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_by_vcodec_sort(self):
 240         formats = [
 241             {'format_id': 'av1-format', 'ext': 'mp4', 'vcodec': 'av1', 'acodec': 'none', 'url': TEST_URL},
 242             {'format_id': 'vp9-hdr-format', 'ext': 'mp4', 'vcodec': 'vp09.02.50.10.01.09.18.09.00', 'acodec': 'none', 'url': TEST_URL},
 243             {'format_id': 'vp9-sdr-format', 'ext': 'mp4', 'vcodec': 'vp09.00.50.08', 'acodec': 'none', 'url': TEST_URL},
 244             {'format_id': 'h265-format', 'ext': 'mp4', 'vcodec': 'h265', 'acodec': 'none', 'url': TEST_URL},
 245         ]
 246         info_dict = _make_result(formats)
 247
 248         ydl = YDL({'format': 'bestvideo', 'format_sort': ['vcodec:vp9.2']})
 249         ydl.process_ie_result(info_dict.copy())
 250         downloaded = ydl.downloaded_info_dicts[0]
 251         self.assertEqual(downloaded['format_id'], 'vp9-hdr-format')
 252
 253         ydl = YDL({'format': 'bestvideo', 'format_sort': ['vcodec:vp9']})
 254         ydl.process_ie_result(info_dict.copy())
 255         downloaded = ydl.downloaded_info_dicts[0]
 256         self.assertEqual(downloaded['format_id'], 'vp9-sdr-format')
 257
 258         ydl = YDL({'format': 'bestvideo', 'format_sort': ['+vcodec:vp9.2']})
 259         ydl.process_ie_result(info_dict.copy())
 260         downloaded = ydl.downloaded_info_dicts[0]
 261         self.assertEqual(downloaded['format_id'], 'vp9-hdr-format')
 262
 263         ydl = YDL({'format': 'bestvideo', 'format_sort': ['+vcodec:vp9']})
 264         ydl.process_ie_result(info_dict.copy())
 265         downloaded = ydl.downloaded_info_dicts[0]
 266         self.assertEqual(downloaded['format_id'], 'vp9-sdr-format')
 267
 268     def test_format_selection_string_ops(self):
 269         formats = [
 270             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 271             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 272         ]
 273         info_dict = _make_result(formats)
 274
 275         # equals (=)
 276         ydl = YDL({'format': '[format_id=abc-cba]'})
 277         ydl.process_ie_result(info_dict.copy())
 278         downloaded = ydl.downloaded_info_dicts[0]
 279         self.assertEqual(downloaded['format_id'], 'abc-cba')
 280
 281         # does not equal (!=)
 282         ydl = YDL({'format': '[format_id!=abc-cba]'})
 283         ydl.process_ie_result(info_dict.copy())
 284         downloaded = ydl.downloaded_info_dicts[0]
 285         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 286
 287         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 288         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 289
 290         # starts with (^=)
 291         ydl = YDL({'format': '[format_id^=abc]'})
 292         ydl.process_ie_result(info_dict.copy())
 293         downloaded = ydl.downloaded_info_dicts[0]
 294         self.assertEqual(downloaded['format_id'], 'abc-cba')
 295
 296         # does not start with (!^=)
 297         ydl = YDL({'format': '[format_id!^=abc]'})
 298         ydl.process_ie_result(info_dict.copy())
 299         downloaded = ydl.downloaded_info_dicts[0]
 300         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 301
 302         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 303         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 304
 305         # ends with ($=)
 306         ydl = YDL({'format': '[format_id$=cba]'})
 307         ydl.process_ie_result(info_dict.copy())
 308         downloaded = ydl.downloaded_info_dicts[0]
 309         self.assertEqual(downloaded['format_id'], 'abc-cba')
 310
 311         # does not end with (!$=)
 312         ydl = YDL({'format': '[format_id!$=cba]'})
 313         ydl.process_ie_result(info_dict.copy())
 314         downloaded = ydl.downloaded_info_dicts[0]
 315         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 316
 317         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 318         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 319
 320         # contains (*=)
 321         ydl = YDL({'format': '[format_id*=bc-cb]'})
 322         ydl.process_ie_result(info_dict.copy())
 323         downloaded = ydl.downloaded_info_dicts[0]
 324         self.assertEqual(downloaded['format_id'], 'abc-cba')
 325
 326         # does not contain (!*=)
 327         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 328         ydl.process_ie_result(info_dict.copy())
 329         downloaded = ydl.downloaded_info_dicts[0]
 330         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 331
 332         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 333         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 334
 335         ydl = YDL({'format': '[format_id!*=-]'})
 336         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 337
 338     def test_youtube_format_selection(self):
 339         # FIXME: Rewrite in accordance with the new format sorting options
 340         return
 341
 342         order = [
 343             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 344             # Apple HTTP Live Streaming
 345             '96', '95', '94', '93', '92', '132', '151',
 346             # 3D
 347             '85', '84', '102', '83', '101', '82', '100',
 348             # Dash video
 349             '137', '248', '136', '247', '135', '246',
 350             '245', '244', '134', '243', '133', '242', '160',
 351             # Dash audio
 352             '141', '172', '140', '171', '139',
 353         ]
 354
 355         def format_info(f_id):
 356             info = YoutubeIE._formats[f_id].copy()
 357
 358             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 359             # and 'vcodec', while in tests such information is incomplete since
 360             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 361             # test_YoutubeDL.test_youtube_format_selection is broken without
 362             # this fix
 363             if 'acodec' in info and 'vcodec' not in info:
 364                 info['vcodec'] = 'none'
 365             elif 'vcodec' in info and 'acodec' not in info:
 366                 info['acodec'] = 'none'
 367
 368             info['format_id'] = f_id
 369             info['url'] = 'url:' + f_id
 370             return info
 371         formats_order = [format_info(f_id) for f_id in order]
 372
 373         info_dict = _make_result(list(formats_order), extractor='youtube')
 374         ydl = YDL({'format': 'bestvideo+bestaudio'})
 375         ydl.sort_formats(info_dict)
 376         ydl.process_ie_result(info_dict)
 377         downloaded = ydl.downloaded_info_dicts[0]
 378         self.assertEqual(downloaded['format_id'], '248+172')
 379         self.assertEqual(downloaded['ext'], 'mp4')
 380
 381         info_dict = _make_result(list(formats_order), extractor='youtube')
 382         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 383         ydl.sort_formats(info_dict)
 384         ydl.process_ie_result(info_dict)
 385         downloaded = ydl.downloaded_info_dicts[0]
 386         self.assertEqual(downloaded['format_id'], '38')
 387
 388         info_dict = _make_result(list(formats_order), extractor='youtube')
 389         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 390         ydl.sort_formats(info_dict)
 391         ydl.process_ie_result(info_dict)
 392         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 393         self.assertEqual(downloaded_ids, ['137', '141'])
 394
 395         info_dict = _make_result(list(formats_order), extractor='youtube')
 396         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 397         ydl.sort_formats(info_dict)
 398         ydl.process_ie_result(info_dict)
