requested_subtitles']
 609         self.assertTrue(subs)
 610         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 611         self.assertFalse(subs['es']['_auto'])
 612         self.assertTrue(subs['pt']['_auto'])
 613
 614         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 615         subs = result['requested_subtitles']
 616         self.assertTrue(subs)
 617         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 618         self.assertTrue(subs['es']['_auto'])
 619         self.assertTrue(subs['pt']['_auto'])
 620
 621     def test_add_extra_info(self):
 622         test_dict = {
 623             'extractor': 'Foo',
 624         }
 625         extra_info = {
 626             'extractor': 'Bar',
 627             'playlist': 'funny videos',
 628         }
 629         YDL.add_extra_info(test_dict, extra_info)
 630         self.assertEqual(test_dict['extractor'], 'Foo')
 631         self.assertEqual(test_dict['playlist'], 'funny videos')
 632
 633     outtmpl_info = {
 634         'id': '1234',
 635         'ext': 'mp4',
 636         'width': None,
 637         'height': 1080,
 638         'filesize': 1024,
 639         'title1': '$PATH',
 640         'title2': '%PATH%',
 641         'title3': 'foo/bar\\test',
 642         'title4': 'foo "bar" test',
 643         'title5': 'áéí 𝐀',
 644         'timestamp': 1618488000,
 645         'duration': 100000,
 646         'playlist_index': 1,
 647         'playlist_autonumber': 2,
 648         '__last_playlist_index': 100,
 649         'n_entries': 10,
 650         'formats': [
 651             {'id': 'id 1', 'height': 1080, 'width': 1920},
 652             {'id': 'id 2', 'height': 720},
 653             {'id': 'id 3'}
 654         ]
 655     }
 656
 657     def test_prepare_outtmpl_and_filename(self):
 658         def test(tmpl, expected, *, info=None, **params):
 659             params['outtmpl'] = tmpl
 660             ydl = FakeYDL(params)
 661             ydl._num_downloads = 1
 662             self.assertEqual(ydl.validate_outtmpl(tmpl), None)
 663
 664             out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info)
 665             fname = ydl.prepare_filename(info or self.outtmpl_info)
 666
 667             if not isinstance(expected, (list, tuple)):
 668                 expected = (expected, expected)
 669             for (name, got), expect in zip((('outtmpl', out), ('filename', fname)), expected):
 670                 if callable(expect):
 671                     self.assertTrue(expect(got), f'Wrong {name} from {tmpl}')
 672                 elif expect is not None:
 673                     self.assertEqual(got, expect, f'Wrong {name} from {tmpl}')
 674
 675         # Side-effects
 676         original_infodict = dict(self.outtmpl_info)
 677         test('foo.bar', 'foo.bar')
 678         original_infodict['epoch'] = self.outtmpl_info.get('epoch')
 679         self.assertTrue(isinstance(original_infodict['epoch'], int))
 680         test('%(epoch)d', int_or_none)
 681         self.assertEqual(original_infodict, self.outtmpl_info)
 682
 683         # Auto-generated fields
 684         test('%(id)s.%(ext)s', '1234.mp4')
 685         test('%(duration_string)s', ('27:46:40', '27-46-40'))
 686         test('%(resolution)s', '1080p')
 687         test('%(playlist_index|)s', '001')
 688         test('%(playlist_index&{}!)s', '1!')
