test/test_YoutubeDL.py

   1 #!/usr/bin/env python3
   2
   3 # Allow direct execution
   4 import os
   5 import sys
   6 import unittest
   7 from unittest.mock import patch
   8
   9 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  10
  11
  12 import contextlib
  13 import copy
  14 import json
  15
  16 from test.helper import FakeYDL, assertRegexpMatches, try_rm
  17 from yt_dlp import YoutubeDL
  18 from yt_dlp.compat import compat_os_name
  19 from yt_dlp.extractor import YoutubeIE
  20 from yt_dlp.extractor.common import InfoExtractor
  21 from yt_dlp.postprocessor.common import PostProcessor
  22 from yt_dlp.utils import (
  23     ExtractorError,
  24     LazyList,
  25     OnDemandPagedList,
  26     int_or_none,
  27     match_filter_func,
  28 )
  29 from yt_dlp.utils.traversal import traverse_obj
  30
  31 TEST_URL = 'http://localhost/sample.mp4'
  32
  33
  34 class YDL(FakeYDL):
  35     def __init__(self, *args, **kwargs):
  36         super().__init__(*args, **kwargs)
  37         self.downloaded_info_dicts = []
  38         self.msgs = []
  39
  40     def process_info(self, info_dict):
  41         self.downloaded_info_dicts.append(info_dict.copy())
  42
  43     def to_screen(self, msg, *args, **kwargs):
  44         self.msgs.append(msg)
  45
  46     def dl(self, *args, **kwargs):
  47         assert False, 'Downloader must not be invoked for test_YoutubeDL'
  48
  49
  50 def _make_result(formats, **kwargs):
  51     res = {
  52         'formats': formats,
  53         'id': 'testid',
  54         'title': 'testttitle',
  55         'extractor': 'testex',
  56         'extractor_key': 'TestEx',
  57         'webpage_url': 'http://example.com/watch?v=shenanigans',
  58     }
  59     res.update(**kwargs)
  60     return res
  61
  62
  63 class TestFormatSelection(unittest.TestCase):
  64     def test_prefer_free_formats(self):
  65         # Same resolution => download webm
  66         ydl = YDL()
  67         ydl.params['prefer_free_formats'] = True
  68         formats = [
  69             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  70             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  71         ]
  72         info_dict = _make_result(formats)
  73         ydl.sort_formats(info_dict)
  74         ydl.process_ie_result(info_dict)
  75         downloaded = ydl.downloaded_info_dicts[0]
  76         self.assertEqual(downloaded['ext'], 'webm')
  77
  78         # Different resolution => download best quality (mp4)
  79         ydl = YDL()
  80         ydl.params['prefer_free_formats'] = True
  81         formats = [
  82             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  83             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  84         ]
  85         info_dict['formats'] = formats
  86         ydl.sort_formats(info_dict)
  87         ydl.process_ie_result(info_dict)
  88         downloaded = ydl.downloaded_info_dicts[0]
  89         self.assertEqual(downloaded['ext'], 'mp4')
  90
  91         # No prefer_free_formats => prefer mp4 and webm
  92         ydl = YDL()
  93         ydl.params['prefer_free_formats'] = False
  94         formats = [
  95             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  96             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  97             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  98         ]
  99         info_dict['formats'] = formats
 100         ydl.sort_formats(info_dict)
 101         ydl.process_ie_result(info_dict)
 102         downloaded = ydl.downloaded_info_dicts[0]
 103         self.assertEqual(downloaded['ext'], 'mp4')
 104
 105         ydl = YDL()
 106         ydl.params['prefer_free_formats'] = False
 107         formats = [
 108             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
 109             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 110         ]
 111         info_dict['formats'] = formats
 112         ydl.sort_formats(info_dict)
 113         ydl.process_ie_result(info_dict)
 114         downloaded = ydl.downloaded_info_dicts[0]
 115         self.assertEqual(downloaded['ext'], 'webm')
 116
 117     def test_format_selection(self):
 118         formats = [
 119             {'format_id': '35', 'ext': 'mp4', 'preference': 0, 'url': TEST_URL},
 120             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 121             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 122             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 123             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 124         ]
 125         info_dict = _make_result(formats)
 126
 127         def test(inp, *expected, multi=False):
 128             ydl = YDL({
 129                 'format': inp,
 130                 'allow_multiple_video_streams': multi,
 131                 'allow_multiple_audio_streams': multi,
 132             })
 133             ydl.process_ie_result(info_dict.copy())
 134             downloaded = [x['format_id'] for x in ydl.downloaded_info_dicts]
 135             self.assertEqual(downloaded, list(expected))
 136
 137         test('20/47', '47')
 138         test('20/71/worst', '35')
 139         test(None, '2')
 140         test('webm/mp4', '47')
 141         test('3gp/40/mp4', '35')
 142         test('example-with-dashes', 'example-with-dashes')
 143         test('all', '2', '47', '45', 'example-with-dashes', '35')
 144         test('mergeall', '2+47+45+example-with-dashes+35', multi=True)
 145         # See: https://github.com/yt-dlp/yt-dlp/pulls/8797
 146         test('7_a/worst', '35')
 147
 148     def test_format_selection_audio(self):
 149         formats = [
 150             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 151             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 152             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 153             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 154         ]
 155         info_dict = _make_result(formats)
 156
 157         ydl = YDL({'format': 'bestaudio'})
 158         ydl.process_ie_result(info_dict.copy())
 159         downloaded = ydl.downloaded_info_dicts[0]
 160         self.assertEqual(downloaded['format_id'], 'audio-high')
 161
 162         ydl = YDL({'format': 'worstaudio'})
 163         ydl.process_ie_result(info_dict.copy())
 164         downloaded = ydl.downloaded_info_dicts[0]
 165         self.assertEqual(downloaded['format_id'], 'audio-low')
 166
 167         formats = [
 168             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 169             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 170         ]
 171         info_dict = _make_result(formats)
 172
 173         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 174         ydl.process_ie_result(info_dict.copy())
 175         downloaded = ydl.downloaded_info_dicts[0]
 176         self.assertEqual(downloaded['format_id'], 'vid-high')
 177
 178     def test_format_selection_audio_exts(self):
 179         formats = [
 180             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 181             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 182             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 184             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 185         ]
 186
 187         info_dict = _make_result(formats)
 188         ydl = YDL({'format': 'best', 'format_sort': ['abr', 'ext']})
 189         ydl.sort_formats(info_dict)
 190         ydl.process_ie_result(copy.deepcopy(info_dict))
 191         downloaded = ydl.downloaded_info_dicts[0]
 192         self.assertEqual(downloaded['format_id'], 'aac-64')
 193
 194         ydl = YDL({'format': 'mp3'})
 195         ydl.sort_formats(info_dict)
 196         ydl.process_ie_result(copy.deepcopy(info_dict))
 197         downloaded = ydl.downloaded_info_dicts[0]
 198         self.assertEqual(downloaded['format_id'], 'mp3-64')
 199
 200         ydl = YDL({'prefer_free_formats': True, 'format_sort': ['abr', 'ext']})
 201         ydl.sort_formats(info_dict)
 202         ydl.process_ie_result(copy.deepcopy(info_dict))
 203         downloaded = ydl.downloaded_info_dicts[0]
 204         self.assertEqual(downloaded['format_id'], 'ogg-64')
 205
 206     def test_format_selection_video(self):
 207         formats = [
 208             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 209             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 210             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 211         ]
 212         info_dict = _make_result(formats)
 213
 214         ydl = YDL({'format': 'bestvideo'})
 215         ydl.process_ie_result(info_dict.copy())
 216         downloaded = ydl.downloaded_info_dicts[0]
 217         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 218
 219         ydl = YDL({'format': 'worstvideo'})
 220         ydl.process_ie_result(info_dict.copy())
 221         downloaded = ydl.downloaded_info_dicts[0]
 222         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 223
 224         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 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         formats = [
 230             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 231         ]
 232         info_dict = _make_result(formats)
 233
