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         # Sanitization options
 765         test('%(title3)s', (None, 'foo⧸bar⧹test'))
 766         test('%(title5)s', (None, 'aei_A'), restrictfilenames=True)
 767         test('%(title3)s', (None, 'foo_bar_test'), windowsfilenames=False, restrictfilenames=True)
 768         if sys.platform != 'win32':
 769             test('%(title3)s', (None, 'foo⧸bar\\test'), windowsfilenames=False)
 770
 771         # ID sanitization
 772         test('%(id)s', '_abcd', info={'id': '_abcd'})
 773         test('%(some_id)s', '_abcd', info={'some_id': '_abcd'})
 774         test('%(formats.0.id)s', '_abcd', info={'formats': [{'id': '_abcd'}]})
 775         test('%(id)s', '-abcd', info={'id': '-abcd'})
 776         test('%(id)s', '.abcd', info={'id': '.abcd'})
 777         test('%(id)s', 'ab__cd', info={'id': 'ab__cd'})
 778         test('%(id)s', ('ab:cd', 'ab：cd'), info={'id': 'ab:cd'})
 779         test('%(id.0)s', '-', info={'id': '--'})
 780
 781         # Invalid templates
 782         self.assertTrue(isinstance(YoutubeDL.validate_outtmpl('%(title)'), ValueError))
 783         test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder='none')
 784         test('%(..)s', 'NA')
 785         test('%(formats.{id)s', 'NA')
 786
 787         # Entire info_dict
 788         def expect_same_infodict(out):
 789             got_dict = json.loads(out)
 790             for info_field, expected in self.outtmpl_info.items():
 791                 self.assertEqual(got_dict.get(info_field), expected, info_field)
 792             return True
 793
 794         test('%()j', (expect_same_infodict, None))
 795
 796         # NA placeholder
 797         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s'
 798         test(NA_TEST_OUTTMPL, 'NA-NA-def-1234.mp4')
 799         test(NA_TEST_OUTTMPL, 'none-none-def-1234.mp4', outtmpl_na_placeholder='none')
 800         test(NA_TEST_OUTTMPL, '--def-1234.mp4', outtmpl_na_placeholder='')
 801         test('%(non_existent.0)s', 'NA')
 802
 803         # String formatting
 804         FMT_TEST_OUTTMPL = '%%(height)%s.%%(ext)s'
 805         test(FMT_TEST_OUTTMPL % 's', '1080.mp4')
 806         test(FMT_TEST_OUTTMPL % 'd', '1080.mp4')
 807         test(FMT_TEST_OUTTMPL % '6d', '  1080.mp4')
 808         test(FMT_TEST_OUTTMPL % '-6d', '1080  .mp4')
 809         test(FMT_TEST_OUTTMPL % '06d', '001080.mp4')
 810         test(FMT_TEST_OUTTMPL % ' 06d', ' 01080.mp4')
 811         test(FMT_TEST_OUTTMPL % '   06d', ' 01080.mp4')
 812         test(FMT_TEST_OUTTMPL % '0 6d', ' 01080.mp4')
 813         test(FMT_TEST_OUTTMPL % '0   6d', ' 01080.mp4')
 814         test(FMT_TEST_OUTTMPL % '   0   6d', ' 01080.mp4')
 815
 816         # Type casting
 817         test('%(id)d', '1234')
 818         test('%(height)c', '1')
 819         test('%(ext)c', 'm')
 820         test('%(id)d %(id)r', "1234 '1234'")
 821         test('%(id)r %(height)r', "'1234' 1080")
 822         test('%(title5)a %(height)a', (R"'\xe1\xe9\xed \U0001d400' 1080", None))
 823         test('%(ext)s-%(ext|def)d', 'mp4-def')
 824         test('%(width|0)04d', '0')
 825         test('a%(width|b)d', 'ab', outtmpl_na_placeholder='none')
 826
 827         FORMATS = self.outtmpl_info['formats']
 828
 829         # Custom type casting
 830         test('%(formats.:.id)l', 'id 1, id 2, id 3')
 831         test('%(formats.:.id)#l', ('id 1\nid 2\nid 3', 'id 1 id 2 id 3'))
 832         test('%(ext)l', 'mp4')
 833         test('%(formats.:.id) 18l', '  id 1, id 2, id 3')
 834         test('%(formats)j', (json.dumps(FORMATS), None))
 835         test('%(formats)#j', (
 836             json.dumps(FORMATS, indent=4),