 689         test('%(playlist_autonumber)s', '02')
 690         test('%(autonumber)s', '00001')
 691         test('%(autonumber+2)03d', '005', autonumber_start=3)
 692         test('%(autonumber)s', '001', autonumber_size=3)
 693
 694         # Escaping %
 695         test('%', '%')
 696         test('%%', '%')
 697         test('%%%%', '%%')
 698         test('%s', '%s')
 699         test('%%%s', '%%s')
 700         test('%d', '%d')
 701         test('%abc%', '%abc%')
 702         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 703         test('%%%(height)s', '%1080')
 704         test('%(width)06d.%(ext)s', 'NA.mp4')
 705         test('%(width)06d.%%(ext)s', 'NA.%(ext)s')
 706         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 707
 708         # ID sanitization
 709         test('%(id)s', '_abcd', info={'id': '_abcd'})
 710         test('%(some_id)s', '_abcd', info={'some_id': '_abcd'})
 711         test('%(formats.0.id)s', '_abcd', info={'formats': [{'id': '_abcd'}]})
 712         test('%(id)s', '-abcd', info={'id': '-abcd'})
 713         test('%(id)s', '.abcd', info={'id': '.abcd'})
 714         test('%(id)s', 'ab__cd', info={'id': 'ab__cd'})
 715         test('%(id)s', ('ab:cd', 'ab：cd'), info={'id': 'ab:cd'})
 716         test('%(id.0)s', '-', info={'id': '--'})
 717
 718         # Invalid templates
 719         self.assertTrue(isinstance(YoutubeDL.validate_outtmpl('%(title)'), ValueError))
 720         test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder='none')
 721         test('%(..)s', 'NA')
 722         test('%(formats.{id)s', 'NA')
 723
 724         # Entire info_dict
 725         def expect_same_infodict(out):
 726             got_dict = json.loads(out)
 727             for info_field, expected in self.outtmpl_info.items():
 728                 self.assertEqual(got_dict.get(info_field), expected, info_field)
 729             return True
 730
 731         test('%()j', (expect_same_infodict, str))
 732
 733         # NA placeholder
 734         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s'
 735         test(NA_TEST_OUTTMPL, 'NA-NA-def-1234.mp4')
 736         test(NA_TEST_OUTTMPL, 'none-none-def-1234.mp4', outtmpl_na_placeholder='none')
 737         test(NA_TEST_OUTTMPL, '--def-1234.mp4', outtmpl_na_placeholder='')
 738         test('%(non_existent.0)s', 'NA')
 739
 740         # String formatting
 741         FMT_TEST_OUTTMPL = '%%(height)%s.%%(ext)s'
 742         test(FMT_TEST_OUTTMPL % 's', '1080.mp4')
 743         test(FMT_TEST_OUTTMPL % 'd', '1080.mp4')
 744         test(FMT_TEST_OUTTMPL % '6d', '  1080.mp4')
 745         test(FMT_TEST_OUTTMPL % '-6d', '1080  .mp4')
 746         test(FMT_TEST_OUTTMPL % '06d', '001080.mp4')
 747         test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4')
 748         test(FMT_TEST_OUTTMPL % '   06d', ' 01080.mp4')
 749         test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4')
 750         test(FMT_TEST_OUTTMPL % '0   6d', ' 01080.mp4')
 751         test(FMT_TEST_OUTTMPL % '   0   6d', ' 01080.mp4')
 752
 753         # Type casting
 754         test('%(id)d', '1234')
 755         test('%(height)c', '1')
 756         test('%(ext)c', 'm')
 757         test('%(id)d %(id)r', "1234 '1234'")
 758         test('%(id)r %(height)r', "'1234' 1080")
 759         test('%(title5)a %(height)a', (R"'\xe1\xe9\xed \U0001d400' 1080", None))
 760         test('%(ext)s-%(ext|def)d', 'mp4-def')
 761         test('%(width|0)04d', '0')
 762         test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none')
 763
 764         FORMATS = self.outtmpl_info['formats']
 765
 766         # Custom type casting
 767         test('%(formats.:.id)l', 'id 1, id 2, id 3')
 768         test('%(formats.:.id)#l', ('id 1\nid 2\nid 3', 'id 1 id 2 id 3'))
 769         test('%(ext)l', 'mp4')
 770         test('%(formats.:.id) 18l', '  id 1, id 2, id 3')
 771         test('%(formats)j', (json.dumps(FORMATS), None))
 772         test('%(formats)#j', (
 773             json.dumps(FORMATS, indent=4),
 774             json.dumps(FORMATS, indent=4).replace(':', '：').replace('"', "＂").replace('\n', ' ')
 775         ))
 776         test('%(title5).3B', 'á')
 777         test('%(title5)U', 'áéí 𝐀')
 778         test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀')
 779         test('%(title5)+U', 'áéí A')
 780         test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A')
 781         test('%(height)D', '1k')