 399         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 400         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 401
 402         info_dict = _make_result(list(formats_order), extractor='youtube')
 403         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 404         ydl.sort_formats(info_dict)
 405         ydl.process_ie_result(info_dict)
 406         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 407         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 408
 409         info_dict = _make_result(list(formats_order), extractor='youtube')
 410         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 411         ydl.sort_formats(info_dict)
 412         ydl.process_ie_result(info_dict)
 413         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 414         self.assertEqual(downloaded_ids, ['248+141'])
 415
 416         for f1, f2 in zip(formats_order, formats_order[1:]):
 417             info_dict = _make_result([f1, f2], extractor='youtube')
 418             ydl = YDL({'format': 'best/bestvideo'})
 419             ydl.sort_formats(info_dict)
 420             ydl.process_ie_result(info_dict)
 421             downloaded = ydl.downloaded_info_dicts[0]
 422             self.assertEqual(downloaded['format_id'], f1['format_id'])
 423
 424             info_dict = _make_result([f2, f1], extractor='youtube')
 425             ydl = YDL({'format': 'best/bestvideo'})
 426             ydl.sort_formats(info_dict)
 427             ydl.process_ie_result(info_dict)
 428             downloaded = ydl.downloaded_info_dicts[0]
 429             self.assertEqual(downloaded['format_id'], f1['format_id'])
 430
 431     def test_audio_only_extractor_format_selection(self):
 432         # For extractors with incomplete formats (all formats are audio-only or
 433         # video-only) best and worst should fallback to corresponding best/worst
 434         # video-only or audio-only formats (as per
 435         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 436         formats = [
 437             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 438             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 439         ]
 440         info_dict = _make_result(formats)
 441
 442         ydl = YDL({'format': 'best'})
 443         ydl.process_ie_result(info_dict.copy())
 444         downloaded = ydl.downloaded_info_dicts[0]
 445         self.assertEqual(downloaded['format_id'], 'high')
 446
 447         ydl = YDL({'format': 'worst'})
 448         ydl.process_ie_result(info_dict.copy())
 449         downloaded = ydl.downloaded_info_dicts[0]
 450         self.assertEqual(downloaded['format_id'], 'low')
 451
 452     def test_format_not_available(self):
 453         formats = [
 454             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 455             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 456         ]
 457         info_dict = _make_result(formats)
 458
 459         # This must fail since complete video-audio format does not match filter
 460         # and extractor does not provide incomplete only formats (i.e. only
 461         # video-only or audio-only).
 462         ydl = YDL({'format': 'best[height>360]'})
 463         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 464
 465     def test_format_selection_issue_10083(self):
 466         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 467         formats = [
 468             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 469             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 470             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 471         ]
 472         info_dict = _make_result(formats)
 473
 474         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 475         ydl.process_ie_result(info_dict.copy())
 476         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 477
 478     def test_invalid_format_specs(self):
 479         def assert_syntax_error(format_spec):
 480             self.assertRaises(SyntaxError, YDL, {'format': format_spec})
 481
 482         assert_syntax_error('bestvideo,,best')
 483         assert_syntax_error('+bestaudio')
 484         assert_syntax_error('bestvideo+')
 485         assert_syntax_error('/')
 486         assert_syntax_error('[720<height]')
 487
 488     def test_format_filtering(self):
 489         formats = [
 490             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 491             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 492             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 493             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 494             {'format_id': 'E', 'filesize': 3000},
 495             {'format_id': 'F'},
 496             {'format_id': 'G', 'filesize': 1000000},
 497         ]
 498         for f in formats:
 499             f['url'] = 'http://_/'
 500             f['ext'] = 'unknown'
 501         info_dict = _make_result(formats, _format_sort_fields=('id', ))
 502
 503         ydl = YDL({'format': 'best[filesize<3000]'})
 504         ydl.process_ie_result(info_dict)
 505         downloaded = ydl.downloaded_info_dicts[0]
 506         self.assertEqual(downloaded['format_id'], 'D')
 507
 508         ydl = YDL({'format': 'best[filesize<=3000]'})
 509         ydl.process_ie_result(info_dict)
 510         downloaded = ydl.downloaded_info_dicts[0]
 511         self.assertEqual(downloaded['format_id'], 'E')
 512
 513         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 514         ydl.process_ie_result(info_dict)
 515         downloaded = ydl.downloaded_info_dicts[0]
 516         self.assertEqual(downloaded['format_id'], 'F')
 517
 518         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 519         ydl.process_ie_result(info_dict)
 520         downloaded = ydl.downloaded_info_dicts[0]
 521         self.assertEqual(downloaded['format_id'], 'B')
 522
 523         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 524         ydl.process_ie_result(info_dict)
 525         downloaded = ydl.downloaded_info_dicts[0]
 526         self.assertEqual(downloaded['format_id'], 'C')
 527
 528         ydl = YDL({'format': '[filesize>?1]'})
 529         ydl.process_ie_result(info_dict)
 530         downloaded = ydl.downloaded_info_dicts[0]
 531         self.assertEqual(downloaded['format_id'], 'G')
 532
 533         ydl = YDL({'format': '[filesize<1M]'})
 534         ydl.process_ie_result(info_dict)
 535         downloaded = ydl.downloaded_info_dicts[0]
 536         self.assertEqual(downloaded['format_id'], 'E')
 537
 538         ydl = YDL({'format': '[filesize<1MiB]'})
 539         ydl.process_ie_result(info_dict)
 540         downloaded = ydl.downloaded_info_dicts[0]
 541         self.assertEqual(downloaded['format_id'], 'G')
 542
 543         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 544         ydl.process_ie_result(info_dict)
 545         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 546         self.assertEqual(downloaded_ids, ['D', 'C', 'B'])
 547
 548         ydl = YDL({'format': 'best[height<40]'})
 549         with contextlib.suppress(ExtractorError):
 550             ydl.process_ie_result(info_dict)
 551         self.assertEqual(ydl.downloaded_info_dicts, [])
 552
 553     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.available', False)
 554     def test_default_format_spec_without_ffmpeg(self):
 555         ydl = YDL({})
 556         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 557
 558         ydl = YDL({'simulate': True})
 559         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 560
 561         ydl = YDL({})
 562         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 563
 564         ydl = YDL({'simulate': True})