 837             json.dumps(FORMATS, indent=4).replace(':', '：').replace('"', '＂').replace('\n', ' '),
 838         ))
 839         test('%(title5).3B', 'á')
 840         test('%(title5)U', 'áéí 𝐀')
 841         test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀')
 842         test('%(title5)+U', 'áéí A')
 843         test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A')
 844         test('%(height)D', '1k')
 845         test('%(filesize)#D', '1Ki')
 846         test('%(height)5.2D', ' 1.08k')
 847         test('%(title4)#S', 'foo_bar_test')
 848         test('%(title4).10S', ('foo ＂bar＂ ', 'foo ＂bar＂' + ('#' if os.name == 'nt' else ' ')))
 849         if os.name == 'nt':
 850             test('%(title4)q', ('"foo ""bar"" test"', None))
 851             test('%(formats.:.id)#q', ('"id 1" "id 2" "id 3"', None))
 852             test('%(formats.0.id)#q', ('"id 1"', None))
 853         else:
 854             test('%(title4)q', ('\'foo "bar" test\'', '\'foo ＂bar＂ test\''))
 855             test('%(formats.:.id)#q', "'id 1' 'id 2' 'id 3'")
 856             test('%(formats.0.id)#q', "'id 1'")
 857
 858         # Internal formatting
 859         test('%(timestamp-1000>%H-%M-%S)s', '11-43-20')
 860         test('%(title|%)s %(title|%%)s', '% %%')
 861         test('%(id+1-height+3)05d', '00158')
 862         test('%(width+100)05d', 'NA')
 863         test('%(filesize*8)d', '8192')
 864         test('%(formats.0) 15s', ('% 15s' % FORMATS[0], None))
 865         test('%(formats.0)r', (repr(FORMATS[0]), None))
 866         test('%(height.0)03d', '001')
 867         test('%(-height.0)04d', '-001')
 868         test('%(formats.-1.id)s', FORMATS[-1]['id'])
 869         test('%(formats.0.id.-1)d', FORMATS[0]['id'][-1])
 870         test('%(formats.3)s', 'NA')
 871         test('%(formats.:2:-1)r', repr(FORMATS[:2:-1]))
 872         test('%(formats.0.id.-1+id)f', '1235.000000')
 873         test('%(formats.0.id.-1+formats.1.id.-1)d', '3')
 874         out = json.dumps([{'id': f['id'], 'height.:2': str(f['height'])[:2]}
 875                           if 'height' in f else {'id': f['id']}
 876                           for f in FORMATS])
 877         test('%(formats.:.{id,height.:2})j', (out, None))
 878         test('%(formats.:.{id,height}.id)l', ', '.join(f['id'] for f in FORMATS))
 879         test('%(.{id,title})j', ('{"id": "1234"}', '{＂id＂： ＂1234＂}'))
 880
 881         # Alternates
 882         test('%(title,id)s', '1234')
 883         test('%(width-100,height+20|def)d', '1100')
 884         test('%(width-100,height+width|def)s', 'def')
 885         test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00')
 886
 887         # Replacement
 888         test('%(id&foo)s.bar', 'foo.bar')
 889         test('%(title&foo)s.bar', 'NA.bar')
 890         test('%(title&foo|baz)s.bar', 'baz.bar')
 891         test('%(x,id&foo|baz)s.bar', 'foo.bar')
 892         test('%(x,title&foo|baz)s.bar', 'baz.bar')
 893         test('%(id&a\nb|)s', ('a\nb', 'a b'))
 894         test('%(id&hi {:>10} {}|)s', 'hi       1234 1234')
 895         test(R'%(id&{0} {}|)s', 'NA')
 896         test(R'%(id&{0.1}|)s', 'NA')
 897         test('%(height&{:,d})S', '1,080')
 898
 899         # Laziness
 900         def gen():
 901             yield from range(5)
 902             raise self.assertTrue(False, 'LazyList should not be evaluated till here')
 903         test('%(key.4)s', '4', info={'key': LazyList(gen())})
 904
 905         # Empty filename
 906         test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4')
 907         # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4'))  # FIXME: ?