 782         test('%(filesize)#D', '1Ki')
 783         test('%(height)5.2D', ' 1.08k')
 784         test('%(title4)#S', 'foo_bar_test')
 785         test('%(title4).10S', ('foo ＂bar＂ ', 'foo ＂bar＂' + ('#' if compat_os_name == 'nt' else ' ')))
 786         if compat_os_name == 'nt':
 787             test('%(title4)q', ('"foo \\"bar\\" test"', "＂foo ⧹＂bar⧹＂ test＂"))
 788             test('%(formats.:.id)#q', ('"id 1" "id 2" "id 3"', '＂id 1＂ ＂id 2＂ ＂id 3＂'))
 789             test('%(formats.0.id)#q', ('"id 1"', '＂id 1＂'))
 790         else:
 791             test('%(title4)q', ('\'foo "bar" test\'', '\'foo ＂bar＂ test\''))
 792             test('%(formats.:.id)#q', "'id 1' 'id 2' 'id 3'")
 793             test('%(formats.0.id)#q', "'id 1'")
 794
 795         # Internal formatting
 796         test('%(timestamp-1000>%H-%M-%S)s', '11-43-20')
 797         test('%(title|%)s %(title|%%)s', '% %%')
 798         test('%(id+1-height+3)05d', '00158')
 799         test('%(width+100)05d', 'NA')
 800         test('%(formats.0) 15s', ('% 15s' % FORMATS[0], None))
 801         test('%(formats.0)r', (repr(FORMATS[0]), None))
 802         test('%(height.0)03d', '001')
 803         test('%(-height.0)04d', '-001')
 804         test('%(formats.-1.id)s', FORMATS[-1]['id'])
 805         test('%(formats.0.id.-1)d', FORMATS[0]['id'][-1])
 806         test('%(formats.3)s', 'NA')
 807         test('%(formats.:2:-1)r', repr(FORMATS[:2:-1]))
 808         test('%(formats.0.id.-1+id)f', '1235.000000')
 809         test('%(formats.0.id.-1+formats.1.id.-1)d', '3')
 810         out = json.dumps([{'id': f['id'], 'height.:2': str(f['height'])[:2]}
 811                           if 'height' in f else {'id': f['id']}
 812                           for f in FORMATS])
 813         test('%(formats.:.{id,height.:2})j', (out, None))
 814         test('%(formats.:.{id,height}.id)l', ', '.join(f['id'] for f in FORMATS))
 815         test('%(.{id,title})j', ('{"id": "1234"}', '{＂id＂： ＂1234＂}'))
 816
 817         # Alternates
 818         test('%(title,id)s', '1234')
 819         test('%(width-100,height+20|def)d', '1100')
 820         test('%(width-100,height+width|def)s', 'def')
 821         test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00')
 822
 823         # Replacement
 824         test('%(id&foo)s.bar', 'foo.bar')
 825         test('%(title&foo)s.bar', 'NA.bar')
 826         test('%(title&foo|baz)s.bar', 'baz.bar')
 827         test('%(x,id&foo|baz)s.bar', 'foo.bar')
 828         test('%(x,title&foo|baz)s.bar', 'baz.bar')
 829         test('%(id&a\nb|)s', ('a\nb', 'a b'))
 830         test('%(id&hi {:>10} {}|)s', 'hi       1234 1234')
 831         test(R'%(id&{0} {}|)s', 'NA')
 832         test(R'%(id&{0.1}|)s', 'NA')
 833         test('%(height&{:,d})S', '1,080')
 834
 835         # Laziness
 836         def gen():
 837             yield from range(5)
 838             raise self.assertTrue(False, 'LazyList should not be evaluated till here')
 839         test('%(key.4)s', '4', info={'key': LazyList(gen())})
 840
 841         # Empty filename
 842         test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4')
 843         # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4'))  # fixme
 844         # test('%(foo|)s', ('', '_'))  # fixme
 845
 846         # Environment variable expansion for prepare_filename
 847         os.environ['__yt_dlp_var'] = 'expanded'
 848         envvar = '%__yt_dlp_var%' if compat_os_name == 'nt' else '$__yt_dlp_var'
 849         test(envvar, (envvar, 'expanded'))
 850         if compat_os_name == 'nt':
 851             test('%s%', ('%s%', '%s%'))
 852             os.environ['s'] = 'expanded'
 853             test('%s%', ('%s%', 'expanded'))  # %s% should be expanded before escaping %s
 854             os.environ['(test)s'] = 'expanded'
 855             test('%(test)s%', ('NA%', 'expanded'))  # Environment should take priority over template
 856
 857         # Path expansion and escaping
 858         test('Hello %(title1)s', 'Hello $PATH')
 859         test('Hello %(title2)s', 'Hello %PATH%')
 860         test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test'))
 861         test('folder/%(title3)s', ('folder/foo/bar\\test', 'folder%sfoo⧸bar⧹test' % os.path.sep))
 862
 863     def test_format_note(self):
 864         ydl = YoutubeDL()
 865         self.assertEqual(ydl._format_note({}), '')
 866         assertRegexpMatches(self, ydl._format_note({
 867             'vbr': 10,
