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.extractor import YoutubeIE
  19 from yt_dlp.extractor.common import InfoExtractor
  20 from yt_dlp.postprocessor.common import PostProcessor
  21 from yt_dlp.utils import (
  22     ExtractorError,
  23     LazyList,
  24     OnDemandPagedList,
  25     int_or_none,
  26     match_filter_func,
  27 )
  28 from yt_dlp.utils.traversal import traverse_obj
  29
  30 TEST_URL = 'http://localhost/sample.mp4'
  31
  32
  33 class YDL(FakeYDL):
  34     def __init__(self, *args, **kwargs):
  35         super().__init__(*args, **kwargs)
  36         self.downloaded_info_dicts = []
  37         self.msgs = []
  38
  39     def process_info(self, info_dict):
  40         self.downloaded_info_dicts.append(info_dict.copy())
  41
  42     def to_screen(self, msg, *args, **kwargs):
  43         self.msgs.append(msg)
  44
  45     def dl(self, *args, **kwargs):
  46         assert False, 'Downloader must not be invoked for test_YoutubeDL'
  47
  48
  49 def _make_result(formats, **kwargs):
  50     res = {
  51         'formats': formats,
  52         'id': 'testid',
  53         'title': 'testttitle',
  54         'extractor': 'testex',
  55         'extractor_key': 'TestEx',
  56         'webpage_url': 'http://example.com/watch?v=shenanigans',
  57     }
  58     res.update(**kwargs)
  59     return res
  60
  61
  62 class TestFormatSelection(unittest.TestCase):
  63     def test_prefer_free_formats(self):
  64         # Same resolution => download webm
  65         ydl = YDL()
  66         ydl.params['prefer_free_formats'] = True
  67         formats = [
  68             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  69             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  70         ]
  71         info_dict = _make_result(formats)
  72         ydl.sort_formats(info_dict)
  73         ydl.process_ie_result(info_dict)
  74         downloaded = ydl.downloaded_info_dicts[0]
  75         self.assertEqual(downloaded['ext'], 'webm')
  76
  77         # Different resolution => download best quality (mp4)
  78         ydl = YDL()
  79         ydl.params['prefer_free_formats'] = True
  80         formats = [
  81             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  82             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  83         ]
  84         info_dict['formats'] = formats
  85         ydl.sort_formats(info_dict)
  86         ydl.process_ie_result(info_dict)
  87         downloaded = ydl.downloaded_info_dicts[0]
  88         self.assertEqual(downloaded['ext'], 'mp4')
  89
  90         # No prefer_free_formats => prefer mp4 and webm
  91         ydl = YDL()
  92         ydl.params['prefer_free_formats'] = False
  93         formats = [
  94             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  95             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  96             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  97         ]
  98         info_dict['formats'] = formats
  99         ydl.sort_formats(info_dict)
 100         ydl.process_ie_result(info_dict)
 101         downloaded = ydl.downloaded_info_dicts[0]
 102         self.assertEqual(downloaded['ext'], 'mp4')
 103
 104         ydl = YDL()
 105         ydl.params['prefer_free_formats'] = False
 106         formats = [
 107             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
 108             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 109         ]
 110         info_dict['formats'] = formats
 111         ydl.sort_formats(info_dict)
 112         ydl.process_ie_result(info_dict)
 113         downloaded = ydl.downloaded_info_dicts[0]
 114         self.assertEqual(downloaded['ext'], 'webm')
 115
 116     def test_format_selection(self):
 117         formats = [
 118             {'format_id': '35', 'ext': 'mp4', 'preference': 0, 'url': TEST_URL},
 119             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 120             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 121             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 122             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 123         ]
 124         info_dict = _make_result(formats)
 125
 126         def test(inp, *expected, multi=False):
 127             ydl = YDL({
 128                 'format': inp,
 129                 'allow_multiple_video_streams': multi,
 130                 'allow_multiple_audio_streams': multi,
 131             })
 132             ydl.process_ie_result(info_dict.copy())
 133             downloaded = [x['format_id'] for x in ydl.downloaded_info_dicts]
 134             self.assertEqual(downloaded, list(expected))
 135
 136         test('20/47', '47')
 137         test('20/71/worst', '35')
 138         test(None, '2')
 139         test('webm/mp4', '47')
 140         test('3gp/40/mp4', '35')
 141         test('example-with-dashes', 'example-with-dashes')
 142         test('all', '2', '47', '45', 'example-with-dashes', '35')
 143         test('mergeall', '2+47+45+example-with-dashes+35', multi=True)
 144         # See: https://github.com/yt-dlp/yt-dlp/pulls/8797
 145         test('7_a/worst', '35')
 146
 147     def test_format_selection_audio(self):
 148         formats = [
 149             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 150             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 151             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 152             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 153         ]
 154         info_dict = _make_result(formats)
 155
 156         ydl = YDL({'format': 'bestaudio'})
 157         ydl.process_ie_result(info_dict.copy())
 158         downloaded = ydl.downloaded_info_dicts[0]
 159         self.assertEqual(downloaded['format_id'], 'audio-high')
 160
 161         ydl = YDL({'format': 'worstaudio'})
 162         ydl.process_ie_result(info_dict.copy())
 163         downloaded = ydl.downloaded_info_dicts[0]
 164         self.assertEqual(downloaded['format_id'], 'audio-low')
 165
 166         formats = [
 167             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 168             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 169         ]
 170         info_dict = _make_result(formats)
 171
 172         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 173         ydl.process_ie_result(info_dict.copy())
 174         downloaded = ydl.downloaded_info_dicts[0]
 175         self.assertEqual(downloaded['format_id'], 'vid-high')
 176
 177     def test_format_selection_audio_exts(self):
 178         formats = [
 179             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 180             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 181             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 182             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 184         ]
 185
 186         info_dict = _make_result(formats)
 187         ydl = YDL({'format': 'best', 'format_sort': ['abr', 'ext']})
 188         ydl.sort_formats(info_dict)
 189         ydl.process_ie_result(copy.deepcopy(info_dict))
 190         downloaded = ydl.downloaded_info_dicts[0]
 191         self.assertEqual(downloaded['format_id'], 'aac-64')
 192
 193         ydl = YDL({'format': 'mp3'})
 194         ydl.sort_formats(info_dict)
 195         ydl.process_ie_result(copy.deepcopy(info_dict))
 196         downloaded = ydl.downloaded_info_dicts[0]
 197         self.assertEqual(downloaded['format_id'], 'mp3-64')
 198
 199         ydl = YDL({'prefer_free_formats': True, 'format_sort': ['abr', 'ext']})
 200         ydl.sort_formats(info_dict)
 201         ydl.process_ie_result(copy.deepcopy(info_dict))
 202         downloaded = ydl.downloaded_info_dicts[0]
 203         self.assertEqual(downloaded['format_id'], 'ogg-64')
 204
 205     def test_format_selection_video(self):
 206         formats = [
 207             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 208             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 209             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 210         ]