 908         # test('%(foo|)s', ('', '_'))  # FIXME: ?
 909
 910         # Environment variable expansion for prepare_filename
 911         os.environ['__yt_dlp_var'] = 'expanded'
 912         envvar = '%__yt_dlp_var%' if os.name == 'nt' else '$__yt_dlp_var'
 913         test(envvar, (envvar, 'expanded'))
 914         if os.name == 'nt':
 915             test('%s%', ('%s%', '%s%'))
 916             os.environ['s'] = 'expanded'
 917             test('%s%', ('%s%', 'expanded'))  # %s% should be expanded before escaping %s
 918             os.environ['(test)s'] = 'expanded'
 919             test('%(test)s%', ('NA%', 'expanded'))  # Environment should take priority over template
 920
 921         # Path expansion and escaping
 922         test('Hello %(title1)s', 'Hello $PATH')
 923         test('Hello %(title2)s', 'Hello %PATH%')
 924         test('%(title3)s', ('foo/bar\\test', 'foo⧸bar⧹test'))
 925         test('folder/%(title3)s', ('folder/foo/bar\\test', f'folder{os.path.sep}foo⧸bar⧹test'))
 926
 927     def test_format_note(self):
 928         ydl = YoutubeDL()
 929         self.assertEqual(ydl._format_note({}), '')
 930         assertRegexpMatches(self, ydl._format_note({
 931             'vbr': 10,
 932         }), r'^\s*10k$')
 933         assertRegexpMatches(self, ydl._format_note({
 934             'fps': 30,
 935         }), r'^30fps$')
 936
 937     def test_postprocessors(self):
 938         filename = 'post-processor-testfile.mp4'
 939         audiofile = filename + '.mp3'
 940
 941         class SimplePP(PostProcessor):
 942             def run(self, info):
 943                 with open(audiofile, 'w') as f:
 944                     f.write('EXAMPLE')
 945                 return [info['filepath']], info
 946
 947         def run_pp(params, pp):
 948             with open(filename, 'w') as f:
 949                 f.write('EXAMPLE')
 950             ydl = YoutubeDL(params)
 951             ydl.add_post_processor(pp())
 952             ydl.post_process(filename, {'filepath': filename})
 953
 954         run_pp({'keepvideo': True}, SimplePP)
 955         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 956         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 957         os.unlink(filename)
 958         os.unlink(audiofile)
 959
 960         run_pp({'keepvideo': False}, SimplePP)
 961         self.assertFalse(os.path.exists(filename), f'{filename} exists')
 962         self.assertTrue(os.path.exists(audiofile), f'{audiofile} doesn\'t exist')
 963         os.unlink(audiofile)
 964
 965         class ModifierPP(PostProcessor):
 966             def run(self, info):
 967                 with open(info['filepath'], 'w') as f:
 968                     f.write('MODIFIED')
 969                 return [], info
 970
 971         run_pp({'keepvideo': False}, ModifierPP)
 972         self.assertTrue(os.path.exists(filename), f'{filename} doesn\'t exist')
 973         os.unlink(filename)
 974
 975     def test_match_filter(self):
 976         first = {
 977             'id': '1',
 978             'url': TEST_URL,
 979             'title': 'one',
 980             'extractor': 'TEST',
 981             'duration': 30,
 982             'filesize': 10 * 1024,
 983             'playlist_id': '42',
 984             'uploader': '變態妍字幕版 太妍 тест',
 985             'creator': "тест ' 123 ' тест--",
 986             'webpage_url': 'http://example.com/watch?v=shenanigans',
 987         }
 988         second = {
 989             'id': '2',
 990             'url': TEST_URL,
 991             'title': 'two',
 992             'extractor': 'TEST',
 993             'duration': 10,
 994             'description': 'foo',
 995             'filesize': 5 * 1024,
 996             'playlist_id': '43',
 997             'uploader': 'тест 123',
 998             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 999         }
1000         videos = [first, second]
1001