 868         }), r'^\s*10k$')
 869         assertRegexpMatches(self, ydl._format_note({
 870             'fps': 30,
 871         }), r'^30fps$')
 872
 873     def test_postprocessors(self):
 874         filename = 'post-processor-testfile.mp4'
 875         audiofile = filename + '.mp3'
 876
 877         class SimplePP(PostProcessor):
 878             def run(self, info):
 879                 with open(audiofile, 'w') as f:
 880                     f.write('EXAMPLE')
 881                 return [info['filepath']], info
 882
 883         def run_pp(params, PP):
 884             with open(filename, 'w') as f:
 885                 f.write('EXAMPLE')
 886             ydl = YoutubeDL(params)
 887             ydl.add_post_processor(PP())
 888             ydl.post_process(filename, {'filepath': filename})
 889
 890         run_pp({'keepvideo': True}, SimplePP)
 891         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 892         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 893         os.unlink(filename)
 894         os.unlink(audiofile)
 895
 896         run_pp({'keepvideo': False}, SimplePP)
 897         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 898         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 899         os.unlink(audiofile)
 900
 901         class ModifierPP(PostProcessor):
 902             def run(self, info):
 903                 with open(info['filepath'], 'w') as f:
 904                     f.write('MODIFIED')
 905                 return [], info
 906
 907         run_pp({'keepvideo': False}, ModifierPP)
 908         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 909         os.unlink(filename)
 910
 911     def test_match_filter(self):
 912         first = {
 913             'id': '1',
 914             'url': TEST_URL,
 915             'title': 'one',
 916             'extractor': 'TEST',
 917             'duration': 30,
 918             'filesize': 10 * 1024,
 919             'playlist_id': '42',
 920             'uploader': "變態妍字幕版 太妍 тест",
 921             'creator': "тест ' 123 ' тест--",
 922             'webpage_url': 'http://example.com/watch?v=shenanigans',
 923         }
 924         second = {
 925             'id': '2',
 926             'url': TEST_URL,
 927             'title': 'two',
 928             'extractor': 'TEST',
 929             'duration': 10,
 930             'description': 'foo',
 931             'filesize': 5 * 1024,
 932             'playlist_id': '43',
 933             'uploader': "тест 123",
 934             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 935         }
 936         videos = [first, second]
 937
 938         def get_videos(filter_=None):
 939             ydl = YDL({'match_filter': filter_, 'simulate': True})
 940             for v in videos:
 941                 ydl.process_ie_result(v, download=True)
 942             return [v['id'] for v in ydl.downloaded_info_dicts]
 943
 944         res = get_videos()
 945         self.assertEqual(res, ['1', '2'])
 946
 947         def f(v, incomplete):
 948             if v['id'] == '1':
 949                 return None
 950             else:
 951                 return 'Video id is not 1'
 952         res = get_videos(f)
 953         self.assertEqual(res, ['1'])
 954
 955         f = match_filter_func('duration < 30')
 956         res = get_videos(f)
 957         self.assertEqual(res, ['2'])
 958
 959         f = match_filter_func('description = foo')
 960         res = get_videos(f)
 961         self.assertEqual(res, ['2'])
 962
 963         f = match_filter_func('description =? foo')
 964         res = get_videos(f)
 965         self.assertEqual(res, ['1', '2'])
 966
 967         f = match_filter_func('filesize > 5KiB')
 968         res = get_videos(f)
 969         self.assertEqual(res, ['1'])
 970
 971         f = match_filter_func('playlist_id = 42')
 972         res = get_videos(f)
 973         self.assertEqual(res, ['1'])
 974
 975         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 976         res = get_videos(f)
 977         self.assertEqual(res, ['1'])
 978
 979         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 980         res = get_videos(f)
 981         self.assertEqual(res, ['2'])
 982
 983         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 984         res = get_videos(f)
 985         self.assertEqual(res, ['1'])
 986
 987         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 988         res = get_videos(f)
