test/test_YoutubeDL.py

   1 #!/usr/bin/env python
   2 # coding: utf-8
   3
   4 from __future__ import unicode_literals
   5
   6 # Allow direct execution
   7 import os
   8 import sys
   9 import unittest
  10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12 import copy
  13
  14 from test.helper import FakeYDL, assertRegexpMatches
  15 from yt_dlp import YoutubeDL
  16 from yt_dlp.compat import compat_str, compat_urllib_error
  17 from yt_dlp.extractor import YoutubeIE
  18 from yt_dlp.extractor.common import InfoExtractor
  19 from yt_dlp.postprocessor.common import PostProcessor
  20 from yt_dlp.utils import ExtractorError, match_filter_func
  21
  22 TEST_URL = 'http://localhost/sample.mp4'
  23
  24
  25 class YDL(FakeYDL):
  26     def __init__(self, *args, **kwargs):
  27         super(YDL, self).__init__(*args, **kwargs)
  28         self.downloaded_info_dicts = []
  29         self.msgs = []
  30
  31     def process_info(self, info_dict):
  32         self.downloaded_info_dicts.append(info_dict)
  33
  34     def to_screen(self, msg):
  35         self.msgs.append(msg)
  36
  37
  38 def _make_result(formats, **kwargs):
  39     res = {
  40         'formats': formats,
  41         'id': 'testid',
  42         'title': 'testttitle',
  43         'extractor': 'testex',
  44         'extractor_key': 'TestEx',
  45         'webpage_url': 'http://example.com/watch?v=shenanigans',
  46     }
  47     res.update(**kwargs)
  48     return res
  49
  50
  51 class TestFormatSelection(unittest.TestCase):
  52     def test_prefer_free_formats(self):
  53         # Same resolution => download webm
  54         ydl = YDL()
  55         ydl.params['prefer_free_formats'] = True
  56         formats = [
  57             {'ext': 'webm', 'height': 460, 'url': TEST_URL},
  58             {'ext': 'mp4', 'height': 460, 'url': TEST_URL},
  59         ]
  60         info_dict = _make_result(formats)
  61         yie = YoutubeIE(ydl)
  62         yie._sort_formats(info_dict['formats'])
  63         ydl.process_ie_result(info_dict)
  64         downloaded = ydl.downloaded_info_dicts[0]
  65         self.assertEqual(downloaded['ext'], 'webm')
  66
  67         # Different resolution => download best quality (mp4)
  68         ydl = YDL()
  69         ydl.params['prefer_free_formats'] = True
  70         formats = [
  71             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  72             {'ext': 'mp4', 'height': 1080, 'url': TEST_URL},
  73         ]
  74         info_dict['formats'] = formats
  75         yie = YoutubeIE(ydl)
  76         yie._sort_formats(info_dict['formats'])
  77         ydl.process_ie_result(info_dict)
  78         downloaded = ydl.downloaded_info_dicts[0]
  79         self.assertEqual(downloaded['ext'], 'mp4')
  80
  81         # No prefer_free_formats => prefer mp4 and webm
  82         ydl = YDL()
  83         ydl.params['prefer_free_formats'] = False
  84         formats = [
  85             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
  86             {'ext': 'mp4', 'height': 720, 'url': TEST_URL},
  87             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
  88         ]
  89         info_dict['formats'] = formats
  90         yie = YoutubeIE(ydl)
  91         yie._sort_formats(info_dict['formats'])
  92         ydl.process_ie_result(info_dict)
  93         downloaded = ydl.downloaded_info_dicts[0]
  94         self.assertEqual(downloaded['ext'], 'mp4')
  95
  96         ydl = YDL()
  97         ydl.params['prefer_free_formats'] = False
  98         formats = [
  99             {'ext': 'flv', 'height': 720, 'url': TEST_URL},
 100             {'ext': 'webm', 'height': 720, 'url': TEST_URL},
 101         ]
 102         info_dict['formats'] = formats
 103         yie = YoutubeIE(ydl)
 104         yie._sort_formats(info_dict['formats'])
 105         ydl.process_ie_result(info_dict)
 106         downloaded = ydl.downloaded_info_dicts[0]
 107         self.assertEqual(downloaded['ext'], 'webm')
 108
 109     def test_format_selection(self):
 110         formats = [
 111             {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 112             {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL},
 113             {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL},
 114             {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL},
 115             {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL},
 116         ]
 117         info_dict = _make_result(formats)
 118
 119         ydl = YDL({'format': '20/47'})
 120         ydl.process_ie_result(info_dict.copy())
 121         downloaded = ydl.downloaded_info_dicts[0]
 122         self.assertEqual(downloaded['format_id'], '47')
 123
 124         ydl = YDL({'format': '20/71/worst'})
 125         ydl.process_ie_result(info_dict.copy())
 126         downloaded = ydl.downloaded_info_dicts[0]
 127         self.assertEqual(downloaded['format_id'], '35')
 128
 129         ydl = YDL()
 130         ydl.process_ie_result(info_dict.copy())
 131         downloaded = ydl.downloaded_info_dicts[0]
 132         self.assertEqual(downloaded['format_id'], '2')
 133
 134         ydl = YDL({'format': 'webm/mp4'})
 135         ydl.process_ie_result(info_dict.copy())
 136         downloaded = ydl.downloaded_info_dicts[0]
 137         self.assertEqual(downloaded['format_id'], '47')
 138
 139         ydl = YDL({'format': '3gp/40/mp4'})
 140         ydl.process_ie_result(info_dict.copy())
 141         downloaded = ydl.downloaded_info_dicts[0]
 142         self.assertEqual(downloaded['format_id'], '35')