 211         info_dict = _make_result(formats)
 212
 213         ydl = YDL({'format': 'bestvideo'})
 214         ydl.process_ie_result(info_dict.copy())
 215         downloaded = ydl.downloaded_info_dicts[0]
 216         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 217
 218         ydl = YDL({'format': 'worstvideo'})
 219         ydl.process_ie_result(info_dict.copy())
 220         downloaded = ydl.downloaded_info_dicts[0]
 221         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 222
 223         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 224         ydl.process_ie_result(info_dict.copy())
 225         downloaded = ydl.downloaded_info_dicts[0]
 226         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 227
 228         formats = [
 229             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 230         ]
 231         info_dict = _make_result(formats)
 232
 233         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 234         ydl.process_ie_result(info_dict.copy())
 235         downloaded = ydl.downloaded_info_dicts[0]
 236         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 237
 238     def test_format_selection_by_vcodec_sort(self):
 239         formats = [
 240             {'format_id': 'av1-format', 'ext': 'mp4', 'vcodec': 'av1', 'acodec': 'none', 'url': TEST_URL},
 241             {'format_id': 'vp9-hdr-format', 'ext': 'mp4', 'vcodec': 'vp09.02.50.10.01.09.18.09.00', 'acodec': 'none', 'url': TEST_URL},
 242             {'format_id': 'vp9-sdr-format', 'ext': 'mp4', 'vcodec': 'vp09.00.50.08', 'acodec': 'none', 'url': TEST_URL},
 243             {'format_id': 'h265-format', 'ext': 'mp4', 'vcodec': 'h265', 'acodec': 'none', 'url': TEST_URL},
 244         ]
 245         info_dict = _make_result(formats)
 246
 247         ydl = YDL({'format': 'bestvideo', 'format_sort': ['vcodec:vp9.2']})
 248         ydl.process_ie_result(info_dict.copy())
 249         downloaded = ydl.downloaded_info_dicts[0]
 250         self.assertEqual(downloaded['format_id'], 'vp9-hdr-format')
 251
 252         ydl = YDL({'format': 'bestvideo', 'format_sort': ['vcodec:vp9']})
 253         ydl.process_ie_result(info_dict.copy())
 254         downloaded = ydl.downloaded_info_dicts[0]
 255         self.assertEqual(downloaded['format_id'], 'vp9-sdr-format')
 256
 257         ydl = YDL({'format': 'bestvideo', 'format_sort': ['+vcodec:vp9.2']})
 258         ydl.process_ie_result(info_dict.copy())
 259         downloaded = ydl.downloaded_info_dicts[0]
 260         self.assertEqual(downloaded['format_id'], 'vp9-hdr-format')
 261
 262         ydl = YDL({'format': 'bestvideo', 'format_sort': ['+vcodec:vp9']})
 263         ydl.process_ie_result(info_dict.copy())
 264         downloaded = ydl.downloaded_info_dicts[0]
 265         self.assertEqual(downloaded['format_id'], 'vp9-sdr-format')
 266
 267     def test_format_selection_string_ops(self):
 268         formats = [
 269             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 270             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 271         ]
 272         info_dict = _make_result(formats)
 273
 274         # equals (=)
 275         ydl = YDL({'format': '[format_id=abc-cba]'})
 276         ydl.process_ie_result(info_dict.copy())
 277         downloaded = ydl.downloaded_info_dicts[0]
 278         self.assertEqual(downloaded['format_id'], 'abc-cba')
 279
 280         # does not equal (!=)
 281         ydl = YDL({'format': '[format_id!=abc-cba]'})
 282         ydl.process_ie_result(info_dict.copy())
 283         downloaded = ydl.downloaded_info_dicts[0]
 284         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 285
 286         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 287         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 288
 289         # starts with (^=)
 290         ydl = YDL({'format': '[format_id^=abc]'})
 291         ydl.process_ie_result(info_dict.copy())
 292         downloaded = ydl.downloaded_info_dicts[0]
 293         self.assertEqual(downloaded['format_id'], 'abc-cba')
 294
 295         # does not start with (!^=)
 296         ydl = YDL({'format': '[format_id!^=abc]'})
 297         ydl.process_ie_result(info_dict.copy())
 298         downloaded = ydl.downloaded_info_dicts[0]
 299         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 300
 301         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 302         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 303
 304         # ends with ($=)
 305         ydl = YDL({'format': '[format_id$=cba]'})
 306         ydl.process_ie_result(info_dict.copy())
 307         downloaded = ydl.downloaded_info_dicts[0]
 308         self.assertEqual(downloaded['format_id'], 'abc-cba')
 309
 310         # does not end with (!$=)
 311         ydl = YDL({'format': '[format_id!$=cba]'})
 312         ydl.process_ie_result(info_dict.copy())
 313         downloaded = ydl.downloaded_info_dicts[0]
 314         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 315
 316         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 317         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 318
 319         # contains (*=)
 320         ydl = YDL({'format': '[format_id*=bc-cb]'})
 321         ydl.process_ie_result(info_dict.copy())
 322         downloaded = ydl.downloaded_info_dicts[0]
 323         self.assertEqual(downloaded['format_id'], 'abc-cba')
 324
 325         # does not contain (!*=)
 326         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 327         ydl.process_ie_result(info_dict.copy())
 328         downloaded = ydl.downloaded_info_dicts[0]
 329         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 330
 331         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 332         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 333
 334         ydl = YDL({'format': '[format_id!*=-]'})
 335         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 336
 337     def test_youtube_format_selection(self):
 338         # FIXME: Rewrite in accordance with the new format sorting options
 339         return
 340
 341         order = [
 342             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 343             # Apple HTTP Live Streaming
 344             '96', '95', '94', '93', '92', '132', '151',
 345             # 3D
 346             '85', '84', '102', '83', '101', '82', '100',
 347             # Dash video
 348             '137', '248', '136', '247', '135', '246',
 349             '245', '244', '134', '243', '133', '242', '160',
 350             # Dash audio
 351             '141', '172', '140', '171', '139',
 352         ]
 353
 354         def format_info(f_id):
 355             info = YoutubeIE._formats[f_id].copy()
 356
 357             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 358             # and 'vcodec', while in tests such information is incomplete since
 359             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 360             # test_YoutubeDL.test_youtube_format_selection is broken without
 361             # this fix
 362             if 'acodec' in info and 'vcodec' not in info:
 363                 info['vcodec'] = 'none'
 364             elif 'vcodec' in info and 'acodec' not in info:
 365                 info['acodec'] = 'none'
 366
 367             info['format_id'] = f_id
 368             info['url'] = 'url:' + f_id
 369             return info
 370         formats_order = [format_info(f_id) for f_id in order]
 371
 372         info_dict = _make_result(list(formats_order), extractor='youtube')
 373         ydl = YDL({'format': 'bestvideo+bestaudio'})
 374         ydl.sort_formats(info_dict)
 375         ydl.process_ie_result(info_dict)
 376         downloaded = ydl.downloaded_info_dicts[0]
 377         self.assertEqual(downloaded['format_id'], '248+172')
 378         self.assertEqual(downloaded['ext'], 'mp4')
 379
 380         info_dict = _make_result(list(formats_order), extractor='youtube')
 381         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 382         ydl.sort_formats(info_dict)