 989         self.assertEqual(res, ['1'])
 990
 991         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 992         res = get_videos(f)
 993         self.assertEqual(res, [])
 994
 995     def test_playlist_items_selection(self):
 996         INDICES, PAGE_SIZE = list(range(1, 11)), 3
 997
 998         def entry(i, evaluated):
 999             evaluated.append(i)
1000             return {
1001                 'id': str(i),
1002                 'title': str(i),
1003                 'url': TEST_URL,
1004             }
1005
1006         def pagedlist_entries(evaluated):
1007             def page_func(n):
1008                 start = PAGE_SIZE * n
1009                 for i in INDICES[start: start + PAGE_SIZE]:
1010                     yield entry(i, evaluated)
1011             return OnDemandPagedList(page_func, PAGE_SIZE)
1012
1013         def page_num(i):
1014             return (i + PAGE_SIZE - 1) // PAGE_SIZE
1015
1016         def generator_entries(evaluated):
1017             for i in INDICES:
1018                 yield entry(i, evaluated)
1019
1020         def list_entries(evaluated):
1021             return list(generator_entries(evaluated))
1022
1023         def lazylist_entries(evaluated):
1024             return LazyList(generator_entries(evaluated))
1025
1026         def get_downloaded_info_dicts(params, entries):
1027             ydl = YDL(params)
1028             ydl.process_ie_result({
1029                 '_type': 'playlist',
1030                 'id': 'test',
1031                 'extractor': 'test:playlist',
1032                 'extractor_key': 'test:playlist',
1033                 'webpage_url': 'http://example.com',
1034                 'entries': entries,
1035             })
1036             return ydl.downloaded_info_dicts
1037
1038         def test_selection(params, expected_ids, evaluate_all=False):
1039             expected_ids = list(expected_ids)
1040             if evaluate_all:
1041                 generator_eval = pagedlist_eval = INDICES
1042             elif not expected_ids:
1043                 generator_eval = pagedlist_eval = []
1044             else:
1045                 generator_eval = INDICES[0: max(expected_ids)]
1046                 pagedlist_eval = INDICES[PAGE_SIZE * page_num(min(expected_ids)) - PAGE_SIZE:
1047                                          PAGE_SIZE * page_num(max(expected_ids))]
1048
1049             for name, func, expected_eval in (
1050                 ('list', list_entries, INDICES),
1051                 ('Generator', generator_entries, generator_eval),
1052                 # ('LazyList', lazylist_entries, generator_eval),  # Generator and LazyList follow the exact same code path
1053                 ('PagedList', pagedlist_entries, pagedlist_eval),
1054             ):
1055                 evaluated = []
1056                 entries = func(evaluated)
1057                 results = [(v['playlist_autonumber'] - 1, (int(v['id']), v['playlist_index']))
1058                            for v in get_downloaded_info_dicts(params, entries)]
1059                 self.assertEqual(results, list(enumerate(zip(expected_ids, expected_ids))), f'Entries of {name} for {params}')
1060                 self.assertEqual(sorted(evaluated), expected_eval, f'Evaluation of {name} for {params}')
1061
1062         test_selection({}, INDICES)
1063         test_selection({'playlistend': 20}, INDICES, True)
1064         test_selection({'playlistend': 2}, INDICES[:2])
1065         test_selection({'playliststart': 11}, [], True)
1066         test_selection({'playliststart': 2}, INDICES[1:])
1067         test_selection({'playlist_items': '2-4'}, INDICES[1:4])
1068         test_selection({'playlist_items': '2,4'}, [2, 4])
1069         test_selection({'playlist_items': '20'}, [], True)
1070         test_selection({'playlist_items': '0'}, [])
1071
1072         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
1073         test_selection({'playlist_items': '2-4,3-4,3'}, [2, 3, 4])
1074         test_selection({'playlist_items': '4,2'}, [4, 2])
1075
1076         # Tests for https://github.com/yt-dlp/yt-dlp/issues/720
1077         # https://github.com/yt-dlp/yt-dlp/issues/302
1078         test_selection({'playlistreverse': True}, INDICES[::-1])
1079         test_selection({'playliststart': 2, 'playlistreverse': True}, INDICES[:0:-1])
1080         test_selection({'playlist_items': '2,4', 'playlistreverse': True}, [4, 2])