1002         def get_videos(filter_=None):
1003             ydl = YDL({'match_filter': filter_, 'simulate': True})
1004             for v in videos:
1005                 ydl.process_ie_result(v.copy(), download=True)
1006             return [v['id'] for v in ydl.downloaded_info_dicts]
1007
1008         res = get_videos()
1009         self.assertEqual(res, ['1', '2'])
1010
1011         def f(v, incomplete):
1012             if v['id'] == '1':
1013                 return None
1014             else:
1015                 return 'Video id is not 1'
1016         res = get_videos(f)
1017         self.assertEqual(res, ['1'])
1018
1019         f = match_filter_func('duration < 30')
1020         res = get_videos(f)
1021         self.assertEqual(res, ['2'])
1022
1023         f = match_filter_func('description = foo')
1024         res = get_videos(f)
1025         self.assertEqual(res, ['2'])
1026
1027         f = match_filter_func('description =? foo')
1028         res = get_videos(f)
1029         self.assertEqual(res, ['1', '2'])
1030
1031         f = match_filter_func('filesize > 5KiB')
1032         res = get_videos(f)
1033         self.assertEqual(res, ['1'])
1034
1035         f = match_filter_func('playlist_id = 42')
1036         res = get_videos(f)
1037         self.assertEqual(res, ['1'])
1038
1039         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
1040         res = get_videos(f)
1041         self.assertEqual(res, ['1'])
1042
1043         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
1044         res = get_videos(f)
1045         self.assertEqual(res, ['2'])
1046
1047         f = match_filter_func('creator = "тест \' 123 \' тест--"')
1048         res = get_videos(f)
1049         self.assertEqual(res, ['1'])
1050
1051         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
1052         res = get_videos(f)
1053         self.assertEqual(res, ['1'])
1054
1055         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
1056         res = get_videos(f)
1057         self.assertEqual(res, [])
1058
1059     def test_playlist_items_selection(self):
1060         INDICES, PAGE_SIZE = list(range(1, 11)), 3
1061
1062         def entry(i, evaluated):
1063             evaluated.append(i)
1064             return {
1065                 'id': str(i),
1066                 'title': str(i),
1067                 'url': TEST_URL,
1068             }
1069
1070         def pagedlist_entries(evaluated):
1071             def page_func(n):
1072                 start = PAGE_SIZE * n
1073                 for i in INDICES[start: start + PAGE_SIZE]:
1074                     yield entry(i, evaluated)
1075             return OnDemandPagedList(page_func, PAGE_SIZE)
1076
1077         def page_num(i):
1078             return (i + PAGE_SIZE - 1) // PAGE_SIZE
1079
1080         def generator_entries(evaluated):
1081             for i in INDICES:
1082                 yield entry(i, evaluated)
1083
1084         def list_entries(evaluated):
1085             return list(generator_entries(evaluated))
1086
1087         def lazylist_entries(evaluated):
1088             return LazyList(generator_entries(evaluated))
1089
1090         def get_downloaded_info_dicts(params, entries):
1091             ydl = YDL(params)
1092             ydl.process_ie_result({
1093                 '_type': 'playlist',
1094                 'id': 'test',
1095                 'extractor': 'test:playlist',
1096                 'extractor_key': 'test:playlist',
1097                 'webpage_url': 'http://example.com',
1098                 'entries': entries,
1099             })
1100             return ydl.downloaded_info_dicts
1101
1102         def test_selection(params, expected_ids, evaluate_all=False):
1103             expected_ids = list(expected_ids)
1104             if evaluate_all:
1105                 generator_eval = pagedlist_eval = INDICES
1106             elif not expected_ids:
1107                 generator_eval = pagedlist_eval = []