 234         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 235         ydl.process_ie_result(info_dict.copy())
 236         downloaded = ydl.downloaded_info_dicts[0]
 237         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 238
 239     def test_format_selection_string_ops(self):
 240         formats = [
 241             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 242             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 243         ]
 244         info_dict = _make_result(formats)
 245
 246         # equals (=)
 247         ydl = YDL({'format': '[format_id=abc-cba]'})
 248         ydl.process_ie_result(info_dict.copy())
 249         downloaded = ydl.downloaded_info_dicts[0]
 250         self.assertEqual(downloaded['format_id'], 'abc-cba')
 251
 252         # does not equal (!=)
 253         ydl = YDL({'format': '[format_id!=abc-cba]'})
 254         ydl.process_ie_result(info_dict.copy())
 255         downloaded = ydl.downloaded_info_dicts[0]
 256         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 257
 258         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 259         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 260
 261         # starts with (^=)
 262         ydl = YDL({'format': '[format_id^=abc]'})
 263         ydl.process_ie_result(info_dict.copy())
 264         downloaded = ydl.downloaded_info_dicts[0]
 265         self.assertEqual(downloaded['format_id'], 'abc-cba')
 266
 267         # does not start with (!^=)
 268         ydl = YDL({'format': '[format_id!^=abc]'})
 269         ydl.process_ie_result(info_dict.copy())
 270         downloaded = ydl.downloaded_info_dicts[0]
 271         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 272
 273         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 274         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 275
 276         # ends with ($=)
 277         ydl = YDL({'format': '[format_id$=cba]'})
 278         ydl.process_ie_result(info_dict.copy())
 279         downloaded = ydl.downloaded_info_dicts[0]
 280         self.assertEqual(downloaded['format_id'], 'abc-cba')
 281
 282         # does not end with (!$=)
 283         ydl = YDL({'format': '[format_id!$=cba]'})
 284         ydl.process_ie_result(info_dict.copy())
 285         downloaded = ydl.downloaded_info_dicts[0]
 286         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 287
 288         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 289         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 290
 291         # contains (*=)
 292         ydl = YDL({'format': '[format_id*=bc-cb]'})
 293         ydl.process_ie_result(info_dict.copy())
 294         downloaded = ydl.downloaded_info_dicts[0]
 295         self.assertEqual(downloaded['format_id'], 'abc-cba')
 296
 297         # does not contain (!*=)
 298         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 299         ydl.process_ie_result(info_dict.copy())
 300         downloaded = ydl.downloaded_info_dicts[0]
 301         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 302
 303         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 304         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 305
 306         ydl = YDL({'format': '[format_id!*=-]'})
 307         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 308
 309     def test_youtube_format_selection(self):
 310         # FIXME: Rewrite in accordance with the new format sorting options
 311         return
 312
 313         order = [
 314             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 315             # Apple HTTP Live Streaming
 316             '96', '95', '94', '93', '92', '132', '151',
 317             # 3D
 318             '85', '84', '102', '83', '101', '82', '100',
 319             # Dash video
 320             '137', '248', '136', '247', '135', '246',
 321             '245', '244', '134', '243', '133', '242', '160',
 322             # Dash audio
 323             '141', '172', '140', '171', '139',
 324         ]
 325
 326         def format_info(f_id):
 327             info = YoutubeIE._formats[f_id].copy()
 328
 329             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 330             # and 'vcodec', while in tests such information is incomplete since
 331             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 332             # test_YoutubeDL.test_youtube_format_selection is broken without
 333             # this fix
 334             if 'acodec' in info and 'vcodec' not in info:
 335                 info['vcodec'] = 'none'
 336             elif 'vcodec' in info and 'acodec' not in info:
 337                 info['acodec'] = 'none'
 338
 339             info['format_id'] = f_id
 340             info['url'] = 'url:' + f_id
 341             return info
 342         formats_order = [format_info(f_id) for f_id in order]
 343
 344         info_dict = _make_result(list(formats_order), extractor='youtube')
 345         ydl = YDL({'format': 'bestvideo+bestaudio'})
 346         ydl.sort_formats(info_dict)
 347         ydl.process_ie_result(info_dict)
 348         downloaded = ydl.downloaded_info_dicts[0]
 349         self.assertEqual(downloaded['format_id'], '248+172')
 350         self.assertEqual(downloaded['ext'], 'mp4')
 351
 352         info_dict = _make_result(list(formats_order), extractor='youtube')
 353         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 354         ydl.sort_formats(info_dict)
 355         ydl.process_ie_result(info_dict)
 356         downloaded = ydl.downloaded_info_dicts[0]
 357         self.assertEqual(downloaded['format_id'], '38')
 358
 359         info_dict = _make_result(list(formats_order), extractor='youtube')
 360         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 361         ydl.sort_formats(info_dict)
 362         ydl.process_ie_result(info_dict)
 363         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 364         self.assertEqual(downloaded_ids, ['137', '141'])
 365
 366         info_dict = _make_result(list(formats_order), extractor='youtube')
 367         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 368         ydl.sort_formats(info_dict)
 369         ydl.process_ie_result(info_dict)
 370         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 371         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 372
 373         info_dict = _make_result(list(formats_order), extractor='youtube')
 374         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 375         ydl.sort_formats(info_dict)
 376         ydl.process_ie_result(info_dict)
 377         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 378         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 379
 380         info_dict = _make_result(list(formats_order), extractor='youtube')
 381         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 382         ydl.sort_formats(info_dict)
 383         ydl.process_ie_result(info_dict)
 384         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 385         self.assertEqual(downloaded_ids, ['248+141'])
 386
 387         for f1, f2 in zip(formats_order, formats_order[1:]):
 388             info_dict = _make_result([f1, f2], extractor='youtube')
 389             ydl = YDL({'format': 'best/bestvideo'})
 390             ydl.sort_formats(info_dict)
 391             ydl.process_ie_result(info_dict)
 392             downloaded = ydl.downloaded_info_dicts[0]
 393             self.assertEqual(downloaded['format_id'], f1['format_id'])
 394
 395             info_dict = _make_result([f2, f1], extractor='youtube')
 396             ydl = YDL({'format': 'best/bestvideo'})
 397             ydl.sort_formats(info_dict)
 398             ydl.process_ie_result(info_dict)
 399             downloaded = ydl.downloaded_info_dicts[0]
 400             self.assertEqual(downloaded['format_id'], f1['format_id'])
 401
 402     def test_audio_only_extractor_format_selection(self):
 403         # For extractors with incomplete formats (all formats are audio-only or
 404         # video-only) best and worst should fallback to corresponding best/worst
 405         # video-only or audio-only formats (as per
 406         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 407         formats = [
 408             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 409             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 410         ]
 411         info_dict = _make_result(formats)
 412
 413         ydl = YDL({'format': 'best'})
 414         ydl.process_ie_result(info_dict.copy())
 415         downloaded = ydl.downloaded_info_dicts[0]
 416         self.assertEqual(downloaded['format_id'], 'high')
 417
 418         ydl = YDL({'format': 'worst'})
 419         ydl.process_ie_result(info_dict.copy())
 420         downloaded = ydl.downloaded_info_dicts[0]
 421         self.assertEqual(downloaded['format_id'], 'low')
 422
 423     def test_format_not_available(self):
 424         formats = [
 425             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 426             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 427         ]
 428         info_dict = _make_result(formats)
 429
 430         # This must fail since complete video-audio format does not match filter
 431         # and extractor does not provide incomplete only formats (i.e. only
 432         # video-only or audio-only).
