test/test_YoutubeDL.py

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