1108             else:
1109                 generator_eval = INDICES[0: max(expected_ids)]
1110                 pagedlist_eval = INDICES[PAGE_SIZE * page_num(min(expected_ids)) - PAGE_SIZE:
1111                                          PAGE_SIZE * page_num(max(expected_ids))]
1112
1113             for name, func, expected_eval in (
1114                 ('list', list_entries, INDICES),
1115                 ('Generator', generator_entries, generator_eval),
1116                 # ('LazyList', lazylist_entries, generator_eval),  # Generator and LazyList follow the exact same code path
1117                 ('PagedList', pagedlist_entries, pagedlist_eval),
1118             ):
1119                 evaluated = []
1120                 entries = func(evaluated)
1121                 results = [(v['playlist_autonumber'] - 1, (int(v['id']), v['playlist_index']))
1122                            for v in get_downloaded_info_dicts(params, entries)]
1123                 self.assertEqual(results, list(enumerate(zip(expected_ids, expected_ids))), f'Entries of {name} for {params}')
1124                 self.assertEqual(sorted(evaluated), expected_eval, f'Evaluation of {name} for {params}')
1125
1126         test_selection({}, INDICES)
1127         test_selection({'playlistend': 20}, INDICES, True)
1128         test_selection({'playlistend': 2}, INDICES[:2])
1129         test_selection({'playliststart': 11}, [], True)
1130         test_selection({'playliststart': 2}, INDICES[1:])
1131         test_selection({'playlist_items': '2-4'}, INDICES[1:4])
1132         test_selection({'playlist_items': '2,4'}, [2, 4])
1133         test_selection({'playlist_items': '20'}, [], True)
1134         test_selection({'playlist_items': '0'}, [])
1135
1136         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
1137         test_selection({'playlist_items': '2-4,3-4,3'}, [2, 3, 4])
1138         test_selection({'playlist_items': '4,2'}, [4, 2])
1139
1140         # Tests for https://github.com/yt-dlp/yt-dlp/issues/720
1141         # https://github.com/yt-dlp/yt-dlp/issues/302
1142         test_selection({'playlistreverse': True}, INDICES[::-1])
1143         test_selection({'playliststart': 2, 'playlistreverse': True}, INDICES[:0:-1])
1144         test_selection({'playlist_items': '2,4', 'playlistreverse': True}, [4, 2])
1145         test_selection({'playlist_items': '4,2'}, [4, 2])
1146
1147         # Tests for --playlist-items start:end:step
1148         test_selection({'playlist_items': ':'}, INDICES, True)
1149         test_selection({'playlist_items': '::1'}, INDICES, True)
1150         test_selection({'playlist_items': '::-1'}, INDICES[::-1], True)
1151         test_selection({'playlist_items': ':6'}, INDICES[:6])
1152         test_selection({'playlist_items': ':-6'}, INDICES[:-5], True)
1153         test_selection({'playlist_items': '-1:6:-2'}, INDICES[:4:-2], True)
1154         test_selection({'playlist_items': '9:-6:-2'}, INDICES[8:3:-2], True)
1155
1156         test_selection({'playlist_items': '1:inf:2'}, INDICES[::2], True)
1157         test_selection({'playlist_items': '-2:inf'}, INDICES[-2:], True)
1158         test_selection({'playlist_items': ':inf:-1'}, [], True)
1159         test_selection({'playlist_items': '0-2:2'}, [2])
1160         test_selection({'playlist_items': '1-:2'}, INDICES[::2], True)
1161         test_selection({'playlist_items': '0--2:2'}, INDICES[1:-1:2], True)
1162
1163         test_selection({'playlist_items': '10::3'}, [10], True)
1164         test_selection({'playlist_items': '-1::3'}, [10], True)
1165         test_selection({'playlist_items': '11::3'}, [], True)
1166         test_selection({'playlist_items': '-15::2'}, INDICES[1::2], True)
1167         test_selection({'playlist_items': '-15::15'}, [], True)
1168
1169     def test_do_not_override_ie_key_in_url_transparent(self):
1170         ydl = YDL()
1171
1172         class Foo1IE(InfoExtractor):