 143
 144         ydl = YDL({'format': 'example-with-dashes'})
 145         ydl.process_ie_result(info_dict.copy())
 146         downloaded = ydl.downloaded_info_dicts[0]
 147         self.assertEqual(downloaded['format_id'], 'example-with-dashes')
 148
 149     def test_format_selection_audio(self):
 150         formats = [
 151             {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 152             {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 153             {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL},
 154             {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL},
 155         ]
 156         info_dict = _make_result(formats)
 157
 158         ydl = YDL({'format': 'bestaudio'})
 159         ydl.process_ie_result(info_dict.copy())
 160         downloaded = ydl.downloaded_info_dicts[0]
 161         self.assertEqual(downloaded['format_id'], 'audio-high')
 162
 163         ydl = YDL({'format': 'worstaudio'})
 164         ydl.process_ie_result(info_dict.copy())
 165         downloaded = ydl.downloaded_info_dicts[0]
 166         self.assertEqual(downloaded['format_id'], 'audio-low')
 167
 168         formats = [
 169             {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL},
 170             {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL},
 171         ]
 172         info_dict = _make_result(formats)
 173
 174         ydl = YDL({'format': 'bestaudio/worstaudio/best'})
 175         ydl.process_ie_result(info_dict.copy())
 176         downloaded = ydl.downloaded_info_dicts[0]
 177         self.assertEqual(downloaded['format_id'], 'vid-high')
 178
 179     def test_format_selection_audio_exts(self):
 180         formats = [
 181             {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 182             {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 183             {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'},
 184             {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 185             {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'},
 186         ]
 187
 188         info_dict = _make_result(formats)
 189         ydl = YDL({'format': 'best'})
 190         ie = YoutubeIE(ydl)
 191         ie._sort_formats(info_dict['formats'])
 192         ydl.process_ie_result(copy.deepcopy(info_dict))
 193         downloaded = ydl.downloaded_info_dicts[0]
 194         self.assertEqual(downloaded['format_id'], 'aac-64')
 195
 196         ydl = YDL({'format': 'mp3'})
 197         ie = YoutubeIE(ydl)
 198         ie._sort_formats(info_dict['formats'])
 199         ydl.process_ie_result(copy.deepcopy(info_dict))
 200         downloaded = ydl.downloaded_info_dicts[0]
 201         self.assertEqual(downloaded['format_id'], 'mp3-64')
 202
 203         ydl = YDL({'prefer_free_formats': True})
 204         ie = YoutubeIE(ydl)
 205         ie._sort_formats(info_dict['formats'])
 206         ydl.process_ie_result(copy.deepcopy(info_dict))
 207         downloaded = ydl.downloaded_info_dicts[0]
 208         self.assertEqual(downloaded['format_id'], 'ogg-64')
 209
 210     def test_format_selection_video(self):
 211         formats = [
 212             {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL},
 213             {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL},
 214             {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL},
 215         ]
 216         info_dict = _make_result(formats)
 217
 218         ydl = YDL({'format': 'bestvideo'})
 219         ydl.process_ie_result(info_dict.copy())
 220         downloaded = ydl.downloaded_info_dicts[0]
 221         self.assertEqual(downloaded['format_id'], 'dash-video-high')
 222
 223         ydl = YDL({'format': 'worstvideo'})
 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         ydl = YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'})
 229         ydl.process_ie_result(info_dict.copy())
 230         downloaded = ydl.downloaded_info_dicts[0]
 231         self.assertEqual(downloaded['format_id'], 'dash-video-low')
 232
 233         formats = [
 234             {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL},
 235         ]
 236         info_dict = _make_result(formats)
 237
 238         ydl = YDL({'format': 'bestvideo[vcodec=avc1.123456]'})
 239         ydl.process_ie_result(info_dict.copy())
 240         downloaded = ydl.downloaded_info_dicts[0]
 241         self.assertEqual(downloaded['format_id'], 'vid-vcodec-dot')
 242
 243     def test_format_selection_string_ops(self):
 244         formats = [
 245             {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL},
 246             {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL},
 247         ]
 248         info_dict = _make_result(formats)
 249
 250         # equals (=)
 251         ydl = YDL({'format': '[format_id=abc-cba]'})
 252         ydl.process_ie_result(info_dict.copy())
 253         downloaded = ydl.downloaded_info_dicts[0]
 254         self.assertEqual(downloaded['format_id'], 'abc-cba')
 255
 256         # does not equal (!=)
 257         ydl = YDL({'format': '[format_id!=abc-cba]'})
 258         ydl.process_ie_result(info_dict.copy())
 259         downloaded = ydl.downloaded_info_dicts[0]
 260         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 261
 262         ydl = YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'})
 263         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 264
 265         # starts with (^=)
 266         ydl = YDL({'format': '[format_id^=abc]'})
 267         ydl.process_ie_result(info_dict.copy())
