4 from __future__
import unicode_literals
6 # Allow direct execution
10 sys
.path
.insert(0, os
.path
.dirname(os
.path
.dirname(os
.path
.abspath(__file__
))))
15 from test
.helper
import FakeYDL
, assertRegexpMatches
16 from yt_dlp
import YoutubeDL
17 from yt_dlp
.compat
import compat_os_name
, compat_setenv
, compat_str
, compat_urllib_error
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 ExtractorError
, int_or_none
, match_filter_func
, LazyList
23 TEST_URL
= 'http://localhost/sample.mp4'
27 def __init__(self
, *args
, **kwargs
):
28 super(YDL
, self
).__init
__(*args
, **kwargs
)
29 self
.downloaded_info_dicts
= []
32 def process_info(self
, info_dict
):
33 info_dict
.pop('__original_infodict', None)
34 self
.downloaded_info_dicts
.append(info_dict
)
36 def to_screen(self
, msg
):
39 def dl(self
, *args
, **kwargs
):
40 assert False, 'Downloader must not be invoked for test_YoutubeDL'
43 def _make_result(formats
, **kwargs
):
47 'title': 'testttitle',
48 'extractor': 'testex',
49 'extractor_key': 'TestEx',
50 'webpage_url': 'http://example.com/watch?v=shenanigans',
56 class TestFormatSelection(unittest
.TestCase
):
57 def test_prefer_free_formats(self
):
58 # Same resolution => download webm
60 ydl
.params
['prefer_free_formats'] = True
62 {'ext': 'webm', 'height': 460, 'url': TEST_URL}
,
63 {'ext': 'mp4', 'height': 460, 'url': TEST_URL}
,
65 info_dict
= _make_result(formats
)
67 yie
._sort
_formats
(info_dict
['formats'])
68 ydl
.process_ie_result(info_dict
)
69 downloaded
= ydl
.downloaded_info_dicts
[0]
70 self
.assertEqual(downloaded
['ext'], 'webm')
72 # Different resolution => download best quality (mp4)
74 ydl
.params
['prefer_free_formats'] = True
76 {'ext': 'webm', 'height': 720, 'url': TEST_URL}
,
77 {'ext': 'mp4', 'height': 1080, 'url': TEST_URL}
,
79 info_dict
['formats'] = formats
81 yie
._sort
_formats
(info_dict
['formats'])
82 ydl
.process_ie_result(info_dict
)
83 downloaded
= ydl
.downloaded_info_dicts
[0]
84 self
.assertEqual(downloaded
['ext'], 'mp4')
86 # No prefer_free_formats => prefer mp4 and webm
88 ydl
.params
['prefer_free_formats'] = False
90 {'ext': 'webm', 'height': 720, 'url': TEST_URL}
,
91 {'ext': 'mp4', 'height': 720, 'url': TEST_URL}
,
92 {'ext': 'flv', 'height': 720, 'url': TEST_URL}
,
94 info_dict
['formats'] = formats
96 yie
._sort
_formats
(info_dict
['formats'])
97 ydl
.process_ie_result(info_dict
)
98 downloaded
= ydl
.downloaded_info_dicts
[0]
99 self
.assertEqual(downloaded
['ext'], 'mp4')
102 ydl
.params
['prefer_free_formats'] = False
104 {'ext': 'flv', 'height': 720, 'url': TEST_URL}
,
105 {'ext': 'webm', 'height': 720, 'url': TEST_URL}
,
107 info_dict
['formats'] = formats
109 yie
._sort
_formats
(info_dict
['formats'])
110 ydl
.process_ie_result(info_dict
)
111 downloaded
= ydl
.downloaded_info_dicts
[0]
112 self
.assertEqual(downloaded
['ext'], 'webm')
114 def test_format_selection(self
):
116 {'format_id': '35', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL}
,
117 {'format_id': 'example-with-dashes', 'ext': 'webm', 'preference': 1, 'url': TEST_URL}
,
118 {'format_id': '45', 'ext': 'webm', 'preference': 2, 'url': TEST_URL}
,
119 {'format_id': '47', 'ext': 'webm', 'preference': 3, 'url': TEST_URL}
,
120 {'format_id': '2', 'ext': 'flv', 'preference': 4, 'url': TEST_URL}
,
122 info_dict
= _make_result(formats
)
124 def test(inp
, *expected
, multi
=False):
127 'allow_multiple_video_streams': multi
,
128 'allow_multiple_audio_streams': multi
,
130 ydl
.process_ie_result(info_dict
.copy())
131 downloaded