1173             _VALID_URL = r'foo1:'
1174
1175             def _real_extract(self, url):
1176                 return {
1177                     '_type': 'url_transparent',
1178                     'url': 'foo2:',
1179                     'ie_key': 'Foo2',
1180                     'title': 'foo1 title',
1181                     'id': 'foo1_id',
1182                 }
1183
1184         class Foo2IE(InfoExtractor):
1185             _VALID_URL = r'foo2:'
1186
1187             def _real_extract(self, url):
1188                 return {
1189                     '_type': 'url',
1190                     'url': 'foo3:',
1191                     'ie_key': 'Foo3',
1192                 }
1193
1194         class Foo3IE(InfoExtractor):
1195             _VALID_URL = r'foo3:'
1196
1197             def _real_extract(self, url):
1198                 return _make_result([{'url': TEST_URL}], title='foo3 title')
1199
1200         ydl.add_info_extractor(Foo1IE(ydl))
1201         ydl.add_info_extractor(Foo2IE(ydl))
1202         ydl.add_info_extractor(Foo3IE(ydl))
1203         ydl.extract_info('foo1:')
1204         downloaded = ydl.downloaded_info_dicts[0]
1205         self.assertEqual(downloaded['url'], TEST_URL)
1206         self.assertEqual(downloaded['title'], 'foo1 title')
1207         self.assertEqual(downloaded['id'], 'testid')
1208         self.assertEqual(downloaded['extractor'], 'testex')
1209         self.assertEqual(downloaded['extractor_key'], 'TestEx')
1210
1211     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
1212     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
1213
1214         class _YDL(YDL):
1215             def __init__(self, *args, **kwargs):
1216                 super().__init__(*args, **kwargs)
1217
1218             def trouble(self, s, tb=None):
1219                 pass
1220
1221         ydl = _YDL({
1222             'format': 'extra',
1223             'ignoreerrors': True,
1224         })
1225
1226         class VideoIE(InfoExtractor):
1227             _VALID_URL = r'video:(?P<id>\d+)'
1228
1229             def _real_extract(self, url):
1230                 video_id = self._match_id(url)
1231                 formats = [{
1232                     'format_id': 'default',
1233                     'url': 'url:',
1234                 }]
1235                 if video_id == '0':
1236                     raise ExtractorError('foo')
1237                 if video_id == '2':
1238                     formats.append({
1239                         'format_id': 'extra',
1240                         'url': TEST_URL,
1241                     })
1242                 return {
1243                     'id': video_id,
1244                     'title': f'Video {video_id}',
1245                     'formats': formats,
1246                 }
1247
1248         class PlaylistIE(InfoExtractor):
1249             _VALID_URL = r'playlist:'
1250
1251             def _entries(self):
1252                 for n in range(3):
1253                     video_id = str(n)
1254                     yield {
1255                         '_type': 'url_transparent',
1256                         'ie_key': VideoIE.ie_key(),
1257                         'id': video_id,
1258                         'url': f'video:{video_id}',
1259                         'title': f'Video Transparent {video_id}',
1260                     }
1261
1262             def _real_extract(self, url):
1263                 return self.playlist_result(self._entries())
1264
1265         ydl.add_info_extractor(VideoIE(ydl))
1266         ydl.add_info_extractor(PlaylistIE(ydl))
1267         info = ydl.extract_info('playlist:')
1268         entries = info['entries']
1269         self.assertEqual(len(entries), 3)
1270         self.assertTrue(entries[0] is None)
1271         self.assertTrue(entries[1] is None)
1272         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1273         downloaded = ydl.downloaded_info_dicts[0]