 433         ydl = YDL({'format': 'best[height>360]'})
 434         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 435
 436     def test_format_selection_issue_10083(self):
 437         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 438         formats = [
 439             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 440             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 441             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 442         ]
 443         info_dict = _make_result(formats)
 444
 445         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 446         ydl.process_ie_result(info_dict.copy())
 447         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 448
 449     def test_invalid_format_specs(self):
 450         def assert_syntax_error(format_spec):
 451             self.assertRaises(SyntaxError, YDL, {'format': format_spec})
 452
 453         assert_syntax_error('bestvideo,,best')
 454         assert_syntax_error('+bestaudio')
 455         assert_syntax_error('bestvideo+')
 456         assert_syntax_error('/')
 457         assert_syntax_error('[720<height]')
 458
 459     def test_format_filtering(self):
 460         formats = [
 461             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 462             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 463             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 464             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 465             {'format_id': 'E', 'filesize': 3000},
 466             {'format_id': 'F'},
 467             {'format_id': 'G', 'filesize': 1000000},
 468         ]
 469         for f in formats:
 470             f['url'] = 'http://_/'
 471             f['ext'] = 'unknown'
 472         info_dict = _make_result(formats, _format_sort_fields=('id', ))
 473
 474         ydl = YDL({'format': 'best[filesize<3000]'})
 475         ydl.process_ie_result(info_dict)
 476         downloaded = ydl.downloaded_info_dicts[0]
 477         self.assertEqual(downloaded['format_id'], 'D')
 478
 479         ydl = YDL({'format': 'best[filesize<=3000]'})
 480         ydl.process_ie_result(info_dict)
 481         downloaded = ydl.downloaded_info_dicts[0]
 482         self.assertEqual(downloaded['format_id'], 'E')
 483
 484         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 485         ydl.process_ie_result(info_dict)
 486         downloaded = ydl.downloaded_info_dicts[0]
 487         self.assertEqual(downloaded['format_id'], 'F')
 488
 489         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 490         ydl.process_ie_result(info_dict)
 491         downloaded = ydl.downloaded_info_dicts[0]
 492         self.assertEqual(downloaded['format_id'], 'B')
 493
 494         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 495         ydl.process_ie_result(info_dict)
 496         downloaded = ydl.downloaded_info_dicts[0]
 497         self.assertEqual(downloaded['format_id'], 'C')
 498
 499         ydl = YDL({'format': '[filesize>?1]'})
 500         ydl.process_ie_result(info_dict)
 501         downloaded = ydl.downloaded_info_dicts[0]
 502         self.assertEqual(downloaded['format_id'], 'G')
 503
 504         ydl = YDL({'format': '[filesize<1M]'})
 505         ydl.process_ie_result(info_dict)
 506         downloaded = ydl.downloaded_info_dicts[0]
 507         self.assertEqual(downloaded['format_id'], 'E')
 508
 509         ydl = YDL({'format': '[filesize<1MiB]'})
 510         ydl.process_ie_result(info_dict)
 511         downloaded = ydl.downloaded_info_dicts[0]
 512         self.assertEqual(downloaded['format_id'], 'G')
 513
 514         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 515         ydl.process_ie_result(info_dict)
 516         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 517         self.assertEqual(downloaded_ids, ['D', 'C', 'B'])
 518
 519         ydl = YDL({'format': 'best[height<40]'})
 520         with contextlib.suppress(ExtractorError):
 521             ydl.process_ie_result(info_dict)
 522         self.assertEqual(ydl.downloaded_info_dicts, [])
 523
 524     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.available', False)
 525     def test_default_format_spec_without_ffmpeg(self):
 526         ydl = YDL({})
 527         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 528
 529         ydl = YDL({'simulate': True})
 530         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 531
 532         ydl = YDL({})
 533         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 534
 535         ydl = YDL({'simulate': True})
 536         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 537
 538         ydl = YDL({'outtmpl': '-'})
 539         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 540
 541         ydl = YDL({})
 542         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 543         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 544
 545     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.available', True)
 546     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.can_merge', lambda _: True)
 547     def test_default_format_spec_with_ffmpeg(self):
 548         ydl = YDL({})
 549         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 550
 551         ydl = YDL({'simulate': True})
 552         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 553
 554         ydl = YDL({})
 555         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 556
 557         ydl = YDL({'simulate': True})
 558         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 559
 560         ydl = YDL({'outtmpl': '-'})
 561         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 562
 563         ydl = YDL({})
 564         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 565         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 566
 567
 568 class TestYoutubeDL(unittest.TestCase):
 569     def test_subtitles(self):
 570         def s_formats(lang, autocaption=False):
 571             return [{
 572                 'ext': ext,
 573                 'url': f'http://localhost/video.{lang}.{ext}',
 574                 '_auto': autocaption,
 575             } for ext in ['vtt', 'srt', 'ass']]
 576         subtitles = {l: s_formats(l) for l in ['en', 'fr', 'es']}
 577         auto_captions = {l: s_formats(l, True) for l in ['it', 'pt', 'es']}
 578         info_dict = {
 579             'id': 'test',
 580             'title': 'Test',
 581             'url': 'http://localhost/video.mp4',
 582             'subtitles': subtitles,
 583             'automatic_captions': auto_captions,
 584             'extractor': 'TEST',