= map(lambda x
: x
['format_id'], ydl
.downloaded_info_dicts
)
132 self
.assertEqual(list(downloaded
), list(expected
))
135 test('20/71/worst', '35')
137 test('webm/mp4', '47')
138 test('3gp/40/mp4', '35')
139 test('example-with-dashes', 'example-with-dashes')
140 test('all', '35', 'example-with-dashes', '45', '47', '2') # Order doesn't actually matter for this
141 test('mergeall', '2+47+45+example-with-dashes+35', multi
=True)
143 def test_format_selection_audio(self
):
145 {'format_id': 'audio-low', 'ext': 'webm', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL}
,
146 {'format_id': 'audio-mid', 'ext': 'webm', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL}
,
147 {'format_id': 'audio-high', 'ext': 'flv', 'preference': 3, 'vcodec': 'none', 'url': TEST_URL}
,
148 {'format_id': 'vid', 'ext': 'mp4', 'preference': 4, 'url': TEST_URL}
,
150 info_dict
= _make_result(formats
)
152 ydl
= YDL({'format': 'bestaudio'}
)
153 ydl
.process_ie_result(info_dict
.copy())
154 downloaded
= ydl
.downloaded_info_dicts
[0]
155 self
.assertEqual(downloaded
['format_id'], 'audio-high')
157 ydl
= YDL({'format': 'worstaudio'}
)
158 ydl
.process_ie_result(info_dict
.copy())
159 downloaded
= ydl
.downloaded_info_dicts
[0]
160 self
.assertEqual(downloaded
['format_id'], 'audio-low')
163 {'format_id': 'vid-low', 'ext': 'mp4', 'preference': 1, 'url': TEST_URL}
,
164 {'format_id': 'vid-high', 'ext': 'mp4', 'preference': 2, 'url': TEST_URL}
,
166 info_dict
= _make_result(formats
)
168 ydl
= YDL({'format': 'bestaudio/worstaudio/best'}
)
169 ydl
.process_ie_result(info_dict
.copy())
170 downloaded
= ydl
.downloaded_info_dicts
[0]
171 self
.assertEqual(downloaded
['format_id'], 'vid-high')
173 def test_format_selection_audio_exts(self
):
175 {'format_id': 'mp3-64', 'ext': 'mp3', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'}
,
176 {'format_id': 'ogg-64', 'ext': 'ogg', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'}
,
177 {'format_id': 'aac-64', 'ext': 'aac', 'abr': 64, 'url': 'http://_', 'vcodec': 'none'}
,
178 {'format_id': 'mp3-32', 'ext': 'mp3', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'}
,
179 {'format_id': 'aac-32', 'ext': 'aac', 'abr': 32, 'url': 'http://_', 'vcodec': 'none'}
,
182 info_dict
= _make_result(formats
)
183 ydl
= YDL({'format': 'best'}
)
185 ie
._sort
_formats
(info_dict
['formats'])
186 ydl
.process_ie_result(copy
.deepcopy(info_dict
))
187 downloaded
= ydl
.downloaded_info_dicts
[0]
188 self
.assertEqual(downloaded
['format_id'], 'aac-64')
190 ydl
= YDL({'format': 'mp3'}
)
192 ie
._sort
_formats
(info_dict
['formats'])
193 ydl
.process_ie_result(copy
.deepcopy(info_dict
))
194 downloaded
= ydl
.downloaded_info_dicts
[0]
195 self
.assertEqual(downloaded
['format_id'], 'mp3-64')
197 ydl
= YDL({'prefer_free_formats': True}
)
199 ie
._sort
_formats
(info_dict
['formats'])
200 ydl
.process_ie_result(copy
.deepcopy(info_dict
))
201 downloaded
= ydl
.downloaded_info_dicts
[0]
202 self
.assertEqual(downloaded
['format_id'], 'ogg-64')
204 def test_format_selection_video(self
):
206 {'format_id': 'dash-video-low', 'ext': 'mp4', 'preference': 1, 'acodec': 'none', 'url': TEST_URL}
,
207 {'format_id': 'dash-video-high', 'ext': 'mp4', 'preference': 2, 'acodec': 'none', 'url': TEST_URL}
,
208 {'format_id': 'vid', 'ext': 'mp4', 'preference': 3, 'url': TEST_URL}
,
210 info_dict
= _make_result(formats
)
212 ydl
= YDL({'format': 'bestvideo'}
)
213 ydl
.process_ie_result(info_dict
.copy())
214 downloaded
= ydl
.downloaded_info_dicts
[0]
215 self
.assertEqual(downloaded