1081         test_selection({'playlist_items': '4,2'}, [4, 2])
1082
1083         # Tests for --playlist-items start:end:step
1084         test_selection({'playlist_items': ':'}, INDICES, True)
1085         test_selection({'playlist_items': '::1'}, INDICES, True)
1086         test_selection({'playlist_items': '::-1'}, INDICES[::-1], True)
1087         test_selection({'playlist_items': ':6'}, INDICES[:6])
1088         test_selection({'playlist_items': ':-6'}, INDICES[:-5], True)
1089         test_selection({'playlist_items': '-1:6:-2'}, INDICES[:4:-2], True)
1090         test_selection({'playlist_items': '9:-6:-2'}, INDICES[8:3:-2], True)
1091
1092         test_selection({'playlist_items': '1:inf:2'}, INDICES[::2], True)
1093         test_selection({'playlist_items': '-2:inf'}, INDICES[-2:], True)
1094         test_selection({'playlist_items': ':inf:-1'}, [], True)
1095         test_selection({'playlist_items': '0-2:2'}, [2])
1096         test_selection({'playlist_items': '1-:2'}, INDICES[::2], True)
1097         test_selection({'playlist_items': '0--2:2'}, INDICES[1:-1:2], True)
1098
1099         test_selection({'playlist_items': '10::3'}, [10], True)
1100         test_selection({'playlist_items': '-1::3'}, [10], True)
1101         test_selection({'playlist_items': '11::3'}, [], True)
1102         test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True)
1103         test_selection({'playlist_items': '-15::15'}, [], True)
1104
1105     def test_do_not_override_ie_key_in_url_transparent(self):
1106         ydl = YDL()
1107
1108         class Foo1IE(InfoExtractor):
1109             _VALID_URL = r'foo1:'
1110
1111             def _real_extract(self, url):
1112                 return {
1113                     '_type': 'url_transparent',
1114                     'url': 'foo2:',
1115                     'ie_key': 'Foo2',
1116                     'title': 'foo1 title',
1117                     'id': 'foo1_id',
1118                 }
1119
1120         class Foo2IE(InfoExtractor):
1121             _VALID_URL = r'foo2:'
1122
1123             def _real_extract(self, url):
1124                 return {
1125                     '_type': 'url',
1126                     'url': 'foo3:',
1127                     'ie_key': 'Foo3',
1128                 }
1129
1130         class Foo3IE(InfoExtractor):
1131             _VALID_URL = r'foo3:'
1132
1133             def _real_extract(self, url):
1134                 return _make_result([{'url': TEST_URL}], title='foo3 title')
1135
1136         ydl.add_info_extractor(Foo1IE(ydl))
1137         ydl.add_info_extractor(Foo2IE(ydl))
1138         ydl.add_info_extractor(Foo3IE(ydl))
1139         ydl.extract_info('foo1:')
1140         downloaded = ydl.downloaded_info_dicts[0]
1141         self.assertEqual(downloaded['url'], TEST_URL)
1142         self.assertEqual(downloaded['title'], 'foo1 title')
1143         self.assertEqual(downloaded['id'], 'testid')
1144         self.assertEqual(downloaded['extractor'], 'testex')
1145         self.assertEqual(downloaded['extractor_key'], 'TestEx')
1146
1147     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
1148     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
1149
1150         class _YDL(YDL):
1151             def __init__(self, *args, **kwargs):
1152                 super().__init__(*args, **kwargs)
1153
1154             def trouble(self, s, tb=None):
1155                 pass
1156
1157         ydl = _YDL({
1158             'format': 'extra',
1159             'ignoreerrors': True,
1160         })
1161
1162         class VideoIE(InfoExtractor):
1163             _VALID_URL = r'video:(?P<id>\d+)'
1164
1165             def _real_extract(self, url):
1166                 video_id = self._match_id(url)
1167                 formats = [{
1168                     'format_id': 'default',
1169                     'url': 'url:',
1170                 }]
1171                 if video_id == '0':
1172                     raise ExtractorError('foo')
1173                 if video_id == '2':
1174                     formats.append({
1175                         'format_id': 'extra',
1176                         'url': TEST_URL,
1177                     })
1178                 return {
1179                     'id': video_id,
1180                     'title': 'Video %s' % video_id,
1181                     'formats': formats,
1182                 }
1183
1184         class PlaylistIE(InfoExtractor):
1185             _VALID_URL = r'playlist:'