 383         ydl.process_ie_result(info_dict)
 384         downloaded = ydl.downloaded_info_dicts[0]
 385         self.assertEqual(downloaded['format_id'], '38')
 386
 387         info_dict = _make_result(list(formats_order), extractor='youtube')
 388         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 389         ydl.sort_formats(info_dict)
 390         ydl.process_ie_result(info_dict)
 391         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 392         self.assertEqual(downloaded_ids, ['137', '141'])
 393
 394         info_dict = _make_result(list(formats_order), extractor='youtube')
 395         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 396         ydl.sort_formats(info_dict)
 397         ydl.process_ie_result(info_dict)
 398         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 399         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 400
 401         info_dict = _make_result(list(formats_order), extractor='youtube')
 402         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 403         ydl.sort_formats(info_dict)
 404         ydl.process_ie_result(info_dict)
 405         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 406         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 407
 408         info_dict = _make_result(list(formats_order), extractor='youtube')
 409         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 410         ydl.sort_formats(info_dict)
 411         ydl.process_ie_result(info_dict)
 412         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 413         self.assertEqual(downloaded_ids, ['248+141'])
 414
 415         for f1, f2 in zip(formats_order, formats_order[1:]):
 416             info_dict = _make_result([f1, f2], extractor='youtube')
 417             ydl = YDL({'format': 'best/bestvideo'})
 418             ydl.sort_formats(info_dict)
 419             ydl.process_ie_result(info_dict)
 420             downloaded = ydl.downloaded_info_dicts[0]
 421             self.assertEqual(downloaded['format_id'], f1['format_id'])
 422
 423             info_dict = _make_result([f2, f1], extractor='youtube')
 424             ydl = YDL({'format': 'best/bestvideo'})
 425             ydl.sort_formats(info_dict)
 426             ydl.process_ie_result(info_dict)
 427             downloaded = ydl.downloaded_info_dicts[0]
 428             self.assertEqual(downloaded['format_id'], f1['format_id'])
 429
 430     def test_audio_only_extractor_format_selection(self):
 431         # For extractors with incomplete formats (all formats are audio-only or
 432         # video-only) best and worst should fallback to corresponding best/worst
 433         # video-only or audio-only formats (as per
 434         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 435         formats = [
 436             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 437             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 438         ]
 439         info_dict = _make_result(formats)
 440
 441         ydl = YDL({'format': 'best'})
 442         ydl.process_ie_result(info_dict.copy())
 443         downloaded = ydl.downloaded_info_dicts[0]
 444         self.assertEqual(downloaded['format_id'], 'high')
 445
 446         ydl = YDL({'format': 'worst'})
 447         ydl.process_ie_result(info_dict.copy())
 448         downloaded = ydl.downloaded_info_dicts[0]
 449         self.assertEqual(downloaded['format_id'], 'low')
 450
 451     def test_format_not_available(self):
 452         formats = [
 453             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 454             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 455         ]
 456         info_dict = _make_result(formats)
 457
 458         # This must fail since complete video-audio format does not match filter
 459         # and extractor does not provide incomplete only formats (i.e. only
 460         # video-only or audio-only).
 461         ydl = YDL({'format': 'best[height>360]'})
 462         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 463
 464     def test_format_selection_issue_10083(self):
 465         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 466         formats = [
 467             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 468             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 469             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 470         ]
 471         info_dict = _make_result(formats)
 472
 473         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 474         ydl.process_ie_result(info_dict.copy())
 475         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 476
 477     def test_invalid_format_specs(self):
 478         def assert_syntax_error(format_spec):
 479             self.assertRaises(SyntaxError, YDL, {'format': format_spec})
 480
 481         assert_syntax_error('bestvideo,,best')
 482         assert_syntax_error('+bestaudio')
 483         assert_syntax_error('bestvideo+')
 484         assert_syntax_error('/')
 485         assert_syntax_error('[720<height]')
 486
 487     def test_format_filtering(self):
 488         formats = [
 489             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 490             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 491             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 492             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 493             {'format_id': 'E', 'filesize': 3000},
 494             {'format_id': 'F'},
 495             {'format_id': 'G', 'filesize': 1000000},
 496         ]
 497         for f in formats:
 498             f['url'] = 'http://_/'
 499             f['ext'] = 'unknown'
 500         info_dict = _make_result(formats, _format_sort_fields=('id', ))
 501
 502         ydl = YDL({'format': 'best[filesize<3000]'})
 503         ydl.process_ie_result(info_dict)
 504         downloaded = ydl.downloaded_info_dicts[0]
 505         self.assertEqual(downloaded['format_id'], 'D')
 506
 507         ydl = YDL({'format': 'best[filesize<=3000]'})
 508         ydl.process_ie_result(info_dict)
 509         downloaded = ydl.downloaded_info_dicts[0]
 510         self.assertEqual(downloaded['format_id'], 'E')
 511
 512         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 513         ydl.process_ie_result(info_dict)
 514         downloaded = ydl.downloaded_info_dicts[0]
 515         self.assertEqual(downloaded['format_id'], 'F')
 516
 517         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 518         ydl.process_ie_result(info_dict)
 519         downloaded = ydl.downloaded_info_dicts[0]
 520         self.assertEqual(downloaded['format_id'], 'B')
 521
 522         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 523         ydl.process_ie_result(info_dict)
 524         downloaded = ydl.downloaded_info_dicts[0]
 525         self.assertEqual(downloaded['format_id'], 'C')
 526
 527         ydl = YDL({'format': '[filesize>?1]'})
 528         ydl.process_ie_result(info_dict)
 529         downloaded = ydl.downloaded_info_dicts[0]
 530         self.assertEqual(downloaded['format_id'], 'G')
 531
 532         ydl = YDL({'format': '[filesize<1M]'})
 533         ydl.process_ie_result(info_dict)
 534         downloaded = ydl.downloaded_info_dicts[0]
 535         self.assertEqual(downloaded['format_id'], 'E')
 536
 537         ydl = YDL({'format': '[filesize<1MiB]'})
 538         ydl.process_ie_result(info_dict)
 539         downloaded = ydl.downloaded_info_dicts[0]
 540         self.assertEqual(downloaded['format_id'], 'G')
 541
 542         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 543         ydl.process_ie_result(info_dict)
 544         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 545         self.assertEqual(downloaded_ids, ['D', 'C', 'B'])
 546
 547         ydl = YDL({'format': 'best[height<40]'})
 548         with contextlib.suppress(ExtractorError):
 549             ydl.process_ie_result(info_dict)
 550         self.assertEqual(ydl.downloaded_info_dicts, [])
 551
 552     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.available', False)