 268         downloaded = ydl.downloaded_info_dicts[0]
 269         self.assertEqual(downloaded['format_id'], 'abc-cba')
 270
 271         # does not start with (!^=)
 272         ydl = YDL({'format': '[format_id!^=abc]'})
 273         ydl.process_ie_result(info_dict.copy())
 274         downloaded = ydl.downloaded_info_dicts[0]
 275         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 276
 277         ydl = YDL({'format': '[format_id!^=abc][format_id!^=zxc]'})
 278         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 279
 280         # ends with ($=)
 281         ydl = YDL({'format': '[format_id$=cba]'})
 282         ydl.process_ie_result(info_dict.copy())
 283         downloaded = ydl.downloaded_info_dicts[0]
 284         self.assertEqual(downloaded['format_id'], 'abc-cba')
 285
 286         # does not end with (!$=)
 287         ydl = YDL({'format': '[format_id!$=cba]'})
 288         ydl.process_ie_result(info_dict.copy())
 289         downloaded = ydl.downloaded_info_dicts[0]
 290         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 291
 292         ydl = YDL({'format': '[format_id!$=cba][format_id!$=cxz]'})
 293         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 294
 295         # contains (*=)
 296         ydl = YDL({'format': '[format_id*=bc-cb]'})
 297         ydl.process_ie_result(info_dict.copy())
 298         downloaded = ydl.downloaded_info_dicts[0]
 299         self.assertEqual(downloaded['format_id'], 'abc-cba')
 300
 301         # does not contain (!*=)
 302         ydl = YDL({'format': '[format_id!*=bc-cb]'})
 303         ydl.process_ie_result(info_dict.copy())
 304         downloaded = ydl.downloaded_info_dicts[0]
 305         self.assertEqual(downloaded['format_id'], 'zxc-cxz')
 306
 307         ydl = YDL({'format': '[format_id!*=abc][format_id!*=zxc]'})
 308         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 309
 310         ydl = YDL({'format': '[format_id!*=-]'})
 311         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 312
 313     def test_youtube_format_selection(self):
 314         # FIXME: Rewrite in accordance with the new format sorting options
 315         return
 316
 317         order = [
 318             '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
 319             # Apple HTTP Live Streaming
 320             '96', '95', '94', '93', '92', '132', '151',
 321             # 3D
 322             '85', '84', '102', '83', '101', '82', '100',
 323             # Dash video
 324             '137', '248', '136', '247', '135', '246',
 325             '245', '244', '134', '243', '133', '242', '160',
 326             # Dash audio
 327             '141', '172', '140', '171', '139',
 328         ]
 329
 330         def format_info(f_id):
 331             info = YoutubeIE._formats[f_id].copy()
 332
 333             # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
 334             # and 'vcodec', while in tests such information is incomplete since
 335             # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
 336             # test_YoutubeDL.test_youtube_format_selection is broken without
 337             # this fix
 338             if 'acodec' in info and 'vcodec' not in info:
 339                 info['vcodec'] = 'none'
 340             elif 'vcodec' in info and 'acodec' not in info:
 341                 info['acodec'] = 'none'
 342
 343             info['format_id'] = f_id
 344             info['url'] = 'url:' + f_id
 345             return info
 346         formats_order = [format_info(f_id) for f_id in order]
 347
 348         info_dict = _make_result(list(formats_order), extractor='youtube')
 349         ydl = YDL({'format': 'bestvideo+bestaudio'})
 350         yie = YoutubeIE(ydl)
 351         yie._sort_formats(info_dict['formats'])
 352         ydl.process_ie_result(info_dict)
 353         downloaded = ydl.downloaded_info_dicts[0]
 354         self.assertEqual(downloaded['format_id'], '248+172')
 355         self.assertEqual(downloaded['ext'], 'mp4')
 356
 357         info_dict = _make_result(list(formats_order), extractor='youtube')
 358         ydl = YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'})
 359         yie = YoutubeIE(ydl)
 360         yie._sort_formats(info_dict['formats'])
 361         ydl.process_ie_result(info_dict)
 362         downloaded = ydl.downloaded_info_dicts[0]
 363         self.assertEqual(downloaded['format_id'], '38')
 364
 365         info_dict = _make_result(list(formats_order), extractor='youtube')
 366         ydl = YDL({'format': 'bestvideo/best,bestaudio'})
 367         yie = YoutubeIE(ydl)
 368         yie._sort_formats(info_dict['formats'])
 369         ydl.process_ie_result(info_dict)
 370         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 371         self.assertEqual(downloaded_ids, ['137', '141'])
 372
 373         info_dict = _make_result(list(formats_order), extractor='youtube')
 374         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'})
 375         yie = YoutubeIE(ydl)
 376         yie._sort_formats(info_dict['formats'])
 377         ydl.process_ie_result(info_dict)
 378         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 379         self.assertEqual(downloaded_ids, ['137+141', '248+141'])
 380
 381         info_dict = _make_result(list(formats_order), extractor='youtube')
 382         ydl = YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'})
 383         yie = YoutubeIE(ydl)
 384         yie._sort_formats(info_dict['formats'])
 385         ydl.process_ie_result(info_dict)