 585             'webpage_url': 'http://example.com/watch?v=shenanigans',
 586         }
 587
 588         def get_info(params={}):
 589             params.setdefault('simulate', True)
 590             ydl = YDL(params)
 591             ydl.report_warning = lambda *args, **kargs: None
 592             return ydl.process_video_result(info_dict, download=False)
 593
 594         result = get_info()
 595         self.assertFalse(result.get('requested_subtitles'))
 596         self.assertEqual(result['subtitles'], subtitles)
 597         self.assertEqual(result['automatic_captions'], auto_captions)
 598
 599         result = get_info({'writesubtitles': True})
 600         subs = result['requested_subtitles']
 601         self.assertTrue(subs)
 602         self.assertEqual(set(subs.keys()), {'en'})
 603         self.assertTrue(subs['en'].get('data') is None)
 604         self.assertEqual(subs['en']['ext'], 'ass')
 605
 606         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 607         subs = result['requested_subtitles']
 608         self.assertEqual(subs['en']['ext'], 'srt')
 609
 610         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 611         subs = result['requested_subtitles']
 612         self.assertTrue(subs)
 613         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 614
 615         result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']})
 616         subs = result['requested_subtitles']
 617         self.assertTrue(subs)
 618         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 619
 620         result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']})
 621         subs = result['requested_subtitles']
 622         self.assertTrue(subs)
 623         self.assertEqual(set(subs.keys()), {'fr'})
 624
 625         result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']})
 626         subs = result['requested_subtitles']
 627         self.assertTrue(subs)
 628         self.assertEqual(set(subs.keys()), {'en'})
 629
 630         result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']})
 631         subs = result['requested_subtitles']
 632         self.assertTrue(subs)
 633         self.assertEqual(set(subs.keys()), {'es', 'en'})
 634
 635         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 636         subs = result['requested_subtitles']
 637         self.assertTrue(subs)
 638         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 639         self.assertFalse(subs['es']['_auto'])
 640         self.assertTrue(subs['pt']['_auto'])
 641
 642         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 643         subs = result['requested_subtitles']
 644         self.assertTrue(subs)
 645         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 646         self.assertTrue(subs['es']['_auto'])
 647         self.assertTrue(subs['pt']['_auto'])
 648
 649     def test_add_extra_info(self):
 650         test_dict = {
 651             'extractor': 'Foo',
 652         }
 653         extra_info = {
 654             'extractor': 'Bar',
 655             'playlist': 'funny videos',
 656         }
 657         YDL.add_extra_info(test_dict, extra_info)
 658         self.assertEqual(test_dict['extractor'], 'Foo')
 659         self.assertEqual(test_dict['playlist'], 'funny videos')
 660
 661     outtmpl_info = {
 662         'id': '1234',
 663         'ext': 'mp4',
 664         'width': None,
 665         'height': 1080,
 666         'filesize': 1024,
 667         'title1': '$PATH',
 668         'title2': '%PATH%',
 669         'title3': 'foo/bar\\test',
 670         'title4': 'foo "bar" test',
 671         'title5': 'áéí 𝐀',
 672         'timestamp': 1618488000,
 673         'duration': 100000,
 674         'playlist_index': 1,
 675         'playlist_autonumber': 2,
 676         '__last_playlist_index': 100,
 677         'n_entries': 10,
 678         'formats': [
 679             {'id': 'id 1', 'height': 1080, 'width': 1920},
 680             {'id': 'id 2', 'height': 720},
 681             {'id': 'id 3'},
 682         ],
 683     }
 684
 685     def test_prepare_outtmpl_and_filename(self):
 686         def test(tmpl, expected, *, info=None, **params):
 687             params['outtmpl'] = tmpl
 688             ydl = FakeYDL(params)
 689             ydl._num_downloads = 1
 690             self.assertEqual(ydl.validate_outtmpl(tmpl), None)
 691
 692             out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info)
 693             fname = ydl.prepare_filename(info or self.outtmpl_info)
 694
 695             if not isinstance(expected, (list, tuple)):
 696                 expected = (expected, expected)
 697             for (name, got), expect in zip((('outtmpl', out), ('filename', fname)), expected):
 698                 if callable(expect):
 699                     self.assertTrue(expect(got), f'Wrong {name} from {tmpl}')
 700                 elif expect is not None:
 701                     self.assertEqual(got, expect, f'Wrong {name} from {tmpl}')
 702
 703         # Side-effects
 704         original_infodict = dict(self.outtmpl_info)
 705         test('foo.bar', 'foo.bar')
 706         original_infodict['epoch'] = self.outtmpl_info.get('epoch')
 707         self.assertTrue(isinstance(original_infodict['epoch'], int))
 708         test('%(epoch)d', int_or_none)
 709         self.assertEqual(original_infodict, self.outtmpl_info)
 710
 711         # Auto-generated fields
 712         test('%(id)s.%(ext)s', '1234.mp4')
 713         test('%(duration_string)s', ('27:46:40', '27-46-40'))
 714         test('%(resolution)s', '1080p')
 715         test('%(playlist_index|)s', '001')
 716         test('%(playlist_index&{}!)s', '1!')
 717         test('%(playlist_autonumber)s', '02')
 718         test('%(autonumber)s', '00001')
 719         test('%(autonumber+2)03d', '005', autonumber_start=3)
 720         test('%(autonumber)s', '001', autonumber_size=3)
 721
 722         # Escaping %
 723         test('%', '%')
 724         test('%%', '%')
 725         test('%%%%', '%%')
 726         test('%s', '%s')
 727         test('%%%s', '%%s')
 728         test('%d', '%d')
 729         test('%abc%', '%abc%')
 730         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 731         test('%%%(height)s', '%1080')
 732         test('%(width)06d.%(ext)s', 'NA.mp4')
 733         test('%(width)06d.%%(ext)s', 'NA.%(ext)s')
 734         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 735
 736         # ID sanitization
 737         test('%(id)s', '_abcd', info={'id': '_abcd'})
 738         test('%(some_id)s', '_abcd', info={'some_id': '_abcd'})
 739         test('%(formats.0.id)s', '_abcd', info={'formats': [{'id': '_abcd'}]})