 565         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 566
 567         ydl = YDL({'outtmpl': '-'})
 568         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 569
 570         ydl = YDL({})
 571         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 572         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 573
 574     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.available', True)
 575     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.can_merge', lambda _: True)
 576     def test_default_format_spec_with_ffmpeg(self):
 577         ydl = YDL({})
 578         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 579
 580         ydl = YDL({'simulate': True})
 581         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 582
 583         ydl = YDL({})
 584         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 585
 586         ydl = YDL({'simulate': True})
 587         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 588
 589         ydl = YDL({'outtmpl': '-'})
 590         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 591
 592         ydl = YDL({})
 593         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 594         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 595
 596
 597 class TestYoutubeDL(unittest.TestCase):
 598     def test_subtitles(self):
 599         def s_formats(lang, autocaption=False):
 600             return [{
 601                 'ext': ext,
 602                 'url': f'http://localhost/video.{lang}.{ext}',
 603                 '_auto': autocaption,
 604             } for ext in ['vtt', 'srt', 'ass']]
 605         subtitles = {l: s_formats(l) for l in ['en', 'fr', 'es']}
 606         auto_captions = {l: s_formats(l, True) for l in ['it', 'pt', 'es']}
 607         info_dict = {
 608             'id': 'test',
 609             'title': 'Test',
 610             'url': 'http://localhost/video.mp4',
 611             'subtitles': subtitles,
 612             'automatic_captions': auto_captions,
 613             'extractor': 'TEST',
 614             'webpage_url': 'http://example.com/watch?v=shenanigans',
 615         }
 616
 617         def get_info(params={}):
 618             params.setdefault('simulate', True)
 619             ydl = YDL(params)
 620             ydl.report_warning = lambda *args, **kargs: None
 621             return ydl.process_video_result(info_dict, download=False)
 622
 623         result = get_info()
 624         self.assertFalse(result.get('requested_subtitles'))
 625         self.assertEqual(result['subtitles'], subtitles)
 626         self.assertEqual(result['automatic_captions'], auto_captions)
 627
 628         result = get_info({'writesubtitles': True})
 629         subs = result['requested_subtitles']
 630         self.assertTrue(subs)
 631         self.assertEqual(set(subs.keys()), {'en'})
 632         self.assertTrue(subs['en'].get('data') is None)
 633         self.assertEqual(subs['en']['ext'], 'ass')
 634
 635         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 636         subs = result['requested_subtitles']
 637         self.assertEqual(subs['en']['ext'], 'srt')
 638
 639         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 640         subs = result['requested_subtitles']
 641         self.assertTrue(subs)
 642         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 643
 644         result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']})
 645         subs = result['requested_subtitles']
 646         self.assertTrue(subs)
 647         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 648
 649         result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']})
 650         subs = result['requested_subtitles']
 651         self.assertTrue(subs)
 652         self.assertEqual(set(subs.keys()), {'fr'})
 653
 654         result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']})
 655         subs = result['requested_subtitles']
 656         self.assertTrue(subs)
 657         self.assertEqual(set(subs.keys()), {'en'})
 658
 659         result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']})
 660         subs = result['requested_subtitles']
 661         self.assertTrue(subs)
 662         self.assertEqual(set(subs.keys()), {'es', 'en'})
 663
 664         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 665         subs = result['requested_subtitles']
 666         self.assertTrue(subs)
 667         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 668         self.assertFalse(subs['es']['_auto'])
 669         self.assertTrue(subs['pt']['_auto'])
 670
 671         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 672         subs = result['requested_subtitles']
 673         self.assertTrue(subs)
 674         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 675         self.assertTrue(subs['es']['_auto'])
 676         self.assertTrue(subs['pt']['_auto'])
 677
 678     def test_add_extra_info(self):
 679         test_dict = {
 680             'extractor': 'Foo',
 681         }
 682         extra_info = {
 683             'extractor': 'Bar',
 684             'playlist': 'funny videos',
 685         }
 686         YDL.add_extra_info(test_dict, extra_info)
 687         self.assertEqual(test_dict['extractor'], 'Foo')
 688         self.assertEqual(test_dict['playlist'], 'funny videos')
 689
 690     outtmpl_info = {
 691         'id': '1234',
 692         'ext': 'mp4',
 693         'width': None,
 694         'height': 1080,
 695         'filesize': 1024,
 696         'title1': '$PATH',
 697         'title2': '%PATH%',
 698         'title3': 'foo/bar\\test',
 699         'title4': 'foo "bar" test',
 700         'title5': 'áéí 𝐀',
 701         'timestamp': 1618488000,
 702         'duration': 100000,
 703         'playlist_index': 1,
 704         'playlist_autonumber': 2,
 705         '__last_playlist_index': 100,
 706         'n_entries': 10,
 707         'formats': [
 708             {'id': 'id 1', 'height': 1080, 'width': 1920},
 709             {'id': 'id 2', 'height': 720},
 710             {'id': 'id 3'},
 711         ],
 712     }
 713
 714     def test_prepare_outtmpl_and_filename(self):
 715         def test(tmpl, expected, *, info=None, **params):
 716             params['outtmpl'] = tmpl
 717             ydl = FakeYDL(params)
 718             ydl._num_downloads = 1
 719             self.assertEqual(ydl.validate_outtmpl(tmpl), None)
 720
 721             out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info)
 722             fname = ydl.prepare_filename(info or self.outtmpl_info)
 723
 724             if not isinstance(expected, (list, tuple)):
 725                 expected = (expected, expected)
 726             for (name, got), expect in zip((('outtmpl', out), ('filename', fname)), expected):
 727                 if callable(expect):
 728                     self.assertTrue(expect(got), f'Wrong {name} from {tmpl}')
 729                 elif expect is not None:
 730                     self.assertEqual(got, expect, f'Wrong {name} from {tmpl}')
 731
 732         # Side-effects
 733         original_infodict = dict(self.outtmpl_info)
 734         test('foo.bar', 'foo.bar')
 735         original_infodict['epoch'] = self.outtmpl_info.get('epoch')
 736         self.assertTrue(isinstance(original_infodict['epoch'], int))
 737         test('%(epoch)d', int_or_none)
 738         self.assertEqual(original_infodict, self.outtmpl_info)
 739
 740         # Auto-generated fields
 741         test('%(id)s.%(ext)s', '1234.mp4')
 742         test('%(duration_string)s', ('27:46:40', '27-46-40'))
 743         test('%(resolution)s', '1080p')
 744         test('%(playlist_index|)s', '001')
 745         test('%(playlist_index&{}!)s', '1!')