['format_id'], 'dash-video-high')
217 ydl
= YDL({'format': 'worstvideo'}
)
218 ydl
.process_ie_result(info_dict
.copy())
219 downloaded
= ydl
.downloaded_info_dicts
[0]
220 self
.assertEqual(downloaded
['format_id'], 'dash-video-low')
222 ydl
= YDL({'format': 'bestvideo[format_id^=dash][format_id$=low]'}
)
223 ydl
.process_ie_result(info_dict
.copy())
224 downloaded
= ydl
.downloaded_info_dicts
[0]
225 self
.assertEqual(downloaded
['format_id'], 'dash-video-low')
228 {'format_id': 'vid-vcodec-dot', 'ext': 'mp4', 'preference': 1, 'vcodec': 'avc1.123456', 'acodec': 'none', 'url': TEST_URL}
,
230 info_dict
= _make_result(formats
)
232 ydl
= YDL({'format': 'bestvideo[vcodec=avc1.123456]'}
)
233 ydl
.process_ie_result(info_dict
.copy())
234 downloaded
= ydl
.downloaded_info_dicts
[0]
235 self
.assertEqual(downloaded
['format_id'], 'vid-vcodec-dot')
237 def test_format_selection_string_ops(self
):
239 {'format_id': 'abc-cba', 'ext': 'mp4', 'url': TEST_URL}
,
240 {'format_id': 'zxc-cxz', 'ext': 'webm', 'url': TEST_URL}
,
242 info_dict
= _make_result(formats
)
245 ydl
= YDL({'format': '[format_id=abc-cba]'}
)
246 ydl
.process_ie_result(info_dict
.copy())
247 downloaded
= ydl
.downloaded_info_dicts
[0]
248 self
.assertEqual(downloaded
['format_id'], 'abc-cba')
250 # does not equal (!=)
251 ydl
= YDL({'format': '[format_id!=abc-cba]'}
)
252 ydl
.process_ie_result(info_dict
.copy())
253 downloaded
= ydl
.downloaded_info_dicts
[0]
254 self
.assertEqual(downloaded
['format_id'], 'zxc-cxz')
256 ydl
= YDL({'format': '[format_id!=abc-cba][format_id!=zxc-cxz]'}
)
257 self
.assertRaises(ExtractorError
, ydl
.process_ie_result
, info_dict
.copy())
260 ydl
= YDL({'format': '[format_id^=abc]'}
)
261 ydl
.process_ie_result(info_dict
.copy())
262 downloaded
= ydl
.downloaded_info_dicts
[0]
263 self
.assertEqual(downloaded
['format_id'], 'abc-cba')
265 # does not start with (!^=)
266 ydl
= YDL({'format': '[format_id!^=abc]'}
)
267 ydl
.process_ie_result(info_dict
.copy())
268 downloaded
= ydl
.downloaded_info_dicts
[0]
269 self
.assertEqual(downloaded
['format_id'], 'zxc-cxz')
271 ydl
= YDL({'format': '[format_id!^=abc][format_id!^=zxc]'}
)
272 self
.assertRaises(ExtractorError
, ydl
.process_ie_result
, info_dict
.copy())
275 ydl
= YDL({'format': '[format_id$=cba]'}
)
276 ydl
.process_ie_result(info_dict
.copy())
277 downloaded
= ydl
.downloaded_info_dicts
[0]
278 self
.assertEqual(downloaded
['format_id'], 'abc-cba')
280 # does not end 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'], 'zxc-cxz')
286 ydl
= YDL({'format': '[format_id!$=cba][format_id!$=cxz]'}
)
287 self
.assertRaises(ExtractorError
, ydl
.process_ie_result
, info_dict
.copy())
290 ydl
= YDL({'format': '[format_id*=bc-cb]'}
)
291 ydl
.process_ie_result(info_dict
.copy())
292 downloaded
= ydl
.downloaded_info_dicts
[0]
293 self
.assertEqual(downloaded
['format_id'], 'abc-cba')
295 # does not contain (!*=)
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'], 'zxc-cxz')
301 ydl
= YDL({'format': '[format_id!*=abc][format_id!*=zxc]'}
)
302 self
.assertRaises(ExtractorError
, ydl
.process_ie_result
, info_dict
.copy())
304 ydl
= YDL({'format': '[format_id!*=-]'}
)
305 self
.assertRaises(ExtractorError
, ydl
.process_ie_result
, info_dict
.copy())
307 def test_youtube_format_selection(self
):
308 # FIXME: Rewrite in accordance with the new format sorting options
312 '38', '37', '46', '22', '45', '35', '44', '18', '34', '43', '6', '5', '17', '36', '13',
313 # Apple HTTP Live Streaming
314 '96', '95', '94', '93', '92', '132', '151',