1186
1187             def _entries(self):
1188                 for n in range(3):
1189                     video_id = str(n)
1190                     yield {
1191                         '_type': 'url_transparent',
1192                         'ie_key': VideoIE.ie_key(),
1193                         'id': video_id,
1194                         'url': 'video:%s' % video_id,
1195                         'title': 'Video Transparent %s' % video_id,
1196                     }
1197
1198             def _real_extract(self, url):
1199                 return self.playlist_result(self._entries())
1200
1201         ydl.add_info_extractor(VideoIE(ydl))
1202         ydl.add_info_extractor(PlaylistIE(ydl))
1203         info = ydl.extract_info('playlist:')
1204         entries = info['entries']
1205         self.assertEqual(len(entries), 3)
1206         self.assertTrue(entries[0] is None)
1207         self.assertTrue(entries[1] is None)
1208         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1209         downloaded = ydl.downloaded_info_dicts[0]
1210         entries[2].pop('requested_downloads', None)
1211         self.assertEqual(entries[2], downloaded)
1212         self.assertEqual(downloaded['url'], TEST_URL)
1213         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1214         self.assertEqual(downloaded['id'], '2')
1215         self.assertEqual(downloaded['extractor'], 'Video')
1216         self.assertEqual(downloaded['extractor_key'], 'Video')
1217
1218     def test_header_cookies(self):
1219         from http.cookiejar import Cookie
1220
1221         ydl = FakeYDL()
1222         ydl.report_warning = lambda *_, **__: None
1223
1224         def cookie(name, value, version=None, domain='', path='', secure=False, expires=None):
1225             return Cookie(
1226                 version or 0, name, value, None, False,
1227                 domain, bool(domain), bool(domain), path, bool(path),
1228                 secure, expires, False, None, None, rest={})
1229
1230         _test_url = 'https://yt.dlp/test'
1231
1232         def test(encoded_cookies, cookies, *, headers=False, round_trip=None, error_re=None):
1233             def _test():
1234                 ydl.cookiejar.clear()
1235                 ydl._load_cookies(encoded_cookies, autoscope=headers)
1236                 if headers:
1237                     ydl._apply_header_cookies(_test_url)
1238                 data = {'url': _test_url}
1239                 ydl._calc_headers(data)
1240                 self.assertCountEqual(
1241                     map(vars, ydl.cookiejar), map(vars, cookies),
1242                     'Extracted cookiejar.Cookie is not the same')
1243                 if not headers:
1244                     self.assertEqual(
1245                         data.get('cookies'), round_trip or encoded_cookies,
1246                         'Cookie is not the same as round trip')
1247                 ydl.__dict__['_YoutubeDL__header_cookies'] = []
1248
1249             with self.subTest(msg=encoded_cookies):
1250                 if not error_re:
1251                     _test()
1252                     return
1253                 with self.assertRaisesRegex(Exception, error_re):
1254                     _test()
1255
1256         test('test=value; Domain=.yt.dlp', [cookie('test', 'value', domain='.yt.dlp')])
1257         test('test=value', [cookie('test', 'value')], error_re=r'Unscoped cookies are not allowed')
1258         test('cookie1=value1; Domain=.yt.dlp; Path=/test; cookie2=value2; Domain=.yt.dlp; Path=/', [
1259             cookie('cookie1', 'value1', domain='.yt.dlp', path='/test'),
1260             cookie('cookie2', 'value2', domain='.yt.dlp', path='/')])
1261         test('test=value; Domain=.yt.dlp; Path=/test; Secure; Expires=9999999999', [
1262             cookie('test', 'value', domain='.yt.dlp', path='/test', secure=True, expires=9999999999)])
1263         test('test="value; "; path=/test; domain=.yt.dlp', [
1264             cookie('test', 'value; ', domain='.yt.dlp', path='/test')],
1265             round_trip='test="value\\073 "; Domain=.yt.dlp; Path=/test')
1266         test('name=; Domain=.yt.dlp', [cookie('name', '', domain='.yt.dlp')],
1267              round_trip='name=""; Domain=.yt.dlp')
1268
1269         test('test=value', [cookie('test', 'value', domain='.yt.dlp')], headers=True)
1270         test('cookie1=value; Domain=.yt.dlp; cookie2=value', [], headers=True, error_re=r'Invalid syntax')