 386         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 387         self.assertEqual(downloaded_ids, ['136+141', '247+141'])
 388
 389         info_dict = _make_result(list(formats_order), extractor='youtube')
 390         ydl = YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'})
 391         yie = YoutubeIE(ydl)
 392         yie._sort_formats(info_dict['formats'])
 393         ydl.process_ie_result(info_dict)
 394         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 395         self.assertEqual(downloaded_ids, ['248+141'])
 396
 397         for f1, f2 in zip(formats_order, formats_order[1:]):
 398             info_dict = _make_result([f1, f2], extractor='youtube')
 399             ydl = YDL({'format': 'best/bestvideo'})
 400             yie = YoutubeIE(ydl)
 401             yie._sort_formats(info_dict['formats'])
 402             ydl.process_ie_result(info_dict)
 403             downloaded = ydl.downloaded_info_dicts[0]
 404             self.assertEqual(downloaded['format_id'], f1['format_id'])
 405
 406             info_dict = _make_result([f2, f1], extractor='youtube')
 407             ydl = YDL({'format': 'best/bestvideo'})
 408             yie = YoutubeIE(ydl)
 409             yie._sort_formats(info_dict['formats'])
 410             ydl.process_ie_result(info_dict)
 411             downloaded = ydl.downloaded_info_dicts[0]
 412             self.assertEqual(downloaded['format_id'], f1['format_id'])
 413
 414     def test_audio_only_extractor_format_selection(self):
 415         # For extractors with incomplete formats (all formats are audio-only or
 416         # video-only) best and worst should fallback to corresponding best/worst
 417         # video-only or audio-only formats (as per
 418         # https://github.com/ytdl-org/youtube-dl/pull/5556)
 419         formats = [
 420             {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL},
 421             {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL},
 422         ]
 423         info_dict = _make_result(formats)
 424
 425         ydl = YDL({'format': 'best'})
 426         ydl.process_ie_result(info_dict.copy())
 427         downloaded = ydl.downloaded_info_dicts[0]
 428         self.assertEqual(downloaded['format_id'], 'high')
 429
 430         ydl = YDL({'format': 'worst'})
 431         ydl.process_ie_result(info_dict.copy())
 432         downloaded = ydl.downloaded_info_dicts[0]
 433         self.assertEqual(downloaded['format_id'], 'low')
 434
 435     def test_format_not_available(self):
 436         formats = [
 437             {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL},
 438             {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 439         ]
 440         info_dict = _make_result(formats)
 441
 442         # This must fail since complete video-audio format does not match filter
 443         # and extractor does not provide incomplete only formats (i.e. only
 444         # video-only or audio-only).
 445         ydl = YDL({'format': 'best[height>360]'})
 446         self.assertRaises(ExtractorError, ydl.process_ie_result, info_dict.copy())
 447
 448     def test_format_selection_issue_10083(self):
 449         # See https://github.com/ytdl-org/youtube-dl/issues/10083
 450         formats = [
 451             {'format_id': 'regular', 'height': 360, 'url': TEST_URL},
 452             {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL},
 453             {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL},
 454         ]
 455         info_dict = _make_result(formats)
 456
 457         ydl = YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'})
 458         ydl.process_ie_result(info_dict.copy())
 459         self.assertEqual(ydl.downloaded_info_dicts[0]['format_id'], 'video+audio')
 460
 461     def test_invalid_format_specs(self):
 462         def assert_syntax_error(format_spec):
 463             ydl = YDL({'format': format_spec})
 464             info_dict = _make_result([{'format_id': 'foo', 'url': TEST_URL}])
 465             self.assertRaises(SyntaxError, ydl.process_ie_result, info_dict)
 466
 467         assert_syntax_error('bestvideo,,best')
 468         assert_syntax_error('+bestaudio')
 469         assert_syntax_error('bestvideo+')
 470         assert_syntax_error('/')
 471
 472     def test_format_filtering(self):
 473         formats = [
 474             {'format_id': 'A', 'filesize': 500, 'width': 1000},
 475             {'format_id': 'B', 'filesize': 1000, 'width': 500},
 476             {'format_id': 'C', 'filesize': 1000, 'width': 400},
 477             {'format_id': 'D', 'filesize': 2000, 'width': 600},
 478             {'format_id': 'E', 'filesize': 3000},
 479             {'format_id': 'F'},
 480             {'format_id': 'G', 'filesize': 1000000},
 481         ]
 482         for f in formats:
 483             f['url'] = 'http://_/'
 484             f['ext'] = 'unknown'
 485         info_dict = _make_result(formats)
 486
 487         ydl = YDL({'format': 'best[filesize<3000]'})
 488         ydl.process_ie_result(info_dict)
 489         downloaded = ydl.downloaded_info_dicts[0]
 490         self.assertEqual(downloaded['format_id'], 'D')
 491
 492         ydl = YDL({'format': 'best[filesize<=3000]'})
 493         ydl.process_ie_result(info_dict)
 494         downloaded = ydl.downloaded_info_dicts[0]
 495         self.assertEqual(downloaded['format_id'], 'E')
 496
 497         ydl = YDL({'format': 'best[filesize <= ? 3000]'})
 498         ydl.process_ie_result(info_dict)
 499         downloaded = ydl.downloaded_info_dicts[0]