316 '85', '84', '102', '83', '101', '82', '100',
318 '137', '248', '136', '247', '135', '246',
319 '245', '244', '134', '243', '133', '242', '160',
321 '141', '172', '140', '171', '139',
324 def format_info(f_id
):
325 info
= YoutubeIE
._formats
[f_id
].copy()
327 # XXX: In real cases InfoExtractor._parse_mpd_formats() fills up 'acodec'
328 # and 'vcodec', while in tests such information is incomplete since
329 # commit a6c2c24479e5f4827ceb06f64d855329c0a6f593
330 # test_YoutubeDL.test_youtube_format_selection is broken without
332 if 'acodec' in info
and 'vcodec' not in info
:
333 info
['vcodec'] = 'none'
334 elif 'vcodec' in info
and 'acodec' not in info
:
335 info
['acodec'] = 'none'
337 info
['format_id'] = f_id
338 info
['url'] = 'url:' + f_id
340 formats_order
= [format_info(f_id
) for f_id
in order
]
342 info_dict
= _make_result(list(formats_order
), extractor
='youtube')
343 ydl
= YDL({'format': 'bestvideo+bestaudio'}
)
345 yie
._sort
_formats
(info_dict
['formats'])
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')
351 info_dict
= _make_result(list(formats_order
), extractor
='youtube')
352 ydl
= YDL({'format': 'bestvideo[height>=999999]+bestaudio/best'}
)
354 yie
._sort
_formats
(info_dict
['formats'])
355 ydl
.process_ie_result(info_dict
)
356 downloaded
= ydl
.downloaded_info_dicts
[0]
357 self
.assertEqual(downloaded
['format_id'], '38')
359 info_dict
= _make_result(list(formats_order
), extractor
='youtube')
360 ydl
= YDL({'format': 'bestvideo/best,bestaudio'}
)
362 yie
._sort
_formats
(info_dict
['formats'])
363 ydl
.process_ie_result(info_dict
)
364 downloaded_ids
= [info
['format_id'] for info
in ydl
.downloaded_info_dicts
]
365 self
.assertEqual(downloaded_ids
, ['137', '141'])
367 info_dict
= _make_result(list(formats_order
), extractor
='youtube')
368 ydl
= YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])+bestaudio'}
)
370 yie
._sort
_formats
(info_dict
['formats'])
371 ydl
.process_ie_result(info_dict
)
372 downloaded_ids
= [info
['format_id'] for info
in ydl
.downloaded_info_dicts
]
373 self
.assertEqual(downloaded_ids
, ['137+141', '248+141'])
375 info_dict
= _make_result(list(formats_order
), extractor
='youtube')
376 ydl
= YDL({'format': '(bestvideo[ext=mp4],bestvideo[ext=webm])[height<=720]+bestaudio'}
)
378 yie
._sort
_formats
(info_dict
['formats'])
379 ydl
.process_ie_result(info_dict
)
380 downloaded_ids
= [info
['format_id'] for info
in ydl
.downloaded_info_dicts
]
381 self
.assertEqual(downloaded_ids
, ['136+141', '247+141'])
383 info_dict
= _make_result(list(formats_order
), extractor
='youtube')
384 ydl
= YDL({'format': '(bestvideo[ext=none]/bestvideo[ext=webm])+bestaudio'}
)
386 yie
._sort
_formats
(info_dict
['formats'])
387 ydl
.process_ie_result(info_dict
)
388 downloaded_ids
= [info
['format_id'] for info
in ydl
.downloaded_info_dicts
]
389 self
.assertEqual(downloaded_ids
, ['248+141'])
391 for f1
, f2
in zip(formats_order
, formats_order
[1:]):
392 info_dict
= _make_result([f1
, f2
], extractor
='youtube')
393 ydl
= YDL({'format': 'best/bestvideo'}
)
395 yie
._sort
_formats
(info_dict
['formats'])
396 ydl
.process_ie_result(info_dict
)
397 downloaded
= ydl
.downloaded_info_dicts
[0]
398 self
.assertEqual(downloaded
['format_id'], f1
['format_id'])
400 info_dict
= _make_result([f2
, f1
], extractor
='youtube')
401 ydl
= YDL({'format': 'best/bestvideo'}
)
403 yie
._sort
_formats
(info_dict
['formats'])
404 ydl
.process_ie_result(info_dict
)
405 downloaded
= ydl
.downloaded_info_dicts
[0]
406 self