 746         test('%(playlist_autonumber)s', '02')
 747         test('%(autonumber)s', '00001')
 748         test('%(autonumber+2)03d', '005', autonumber_start=3)
 749         test('%(autonumber)s', '001', autonumber_size=3)
 750
 751         # Escaping %
 752         test('%', '%')
 753         test('%%', '%')
 754         test('%%%%', '%%')
 755         test('%s', '%s')
 756         test('%%%s', '%%s')
 757         test('%d', '%d')
 758         test('%abc%', '%abc%')
 759         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 760         test('%%%(height)s', '%1080')
 761         test('%(width)06d.%(ext)s', 'NA.mp4')
 762         test('%(width)06d.%%(ext)s', 'NA.%(ext)s')
 763         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 764
 765         # ID sanitization
 766         test('%(id)s', '_abcd', info={'id': '_abcd'})
 767         test('%(some_id)s', '_abcd', info={'some_id': '_abcd'})
 768         test('%(formats.0.id)s', '_abcd', info={'formats': [{'id': '_abcd'}]})
 769         test('%(id)s', '-abcd', info={'id': '-abcd'})
 770         test('%(id)s', '.abcd', info={'id': '.abcd'})
 771         test('%(id)s', 'ab__cd', info={'id': 'ab__cd'})
 772         test('%(id)s', ('ab:cd', 'ab：cd'), info={'id': 'ab:cd'})
 773         test('%(id.0)s', '-', info={'id': '--'})
 774
 775         # Invalid templates
 776         self.assertTrue(isinstance(YoutubeDL.validate_outtmpl('%(title)'), ValueError))
 777         test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder='none')
 778         test('%(..)s', 'NA')
 779         test('%(formats.{id)s', 'NA')
 780
 781         # Entire info_dict
 782         def expect_same_infodict(out):
 783             got_dict = json.loads(out)
 784             for info_field, expected in self.outtmpl_info.items():
 785                 self.assertEqual(got_dict.get(info_field), expected, info_field)
 786             return True
 787
 788         test('%()j', (expect_same_infodict, None))
 789
 790         # NA placeholder
 791         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s'
 792         test(NA_TEST_OUTTMPL, 'NA-NA-def-1234.mp4')
 793         test(NA_TEST_OUTTMPL, 'none-none-def-1234.mp4', outtmpl_na_placeholder='none')
 794         test(NA_TEST_OUTTMPL, '--def-1234.mp4', outtmpl_na_placeholder='')
 795         test('%(non_existent.0)s', 'NA')
 796
 797         # String formatting
 798         FMT_TEST_OUTTMPL = '%%(height)%s.%%(ext)s'
 799         test(FMT_TEST_OUTTMPL % 's', '1080.mp4')
 800         test(FMT_TEST_OUTTMPL % 'd', '1080.mp4')
 801         test(FMT_TEST_OUTTMPL % '6d', '  1080.mp4')
 802         test(FMT_TEST_OUTTMPL % '-6d', '1080  .mp4')
 803         test(FMT_TEST_OUTTMPL % '06d', '001080.mp4')
 804         test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4')
 805         test(FMT_TEST_OUTTMPL % '   06d', ' 01080.mp4')
 806         test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4')
 807         test(FMT_TEST_OUTTMPL % '0   6d', ' 01080.mp4')
 808         test(FMT_TEST_OUTTMPL % '   0   6d', ' 01080.mp4')
 809
 810         # Type casting
 811         test('%(id)d', '1234')
 812         test('%(height)c', '1')
 813         test('%(ext)c', 'm')
 814         test('%(id)d %(id)r', "1234 '1234'")
 815         test('%(id)r %(height)r', "'1234' 1080")
 816         test('%(title5)a %(height)a', (R"'\xe1\xe9\xed \U0001d400' 1080", None))
 817         test('%(ext)s-%(ext|def)d', 'mp4-def')
 818         test('%(width|0)04d', '0')
 819         test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none')
 820
 821         FORMATS = self.outtmpl_info['formats']
 822
 823         # Custom type casting
 824         test('%(formats.:.id)l', 'id 1, id 2, id 3')
 825         test('%(formats.:.id)#l', ('id 1\nid 2\nid 3', 'id 1 id 2 id 3'))
 826         test('%(ext)l', 'mp4')
 827         test('%(formats.:.id) 18l', '  id 1, id 2, id 3')
 828         test('%(formats)j', (json.dumps(FORMATS), None))
 829         test('%(formats)#j', (
 830             json.dumps(FORMATS, indent=4),
 831             json.dumps(FORMATS, indent=4).replace(':', '：').replace('"', '＂').replace('\n', ' '),
 832         ))
 833         test('%(title5).3B', 'á')
 834         test('%(title5)U', 'áéí 𝐀')
 835         test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀')
 836         test('%(title5)+U', 'áéí A')
 837         test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A')
 838         test('%(height)D', '1k')
 839         test('%(filesize)#D', '1Ki')
 840         test('%(height)5.2D', ' 1.08k')
 841         test('%(title4)#S', 'foo_bar_test')
 842         test('%(title4).10S', ('foo ＂bar＂ ', 'foo ＂bar＂' + ('#' if compat_os_name == 'nt' else ' ')))
 843         if compat_os_name == 'nt':
 844             test('%(title4)q', ('"foo ""bar"" test"', None))
 845             test('%(formats.:.id)#q', ('"id 1" "id 2" "id 3"', None))
 846             test('%(formats.0.id)#q', ('"id 1"', None))
 847         else:
 848             test('%(title4)q', ('\'foo "bar" test\'', '\'foo ＂bar＂ test\''))
 849             test('%(formats.:.id)#q', "'id 1' 'id 2' 'id 3'")
 850             test('%(formats.0.id)#q', "'id 1'")
 851
 852         # Internal formatting
 853         test('%(timestamp-1000>%H-%M-%S)s', '11-43-20')
 854         test('%(title|%)s %(title|%%)s', '% %%')
 855         test('%(id+1-height+3)05d', '00158')
 856         test('%(width+100)05d', 'NA')
 857         test('%(filesize*8)d', '8192')
 858         test('%(formats.0) 15s', ('% 15s' % FORMATS[0], None))
 859         test('%(formats.0)r', (repr(FORMATS[0]), None))
 860         test('%(height.0)03d', '001')
 861         test('%(-height.0)04d', '-001')
 862         test('%(formats.-1.id)s', FORMATS[-1]['id'])
 863         test('%(formats.0.id.-1)d', FORMATS[0]['id'][-1])
 864         test('%(formats.3)s', 'NA')
 865         test('%(formats.:2:-1)r', repr(FORMATS[:2:-1]))
 866         test('%(formats.0.id.-1+id)f', '1235.000000')
 867         test('%(formats.0.id.-1+formats.1.id.-1)d', '3')
 868         out = json.dumps([{'id': f['id'], 'height.:2': str(f['height'])[:2]}
 869                           if 'height' in f else {'id': f['id']}
 870                           for f in FORMATS])
 871         test('%(formats.:.{id,height.:2})j', (out, None))