1274         entries[2].pop('requested_downloads', None)
1275         self.assertEqual(entries[2], downloaded)
1276         self.assertEqual(downloaded['url'], TEST_URL)
1277         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1278         self.assertEqual(downloaded['id'], '2')
1279         self.assertEqual(downloaded['extractor'], 'Video')
1280         self.assertEqual(downloaded['extractor_key'], 'Video')
1281
1282     def test_header_cookies(self):
1283         from http.cookiejar import Cookie
1284
1285         ydl = FakeYDL()
1286         ydl.report_warning = lambda *_, **__: None
1287
1288         def cookie(name, value, version=None, domain='', path='', secure=False, expires=None):
1289             return Cookie(
1290                 version or 0, name, value, None, False,
1291                 domain, bool(domain), bool(domain), path, bool(path),
1292                 secure, expires, False, None, None, rest={})
1293
1294         _test_url = 'https://yt.dlp/test'
1295
1296         def test(encoded_cookies, cookies, *, headers=False, round_trip=None, error_re=None):
1297             def _test():
1298                 ydl.cookiejar.clear()
1299                 ydl._load_cookies(encoded_cookies, autoscope=headers)
1300                 if headers:
1301                     ydl._apply_header_cookies(_test_url)
1302                 data = {'url': _test_url}
1303                 ydl._calc_headers(data)
1304                 self.assertCountEqual(
1305                     map(vars, ydl.cookiejar), map(vars, cookies),
1306                     'Extracted cookiejar.Cookie is not the same')
1307                 if not headers:
1308                     self.assertEqual(
1309                         data.get('cookies'), round_trip or encoded_cookies,
1310                         'Cookie is not the same as round trip')
1311                 ydl.__dict__['_YoutubeDL__header_cookies'] = []
1312
1313             with self.subTest(msg=encoded_cookies):
1314                 if not error_re:
1315                     _test()
1316                     return
1317                 with self.assertRaisesRegex(Exception, error_re):
1318                     _test()
1319
1320         test('test=value; Domain=.yt.dlp', [cookie('test', 'value', domain='.yt.dlp')])
1321         test('test=value', [cookie('test', 'value')], error_re=r'Unscoped cookies are not allowed')
1322         test('cookie1=value1; Domain=.yt.dlp; Path=/test; cookie2=value2; Domain=.yt.dlp; Path=/', [
1323             cookie('cookie1', 'value1', domain='.yt.dlp', path='/test'),
1324             cookie('cookie2', 'value2', domain='.yt.dlp', path='/')])
1325         test('test=value; Domain=.yt.dlp; Path=/test; Secure; Expires=9999999999', [
1326             cookie('test', 'value', domain='.yt.dlp', path='/test', secure=True, expires=9999999999)])
1327         test('test="value; "; path=/test; domain=.yt.dlp', [
1328             cookie('test', 'value; ', domain='.yt.dlp', path='/test')],
1329             round_trip='test="value\\073 "; Domain=.yt.dlp; Path=/test')
1330         test('name=; Domain=.yt.dlp', [cookie('name', '', domain='.yt.dlp')],
1331              round_trip='name=""; Domain=.yt.dlp')
1332
1333         test('test=value', [cookie('test', 'value', domain='.yt.dlp')], headers=True)
1334         test('cookie1=value; Domain=.yt.dlp; cookie2=value', [], headers=True, error_re=r'Invalid syntax')
1335         ydl.deprecated_feature = ydl.report_error
1336         test('test=value', [], headers=True, error_re=r'Passing cookies as a header is a potential security risk')
1337
1338     def test_infojson_cookies(self):
1339         TEST_FILE = 'test_infojson_cookies.info.json'
1340         TEST_URL = 'https://example.com/example.mp4'
1341         COOKIES = 'a=b; Domain=.example.com; c=d; Domain=.example.com'
1342         COOKIE_HEADER = {'Cookie': 'a=b; c=d'}
1343
1344         ydl = FakeYDL()