 553     def test_default_format_spec_without_ffmpeg(self):
 554         ydl = YDL({})
 555         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 556
 557         ydl = YDL({'simulate': True})
 558         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 559
 560         ydl = YDL({})
 561         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 562
 563         ydl = YDL({'simulate': True})
 564         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 565
 566         ydl = YDL({'outtmpl': '-'})
 567         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 568
 569         ydl = YDL({})
 570         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 571         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 572
 573     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.available', True)
 574     @patch('yt_dlp.postprocessor.ffmpeg.FFmpegMergerPP.can_merge', lambda _: True)
 575     def test_default_format_spec_with_ffmpeg(self):
 576         ydl = YDL({})
 577         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 578
 579         ydl = YDL({'simulate': True})
 580         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 581
 582         ydl = YDL({})
 583         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 584
 585         ydl = YDL({'simulate': True})
 586         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 587
 588         ydl = YDL({'outtmpl': '-'})
 589         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 590
 591         ydl = YDL({})
 592         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 593         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 594
 595
 596 class TestYoutubeDL(unittest.TestCase):
 597     def test_subtitles(self):
 598         def s_formats(lang, autocaption=False):
 599             return [{
 600                 'ext': ext,
 601                 'url': f'http://localhost/video.{lang}.{ext}',
 602                 '_auto': autocaption,
 603             } for ext in ['vtt', 'srt', 'ass']]
 604         subtitles = {l: s_formats(l) for l in ['en', 'fr', 'es']}
 605         auto_captions = {l: s_formats(l, True) for l in ['it', 'pt', 'es']}
 606         info_dict = {
 607             'id': 'test',
 608             'title': 'Test',
 609             'url': 'http://localhost/video.mp4',
 610             'subtitles': subtitles,
 611             'automatic_captions': auto_captions,
 612             'extractor': 'TEST',
 613             'webpage_url': 'http://example.com/watch?v=shenanigans',
 614         }
 615
 616         def get_info(params={}):
 617             params.setdefault('simulate', True)
 618             ydl = YDL(params)
 619             ydl.report_warning = lambda *args, **kargs: None
 620             return ydl.process_video_result(info_dict, download=False)
 621
 622         result = get_info()
 623         self.assertFalse(result.get('requested_subtitles'))
 624         self.assertEqual(result['subtitles'], subtitles)
 625         self.assertEqual(result['automatic_captions'], auto_captions)
 626
 627         result = get_info({'writesubtitles': True})
 628         subs = result['requested_subtitles']
 629         self.assertTrue(subs)
 630         self.assertEqual(set(subs.keys()), {'en'})
 631         self.assertTrue(subs['en'].get('data') is None)
 632         self.assertEqual(subs['en']['ext'], 'ass')
 633
 634         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 635         subs = result['requested_subtitles']
 636         self.assertEqual(subs['en']['ext'], 'srt')
 637
 638         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 639         subs = result['requested_subtitles']
 640         self.assertTrue(subs)
 641         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 642
 643         result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']})
 644         subs = result['requested_subtitles']
 645         self.assertTrue(subs)
 646         self.assertEqual(set(subs.keys()), {'es', 'fr'})
 647
 648         result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']})
 649         subs = result['requested_subtitles']
 650         self.assertTrue(subs)
 651         self.assertEqual(set(subs.keys()), {'fr'})
 652
 653         result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']})
 654         subs = result['requested_subtitles']
 655         self.assertTrue(subs)
 656         self.assertEqual(set(subs.keys()), {'en'})
 657
 658         result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']})
 659         subs = result['requested_subtitles']
 660         self.assertTrue(subs)
 661         self.assertEqual(set(subs.keys()), {'es', 'en'})
 662
 663         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 664         subs = result['requested_subtitles']
 665         self.assertTrue(subs)
 666         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 667         self.assertFalse(subs['es']['_auto'])
 668         self.assertTrue(subs['pt']['_auto'])
 669
 670         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 671         subs = result['requested_subtitles']
 672         self.assertTrue(subs)
 673         self.assertEqual(set(subs.keys()), {'es', 'pt'})
 674         self.assertTrue(subs['es']['_auto'])
 675         self.assertTrue(subs['pt']['_auto'])
 676
 677     def test_add_extra_info(self):
 678         test_dict = {
 679             'extractor': 'Foo',
 680         }
 681         extra_info = {
 682             'extractor': 'Bar',
 683             'playlist': 'funny videos',
 684         }
 685         YDL.add_extra_info(test_dict, extra_info)
 686         self.assertEqual(test_dict['extractor'], 'Foo')
 687         self.assertEqual(test_dict['playlist'], 'funny videos')
 688
 689     outtmpl_info = {
 690         'id': '1234',
 691         'ext': 'mp4',
 692         'width': None,
 693         'height': 1080,
 694         'filesize': 1024,
 695         'title1': '$PATH',
 696         'title2': '%PATH%',
 697         'title3': 'foo/bar\\test',
 698         'title4': 'foo "bar" test',
 699         'title5': 'áéí 𝐀',
 700         'timestamp': 1618488000,
 701         'duration': 100000,
 702         'playlist_index': 1,
 703         'playlist_autonumber': 2,
 704         '__last_playlist_index': 100,
 705         'n_entries': 10,
 706         'formats': [
 707             {'id': 'id 1', 'height': 1080, 'width': 1920},
 708             {'id': 'id 2', 'height': 720},
 709             {'id': 'id 3'},
 710         ],
 711     }
 712
 713     def test_prepare_outtmpl_and_filename(self):
 714         def test(tmpl, expected, *, info=None, **params):
 715             params['outtmpl'] = tmpl
 716             ydl = FakeYDL(params)
 717             ydl._num_downloads = 1
 718             self.assertEqual(ydl.validate_outtmpl(tmpl), None)
 719
 720             out = ydl.evaluate_outtmpl(tmpl, info or self.outtmpl_info)
 721             fname = ydl.prepare_filename(info or self.outtmpl_info)
 722
 723             if not isinstance(expected, (list, tuple)):
 724                 expected = (expected, expected)
 725             for (name, got), expect in zip((('outtmpl', out), ('filename', fname)), expected):
 726                 if callable(expect):
 727                     self.assertTrue(expect(got), f'Wrong {name} from {tmpl}')
 728                 elif expect is not None:
 729                     self.assertEqual(got, expect, f'Wrong {name} from {tmpl}')
 730
 731         # Side-effects
 732         original_infodict = dict(self.outtmpl_info)
 733         test('foo.bar', 'foo.bar')
 734         original_infodict['epoch'] = self.outtmpl_info.get('epoch')
 735         self.assertTrue(isinstance(original_infodict['epoch'], int))
 736         test('%(epoch)d', int_or_none)
 737         self.assertEqual(original_infodict, self.outtmpl_info)
 738
 739         # Auto-generated fields
 740         test('%(id)s.%(ext)s', '1234.mp4')
 741         test('%(duration_string)s', ('27:46:40', '27-46-40'))
 742         test('%(resolution)s', '1080p')
 743         test('%(playlist_index|)s', '001')
 744         test('%(playlist_index&{}!)s', '1!')