 500         self.assertEqual(downloaded['format_id'], 'F')
 501
 502         ydl = YDL({'format': 'best [filesize = 1000] [width>450]'})
 503         ydl.process_ie_result(info_dict)
 504         downloaded = ydl.downloaded_info_dicts[0]
 505         self.assertEqual(downloaded['format_id'], 'B')
 506
 507         ydl = YDL({'format': 'best [filesize = 1000] [width!=450]'})
 508         ydl.process_ie_result(info_dict)
 509         downloaded = ydl.downloaded_info_dicts[0]
 510         self.assertEqual(downloaded['format_id'], 'C')
 511
 512         ydl = YDL({'format': '[filesize>?1]'})
 513         ydl.process_ie_result(info_dict)
 514         downloaded = ydl.downloaded_info_dicts[0]
 515         self.assertEqual(downloaded['format_id'], 'G')
 516
 517         ydl = YDL({'format': '[filesize<1M]'})
 518         ydl.process_ie_result(info_dict)
 519         downloaded = ydl.downloaded_info_dicts[0]
 520         self.assertEqual(downloaded['format_id'], 'E')
 521
 522         ydl = YDL({'format': '[filesize<1MiB]'})
 523         ydl.process_ie_result(info_dict)
 524         downloaded = ydl.downloaded_info_dicts[0]
 525         self.assertEqual(downloaded['format_id'], 'G')
 526
 527         ydl = YDL({'format': 'all[width>=400][width<=600]'})
 528         ydl.process_ie_result(info_dict)
 529         downloaded_ids = [info['format_id'] for info in ydl.downloaded_info_dicts]
 530         self.assertEqual(downloaded_ids, ['B', 'C', 'D'])
 531
 532         ydl = YDL({'format': 'best[height<40]'})
 533         try:
 534             ydl.process_ie_result(info_dict)
 535         except ExtractorError:
 536             pass
 537         self.assertEqual(ydl.downloaded_info_dicts, [])
 538
 539     def test_default_format_spec(self):
 540         ydl = YDL({'simulate': True})
 541         self.assertEqual(ydl._default_format_spec({}), 'bestvideo*+bestaudio/best')
 542
 543         ydl = YDL({})
 544         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 545
 546         ydl = YDL({'simulate': True})
 547         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'bestvideo*+bestaudio/best')
 548
 549         ydl = YDL({'outtmpl': '-'})
 550         self.assertEqual(ydl._default_format_spec({}), 'best/bestvideo+bestaudio')
 551
 552         ydl = YDL({})
 553         self.assertEqual(ydl._default_format_spec({}, download=False), 'bestvideo*+bestaudio/best')
 554         self.assertEqual(ydl._default_format_spec({'is_live': True}), 'best/bestvideo+bestaudio')
 555
 556
 557 class TestYoutubeDL(unittest.TestCase):
 558     def test_subtitles(self):
 559         def s_formats(lang, autocaption=False):
 560             return [{
 561                 'ext': ext,
 562                 'url': 'http://localhost/video.%s.%s' % (lang, ext),
 563                 '_auto': autocaption,
 564             } for ext in ['vtt', 'srt', 'ass']]
 565         subtitles = dict((l, s_formats(l)) for l in ['en', 'fr', 'es'])
 566         auto_captions = dict((l, s_formats(l, True)) for l in ['it', 'pt', 'es'])
 567         info_dict = {
 568             'id': 'test',
 569             'title': 'Test',
 570             'url': 'http://localhost/video.mp4',
 571             'subtitles': subtitles,
 572             'automatic_captions': auto_captions,
 573             'extractor': 'TEST',
 574             'webpage_url': 'http://example.com/watch?v=shenanigans',
 575         }
 576
 577         def get_info(params={}):
 578             params.setdefault('simulate', True)
 579             ydl = YDL(params)
 580             ydl.report_warning = lambda *args, **kargs: None
 581             return ydl.process_video_result(info_dict, download=False)
 582
 583         result = get_info()
 584         self.assertFalse(result.get('requested_subtitles'))
 585         self.assertEqual(result['subtitles'], subtitles)
 586         self.assertEqual(result['automatic_captions'], auto_captions)
 587
 588         result = get_info({'writesubtitles': True})
 589         subs = result['requested_subtitles']
 590         self.assertTrue(subs)
 591         self.assertEqual(set(subs.keys()), set(['en']))
 592         self.assertTrue(subs['en'].get('data') is None)
 593         self.assertEqual(subs['en']['ext'], 'ass')
 594
 595         result = get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'})
 596         subs = result['requested_subtitles']
 597         self.assertEqual(subs['en']['ext'], 'srt')
 598
 599         result = get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']})
 600         subs = result['requested_subtitles']
 601         self.assertTrue(subs)
 602         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 603
 604         result = get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']})
 605         subs = result['requested_subtitles']
 606         self.assertTrue(subs)
 607         self.assertEqual(set(subs.keys()), set(['es', 'fr']))
 608
 609         result = get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']})
 610         subs = result['requested_subtitles']
 611         self.assertTrue(subs)
 612         self.assertEqual(set(subs.keys()), set(['fr']))
 613
 614         result = get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']})
 615         subs = result['requested_subtitles']
 616         self.assertTrue(subs)
 617         self.assertEqual(set(subs.keys()), set(['en']))
 618
 619         result = get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']})
 620         subs = result['requested_subtitles']
 621         self.assertTrue(subs)