1271         ydl.deprecated_feature = ydl.report_error
1272         test('test=value', [], headers=True, error_re=r'Passing cookies as a header is a potential security risk')
1273
1274     def test_infojson_cookies(self):
1275         TEST_FILE = 'test_infojson_cookies.info.json'
1276         TEST_URL = 'https://example.com/example.mp4'
1277         COOKIES = 'a=b; Domain=.example.com; c=d; Domain=.example.com'
1278         COOKIE_HEADER = {'Cookie': 'a=b; c=d'}
1279
1280         ydl = FakeYDL()
1281         ydl.process_info = lambda x: ydl._write_info_json('test', x, TEST_FILE)
1282
1283         def make_info(info_header_cookies=False, fmts_header_cookies=False, cookies_field=False):
1284             fmt = {'url': TEST_URL}
1285             if fmts_header_cookies:
1286                 fmt['http_headers'] = COOKIE_HEADER
1287             if cookies_field:
1288                 fmt['cookies'] = COOKIES
1289             return _make_result([fmt], http_headers=COOKIE_HEADER if info_header_cookies else None)
1290
1291         def test(initial_info, note):
1292             result = {}
1293             result['processed'] = ydl.process_ie_result(initial_info)
1294             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1295                             msg=f'No cookies set in cookiejar after initial process when {note}')
1296             ydl.cookiejar.clear()
1297             with open(TEST_FILE) as infojson:
1298                 result['loaded'] = ydl.sanitize_info(json.load(infojson), True)
1299             result['final'] = ydl.process_ie_result(result['loaded'].copy(), download=False)
1300             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1301                             msg=f'No cookies set in cookiejar after final process when {note}')
1302             ydl.cookiejar.clear()
1303             for key in ('processed', 'loaded', 'final'):
1304                 info = result[key]
1305                 self.assertIsNone(
1306                     traverse_obj(info, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False),
1307                     msg=f'Cookie header not removed in {key} result when {note}')
1308                 self.assertEqual(
1309                     traverse_obj(info, ((None, ('formats', 0)), 'cookies'), get_all=False), COOKIES,
1310                     msg=f'No cookies field found in {key} result when {note}')
1311
1312         test({'url': TEST_URL, 'http_headers': COOKIE_HEADER, 'id': '1', 'title': 'x'}, 'no formats field')
1313         test(make_info(info_header_cookies=True), 'info_dict header cokies')
1314         test(make_info(fmts_header_cookies=True), 'format header cookies')
1315         test(make_info(info_header_cookies=True, fmts_header_cookies=True), 'info_dict and format header cookies')
1316         test(make_info(info_header_cookies=True, fmts_header_cookies=True, cookies_field=True), 'all cookies fields')
1317         test(make_info(cookies_field=True), 'cookies format field')
1318         test({'url': TEST_URL, 'cookies': COOKIES, 'id': '1', 'title': 'x'}, 'info_dict cookies field only')
1319
1320         try_rm(TEST_FILE)
1321
1322     def test_add_headers_cookie(self):
1323         def check_for_cookie_header(result):
1324             return traverse_obj(result, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False)
1325
1326         ydl = FakeYDL({'http_headers': {'Cookie': 'a=b'}})
1327         ydl._apply_header_cookies(_make_result([])['webpage_url'])  # Scope to input webpage URL: .example.com
1328
1329         fmt = {'url': 'https://example.com/video.mp4'}
1330         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1331         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies in result info_dict')
1332         self.assertEqual(result.get('cookies'), 'a=b; Domain=.example.com', msg='No cookies were set in cookies field')
1333         self.assertIn('a=b', ydl.cookiejar.get_cookie_header(fmt['url']), msg='No cookies were set in cookiejar')
1334
1335         fmt = {'url': 'https://wrong.com/video.mp4'}
1336         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1337         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies for wrong domain')
1338         self.assertFalse(result.get('cookies'), msg='Cookies set in cookies field for wrong domain')
1339         self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain')
1340
1341
1342 if __name__ == '__main__':
1343     unittest.main()