 740         test('%(id)s', '-abcd', info={'id': '-abcd'})
 741         test('%(id)s', '.abcd', info={'id': '.abcd'})
 742         test('%(id)s', 'ab__cd', info={'id': 'ab__cd'})
 743         test('%(id)s', ('ab:cd', 'ab：cd'), info={'id': 'ab:cd'})
 744         test('%(id.0)s', '-', info={'id': '--'})
 745
 746         # Invalid templates
 747         self.assertTrue(isinstance(YoutubeDL.validate_outtmpl('%(title)'), ValueError))
 748         test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder='none')
 749         test('%(..)s', 'NA')
 750         test('%(formats.{id)s', 'NA')
 751
 752         # Entire info_dict
 753         def expect_same_infodict(out):
 754             got_dict = json.loads(out)
 755             for info_field, expected in self.outtmpl_info.items():
 756                 self.assertEqual(got_dict.get(info_field), expected, info_field)
 757             return True
 758
 759         test('%()j', (expect_same_infodict, None))
 760
 761         # NA placeholder
 762         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s'
 763         test(NA_TEST_OUTTMPL, 'NA-NA-def-1234.mp4')
 764         test(NA_TEST_OUTTMPL, 'none-none-def-1234.mp4', outtmpl_na_placeholder='none')
 765         test(NA_TEST_OUTTMPL, '--def-1234.mp4', outtmpl_na_placeholder='')
 766         test('%(non_existent.0)s', 'NA')
 767
 768         # String formatting
 769         FMT_TEST_OUTTMPL = '%%(height)%s.%%(ext)s'
 770         test(FMT_TEST_OUTTMPL % 's', '1080.mp4')
 771         test(FMT_TEST_OUTTMPL % 'd', '1080.mp4')
 772         test(FMT_TEST_OUTTMPL % '6d', '  1080.mp4')
 773         test(FMT_TEST_OUTTMPL % '-6d', '1080  .mp4')
 774         test(FMT_TEST_OUTTMPL % '06d', '001080.mp4')
 775         test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4')
 776         test(FMT_TEST_OUTTMPL % '   06d', ' 01080.mp4')
 777         test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4')
 778         test(FMT_TEST_OUTTMPL % '0   6d', ' 01080.mp4')
 779         test(FMT_TEST_OUTTMPL % '   0   6d', ' 01080.mp4')
 780
 781         # Type casting
 782         test('%(id)d', '1234')
 783         test('%(height)c', '1')
 784         test('%(ext)c', 'm')
 785         test('%(id)d %(id)r', "1234 '1234'")
 786         test('%(id)r %(height)r', "'1234' 1080")
 787         test('%(title5)a %(height)a', (R"'\xe1\xe9\xed \U0001d400' 1080", None))
 788         test('%(ext)s-%(ext|def)d', 'mp4-def')
 789         test('%(width|0)04d', '0')
 790         test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none')
 791
 792         FORMATS = self.outtmpl_info['formats']
 793
 794         # Custom type casting
 795         test('%(formats.:.id)l', 'id 1, id 2, id 3')
 796         test('%(formats.:.id)#l', ('id 1\nid 2\nid 3', 'id 1 id 2 id 3'))
 797         test('%(ext)l', 'mp4')
 798         test('%(formats.:.id) 18l', '  id 1, id 2, id 3')
 799         test('%(formats)j', (json.dumps(FORMATS), None))
 800         test('%(formats)#j', (
 801             json.dumps(FORMATS, indent=4),
 802             json.dumps(FORMATS, indent=4).replace(':', '：').replace('"', '＂').replace('\n', ' '),
 803         ))
 804         test('%(title5).3B', 'á')
 805         test('%(title5)U', 'áéí 𝐀')
 806         test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀')
 807         test('%(title5)+U', 'áéí A')
 808         test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A')
 809         test('%(height)D', '1k')
 810         test('%(filesize)#D', '1Ki')
 811         test('%(height)5.2D', ' 1.08k')
 812         test('%(title4)#S', 'foo_bar_test')
 813         test('%(title4).10S', ('foo ＂bar＂ ', 'foo ＂bar＂' + ('#' if compat_os_name == 'nt' else ' ')))
 814         if compat_os_name == 'nt':
 815             test('%(title4)q', ('"foo ""bar"" test"', None))
 816             test('%(formats.:.id)#q', ('"id 1" "id 2" "id 3"', None))
 817             test('%(formats.0.id)#q', ('"id 1"', None))
 818         else:
 819             test('%(title4)q', ('\'foo "bar" test\'', '\'foo ＂bar＂ test\''))
 820             test('%(formats.:.id)#q', "'id 1' 'id 2' 'id 3'")
 821             test('%(formats.0.id)#q', "'id 1'")
 822
 823         # Internal formatting
 824         test('%(timestamp-1000>%H-%M-%S)s', '11-43-20')
 825         test('%(title|%)s %(title|%%)s', '% %%')
 826         test('%(id+1-height+3)05d', '00158')
 827         test('%(width+100)05d', 'NA')
 828         test('%(filesize*8)d', '8192')
 829         test('%(formats.0) 15s', ('% 15s' % FORMATS[0], None))
 830         test('%(formats.0)r', (repr(FORMATS[0]), None))
 831         test('%(height.0)03d', '001')
 832         test('%(-height.0)04d', '-001')
 833         test('%(formats.-1.id)s', FORMATS[-1]['id'])
 834         test('%(formats.0.id.-1)d', FORMATS[0]['id'][-1])
 835         test('%(formats.3)s', 'NA')
 836         test('%(formats.:2:-1)r', repr(FORMATS[:2:-1]))
 837         test('%(formats.0.id.-1+id)f', '1235.000000')
 838         test('%(formats.0.id.-1+formats.1.id.-1)d', '3')
 839         out = json.dumps([{'id': f['id'], 'height.:2': str(f['height'])[:2]}
 840                           if 'height' in f else {'id': f['id']}
 841                           for f in FORMATS])
 842         test('%(formats.:.{id,height.:2})j', (out, None))
 843         test('%(formats.:.{id,height}.id)l', ', '.join(f['id'] for f in FORMATS))
 844         test('%(.{id,title})j', ('{"id": "1234"}', '{＂id＂： ＂1234＂}'))
 845
 846         # Alternates
 847         test('%(title,id)s', '1234')
 848         test('%(width-100,height+20|def)d', '1100')
 849         test('%(width-100,height+width|def)s', 'def')
 850         test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00')
 851
 852         # Replacement
 853         test('%(id&foo)s.bar', 'foo.bar')
 854         test('%(title&foo)s.bar', 'NA.bar')
 855         test('%(title&foo|baz)s.bar', 'baz.bar')
 856         test('%(x,id&foo|baz)s.bar', 'foo.bar')
 857         test('%(x,title&foo|baz)s.bar', 'baz.bar')
 858         test('%(id&a\nb|)s', ('a\nb', 'a b'))
 859         test('%(id&hi {:>10} {}|)s', 'hi       1234 1234')
 860         test(R'%(id&{0} {}|)s', 'NA')
 861         test(R'%(id&{0.1}|)s', 'NA')
 862         test('%(height&{:,d})S', '1,080')
 863
 864         # Laziness
 865         def gen():
 866             yield from range(5)
 867             raise self.assertTrue(False, 'LazyList should not be evaluated till here')
 868         test('%(key.4)s', '4', info={'key': LazyList(gen())})
 869
 870         # Empty filename
 871         test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4')
 872         # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4'))  # FIXME: ?
 873         # test('%(foo|)s', ('', '_'))  # FIXME: ?