 622         self.assertEqual(set(subs.keys()), set(['es', 'en']))
 623
 624         result = get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 625         subs = result['requested_subtitles']
 626         self.assertTrue(subs)
 627         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 628         self.assertFalse(subs['es']['_auto'])
 629         self.assertTrue(subs['pt']['_auto'])
 630
 631         result = get_info({'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']})
 632         subs = result['requested_subtitles']
 633         self.assertTrue(subs)
 634         self.assertEqual(set(subs.keys()), set(['es', 'pt']))
 635         self.assertTrue(subs['es']['_auto'])
 636         self.assertTrue(subs['pt']['_auto'])
 637
 638     def test_add_extra_info(self):
 639         test_dict = {
 640             'extractor': 'Foo',
 641         }
 642         extra_info = {
 643             'extractor': 'Bar',
 644             'playlist': 'funny videos',
 645         }
 646         YDL.add_extra_info(test_dict, extra_info)
 647         self.assertEqual(test_dict['extractor'], 'Foo')
 648         self.assertEqual(test_dict['playlist'], 'funny videos')
 649
 650     def test_prepare_filename(self):
 651         info = {
 652             'id': '1234',
 653             'ext': 'mp4',
 654             'width': None,
 655             'height': 1080,
 656             'title1': '$PATH',
 657             'title2': '%PATH%',
 658         }
 659
 660         def fname(templ, na_placeholder='NA'):
 661             params = {'outtmpl': templ}
 662             if na_placeholder != 'NA':
 663                 params['outtmpl_na_placeholder'] = na_placeholder
 664             ydl = YoutubeDL(params)
 665             return ydl.prepare_filename(info)
 666         self.assertEqual(fname('%(id)s.%(ext)s'), '1234.mp4')
 667         self.assertEqual(fname('%(id)s-%(width)s.%(ext)s'), '1234-NA.mp4')
 668         NA_TEST_OUTTMPL = '%(uploader_date)s-%(width)d-%(id)s.%(ext)s'
 669         # Replace missing fields with 'NA' by default
 670         self.assertEqual(fname(NA_TEST_OUTTMPL), 'NA-NA-1234.mp4')
 671         # Or by provided placeholder
 672         self.assertEqual(fname(NA_TEST_OUTTMPL, na_placeholder='none'), 'none-none-1234.mp4')
 673         self.assertEqual(fname(NA_TEST_OUTTMPL, na_placeholder=''), '--1234.mp4')
 674         self.assertEqual(fname('%(height)d.%(ext)s'), '1080.mp4')
 675         self.assertEqual(fname('%(height)6d.%(ext)s'), '  1080.mp4')
 676         self.assertEqual(fname('%(height)-6d.%(ext)s'), '1080  .mp4')
 677         self.assertEqual(fname('%(height)06d.%(ext)s'), '001080.mp4')
 678         self.assertEqual(fname('%(height) 06d.%(ext)s'), ' 01080.mp4')
 679         self.assertEqual(fname('%(height)   06d.%(ext)s'), ' 01080.mp4')
 680         self.assertEqual(fname('%(height)0 6d.%(ext)s'), ' 01080.mp4')
 681         self.assertEqual(fname('%(height)0   6d.%(ext)s'), ' 01080.mp4')
 682         self.assertEqual(fname('%(height)   0   6d.%(ext)s'), ' 01080.mp4')
 683         self.assertEqual(fname('%%'), '%')
 684         self.assertEqual(fname('%%%%'), '%%')
 685         self.assertEqual(fname('%%(height)06d.%(ext)s'), '%(height)06d.mp4')
 686         self.assertEqual(fname('%(width)06d.%(ext)s'), 'NA.mp4')
 687         self.assertEqual(fname('%(width)06d.%%(ext)s'), 'NA.%(ext)s')
 688         self.assertEqual(fname('%%(width)06d.%(ext)s'), '%(width)06d.mp4')
 689         self.assertEqual(fname('Hello %(title1)s'), 'Hello $PATH')
 690         self.assertEqual(fname('Hello %(title2)s'), 'Hello %PATH%')
 691
 692     def test_format_note(self):
 693         ydl = YoutubeDL()
 694         self.assertEqual(ydl._format_note({}), '')
 695         assertRegexpMatches(self, ydl._format_note({
 696             'vbr': 10,
 697         }), r'^\s*10k$')
 698         assertRegexpMatches(self, ydl._format_note({
 699             'fps': 30,
 700         }), r'^30fps$')
 701
 702     def test_postprocessors(self):
 703         filename = 'post-processor-testfile.mp4'
 704         audiofile = filename + '.mp3'
 705
 706         class SimplePP(PostProcessor):
 707             def run(self, info):
 708                 with open(audiofile, 'wt') as f:
 709                     f.write('EXAMPLE')
 710                 return [info['filepath']], info
 711
 712         def run_pp(params, PP):
 713             with open(filename, 'wt') as f:
 714                 f.write('EXAMPLE')
 715             ydl = YoutubeDL(params)
 716             ydl.add_post_processor(PP())
 717             ydl.post_process(filename, {'filepath': filename})
 718
 719         run_pp({'keepvideo': True}, SimplePP)
 720         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 721         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 722         os.unlink(filename)
 723         os.unlink(audiofile)
 724
 725         run_pp({'keepvideo': False}, SimplePP)
 726         self.assertFalse(os.path.exists(filename), '%s exists' % filename)
 727         self.assertTrue(os.path.exists(audiofile), '%s doesn\'t exist' % audiofile)
 728         os.unlink(audiofile)
 729
 730         class ModifierPP(PostProcessor):
 731             def run(self, info):
 732                 with open(info['filepath'], 'wt') as f:
 733                     f.write('MODIFIED')
 734                 return [], info
 735
 736         run_pp({'keepvideo': False}, ModifierPP)
 737         self.assertTrue(os.path.exists(filename), '%s doesn\'t exist' % filename)
 738         os.unlink(filename)
 739
 740     def test_match_filter(self):
 741         class FilterYDL(YDL):