 872         test('%(formats.:.{id,height}.id)l', ', '.join(f['id'] for f in FORMATS))
 873         test('%(.{id,title})j', ('{"id": "1234"}', '{＂id＂： ＂1234＂}'))
 874
 875         # Alternates
 876         test('%(title,id)s', '1234')
 877         test('%(width-100,height+20|def)d', '1100')
 878         test('%(width-100,height+width|def)s', 'def')
 879         test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00')
 880
 881         # Replacement
 882         test('%(id&foo)s.bar', 'foo.bar')
 883         test('%(title&foo)s.bar', 'NA.bar')
 884         test('%(title&foo|baz)s.bar', 'baz.bar')
 885         test('%(x,id&foo|baz)s.bar', 'foo.bar')
 886         test('%(x,title&foo|baz)s.bar', 'baz.bar')
 887         test('%(id&a\nb|)s', ('a\nb', 'a b'))
 888         test('%(id&hi {:>10} {}|)s', 'hi       1234 1234')
 889         test(R'%(id&{0} {}|)s', 'NA')
 890         test(R'%(id&{0.1}|)s', 'NA')
 891         test('%(height&{:,d})S', '1,080')
 892
 893         # Laziness
 894         def gen():
 895             yield from range(5)
 896             raise self.assertTrue(False, 'LazyList should not be evaluated till here')
 897         test('%(key.4)s', '4', info={'key': LazyList(gen())})
 898
 899         # Empty filename
 900         test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4')
 901         # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4'))  # FIXME: ?
 902         # test('%(foo|)s', ('', '_'))  # FIXME: ?
 903
 904         # Environment variable expansion for prepare_filename
 905         os.environ['__yt_dlp_var'] = 'expanded'
 906         envvar = '%__yt_dlp_var%' if compat_os_name == 'nt' else '$__yt_dlp_var'
 907         test(envvar, (envvar, 'expanded'))
 908         if compat_os_name == 'nt':
 909             test('%s%', ('%s%', '%s%'))
 910             os.environ['s'] = 'expanded'
 911             test('%s%', ('%s%', 'expanded'))  # %s% should be expanded before escaping %s
 912             os.environ['(test)s'] = 'expanded'
 913             test('%(test)s%', ('NA%', 'expanded'))  # Environment should take priority over template
 914
 915         # Path expansion and escaping
 916         test('Hello %(title1)s', 'Hello $PATH')
 917         test('Hello %(title2)s', 'Hello %PATH%')
 918         test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test'))
 919         test('folder/%(title3)s', ('folder/foo/bar\\test', f'folder{os.path.sep}foo⧸bar⧹test'))
 920
 921     def test_format_note(self):
 922         ydl = YoutubeDL()
 923         self.assertEqual(ydl._format_note({}), '')
 924         assertRegexpMatches(self, ydl._format_note({
 925             'vbr': 10,
 926         }), r'^\s*10k$')
 927         assertRegexpMatches(self, ydl._format_note({
 928             'fps': 30,
 929         }), r'^30fps$')
 930
 931     def test_postprocessors(self):
 932         filename = 'post-processor-testfile.mp4'
 933         audiofile = filename + '.mp3'
 934
 935         class SimplePP(PostProcessor):
 936             def run(self, info):
 937                 with open(audiofile, 'w') as f:
 938                     f.write('EXAMPLE')
 939                 return [info['filepath']], info
 940
 941         def run_pp(params, pp):
 942             with open(filename, 'w') as f:
 943                 f.write('EXAMPLE')
 944             ydl = YoutubeDL(params)
 945             ydl.add_post_processor(pp())
 946             ydl.post_process(filename, {'filepath': filename})
 947
 948         run_pp({'keepvideo': True}, SimplePP)
 949         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 950         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 951         os.unlink(filename)
 952         os.unlink(audiofile)
 953
 954         run_pp({'keepvideo': False}, SimplePP)
 955         self.assertFalse(os.path.exists(filename), f'{filename} exists')
 956         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 957         os.unlink(audiofile)
 958
 959         class ModifierPP(PostProcessor):
 960             def run(self, info):
 961                 with open(info['filepath'], 'w') as f:
 962                     f.write('MODIFIED')
 963                 return [], info
 964
 965         run_pp({'keepvideo': False}, ModifierPP)
 966         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 967         os.unlink(filename)
 968
 969     def test_match_filter(self):
 970         first = {
 971             'id': '1',
 972             'url': TEST_URL,
 973             'title': 'one',
 974             'extractor': 'TEST',
 975             'duration': 30,
 976             'filesize': 10 * 1024,
 977             'playlist_id': '42',
 978             'uploader': '變態妍字幕版 太妍 тест',
 979             'creator': "тест ' 123 ' тест--",
 980             'webpage_url': 'http://example.com/watch?v=shenanigans',
 981         }
 982         second = {
 983             'id': '2',
 984             'url': TEST_URL,
 985             'title': 'two',
 986             'extractor': 'TEST',
 987             'duration': 10,
 988             'description': 'foo',
 989             'filesize': 5 * 1024,
 990             'playlist_id': '43',
 991             'uploader': 'тест 123',
 992             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 993         }
 994         videos = [first, second]
 995
 996         def get_videos(filter_=None):
 997             ydl = YDL({'match_filter': filter_, 'simulate': True})
 998             for v in videos:
 999                 ydl.process_ie_result(v.copy(), download=True)
1000             return [v['id'] for v in ydl.downloaded_info_dicts]
1001
1002         res = get_videos()
1003         self.assertEqual(res, ['1', '2'])
1004
1005         def f(v, incomplete):
1006             if v['id'] == '1':
1007                 return None
1008             else:
1009                 return 'Video id is not 1'
1010         res = get_videos(f)
1011         self.assertEqual(res, ['1'])
1012
1013         f = match_filter_func('duration < 30')
1014         res = get_videos(f)
1015         self.assertEqual(res, ['2'])
1016
1017         f = match_filter_func('description = foo')
1018         res = get_videos(f)
1019         self.assertEqual(res, ['2'])
1020