 745         test('%(playlist_autonumber)s', '02')
 746         test('%(autonumber)s', '00001')
 747         test('%(autonumber+2)03d', '005', autonumber_start=3)
 748         test('%(autonumber)s', '001', autonumber_size=3)
 749
 750         # Escaping %
 751         test('%', '%')
 752         test('%%', '%')
 753         test('%%%%', '%%')
 754         test('%s', '%s')
 755         test('%%%s', '%%s')
 756         test('%d', '%d')
 757         test('%abc%', '%abc%')
 758         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 759         test('%%%(height)s', '%1080')
 760         test('%(width)06d.%(ext)s', 'NA.mp4')
 761         test('%(width)06d.%%(ext)s', 'NA.%(ext)s')
 762         test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
 763
 764         # ID sanitization
 765         test('%(id)s', '_abcd', info={'id': '_abcd'})
 766         test('%(some_id)s', '_abcd', info={'some_id': '_abcd'})
 767         test('%(formats.0.id)s', '_abcd', info={'formats': [{'id': '_abcd'}]})
 768         test('%(id)s', '-abcd', info={'id': '-abcd'})
 769         test('%(id)s', '.abcd', info={'id': '.abcd'})
 770         test('%(id)s', 'ab__cd', info={'id': 'ab__cd'})
 771         test('%(id)s', ('ab:cd', 'ab：cd'), info={'id': 'ab:cd'})
 772         test('%(id.0)s', '-', info={'id': '--'})
 773
 774         # Invalid templates
 775         self.assertTrue(isinstance(YoutubeDL.validate_outtmpl('%(title)'), ValueError))
 776         test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder='none')
 777         test('%(..)s', 'NA')
 778         test('%(formats.{id)s', 'NA')
 779
 780         # Entire info_dict
 781         def expect_same_infodict(out):
 782             got_dict = json.loads(out)
 783             for info_field, expected in self.outtmpl_info.items():
 784                 self.assertEqual(got_dict.get(info_field), expected, info_field)
 785             return True
 786
 787         test('%()j', (expect_same_infodict, None))
 788
 789         # NA placeholder
 790         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s'
 791         test(NA_TEST_OUTTMPL, 'NA-NA-def-1234.mp4')
 792         test(NA_TEST_OUTTMPL, 'none-none-def-1234.mp4', outtmpl_na_placeholder='none')
 793         test(NA_TEST_OUTTMPL, '--def-1234.mp4', outtmpl_na_placeholder='')
 794         test('%(non_existent.0)s', 'NA')
 795
 796         # String formatting
 797         FMT_TEST_OUTTMPL = '%%(height)%s.%%(ext)s'
 798         test(FMT_TEST_OUTTMPL % 's', '1080.mp4')
 799         test(FMT_TEST_OUTTMPL % 'd', '1080.mp4')
 800         test(FMT_TEST_OUTTMPL % '6d', '  1080.mp4')
 801         test(FMT_TEST_OUTTMPL % '-6d', '1080  .mp4')
 802         test(FMT_TEST_OUTTMPL % '06d', '001080.mp4')
 803         test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4')
 804         test(FMT_TEST_OUTTMPL % '   06d', ' 01080.mp4')
 805         test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4')
 806         test(FMT_TEST_OUTTMPL % '0   6d', ' 01080.mp4')
 807         test(FMT_TEST_OUTTMPL % '   0   6d', ' 01080.mp4')
 808
 809         # Type casting
 810         test('%(id)d', '1234')
 811         test('%(height)c', '1')
 812         test('%(ext)c', 'm')
 813         test('%(id)d %(id)r', "1234 '1234'")
 814         test('%(id)r %(height)r', "'1234' 1080")
 815         test('%(title5)a %(height)a', (R"'\xe1\xe9\xed \U0001d400' 1080", None))
 816         test('%(ext)s-%(ext|def)d', 'mp4-def')
 817         test('%(width|0)04d', '0')
 818         test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none')
 819
 820         FORMATS = self.outtmpl_info['formats']
 821
 822         # Custom type casting
 823         test('%(formats.:.id)l', 'id 1, id 2, id 3')
 824         test('%(formats.:.id)#l', ('id 1\nid 2\nid 3', 'id 1 id 2 id 3'))
 825         test('%(ext)l', 'mp4')
 826         test('%(formats.:.id) 18l', '  id 1, id 2, id 3')
 827         test('%(formats)j', (json.dumps(FORMATS), None))
 828         test('%(formats)#j', (
 829             json.dumps(FORMATS, indent=4),
 830             json.dumps(FORMATS, indent=4).replace(':', '：').replace('"', '＂').replace('\n', ' '),
 831         ))
 832         test('%(title5).3B', 'á')
 833         test('%(title5)U', 'áéí 𝐀')
 834         test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀')
 835         test('%(title5)+U', 'áéí A')
 836         test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A')
 837         test('%(height)D', '1k')
 838         test('%(filesize)#D', '1Ki')
 839         test('%(height)5.2D', ' 1.08k')
 840         test('%(title4)#S', 'foo_bar_test')
 841         test('%(title4).10S', ('foo ＂bar＂ ', 'foo ＂bar＂' + ('#' if os.name == 'nt' else ' ')))
 842         if os.name == 'nt':
 843             test('%(title4)q', ('"foo ""bar"" test"', None))
 844             test('%(formats.:.id)#q', ('"id 1" "id 2" "id 3"', None))
 845             test('%(formats.0.id)#q', ('"id 1"', None))
 846         else:
 847             test('%(title4)q', ('\'foo "bar" test\'', '\'foo ＂bar＂ test\''))
 848             test('%(formats.:.id)#q', "'id 1' 'id 2' 'id 3'")
 849             test('%(formats.0.id)#q', "'id 1'")
 850
 851         # Internal formatting
 852         test('%(timestamp-1000>%H-%M-%S)s', '11-43-20')
 853         test('%(title|%)s %(title|%%)s', '% %%')
 854         test('%(id+1-height+3)05d', '00158')
 855         test('%(width+100)05d', 'NA')
 856         test('%(filesize*8)d', '8192')
 857         test('%(formats.0) 15s', ('% 15s' % FORMATS[0], None))
 858         test('%(formats.0)r', (repr(FORMATS[0]), None))
 859         test('%(height.0)03d', '001')
 860         test('%(-height.0)04d', '-001')
 861         test('%(formats.-1.id)s', FORMATS[-1]['id'])
 862         test('%(formats.0.id.-1)d', FORMATS[0]['id'][-1])
 863         test('%(formats.3)s', 'NA')
 864         test('%(formats.:2:-1)r', repr(FORMATS[:2:-1]))
 865         test('%(formats.0.id.-1+id)f', '1235.000000')
 866         test('%(formats.0.id.-1+formats.1.id.-1)d', '3')
 867         out = json.dumps([{'id': f['id'], 'height.:2': str(f['height'])[:2]}
 868                           if 'height' in f else {'id': f['id']}
 869                           for f in FORMATS])
 870         test('%(formats.:.{id,height.:2})j', (out, None))
 871         test('%(formats.:.{id,height}.id)l', ', '.join(f['id'] for f in FORMATS))
 872         test('%(.{id,title})j', ('{"id": "1234"}', '{＂id＂： ＂1234＂}'))
 873
 874         # Alternates
 875         test('%(title,id)s', '1234')
 876         test('%(width-100,height+20|def)d', '1100')
 877         test('%(width-100,height+width|def)s', 'def')
 878         test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00')
 879
 880         # Replacement
 881         test('%(id&foo)s.bar', 'foo.bar')
 882         test('%(title&foo)s.bar', 'NA.bar')
 883         test('%(title&foo|baz)s.bar', 'baz.bar')
 884         test('%(x,id&foo|baz)s.bar', 'foo.bar')
 885         test('%(x,title&foo|baz)s.bar', 'baz.bar')
 886         test('%(id&a\nb|)s', ('a\nb', 'a b'))
 887         test('%(id&hi {:>10} {}|)s', 'hi       1234 1234')
 888         test(R'%(id&{0} {}|)s', 'NA')
 889         test(R'%(id&{0.1}|)s', 'NA')
 890         test('%(height&{:,d})S', '1,080')
 891
 892         # Laziness
 893         def gen():
 894             yield from range(5)
 895             raise self.assertTrue(False, 'LazyList should not be evaluated till here')
 896         test('%(key.4)s', '4', info={'key': LazyList(gen())})
 897
 898         # Empty filename
 899         test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4')
 900         # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4'))  # FIXME: ?
 901         # test('%(foo|)s', ('', '_'))  # FIXME: ?