 742             def __init__(self, *args, **kwargs):
 743                 super(FilterYDL, self).__init__(*args, **kwargs)
 744                 self.params['simulate'] = True
 745
 746             def process_info(self, info_dict):
 747                 super(YDL, self).process_info(info_dict)
 748
 749             def _match_entry(self, info_dict, incomplete):
 750                 res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
 751                 if res is None:
 752                     self.downloaded_info_dicts.append(info_dict)
 753                 return res
 754
 755         first = {
 756             'id': '1',
 757             'url': TEST_URL,
 758             'title': 'one',
 759             'extractor': 'TEST',
 760             'duration': 30,
 761             'filesize': 10 * 1024,
 762             'playlist_id': '42',
 763             'uploader': "變態妍字幕版 太妍 тест",
 764             'creator': "тест ' 123 ' тест--",
 765             'webpage_url': 'http://example.com/watch?v=shenanigans',
 766         }
 767         second = {
 768             'id': '2',
 769             'url': TEST_URL,
 770             'title': 'two',
 771             'extractor': 'TEST',
 772             'duration': 10,
 773             'description': 'foo',
 774             'filesize': 5 * 1024,
 775             'playlist_id': '43',
 776             'uploader': "тест 123",
 777             'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
 778         }
 779         videos = [first, second]
 780
 781         def get_videos(filter_=None):
 782             ydl = FilterYDL({'match_filter': filter_})
 783             for v in videos:
 784                 ydl.process_ie_result(v, download=True)
 785             return [v['id'] for v in ydl.downloaded_info_dicts]
 786
 787         res = get_videos()
 788         self.assertEqual(res, ['1', '2'])
 789
 790         def f(v):
 791             if v['id'] == '1':
 792                 return None
 793             else:
 794                 return 'Video id is not 1'
 795         res = get_videos(f)
 796         self.assertEqual(res, ['1'])
 797
 798         f = match_filter_func('duration < 30')
 799         res = get_videos(f)
 800         self.assertEqual(res, ['2'])
 801
 802         f = match_filter_func('description = foo')
 803         res = get_videos(f)
 804         self.assertEqual(res, ['2'])
 805
 806         f = match_filter_func('description =? foo')
 807         res = get_videos(f)
 808         self.assertEqual(res, ['1', '2'])
 809
 810         f = match_filter_func('filesize > 5KiB')
 811         res = get_videos(f)
 812         self.assertEqual(res, ['1'])
 813
 814         f = match_filter_func('playlist_id = 42')
 815         res = get_videos(f)
 816         self.assertEqual(res, ['1'])
 817
 818         f = match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
 819         res = get_videos(f)
 820         self.assertEqual(res, ['1'])
 821
 822         f = match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
 823         res = get_videos(f)
 824         self.assertEqual(res, ['2'])
 825
 826         f = match_filter_func('creator = "тест \' 123 \' тест--"')
 827         res = get_videos(f)
 828         self.assertEqual(res, ['1'])
 829
 830         f = match_filter_func("creator = 'тест \\' 123 \\' тест--'")
 831         res = get_videos(f)
 832         self.assertEqual(res, ['1'])
 833
 834         f = match_filter_func(r"creator = 'тест \' 123 \' тест--' & duration > 30")
 835         res = get_videos(f)
 836         self.assertEqual(res, [])
 837
 838     def test_playlist_items_selection(self):
 839         entries = [{
 840             'id': compat_str(i),
 841             'title': compat_str(i),
 842             'url': TEST_URL,
 843         } for i in range(1, 5)]
 844         playlist = {
 845             '_type': 'playlist',
 846             'id': 'test',
 847             'entries': entries,
 848             'extractor': 'test:playlist',
 849             'extractor_key': 'test:playlist',
 850             'webpage_url': 'http://example.com',
 851         }
 852
 853         def get_downloaded_info_dicts(params):
 854             ydl = YDL(params)
 855             # make a deep copy because the dictionary and nested entries
 856             # can be modified
 857             ydl.process_ie_result(copy.deepcopy(playlist))
 858             return ydl.downloaded_info_dicts
 859
 860         def get_ids(params):
 861             return [int(v['id']) for v in get_downloaded_info_dicts(params)]
 862
 863         result = get_ids({})
 864         self.assertEqual(result, [1, 2, 3, 4])
 865
 866         result = get_ids({'playlistend': 10})
 867         self.assertEqual(result, [1, 2, 3, 4])
 868
 869         result = get_ids({'playlistend': 2})
 870         self.assertEqual(result, [1, 2])
 871
 872         result = get_ids({'playliststart': 10})
 873         self.assertEqual(result, [])
 874
 875         result = get_ids({'playliststart': 2})
 876         self.assertEqual(result, [2, 3, 4])
 877
 878         result = get_ids({'playlist_items': '2-4'})
 879         self.assertEqual(result, [2, 3, 4])
 880
 881         result = get_ids({'playlist_items': '2,4'})
 882         self.assertEqual(result, [2, 4])
 883
 884         result = get_ids({'playlist_items': '10'})
 885         self.assertEqual(result, [])
 886
 887         result = get_ids({'playlist_items': '3-10'})
 888         self.assertEqual(result, [3, 4])
 889
 890         result = get_ids({'playlist_items': '2-4,3-4,3'})
 891         self.assertEqual(result, [2, 3, 4])