1021         f = match_filter_func('description =? foo')
1022         res = get_videos(f)
1023         self.assertEqual(res, ['1', '2'])
1024
1025         f = match_filter_func('filesize > 5KiB')
1026         res = get_videos(f)
1027         self.assertEqual(res, ['1'])
1028
1029         f = match_filter_func('playlist_id = 42')
1030         res = get_videos(f)
1031         self.assertEqual(res, ['1'])
1032
1033         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
1034         res = get_videos(f)
1035         self.assertEqual(res, ['1'])
1036
1037         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
1038         res = get_videos(f)
1039         self.assertEqual(res, ['2'])
1040
1041         f = match_filter_func('creator = "тест \' 123 \' тест--"')
1042         res = get_videos(f)
1043         self.assertEqual(res, ['1'])
1044
1045         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
1046         res = get_videos(f)
1047         self.assertEqual(res, ['1'])
1048
1049         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
1050         res = get_videos(f)
1051         self.assertEqual(res, [])
1052
1053     def test_playlist_items_selection(self):
1054         INDICES, PAGE_SIZE = list(range(1, 11)), 3
1055
1056         def entry(i, evaluated):
1057             evaluated.append(i)
1058             return {
1059                 'id': str(i),
1060                 'title': str(i),
1061                 'url': TEST_URL,
1062             }
1063
1064         def pagedlist_entries(evaluated):
1065             def page_func(n):
1066                 start = PAGE_SIZE * n
1067                 for i in INDICES[start: start + PAGE_SIZE]:
1068                     yield entry(i, evaluated)
1069             return OnDemandPagedList(page_func, PAGE_SIZE)
1070
1071         def page_num(i):
1072             return (i + PAGE_SIZE - 1) // PAGE_SIZE
1073
1074         def generator_entries(evaluated):
1075             for i in INDICES:
1076                 yield entry(i, evaluated)
1077
1078         def list_entries(evaluated):
1079             return list(generator_entries(evaluated))
1080
1081         def lazylist_entries(evaluated):
1082             return LazyList(generator_entries(evaluated))
1083
1084         def get_downloaded_info_dicts(params, entries):
1085             ydl = YDL(params)
1086             ydl.process_ie_result({
1087                 '_type': 'playlist',
1088                 'id': 'test',
1089                 'extractor': 'test:playlist',
1090                 'extractor_key': 'test:playlist',
1091                 'webpage_url': 'http://example.com',
1092                 'entries': entries,
1093             })
1094             return ydl.downloaded_info_dicts
1095
1096         def test_selection(params, expected_ids, evaluate_all=False):
1097             expected_ids = list(expected_ids)
1098             if evaluate_all:
1099                 generator_eval = pagedlist_eval = INDICES
1100             elif not expected_ids:
1101                 generator_eval = pagedlist_eval = []
1102             else:
1103                 generator_eval = INDICES[0: max(expected_ids)]
1104                 pagedlist_eval = INDICES[PAGE_SIZE * page_num(min(expected_ids)) - PAGE_SIZE:
1105                                          PAGE_SIZE * page_num(max(expected_ids))]
1106
1107             for name, func, expected_eval in (
1108                 ('list', list_entries, INDICES),
1109                 ('Generator', generator_entries, generator_eval),
1110                 # ('LazyList', lazylist_entries, generator_eval),  # Generator and LazyList follow the exact same code path
1111                 ('PagedList', pagedlist_entries, pagedlist_eval),
1112             ):
1113                 evaluated = []
1114                 entries = func(evaluated)
1115                 results = [(v['playlist_autonumber'] - 1, (int(v['id']), v['playlist_index']))
1116                            for v in get_downloaded_info_dicts(params, entries)]
1117                 self.assertEqual(results, list(enumerate(zip(expected_ids, expected_ids))), f'Entries of {name} for {params}')
1118                 self.assertEqual(sorted(evaluated), expected_eval, f'Evaluation of {name} for {params}')
1119
1120         test_selection({}, INDICES)
1121         test_selection({'playlistend': 20}, INDICES, True)
1122         test_selection({'playlistend': 2}, INDICES[:2])
1123         test_selection({'playliststart': 11}, [], True)
1124         test_selection({'playliststart': 2}, INDICES[1:])
1125         test_selection({'playlist_items': '2-4'}, INDICES[1:4])
1126         test_selection({'playlist_items': '2,4'}, [2, 4])
1127         test_selection({'playlist_items': '20'}, [], True)
1128         test_selection({'playlist_items': '0'}, [])
1129
1130         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
1131         test_selection({'playlist_items': '2-4,3-4,3'}, [2, 3, 4])
1132         test_selection({'playlist_items': '4,2'}, [4, 2])
1133
1134         # Tests for https://github.com/yt-dlp/yt-dlp/issues/720
1135         # https://github.com/yt-dlp/yt-dlp/issues/302
1136         test_selection({'playlistreverse': True}, INDICES[::-1])
1137         test_selection({'playliststart': 2, 'playlistreverse': True}, INDICES[:0:-1])
1138         test_selection({'playlist_items': '2,4', 'playlistreverse': True}, [4, 2])
1139         test_selection({'playlist_items': '4,2'}, [4, 2])
1140
1141         # Tests for --playlist-items start:end:step
1142         test_selection({'playlist_items': ':'}, INDICES, True)
1143         test_selection({'playlist_items': '::1'}, INDICES, True)
1144         test_selection({'playlist_items': '::-1'}, INDICES[::-1], True)
1145         test_selection({'playlist_items': ':6'}, INDICES[:6])
1146         test_selection({'playlist_items': ':-6'}, INDICES[:-5], True)
1147         test_selection({'playlist_items': '-1:6:-2'}, INDICES[:4:-2], True)
1148         test_selection({'playlist_items': '9:-6:-2'}, INDICES[8:3:-2], True)
1149
1150         test_selection({'playlist_items': '1:inf:2'}, INDICES[::2], True)
1151         test_selection({'playlist_items': '-2:inf'}, INDICES[-2:], True)
1152         test_selection({'playlist_items': ':inf:-1'}, [], True)