 902
 903         # Environment variable expansion for prepare_filename
 904         os.environ['__yt_dlp_var'] = 'expanded'
 905         envvar = '%__yt_dlp_var%' if os.name == 'nt' else '$__yt_dlp_var'
 906         test(envvar, (envvar, 'expanded'))
 907         if os.name == 'nt':
 908             test('%s%', ('%s%', '%s%'))
 909             os.environ['s'] = 'expanded'
 910             test('%s%', ('%s%', 'expanded'))  # %s% should be expanded before escaping %s
 911             os.environ['(test)s'] = 'expanded'
 912             test('%(test)s%', ('NA%', 'expanded'))  # Environment should take priority over template
 913
 914         # Path expansion and escaping
 915         test('Hello %(title1)s', 'Hello $PATH')
 916         test('Hello %(title2)s', 'Hello %PATH%')
 917         test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test'))
 918         test('folder/%(title3)s', ('folder/foo/bar\\test', f'folder{os.path.sep}foo⧸bar⧹test'))
 919
 920     def test_format_note(self):
 921         ydl = YoutubeDL()
 922         self.assertEqual(ydl._format_note({}), '')
 923         assertRegexpMatches(self, ydl._format_note({
 924             'vbr': 10,
 925         }), r'^\s*10k$')
 926         assertRegexpMatches(self, ydl._format_note({
 927             'fps': 30,
 928         }), r'^30fps$')
 929
 930     def test_postprocessors(self):
 931         filename = 'post-processor-testfile.mp4'
 932         audiofile = filename + '.mp3'
 933
 934         class SimplePP(PostProcessor):
 935             def run(self, info):
 936                 with open(audiofile, 'w') as f:
 937                     f.write('EXAMPLE')
 938                 return [info['filepath']], info
 939
 940         def run_pp(params, pp):
 941             with open(filename, 'w') as f:
 942                 f.write('EXAMPLE')
 943             ydl = YoutubeDL(params)
 944             ydl.add_post_processor(pp())
 945             ydl.post_process(filename, {'filepath': filename})
 946
 947         run_pp({'keepvideo': True}, SimplePP)
 948         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 949         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 950         os.unlink(filename)
 951         os.unlink(audiofile)
 952
 953         run_pp({'keepvideo': False}, SimplePP)
 954         self.assertFalse(os.path.exists(filename), f'{filename} exists')
 955         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 956         os.unlink(audiofile)
 957
 958         class ModifierPP(PostProcessor):
 959             def run(self, info):
 960                 with open(info['filepath'], 'w') as f:
 961                     f.write('MODIFIED')
 962                 return [], info
 963
 964         run_pp({'keepvideo': False}, ModifierPP)
 965         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 966         os.unlink(filename)
 967
 968     def test_match_filter(self):
 969         first = {
 970             'id': '1',
 971             'url': TEST_URL,
 972             'title': 'one',
 973             'extractor': 'TEST',
 974             'duration': 30,
 975             'filesize': 10 * 1024,
 976             'playlist_id': '42',
 977             'uploader': '變態妍字幕版 太妍 тест',
 978             'creator': "тест ' 123 ' тест--",
 979             'webpage_url': 'http://example.com/watch?v=shenanigans',
 980         }
 981         second = {
 982             'id': '2',
 983             'url': TEST_URL,
 984             'title': 'two',
 985             'extractor': 'TEST',
 986             'duration': 10,
 987             'description': 'foo',
 988             'filesize': 5 * 1024,
 989             'playlist_id': '43',
 990             'uploader': 'тест 123',
 991             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 992         }
 993         videos = [first, second]
 994
 995         def get_videos(filter_=None):
 996             ydl = YDL({'match_filter': filter_, 'simulate': True})
 997             for v in videos:
 998                 ydl.process_ie_result(v.copy(), download=True)
 999             return [v['id'] for v in ydl.downloaded_info_dicts]
1000
1001         res = get_videos()
1002         self.assertEqual(res, ['1', '2'])
1003
1004         def f(v, incomplete):
1005             if v['id'] == '1':
1006                 return None
1007             else:
1008                 return 'Video id is not 1'
1009         res = get_videos(f)
1010         self.assertEqual(res, ['1'])
1011
1012         f = match_filter_func('duration < 30')
1013         res = get_videos(f)
1014         self.assertEqual(res, ['2'])
1015
1016         f = match_filter_func('description = foo')
1017         res = get_videos(f)
1018         self.assertEqual(res, ['2'])
1019
1020         f = match_filter_func('description =? foo')
1021         res = get_videos(f)
1022         self.assertEqual(res, ['1', '2'])
1023
1024         f = match_filter_func('filesize > 5KiB')
1025         res = get_videos(f)
1026         self.assertEqual(res, ['1'])
1027
1028         f = match_filter_func('playlist_id = 42')
1029         res = get_videos(f)
1030         self.assertEqual(res, ['1'])
1031
1032         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
1033         res = get_videos(f)
1034         self.assertEqual(res, ['1'])
1035
1036         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
1037         res = get_videos(f)
1038         self.assertEqual(res, ['2'])
1039
1040         f = match_filter_func('creator = "тест \' 123 \' тест--"')
1041         res = get_videos(f)
1042         self.assertEqual(res, ['1'])
1043
1044         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
1045         res = get_videos(f)
1046         self.assertEqual(res, ['1'])
1047
1048         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
1049         res = get_videos(f)
1050         self.assertEqual(res, [])
1051
1052     def test_playlist_items_selection(self):
1053         INDICES, PAGE_SIZE = list(range(1, 11)), 3
1054
1055         def entry(i, evaluated):
1056             evaluated.append(i)
1057             return {
1058                 'id': str(i),
1059                 'title': str(i),
1060                 'url': TEST_URL,
1061             }
1062
1063         def pagedlist_entries(evaluated):
1064             def page_func(n):
1065                 start = PAGE_SIZE * n
1066                 for i in INDICES[start: start + PAGE_SIZE]:
1067                     yield entry(i, evaluated)
1068             return OnDemandPagedList(page_func, PAGE_SIZE)
1069
1070         def page_num(i):
1071             return (i + PAGE_SIZE - 1) // PAGE_SIZE
1072
1073         def generator_entries(evaluated):
1074             for i in INDICES:
1075                 yield entry(i, evaluated)
1076
1077         def list_entries(evaluated):
1078             return list(generator_entries(evaluated))
1079
1080         def lazylist_entries(evaluated):
1081             return LazyList(generator_entries(evaluated))
1082
1083         def get_downloaded_info_dicts(params, entries):
1084             ydl = YDL(params)
1085             ydl.process_ie_result({
1086                 '_type': 'playlist',
1087                 'id': 'test',
1088                 'extractor': 'test:playlist',
1089                 'extractor_key': 'test:playlist',
1090                 'webpage_url': 'http://example.com',
1091                 'entries': entries,
1092             })
1093             return ydl.downloaded_info_dicts
1094
1095         def test_selection(params, expected_ids, evaluate_all=False):
1096             expected_ids = list(expected_ids)
1097             if evaluate_all:
1098                 generator_eval = pagedlist_eval = INDICES
1099             elif not expected_ids:
1100                 generator_eval = pagedlist_eval = []
1101             else:
1102                 generator_eval = INDICES[0: max(expected_ids)]
1103                 pagedlist_eval = INDICES[PAGE_SIZE * page_num(min(expected_ids)) - PAGE_SIZE:
1104                                          PAGE_SIZE * page_num(max(expected_ids))]
1105
1106             for name, func, expected_eval in (
1107                 ('list', list_entries, INDICES),
1108                 ('Generator', generator_entries, generator_eval),
1109                 # ('LazyList', lazylist_entries, generator_eval),  # Generator and LazyList follow the exact same code path
1110                 ('PagedList', pagedlist_entries, pagedlist_eval),
1111             ):
1112                 evaluated = []
1113                 entries = func(evaluated)
1114                 results = [(v['playlist_autonumber'] - 1, (int(v['id']), v['playlist_index']))
1115                            for v in get_downloaded_info_dicts(params, entries)]
1116                 self.assertEqual(results, list(enumerate(zip(expected_ids, expected_ids))), f'Entries of {name} for {params}')
1117                 self.assertEqual(sorted(evaluated), expected_eval, f'Evaluation of {name} for {params}')
1118
1119         test_selection({}, INDICES)
1120         test_selection({'playlistend': 20}, INDICES, True)
1121         test_selection({'playlistend': 2}, INDICES[:2])
1122         test_selection({'playliststart': 11}, [], True)
1123         test_selection({'playliststart': 2}, INDICES[1:])
1124         test_selection({'playlist_items': '2-4'}, INDICES[1:4])
1125         test_selection({'playlist_items': '2,4'}, [2, 4])
1126         test_selection({'playlist_items': '20'}, [], True)
1127         test_selection({'playlist_items': '0'}, [])
1128
1129         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
1130         test_selection({'playlist_items': '2-4,3-4,3'}, [2, 3, 4])
1131         test_selection({'playlist_items': '4,2'}, [4, 2])
1132
1133         # Tests for https://github.com/yt-dlp/yt-dlp/issues/720
1134         # https://github.com/yt-dlp/yt-dlp/issues/302
1135         test_selection({'playlistreverse': True}, INDICES[::-1])
1136         test_selection({'playliststart': 2, 'playlistreverse': True}, INDICES[:0:-1])
1137         test_selection({'playlist_items': '2,4', 'playlistreverse': True}, [4, 2])
1138         test_selection({'playlist_items': '4,2'}, [4, 2])
1139
1140         # Tests for --playlist-items start:end:step
1141         test_selection({'playlist_items': ':'}, INDICES, True)
1142         test_selection({'playlist_items': '::1'}, INDICES, True)
1143         test_selection({'playlist_items': '::-1'}, INDICES[::-1], True)
1144         test_selection({'playlist_items': ':6'}, INDICES[:6])
1145         test_selection({'playlist_items': ':-6'}, INDICES[:-5], True)
1146         test_selection({'playlist_items': '-1:6:-2'}, INDICES[:4:-2], True)
1147         test_selection({'playlist_items': '9:-6:-2'}, INDICES[8:3:-2], True)
1148
1149         test_selection({'playlist_items': '1:inf:2'}, INDICES[::2], True)
1150         test_selection({'playlist_items': '-2:inf'}, INDICES[-2:], True)
1151         test_selection({'playlist_items': ':inf:-1'}, [], True)
1152         test_selection({'playlist_items': '0-2:2'}, [2])
1153         test_selection({'playlist_items': '1-:2'}, INDICES[::2], True)
1154         test_selection({'playlist_items': '0--2:2'}, INDICES[1:-1:2], True)
1155
1156         test_selection({'playlist_items': '10::3'}, [10], True)
1157         test_selection({'playlist_items': '-1::3'}, [10], True)
1158         test_selection({'playlist_items': '11::3'}, [], True)
1159         test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True)
1160         test_selection({'playlist_items': '-15::15'}, [], True)
1161
1162     def test_do_not_override_ie_key_in_url_transparent(self):
1163         ydl = YDL()
1164
1165         class Foo1IE(InfoExtractor):
1166             _VALID_URL = r'foo1:'
1167
1168             def _real_extract(self, url):
1169                 return {
1170                     '_type': 'url_transparent',
1171                     'url': 'foo2:',
1172                     'ie_key': 'Foo2',
1173                     'title': 'foo1 title',
1174                     'id': 'foo1_id',
1175                 }
1176
1177         class Foo2IE(InfoExtractor):
1178             _VALID_URL = r'foo2:'
1179
1180             def _real_extract(self, url):
1181                 return {
1182                     '_type': 'url',
1183                     'url': 'foo3:',
1184                     'ie_key': 'Foo3',
1185                 }
1186
1187         class Foo3IE(InfoExtractor):
1188             _VALID_URL = r'foo3:'
1189
1190             def _real_extract(self, url):
1191                 return _make_result([{'url': TEST_URL}], title='foo3 title')
1192
1193         ydl.add_info_extractor(Foo1IE(ydl))
1194         ydl.add_info_extractor(Foo2IE(ydl))
1195         ydl.add_info_extractor(Foo3IE(ydl))
1196         ydl.extract_info('foo1:')
1197         downloaded = ydl.downloaded_info_dicts[0]
1198         self.assertEqual(downloaded['url'], TEST_URL)
1199         self.assertEqual(downloaded['title'], 'foo1 title')
1200         self.assertEqual(downloaded['id'], 'testid')
1201         self.assertEqual(downloaded['extractor'], 'testex')
1202         self.assertEqual(downloaded['extractor_key'], 'TestEx')
1203
1204     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
1205     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
1206
1207         class _YDL(YDL):
1208             def __init__(self, *args, **kwargs):
1209                 super().__init__(*args, **kwargs)
1210
1211             def trouble(self, s, tb=None):
1212                 pass
1213
1214         ydl = _YDL({
1215             'format': 'extra',
1216             'ignoreerrors': True,
1217         })
1218
1219         class VideoIE(InfoExtractor):
1220             _VALID_URL = r'video:(?P<id>\d+)'
1221
1222             def _real_extract(self, url):
1223                 video_id = self._match_id(url)
1224                 formats = [{
1225                     'format_id': 'default',
1226                     'url': 'url:',
1227                 }]
1228                 if video_id == '0':
1229                     raise ExtractorError('foo')
1230                 if video_id == '2':
1231                     formats.append({
1232                         'format_id': 'extra',
1233                         'url': TEST_URL,
1234                     })
1235                 return {
1236                     'id': video_id,
1237                     'title': f'Video {video_id}',
1238                     'formats': formats,
1239                 }
1240
1241         class PlaylistIE(InfoExtractor):
1242             _VALID_URL = r'playlist:'
1243
1244             def _entries(self):
1245                 for n in range(3):
1246                     video_id = str(n)
1247                     yield {
1248                         '_type': 'url_transparent',
1249                         'ie_key': VideoIE.ie_key(),
1250                         'id': video_id,
1251                         'url': f'video:{video_id}',
1252                         'title': f'Video Transparent {video_id}',
1253                     }
1254
1255             def _real_extract(self, url):
1256                 return self.playlist_result(self._entries())
1257
1258         ydl.add_info_extractor(VideoIE(ydl))
1259         ydl.add_info_extractor(PlaylistIE(ydl))
1260         info = ydl.extract_info('playlist:')
1261         entries = info['entries']
1262         self.assertEqual(len(entries), 3)
1263         self.assertTrue(entries[0] is None)
1264         self.assertTrue(entries[1] is None)
1265         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1266         downloaded = ydl.downloaded_info_dicts[0]