 892
 893         # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
 894         # @{
 895         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 896         self.assertEqual(result[0]['playlist_index'], 2)
 897         self.assertEqual(result[1]['playlist_index'], 3)
 898
 899         result = get_downloaded_info_dicts({'playlist_items': '2-4,3-4,3'})
 900         self.assertEqual(result[0]['playlist_index'], 2)
 901         self.assertEqual(result[1]['playlist_index'], 3)
 902         self.assertEqual(result[2]['playlist_index'], 4)
 903
 904         result = get_downloaded_info_dicts({'playlist_items': '4,2'})
 905         self.assertEqual(result[0]['playlist_index'], 4)
 906         self.assertEqual(result[1]['playlist_index'], 2)
 907         # @}
 908
 909     def test_urlopen_no_file_protocol(self):
 910         # see https://github.com/ytdl-org/youtube-dl/issues/8227
 911         ydl = YDL()
 912         self.assertRaises(compat_urllib_error.URLError, ydl.urlopen, 'file:///etc/passwd')
 913
 914     def test_do_not_override_ie_key_in_url_transparent(self):
 915         ydl = YDL()
 916
 917         class Foo1IE(InfoExtractor):
 918             _VALID_URL = r'foo1:'
 919
 920             def _real_extract(self, url):
 921                 return {
 922                     '_type': 'url_transparent',
 923                     'url': 'foo2:',
 924                     'ie_key': 'Foo2',
 925                     'title': 'foo1 title',
 926                     'id': 'foo1_id',
 927                 }
 928
 929         class Foo2IE(InfoExtractor):
 930             _VALID_URL = r'foo2:'
 931
 932             def _real_extract(self, url):
 933                 return {
 934                     '_type': 'url',
 935                     'url': 'foo3:',
 936                     'ie_key': 'Foo3',
 937                 }
 938
 939         class Foo3IE(InfoExtractor):
 940             _VALID_URL = r'foo3:'
 941
 942             def _real_extract(self, url):
 943                 return _make_result([{'url': TEST_URL}], title='foo3 title')
 944
 945         ydl.add_info_extractor(Foo1IE(ydl))
 946         ydl.add_info_extractor(Foo2IE(ydl))
 947         ydl.add_info_extractor(Foo3IE(ydl))
 948         ydl.extract_info('foo1:')
 949         downloaded = ydl.downloaded_info_dicts[0]
 950         self.assertEqual(downloaded['url'], TEST_URL)
 951         self.assertEqual(downloaded['title'], 'foo1 title')
 952         self.assertEqual(downloaded['id'], 'testid')
 953         self.assertEqual(downloaded['extractor'], 'testex')
 954         self.assertEqual(downloaded['extractor_key'], 'TestEx')
 955
 956     # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
 957     def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self):
 958
 959         class _YDL(YDL):
 960             def __init__(self, *args, **kwargs):
 961                 super(_YDL, self).__init__(*args, **kwargs)
 962
 963             def trouble(self, s, tb=None):
 964                 pass
 965
 966         ydl = _YDL({
 967             'format': 'extra',
 968             'ignoreerrors': True,
 969         })
 970
 971         class VideoIE(InfoExtractor):
 972             _VALID_URL = r'video:(?P<id>\d+)'
 973
 974             def _real_extract(self, url):
 975                 video_id = self._match_id(url)
 976                 formats = [{
 977                     'format_id': 'default',
 978                     'url': 'url:',
 979                 }]
 980                 if video_id == '0':
 981                     raise ExtractorError('foo')
 982                 if video_id == '2':
 983                     formats.append({
 984                         'format_id': 'extra',
 985                         'url': TEST_URL,
 986                     })
 987                 return {
 988                     'id': video_id,
 989                     'title': 'Video %s' % video_id,
 990                     'formats': formats,
 991                 }
 992
 993         class PlaylistIE(InfoExtractor):
 994             _VALID_URL = r'playlist:'
 995
 996             def _entries(self):
 997                 for n in range(3):
 998                     video_id = compat_str(n)
 999                     yield {
1000                         '_type': 'url_transparent',
1001                         'ie_key': VideoIE.ie_key(),
1002                         'id': video_id,
1003                         'url': 'video:%s' % video_id,
1004                         'title': 'Video Transparent %s' % video_id,
1005                     }
1006
1007             def _real_extract(self, url):
1008                 return self.playlist_result(self._entries())
1009
1010         ydl.add_info_extractor(VideoIE(ydl))
1011         ydl.add_info_extractor(PlaylistIE(ydl))
1012         info = ydl.extract_info('playlist:')
1013         entries = info['entries']
1014         self.assertEqual(len(entries), 3)
1015         self.assertTrue(entries[0] is None)
1016         self.assertTrue(entries[1] is None)
1017         self.assertEqual(len(ydl.downloaded_info_dicts), 1)
1018         downloaded = ydl.downloaded_info_dicts[0]
1019         self.assertEqual(entries[2], downloaded)
1020         self.assertEqual(downloaded['url'], TEST_URL)
1021         self.assertEqual(downloaded['title'], 'Video Transparent 2')
1022         self.assertEqual(downloaded['id'], '2')
1023         self.assertEqual(downloaded['extractor'], 'Video')
1024         self.assertEqual(downloaded['extractor_key'], 'Video')
1025
1026
1027 if __name__ == '__main__':
1028     unittest.main()