1153         test_selection({'playlist_items': '0-2:2'}, [2])
1154         test_selection({'playlist_items': '1-:2'}, INDICES[::2], True)
1155         test_selection({'playlist_items': '0--2:2'}, INDICES[1:-1:2], True)
1156
1157         test_selection({'playlist_items': '10::3'}, [10], True)
1158         test_selection({'playlist_items': '-1::3'}, [10], True)
1159         test_selection({'playlist_items': '11::3'}, [], True)
1160         test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True)
1161         test_selection({'playlist_items': '-15::15'}, [], True)
1162
1163     def test_do_not_override_ie_key_in_url_transparent(self):
1164         ydl = YDL()
1165
1166         class Foo1IE(InfoExtractor):
1167             _VALID_URL = r'foo1:'
1168
1169             def _real_extract(self, url):
1170                 return {
1171                     '_type': 'url_transparent',
1172                     'url': 'foo2:',
1173                     'ie_key': 'Foo2',
1174                     'title': 'foo1 title',
1175                     'id': 'foo1_id',
1176                 }
1177
1178         class Foo2IE(InfoExtractor):
1179             _VALID_URL = r'foo2:'
1180
1181             def _real_extract(self, url):
1182                 return {
1183                     '_type': 'url',
1184                     'url': 'foo3:',
1185                     'ie_key': 'Foo3',
1186                 }
1187
1188         class Foo3IE(InfoExtractor):
1189             _VALID_URL = r'foo3:'
1190
1191             def _real_extract(self, url):
1192                 return _make_result([{'url': TEST_URL}], title='foo3 title')
1193
1194         ydl.add_info_extractor(Foo1IE(ydl))
1195         ydl.add_info_extractor(Foo2IE(ydl))
1196         ydl.add_info_extractor(Foo3IE(ydl))
1197         ydl.extract_info('foo1:')
1198         downloaded = ydl.downloaded_info_dicts[0]
1199         self.assertEqual(downloaded['url'], TEST_URL)
1200         self.assertEqual(downloaded['title'], 'foo1 title')
1201         self.assertEqual(downloaded['id'], 'testid')
1202         self.assertEqual(downloaded['extractor'], 'testex')
1203         self.assertEqual(downloaded['extractor_key'], 'TestEx')
1204
1205     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
1206     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
1207
1208         class _YDL(YDL):
1209             def __init__(self, *args, **kwargs):
1210                 super().__init__(*args, **kwargs)
1211
1212             def trouble(self, s, tb=None):
1213                 pass
1214
1215         ydl = _YDL({
1216             'format': 'extra',
1217             'ignoreerrors': True,
1218         })
1219
1220         class VideoIE(InfoExtractor):
1221             _VALID_URL = r'video:(?P<id>\d+)'
1222
1223             def _real_extract(self, url):
1224                 video_id = self._match_id(url)
1225                 formats = [{
1226                     'format_id': 'default',
1227                     'url': 'url:',
1228                 }]
1229                 if video_id == '0':
1230                     raise ExtractorError('foo')
1231                 if video_id == '2':
1232                     formats.append({
1233                         'format_id': 'extra',
1234                         'url': TEST_URL,
1235                     })
1236                 return {
1237                     'id': video_id,
1238                     'title': f'Video {video_id}',
1239                     'formats': formats,
1240                 }
1241
1242         class PlaylistIE(InfoExtractor):
1243             _VALID_URL = r'playlist:'
1244
1245             def _entries(self):
1246                 for n in range(3):
1247                     video_id = str(n)
1248                     yield {
1249                         '_type': 'url_transparent',
1250                         'ie_key': VideoIE.ie_key(),
1251                         'id': video_id,
1252                         'url': f'video:{video_id}',
1253                         'title': f'Video Transparent {video_id}',
1254                     }
1255
1256             def _real_extract(self, url):
1257                 return self.playlist_result(self._entries())
1258
1259         ydl.add_info_extractor(VideoIE(ydl))
1260         ydl.add_info_extractor(PlaylistIE(ydl))
1261         info = ydl.extract_info('playlist:')
1262         entries = info['entries']
1263         self.assertEqual(len(entries), 3)
1264         self.assertTrue(entries[0] is None)
1265         self.assertTrue(entries[1] is None)
1266         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1267         downloaded = ydl.downloaded_info_dicts[0]
1268         entries[2].pop('requested_downloads', None)
1269         self.assertEqual(entries[2], downloaded)
1270         self.assertEqual(downloaded['url'], TEST_URL)
1271         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1272         self.assertEqual(downloaded['id'], '2')
1273         self.assertEqual(downloaded['extractor'], 'Video')
1274         self.assertEqual(downloaded['extractor_key'], 'Video')
1275
1276     def test_header_cookies(self):
1277         from http.cookiejar import Cookie
1278
1279         ydl = FakeYDL()
1280         ydl.report_warning = lambda *_, **__: None
1281
1282         def cookie(name, value, version=None, domain='', path='', secure=False, expires=None):
1283             return Cookie(
1284                 version or 0, name, value, None, False,
1285                 domain, bool(domain), bool(domain), path, bool(path),
1286                 secure, expires, False, None, None, rest={})
1287
1288         _test_url = 'https://yt.dlp/test'
1289
1290         def test(encoded_cookies, cookies, *, headers=False, round_trip=None, error_re=None):
1291             def _test():
1292                 ydl.cookiejar.clear()
1293                 ydl._load_cookies(encoded_cookies, autoscope=headers)
1294                 if headers:
1295                     ydl._apply_header_cookies(_test_url)
1296                 data = {'url': _test_url}
1297                 ydl._calc_headers(data)
1298                 self.assertCountEqual(
1299                     map(vars, ydl.cookiejar), map(vars, cookies),
1300                     'Extracted cookiejar.Cookie is not the same')