1267         entries[2].pop('requested_downloads', None)
1268         self.assertEqual(entries[2], downloaded)
1269         self.assertEqual(downloaded['url'], TEST_URL)
1270         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1271         self.assertEqual(downloaded['id'], '2')
1272         self.assertEqual(downloaded['extractor'], 'Video')
1273         self.assertEqual(downloaded['extractor_key'], 'Video')
1274
1275     def test_header_cookies(self):
1276         from http.cookiejar import Cookie
1277
1278         ydl = FakeYDL()
1279         ydl.report_warning = lambda *_, **__: None
1280
1281         def cookie(name, value, version=None, domain='', path='', secure=False, expires=None):
1282             return Cookie(
1283                 version or 0, name, value, None, False,
1284                 domain, bool(domain), bool(domain), path, bool(path),
1285                 secure, expires, False, None, None, rest={})
1286
1287         _test_url = 'https://yt.dlp/test'
1288
1289         def test(encoded_cookies, cookies, *, headers=False, round_trip=None, error_re=None):
1290             def _test():
1291                 ydl.cookiejar.clear()
1292                 ydl._load_cookies(encoded_cookies, autoscope=headers)
1293                 if headers:
1294                     ydl._apply_header_cookies(_test_url)
1295                 data = {'url': _test_url}
1296                 ydl._calc_headers(data)
1297                 self.assertCountEqual(
1298                     map(vars, ydl.cookiejar), map(vars, cookies),
1299                     'Extracted cookiejar.Cookie is not the same')
1300                 if not headers:
1301                     self.assertEqual(
1302                         data.get('cookies'), round_trip or encoded_cookies,
1303                         'Cookie is not the same as round trip')
1304                 ydl.__dict__['_YoutubeDL__header_cookies'] = []
1305
1306             with self.subTest(msg=encoded_cookies):
1307                 if not error_re:
1308                     _test()
1309                     return
1310                 with self.assertRaisesRegex(Exception, error_re):
1311                     _test()
1312
1313         test('test=value; Domain=.yt.dlp', [cookie('test', 'value', domain='.yt.dlp')])
1314         test('test=value', [cookie('test', 'value')], error_re=r'Unscoped cookies are not allowed')
1315         test('cookie1=value1; Domain=.yt.dlp; Path=/test; cookie2=value2; Domain=.yt.dlp; Path=/', [
1316             cookie('cookie1', 'value1', domain='.yt.dlp', path='/test'),
1317             cookie('cookie2', 'value2', domain='.yt.dlp', path='/')])
1318         test('test=value; Domain=.yt.dlp; Path=/test; Secure; Expires=9999999999', [
1319             cookie('test', 'value', domain='.yt.dlp', path='/test', secure=True, expires=9999999999)])
1320         test('test="value; "; path=/test; domain=.yt.dlp', [
1321             cookie('test', 'value; ', domain='.yt.dlp', path='/test')],
1322             round_trip='test="value\\073 "; Domain=.yt.dlp; Path=/test')
1323         test('name=; Domain=.yt.dlp', [cookie('name', '', domain='.yt.dlp')],
1324              round_trip='name=""; Domain=.yt.dlp')
1325
1326         test('test=value', [cookie('test', 'value', domain='.yt.dlp')], headers=True)
1327         test('cookie1=value; Domain=.yt.dlp; cookie2=value', [], headers=True, error_re=r'Invalid syntax')
1328         ydl.deprecated_feature = ydl.report_error
1329         test('test=value', [], headers=True, error_re=r'Passing cookies as a header is a potential security risk')
1330
1331     def test_infojson_cookies(self):
1332         TEST_FILE = 'test_infojson_cookies.info.json'
1333         TEST_URL = 'https://example.com/example.mp4'
1334         COOKIES = 'a=b; Domain=.example.com; c=d; Domain=.example.com'
1335         COOKIE_HEADER = {'Cookie': 'a=b; c=d'}
1336
1337         ydl = FakeYDL()
1338         ydl.process_info = lambda x: ydl._write_info_json('test', x, TEST_FILE)
1339
1340         def make_info(info_header_cookies=False, fmts_header_cookies=False, cookies_field=False):
1341             fmt = {'url': TEST_URL}
1342             if fmts_header_cookies:
1343                 fmt['http_headers'] = COOKIE_HEADER
1344             if cookies_field:
1345                 fmt['cookies'] = COOKIES
1346             return _make_result([fmt], http_headers=COOKIE_HEADER if info_header_cookies else None)
1347
1348         def test(initial_info, note):
1349             result = {}
1350             result['processed'] = ydl.process_ie_result(initial_info)
1351             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1352                             msg=f'No cookies set in cookiejar after initial process when {note}')
1353             ydl.cookiejar.clear()
1354             with open(TEST_FILE) as infojson:
1355                 result['loaded'] = ydl.sanitize_info(json.load(infojson), True)
1356             result['final'] = ydl.process_ie_result(result['loaded'].copy(), download=False)
1357             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1358                             msg=f'No cookies set in cookiejar after final process when {note}')
1359             ydl.cookiejar.clear()
1360             for key in ('processed', 'loaded', 'final'):
1361                 info = result[key]
1362                 self.assertIsNone(
1363                     traverse_obj(info, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False),
1364                     msg=f'Cookie header not removed in {key} result when {note}')
1365                 self.assertEqual(
1366                     traverse_obj(info, ((None, ('formats', 0)), 'cookies'), get_all=False), COOKIES,
1367                     msg=f'No cookies field found in {key} result when {note}')
1368
1369         test({'url': TEST_URL, 'http_headers': COOKIE_HEADER, 'id': '1', 'title': 'x'}, 'no formats field')
1370         test(make_info(info_header_cookies=True), 'info_dict header cokies')
1371         test(make_info(fmts_header_cookies=True), 'format header cookies')
1372         test(make_info(info_header_cookies=True, fmts_header_cookies=True), 'info_dict and format header cookies')
1373         test(make_info(info_header_cookies=True, fmts_header_cookies=True, cookies_field=True), 'all cookies fields')
1374         test(make_info(cookies_field=True), 'cookies format field')
1375         test({'url': TEST_URL, 'cookies': COOKIES, 'id': '1', 'title': 'x'}, 'info_dict cookies field only')
1376
1377         try_rm(TEST_FILE)
1378
1379     def test_add_headers_cookie(self):
1380         def check_for_cookie_header(result):
1381             return traverse_obj(result, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False)
1382
1383         ydl = FakeYDL({'http_headers': {'Cookie': 'a=b'}})
1384         ydl._apply_header_cookies(_make_result([])['webpage_url'])  # Scope to input webpage URL: .example.com
1385
1386         fmt = {'url': 'https://example.com/video.mp4'}
1387         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1388         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies in result info_dict')
1389         self.assertEqual(result.get('cookies'), 'a=b; Domain=.example.com', msg='No cookies were set in cookies field')
1390         self.assertIn('a=b', ydl.cookiejar.get_cookie_header(fmt['url']), msg='No cookies were set in cookiejar')
1391
1392         fmt = {'url': 'https://wrong.com/video.mp4'}
1393         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1394         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies for wrong domain')
1395         self.assertFalse(result.get('cookies'), msg='Cookies set in cookies field for wrong domain')
1396         self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain')
1397
1398
1399 if __name__ == '__main__':
1400     unittest.main()