 874
 875         # Environment variable expansion for prepare_filename
 876         os.environ['__yt_dlp_var'] = 'expanded'
 877         envvar = '%__yt_dlp_var%' if compat_os_name == 'nt' else '$__yt_dlp_var'
 878         test(envvar, (envvar, 'expanded'))
 879         if compat_os_name == 'nt':
 880             test('%s%', ('%s%', '%s%'))
 881             os.environ['s'] = 'expanded'
 882             test('%s%', ('%s%', 'expanded'))  # %s% should be expanded before escaping %s
 883             os.environ['(test)s'] = 'expanded'
 884             test('%(test)s%', ('NA%', 'expanded'))  # Environment should take priority over template
 885
 886         # Path expansion and escaping
 887         test('Hello %(title1)s', 'Hello $PATH')
 888         test('Hello %(title2)s', 'Hello %PATH%')
 889         test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test'))
 890         test('folder/%(title3)s', ('folder/foo/bar\\test', f'folder{os.path.sep}foo⧸bar⧹test'))
 891
 892     def test_format_note(self):
 893         ydl = YoutubeDL()
 894         self.assertEqual(ydl._format_note({}), '')
 895         assertRegexpMatches(self, ydl._format_note({
 896             'vbr': 10,
 897         }), r'^\s*10k$')
 898         assertRegexpMatches(self, ydl._format_note({
 899             'fps': 30,
 900         }), r'^30fps$')
 901
 902     def test_postprocessors(self):
 903         filename = 'post-processor-testfile.mp4'
 904         audiofile = filename + '.mp3'
 905
 906         class SimplePP(PostProcessor):
 907             def run(self, info):
 908                 with open(audiofile, 'w') as f:
 909                     f.write('EXAMPLE')
 910                 return [info['filepath']], info
 911
 912         def run_pp(params, pp):
 913             with open(filename, 'w') as f:
 914                 f.write('EXAMPLE')
 915             ydl = YoutubeDL(params)
 916             ydl.add_post_processor(pp())
 917             ydl.post_process(filename, {'filepath': filename})
 918
 919         run_pp({'keepvideo': True}, SimplePP)
 920         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 921         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 922         os.unlink(filename)
 923         os.unlink(audiofile)
 924
 925         run_pp({'keepvideo': False}, SimplePP)
 926         self.assertFalse(os.path.exists(filename), f'{filename} exists')
 927         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 928         os.unlink(audiofile)
 929
 930         class ModifierPP(PostProcessor):
 931             def run(self, info):
 932                 with open(info['filepath'], 'w') as f:
 933                     f.write('MODIFIED')
 934                 return [], info
 935
 936         run_pp({'keepvideo': False}, ModifierPP)
 937         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 938         os.unlink(filename)
 939
 940     def test_match_filter(self):
 941         first = {
 942             'id': '1',
 943             'url': TEST_URL,
 944             'title': 'one',
 945             'extractor': 'TEST',
 946             'duration': 30,
 947             'filesize': 10 * 1024,
 948             'playlist_id': '42',
 949             'uploader': '變態妍字幕版 太妍 тест',
 950             'creator': "тест ' 123 ' тест--",
 951             'webpage_url': 'http://example.com/watch?v=shenanigans',
 952         }
 953         second = {
 954             'id': '2',
 955             'url': TEST_URL,
 956             'title': 'two',
 957             'extractor': 'TEST',
 958             'duration': 10,
 959             'description': 'foo',
 960             'filesize': 5 * 1024,
 961             'playlist_id': '43',
 962             'uploader': 'тест 123',
 963             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 964         }
 965         videos = [first, second]
 966
 967         def get_videos(filter_=None):
 968             ydl = YDL({'match_filter': filter_, 'simulate': True})
 969             for v in videos:
 970                 ydl.process_ie_result(v.copy(), download=True)
 971             return [v['id'] for v in ydl.downloaded_info_dicts]
 972
 973         res = get_videos()
 974         self.assertEqual(res, ['1', '2'])
 975
 976         def f(v, incomplete):
 977             if v['id'] == '1':
 978                 return None
 979             else:
 980                 return 'Video id is not 1'
 981         res = get_videos(f)
 982         self.assertEqual(res, ['1'])
 983
 984         f = match_filter_func('duration < 30')
 985         res = get_videos(f)
 986         self.assertEqual(res, ['2'])
 987
 988         f = match_filter_func('description = foo')
 989         res = get_videos(f)
 990         self.assertEqual(res, ['2'])
 991
 992         f = match_filter_func('description =? foo')
 993         res = get_videos(f)
 994         self.assertEqual(res, ['1', '2'])
 995
 996         f = match_filter_func('filesize > 5KiB')
 997         res = get_videos(f)
 998         self.assertEqual(res, ['1'])
 999
1000         f = match_filter_func('playlist_id = 42')
1001         res = get_videos(f)
1002         self.assertEqual(res, ['1'])
1003
1004         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
1005         res = get_videos(f)
1006         self.assertEqual(res, ['1'])
1007
1008         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
1009         res = get_videos(f)
1010         self.assertEqual(res, ['2'])
1011
1012         f = match_filter_func('creator = "тест \' 123 \' тест--"')
1013         res = get_videos(f)
1014         self.assertEqual(res, ['1'])
1015
1016         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
1017         res = get_videos(f)
1018         self.assertEqual(res, ['1'])
1019
1020         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
1021         res = get_videos(f)
1022         self.assertEqual(res, [])
1023
1024     def test_playlist_items_selection(self):
1025         INDICES, PAGE_SIZE = list(range(1, 11)), 3
1026
1027         def entry(i, evaluated):
1028             evaluated.append(i)
1029             return {
1030                 'id': str(i),
1031                 'title': str(i),
1032                 'url': TEST_URL,
1033             }
1034
1035         def pagedlist_entries(evaluated):
1036             def page_func(n):
1037                 start = PAGE_SIZE * n
1038                 for i in INDICES[start: start + PAGE_SIZE]:
1039                     yield entry(i, evaluated)
1040             return OnDemandPagedList(page_func, PAGE_SIZE)
1041
1042         def page_num(i):
1043             return (i + PAGE_SIZE - 1) // PAGE_SIZE
1044
1045         def generator_entries(evaluated):
1046             for i in INDICES:
1047                 yield entry(i, evaluated)
1048
1049         def list_entries(evaluated):
1050             return list(generator_entries(evaluated))
1051
1052         def lazylist_entries(evaluated):
1053             return LazyList(generator_entries(evaluated))
1054
1055         def get_downloaded_info_dicts(params, entries):
1056             ydl = YDL(params)
1057             ydl.process_ie_result({
1058                 '_type': 'playlist',
1059                 'id': 'test',
1060                 'extractor': 'test:playlist',
1061                 'extractor_key': 'test:playlist',
1062                 'webpage_url': 'http://example.com',
1063                 'entries': entries,
1064             })
1065             return ydl.downloaded_info_dicts
1066
1067         def test_selection(params, expected_ids, evaluate_all=False):
1068             expected_ids = list(expected_ids)
1069             if evaluate_all:
1070                 generator_eval = pagedlist_eval = INDICES
1071             elif not expected_ids:
1072                 generator_eval = pagedlist_eval = []
1073             else:
1074                 generator_eval = INDICES[0: max(expected_ids)]
1075                 pagedlist_eval = INDICES[PAGE_SIZE * page_num(min(expected_ids)) - PAGE_SIZE:
1076                                          PAGE_SIZE * page_num(max(expected_ids))]
1077
1078             for name, func, expected_eval in (
1079                 ('list', list_entries, INDICES),
1080                 ('Generator', generator_entries, generator_eval),
1081                 # ('LazyList', lazylist_entries, generator_eval),  # Generator and LazyList follow the exact same code path
1082                 ('PagedList', pagedlist_entries, pagedlist_eval),
1083             ):
1084                 evaluated = []
1085                 entries = func(evaluated)
1086                 results = [(v['playlist_autonumber'] - 1, (int(v['id']), v['playlist_index']))
1087                            for v in get_downloaded_info_dicts(params, entries)]
1088                 self.assertEqual(results, list(enumerate(zip(expected_ids, expected_ids))), f'Entries of {name} for {params}')
1089                 self.assertEqual(sorted(evaluated), expected_eval, f'Evaluation of {name} for {params}')
1090
1091         test_selection({}, INDICES)
1092         test_selection({'playlistend': 20}, INDICES, True)
1093         test_selection({'playlistend': 2}, INDICES[:2])
1094         test_selection({'playliststart': 11}, [], True)
1095         test_selection({'playliststart': 2}, INDICES[1:])
1096         test_selection({'playlist_items': '2-4'}, INDICES[1:4])
1097         test_selection({'playlist_items': '2,4'}, [2, 4])
1098         test_selection({'playlist_items': '20'}, [], True)
1099         test_selection({'playlist_items': '0'}, [])
1100
1101         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
1102         test_selection({'playlist_items': '2-4,3-4,3'}, [2, 3, 4])
1103         test_selection({'playlist_items': '4,2'}, [4, 2])
1104
1105         # Tests for https://github.com/yt-dlp/yt-dlp/issues/720
1106         # https://github.com/yt-dlp/yt-dlp/issues/302
1107         test_selection({'playlistreverse': True}, INDICES[::-1])
1108         test_selection({'playliststart': 2, 'playlistreverse': True}, INDICES[:0:-1])
1109         test_selection({'playlist_items': '2,4', 'playlistreverse': True}, [4, 2])
1110         test_selection({'playlist_items': '4,2'}, [4, 2])
1111
1112         # Tests for --playlist-items start:end:step
1113         test_selection({'playlist_items': ':'}, INDICES, True)
1114         test_selection({'playlist_items': '::1'}, INDICES, True)
1115         test_selection({'playlist_items': '::-1'}, INDICES[::-1], True)
1116         test_selection({'playlist_items': ':6'}, INDICES[:6])
1117         test_selection({'playlist_items': ':-6'}, INDICES[:-5], True)
1118         test_selection({'playlist_items': '-1:6:-2'}, INDICES[:4:-2], True)
1119         test_selection({'playlist_items': '9:-6:-2'}, INDICES[8:3:-2], True)
1120
1121         test_selection({'playlist_items': '1:inf:2'}, INDICES[::2], True)
1122         test_selection({'playlist_items': '-2:inf'}, INDICES[-2:], True)
1123         test_selection({'playlist_items': ':inf:-1'}, [], True)
1124         test_selection({'playlist_items': '0-2:2'}, [2])
1125         test_selection({'playlist_items': '1-:2'}, INDICES[::2], True)
1126         test_selection({'playlist_items': '0--2:2'}, INDICES[1:-1:2], True)
1127
1128         test_selection({'playlist_items': '10::3'}, [10], True)
1129         test_selection({'playlist_items': '-1::3'}, [10], True)
1130         test_selection({'playlist_items': '11::3'}, [], True)
1131         test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True)
1132         test_selection({'playlist_items': '-15::15'}, [], True)
1133
1134     def test_do_not_override_ie_key_in_url_transparent(self):
1135         ydl = YDL()
1136
1137         class Foo1IE(InfoExtractor):
1138             _VALID_URL = r'foo1:'
1139
1140             def _real_extract(self, url):
1141                 return {
1142                     '_type': 'url_transparent',
1143                     'url': 'foo2:',
1144                     'ie_key': 'Foo2',
1145                     'title': 'foo1 title',
1146                     'id': 'foo1_id',
1147                 }
1148
1149         class Foo2IE(InfoExtractor):
1150             _VALID_URL = r'foo2:'
1151
1152             def _real_extract(self, url):
1153                 return {
1154                     '_type': 'url',
1155                     'url': 'foo3:',
1156                     'ie_key': 'Foo3',
1157                 }
1158
1159         class Foo3IE(InfoExtractor):
1160             _VALID_URL = r'foo3:'
1161
1162             def _real_extract(self, url):
1163                 return _make_result([{'url': TEST_URL}], title='foo3 title')
1164
1165         ydl.add_info_extractor(Foo1IE(ydl))
1166         ydl.add_info_extractor(Foo2IE(ydl))
1167         ydl.add_info_extractor(Foo3IE(ydl))
1168         ydl.extract_info('foo1:')
1169         downloaded = ydl.downloaded_info_dicts[0]
1170         self.assertEqual(downloaded['url'], TEST_URL)
1171         self.assertEqual(downloaded['title'], 'foo1 title')
1172         self.assertEqual(downloaded['id'], 'testid')
1173         self.assertEqual(downloaded['extractor'], 'testex')
1174         self.assertEqual(downloaded['extractor_key'], 'TestEx')
1175
1176     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
1177     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
1178
1179         class _YDL(YDL):
1180             def __init__(self, *args, **kwargs):
1181                 super().__init__(*args, **kwargs)
1182
1183             def trouble(self, s, tb=None):
1184                 pass
1185
1186         ydl = _YDL({
1187             'format': 'extra',
1188             'ignoreerrors': True,
1189         })
1190
1191         class VideoIE(InfoExtractor):
1192             _VALID_URL = r'video:(?P<id>\d+)'
1193
1194             def _real_extract(self, url):
1195                 video_id = self._match_id(url)
1196                 formats = [{
1197                     'format_id': 'default',
1198                     'url': 'url:',
1199                 }]
1200                 if video_id == '0':
1201                     raise ExtractorError('foo')
1202                 if video_id == '2':
1203                     formats.append({
1204                         'format_id': 'extra',
1205                         'url': TEST_URL,
1206                     })
1207                 return {
1208                     'id': video_id,
1209                     'title': f'Video {video_id}',
1210                     'formats': formats,
1211                 }
1212
1213         class PlaylistIE(InfoExtractor):
1214             _VALID_URL = r'playlist:'
1215
1216             def _entries(self):
1217                 for n in range(3):
1218                     video_id = str(n)
1219                     yield {
1220                         '_type': 'url_transparent',
1221                         'ie_key': VideoIE.ie_key(),
1222                         'id': video_id,
1223                         'url': f'video:{video_id}',
1224                         'title': f'Video Transparent {video_id}',
1225                     }
1226
1227             def _real_extract(self, url):
1228                 return self.playlist_result(self._entries())
1229
1230         ydl.add_info_extractor(VideoIE(ydl))
1231         ydl.add_info_extractor(PlaylistIE(ydl))
1232         info = ydl.extract_info('playlist:')
1233         entries = info['entries']
1234         self.assertEqual(len(entries), 3)
1235         self.assertTrue(entries[0] is None)
1236         self.assertTrue(entries[1] is None)
1237         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1238         downloaded = ydl.downloaded_info_dicts[0]