1301                 if not headers:
1302                     self.assertEqual(
1303                         data.get('cookies'), round_trip or encoded_cookies,
1304                         'Cookie is not the same as round trip')
1305                 ydl.__dict__['_YoutubeDL__header_cookies'] = []
1306
1307             with self.subTest(msg=encoded_cookies):
1308                 if not error_re:
1309                     _test()
1310                     return
1311                 with self.assertRaisesRegex(Exception, error_re):
1312                     _test()
1313
1314         test('test=value; Domain=.yt.dlp', [cookie('test', 'value', domain='.yt.dlp')])
1315         test('test=value', [cookie('test', 'value')], error_re=r'Unscoped cookies are not allowed')
1316         test('cookie1=value1; Domain=.yt.dlp; Path=/test; cookie2=value2; Domain=.yt.dlp; Path=/', [
1317             cookie('cookie1', 'value1', domain='.yt.dlp', path='/test'),
1318             cookie('cookie2', 'value2', domain='.yt.dlp', path='/')])
1319         test('test=value; Domain=.yt.dlp; Path=/test; Secure; Expires=9999999999', [
1320             cookie('test', 'value', domain='.yt.dlp', path='/test', secure=True, expires=9999999999)])
1321         test('test="value; "; path=/test; domain=.yt.dlp', [
1322             cookie('test', 'value; ', domain='.yt.dlp', path='/test')],
1323             round_trip='test="value\\073 "; Domain=.yt.dlp; Path=/test')
1324         test('name=; Domain=.yt.dlp', [cookie('name', '', domain='.yt.dlp')],
1325              round_trip='name=""; Domain=.yt.dlp')
1326
1327         test('test=value', [cookie('test', 'value', domain='.yt.dlp')], headers=True)
1328         test('cookie1=value; Domain=.yt.dlp; cookie2=value', [], headers=True, error_re=r'Invalid syntax')
1329         ydl.deprecated_feature = ydl.report_error
1330         test('test=value', [], headers=True, error_re=r'Passing cookies as a header is a potential security risk')
1331
1332     def test_infojson_cookies(self):
1333         TEST_FILE = 'test_infojson_cookies.info.json'
1334         TEST_URL = 'https://example.com/example.mp4'
1335         COOKIES = 'a=b; Domain=.example.com; c=d; Domain=.example.com'
1336         COOKIE_HEADER = {'Cookie': 'a=b; c=d'}
1337
1338         ydl = FakeYDL()
1339         ydl.process_info = lambda x: ydl._write_info_json('test', x, TEST_FILE)
1340
1341         def make_info(info_header_cookies=False, fmts_header_cookies=False, cookies_field=False):
1342             fmt = {'url': TEST_URL}
1343             if fmts_header_cookies:
1344                 fmt['http_headers'] = COOKIE_HEADER
1345             if cookies_field:
1346                 fmt['cookies'] = COOKIES
1347             return _make_result([fmt], http_headers=COOKIE_HEADER if info_header_cookies else None)
1348
1349         def test(initial_info, note):
1350             result = {}
1351             result['processed'] = ydl.process_ie_result(initial_info)
1352             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1353                             msg=f'No cookies set in cookiejar after initial process when {note}')
1354             ydl.cookiejar.clear()
1355             with open(TEST_FILE) as infojson:
1356                 result['loaded'] = ydl.sanitize_info(json.load(infojson), True)
1357             result['final'] = ydl.process_ie_result(result['loaded'].copy(), download=False)
1358             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1359                             msg=f'No cookies set in cookiejar after final process when {note}')
1360             ydl.cookiejar.clear()
1361             for key in ('processed', 'loaded', 'final'):
1362                 info = result[key]
1363                 self.assertIsNone(
1364                     traverse_obj(info, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False),
1365                     msg=f'Cookie header not removed in {key} result when {note}')
1366                 self.assertEqual(
1367                     traverse_obj(info, ((None, ('formats', 0)), 'cookies'), get_all=False), COOKIES,
1368                     msg=f'No cookies field found in {key} result when {note}')
1369
1370         test({'url': TEST_URL, 'http_headers': COOKIE_HEADER, 'id': '1', 'title': 'x'}, 'no formats field')
1371         test(make_info(info_header_cookies=True), 'info_dict header cokies')
1372         test(make_info(fmts_header_cookies=True), 'format header cookies')
1373         test(make_info(info_header_cookies=True, fmts_header_cookies=True), 'info_dict and format header cookies')
1374         test(make_info(info_header_cookies=True, fmts_header_cookies=True, cookies_field=True), 'all cookies fields')
1375         test(make_info(cookies_field=True), 'cookies format field')
1376         test({'url': TEST_URL, 'cookies': COOKIES, 'id': '1', 'title': 'x'}, 'info_dict cookies field only')
1377
1378         try_rm(TEST_FILE)
1379
1380     def test_add_headers_cookie(self):
1381         def check_for_cookie_header(result):
1382             return traverse_obj(result, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False)
1383
1384         ydl = FakeYDL({'http_headers': {'Cookie': 'a=b'}})
1385         ydl._apply_header_cookies(_make_result([])['webpage_url'])  # Scope to input webpage URL: .example.com
1386
1387         fmt = {'url': 'https://example.com/video.mp4'}
1388         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1389         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies in result info_dict')
1390         self.assertEqual(result.get('cookies'), 'a=b; Domain=.example.com', msg='No cookies were set in cookies field')
1391         self.assertIn('a=b', ydl.cookiejar.get_cookie_header(fmt['url']), msg='No cookies were set in cookiejar')
1392
1393         fmt = {'url': 'https://wrong.com/video.mp4'}
1394         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1395         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies for wrong domain')
1396         self.assertFalse(result.get('cookies'), msg='Cookies set in cookies field for wrong domain')
1397         self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain')
1398
1399
1400 if __name__ == '__main__':
1401     unittest.main()