1345         ydl.process_info = lambda x: ydl._write_info_json('test', x, TEST_FILE)
1346
1347         def make_info(info_header_cookies=False, fmts_header_cookies=False, cookies_field=False):
1348             fmt = {'url': TEST_URL}
1349             if fmts_header_cookies:
1350                 fmt['http_headers'] = COOKIE_HEADER
1351             if cookies_field:
1352                 fmt['cookies'] = COOKIES
1353             return _make_result([fmt], http_headers=COOKIE_HEADER if info_header_cookies else None)
1354
1355         def test(initial_info, note):
1356             result = {}
1357             result['processed'] = ydl.process_ie_result(initial_info)
1358             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1359                             msg=f'No cookies set in cookiejar after initial process when {note}')
1360             ydl.cookiejar.clear()
1361             with open(TEST_FILE) as infojson:
1362                 result['loaded'] = ydl.sanitize_info(json.load(infojson), True)
1363             result['final'] = ydl.process_ie_result(result['loaded'].copy(), download=False)
1364             self.assertTrue(ydl.cookiejar.get_cookies_for_url(TEST_URL),
1365                             msg=f'No cookies set in cookiejar after final process when {note}')
1366             ydl.cookiejar.clear()
1367             for key in ('processed', 'loaded', 'final'):
1368                 info = result[key]
1369                 self.assertIsNone(
1370                     traverse_obj(info, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False),
1371                     msg=f'Cookie header not removed in {key} result when {note}')
1372                 self.assertEqual(
1373                     traverse_obj(info, ((None, ('formats', 0)), 'cookies'), get_all=False), COOKIES,
1374                     msg=f'No cookies field found in {key} result when {note}')
1375
1376         test({'url': TEST_URL, 'http_headers': COOKIE_HEADER, 'id': '1', 'title': 'x'}, 'no formats field')
1377         test(make_info(info_header_cookies=True), 'info_dict header cokies')
1378         test(make_info(fmts_header_cookies=True), 'format header cookies')
1379         test(make_info(info_header_cookies=True, fmts_header_cookies=True), 'info_dict and format header cookies')
1380         test(make_info(info_header_cookies=True, fmts_header_cookies=True, cookies_field=True), 'all cookies fields')
1381         test(make_info(cookies_field=True), 'cookies format field')
1382         test({'url': TEST_URL, 'cookies': COOKIES, 'id': '1', 'title': 'x'}, 'info_dict cookies field only')
1383
1384         try_rm(TEST_FILE)
1385
1386     def test_add_headers_cookie(self):
1387         def check_for_cookie_header(result):
1388             return traverse_obj(result, ((None, ('formats', 0)), 'http_headers', 'Cookie'), casesense=False, get_all=False)
1389
1390         ydl = FakeYDL({'http_headers': {'Cookie': 'a=b'}})
1391         ydl._apply_header_cookies(_make_result([])['webpage_url'])  # Scope to input webpage URL: .example.com
1392
1393         fmt = {'url': 'https://example.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 in result info_dict')
1396         self.assertEqual(result.get('cookies'), 'a=b; Domain=.example.com', msg='No cookies were set in cookies field')
1397         self.assertIn('a=b', ydl.cookiejar.get_cookie_header(fmt['url']), msg='No cookies were set in cookiejar')
1398
1399         fmt = {'url': 'https://wrong.com/video.mp4'}
1400         result = ydl.process_ie_result(_make_result([fmt]), download=False)
1401         self.assertIsNone(check_for_cookie_header(result), msg='http_headers cookies for wrong domain')
1402         self.assertFalse(result.get('cookies'), msg='Cookies set in cookies field for wrong domain')
1403         self.assertFalse(ydl.cookiejar.get_cookie_header(fmt['url']), msg='Cookies set in cookiejar for wrong domain')
1404
1405
1406 if __name__ == '__main__':
1407     unittest.main()