1239         entries[2].pop('requested_downloads', None)
1240         self.assertEqual(entries[2], downloaded)
1241         self.assertEqual(downloaded['url'], TEST_URL)
1242         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1243         self.assertEqual(downloaded['id'], '2')
1244         self.assertEqual(downloaded['extractor'], 'Video')
1245         self.assertEqual(downloaded['extractor_key'], 'Video')
1246
1247     def test_header_cookies(self):
1248         from http.cookiejar import Cookie
1249
1250         ydl = FakeYDL()
1251         ydl.report_warning = lambda *_, **__: None
1252
1253         def cookie(name, value, version=None, domain='', path='', secure=False, expires=None):
1254             return Cookie(
1255                 version or 0, name, value, None, False,
1256                 domain, bool(domain), bool(domain), path, bool(path),
1257                 secure, expires, False, None, None, rest={})
1258
1259         _test_url = 'https://yt.dlp/test'
1260
1261         def test(encoded_cookies, cookies, *, headers=False, round_trip=None, error_re=None):
1262             def _test():
1263                 ydl.cookiejar.clear()
1264                 ydl._load_cookies(encoded_cookies, autoscope=headers)
1265                 if headers:
1266                     ydl._apply_header_cookies(_test_url)
1267                 data = {'url': _test_url}
1268                 ydl._calc_headers(data)
1269                 self.assertCountEqual(
1270                     map(vars, ydl.cookiejar), map(vars, cookies),
1271                     'Extracted cookiejar.Cookie is not the same')
1272                 if not headers:
1273                     self.assertEqual(
1274                         data.get('cookies'), round_trip or encoded_cookies,
1275                         'Cookie is not the same as round trip')
1276                 ydl.__dict__['_YoutubeDL__header_cookies'] = []
1277
1278             with self.subTest(msg=encoded_cookies):
1279                 if not error_re:
1280                     _test()
1281                     return
1282                 with self.assertRaisesRegex(Exception, error_re):
1283                     _test()
1284
1285         test('test=value; Domain=.yt.dlp', [cookie('test', 'value', domain='.yt.dlp')])
1286         test('test=value', [cookie('test', 'value')], error_re=r'Unscoped cookies are not allowed')
1287         test('cookie1=value1; Domain=.yt.dlp; Path=/test; cookie2=value2; Domain=.yt.dlp; Path=/', [
1288             cookie('cookie1', 'value1', domain='.yt.dlp', path='/test'),
1289             cookie('cookie2', 'value2', domain='.yt.dlp', path='/')])
1290         test('test=value; Domain=.yt.dlp; Path=/test; Secure; Expires=9999999999', [
1291             cookie('test', 'value', domain='.yt.dlp', path='/test', secure=True, expires=9999999999)])
1292         test('test="value; "; path=/test; domain=.yt.dlp', [
1293             cookie('test', 'value; ', domain='.yt.dlp', path='/test')],
1294             round_trip='test="value\\073 "; Domain=.yt.dlp; Path=/test')
1295         test('name=; Domain=.yt.dlp', [cookie('name', '', domain='.yt.dlp')],
1296              round_trip='name=""; Domain=.yt.dlp')
1297
1298         test('test=value', [cookie('test', 'value', domain='.yt.dlp')], headers=True)
1299         test('cookie1=value; Domain=.yt.dlp; cookie2=value', [], headers=True, error_re=r'Invalid syntax')
1300         ydl.deprecated_feature = ydl.report_error
1301         test('test=value', [], headers=True, error_re=r'Passing cookies as a header is a potential security risk')
1302
1303     def test_infojson_cookies(self):
1304         TEST_FILE = 'test_infojson_cookies.info.json'
1305         TEST_URL = 'https://example.com/example.mp4'
1306         COOKIES = 'a=b; Domain=.example.com; c=d; Domain=.example.com'
1307         COOKIE_HEADER = {'Cookie': 'a=b; c=d'}
1308
1309         ydl = FakeYDL()
1310         ydl.process_info = lambda x: ydl._write_info_json('test', x, TEST_FILE)
1311
1312         def make_info(info_header_cookies=False, fmts_header_cookies=False, cookies_field=False):
1313             fmt = {'url': TEST_URL}
1314             if fmts_header_cookies:
1315                 fmt['http_headers'] = COOKIE_HEADER
1316             if cookies_field:
1317                 fmt['cookies'] = COOKIES
1318             return _make_result([fmt], http_headers=COOKIE_HEADER if info_header_cookies else None)
1319
1320         def test(initial_info, note):
1321             result = {}
1322             result['processed'] = ydl.process_ie_result(initial_info)
1323             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1324                             msg=f'No cookies set in cookiejar after initial process when {note}')
1325             ydl.cookiejar.clear()
1326             with open(TEST_FILE) as infojson:
1327                 result['loaded'] = ydl.sanitize_info(json.load(infojson), True)
1328             result['final'] = ydl.process_ie_result(result['loaded'].copy(), download=False)
1329             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1330                             msg=f'No cookies set in cookiejar after final process when {note}')
1331             ydl.cookiejar.clear()
1332             for key in ('processed', 'loaded', 'final'):
1333                 info = result[key]
1334                 self.assertIsNone(
1335                     traverse_obj(info, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False),
1336                     msg=f'Cookie header not removed in {key} result when {note}')
1337                 self.assertEqual(
1338                     traverse_obj(info, ((None, ('formats', 0)), 'cookies'), get_all=False), COOKIES,
1339                     msg=f'No cookies field found in {key} result when {note}')
1340
1341         test({'url': TEST_URL, 'http_headers': COOKIE_HEADER, 'id': '1', 'title': 'x'}, 'no formats field')
1342         test(make_info(info_header_cookies=True), 'info_dict header cokies')
1343         test(make_info(fmts_header_cookies=True), 'format header cookies')
1344         test(make_info(info_header_cookies=True, fmts_header_cookies=True), 'info_dict and format header cookies')
1345         test(make_info(info_header_cookies=True, fmts_header_cookies=True, cookies_field=True), 'all cookies fields')
1346         test(make_info(cookies_field=True), 'cookies format field')
1347         test({'url': TEST_URL, 'cookies': COOKIES, 'id': '1', 'title': 'x'}, 'info_dict cookies field only')
1348
1349         try_rm(TEST_FILE)
1350
1351     def test_add_headers_cookie(self):
1352         def check_for_cookie_header(result):
1353             return traverse_obj(result, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False)
1354
1355         ydl = FakeYDL({'http_headers': {'Cookie': 'a=b'}})
1356         ydl._apply_header_cookies(_make_result([])['webpage_url'])  # Scope to input webpage URL: .example.com
1357
1358         fmt = {'url': 'https://example.com/video.mp4'}
1359         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1360         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies in result info_dict')
1361         self.assertEqual(result.get('cookies'), 'a=b; Domain=.example.com', msg='No cookies were set in cookies field')
1362         self.assertIn('a=b', ydl.cookiejar.get_cookie_header(fmt['url']), msg='No cookies were set in cookiejar')
1363
1364         fmt = {'url': 'https://wrong.com/video.mp4'}
1365         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1366         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies for wrong domain')
1367         self.assertFalse(result.get('cookies'), msg='Cookies set in cookies field for wrong domain')
1368         self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain')
1369
1370
1371 if __name__ == '__main__':
1372     unittest.main()