.assertEqual(downloaded
['format_id'], f1
['format_id'])
408 def test_audio_only_extractor_format_selection(self
):
409 # For extractors with incomplete formats (all formats are audio-only or
410 # video-only) best and worst should fallback to corresponding best/worst
411 # video-only or audio-only formats (as per
412 # https://github.com/ytdl-org/youtube-dl/pull/5556)
414 {'format_id': 'low', 'ext': 'mp3', 'preference': 1, 'vcodec': 'none', 'url': TEST_URL}
,
415 {'format_id': 'high', 'ext': 'mp3', 'preference': 2, 'vcodec': 'none', 'url': TEST_URL}
,
417 info_dict
= _make_result(formats
)
419 ydl
= YDL({'format': 'best'}
)
420 ydl
.process_ie_result(info_dict
.copy())
421 downloaded
= ydl
.downloaded_info_dicts
[0]
422 self
.assertEqual(downloaded
['format_id'], 'high')
424 ydl
= YDL({'format': 'worst'}
)
425 ydl
.process_ie_result(info_dict
.copy())
426 downloaded
= ydl
.downloaded_info_dicts
[0]
427 self
.assertEqual(downloaded
['format_id'], 'low')
429 def test_format_not_available(self
):
431 {'format_id': 'regular', 'ext': 'mp4', 'height': 360, 'url': TEST_URL}
,
432 {'format_id': 'video', 'ext': 'mp4', 'height': 720, 'acodec': 'none', 'url': TEST_URL}
,
434 info_dict
= _make_result(formats
)
436 # This must fail since complete video-audio format does not match filter
437 # and extractor does not provide incomplete only formats (i.e. only
438 # video-only or audio-only).
439 ydl
= YDL({'format': 'best[height>360]'}
)
440 self
.assertRaises(ExtractorError
, ydl
.process_ie_result
, info_dict
.copy())
442 def test_format_selection_issue_10083(self
):
443 # See https://github.com/ytdl-org/youtube-dl/issues/10083
445 {'format_id': 'regular', 'height': 360, 'url': TEST_URL}
,
446 {'format_id': 'video', 'height': 720, 'acodec': 'none', 'url': TEST_URL}
,
447 {'format_id': 'audio', 'vcodec': 'none', 'url': TEST_URL}
,
449 info_dict
= _make_result(formats
)
451 ydl
= YDL({'format': 'best[height>360]/bestvideo[height>360]+bestaudio'}
)
452 ydl
.process_ie_result(info_dict
.copy())
453 self
.assertEqual(ydl
.downloaded_info_dicts
[0]['format_id'], 'video+audio')
455 def test_invalid_format_specs(self
):
456 def assert_syntax_error(format_spec
):
457 self
.assertRaises(SyntaxError, YDL
, {'format': format_spec}
)
459 assert_syntax_error('bestvideo,,best')
460 assert_syntax_error('+bestaudio')
461 assert_syntax_error('bestvideo+')
462 assert_syntax_error('/')
463 assert_syntax_error('[720<height]')
465 def test_format_filtering(self
):
467 {'format_id': 'A', 'filesize': 500, 'width': 1000}
,
468 {'format_id': 'B', 'filesize': 1000, 'width': 500}
,
469 {'format_id': 'C', 'filesize': 1000, 'width': 400}
,
470 {'format_id': 'D', 'filesize': 2000, 'width': 600}
,
471 {'format_id': 'E', 'filesize': 3000}
,
473 {'format_id': 'G', 'filesize': 1000000}
,
476 f
['url'] = 'http://_/'
478 info_dict
= _make_result(formats
)
480 ydl
= YDL({'format': 'best[filesize<3000]'}
)
481 ydl
.process_ie_result(info_dict
)
482 downloaded
= ydl
.downloaded_info_dicts
[0]
483 self
.assertEqual(downloaded
['format_id'], 'D')
485 ydl
= YDL({'format': 'best[filesize<=3000]'}
)
486 ydl
.process_ie_result(info_dict
)
487 downloaded
= ydl
.downloaded_info_dicts
[0]
488 self
.assertEqual(downloaded
['format_id'], 'E')
490 ydl
= YDL({'format': 'best[filesize <= ? 3000]'}
)
491 ydl
.process_ie_result(info_dict
)
492 downloaded
= ydl
.downloaded_info_dicts
[0]
493 self
.assertEqual(downloaded
['format_id'], 'F')
495 ydl
= YDL({'format': 'best [filesize = 1000] [width>450]'}
)
496 ydl
.process_ie_result(info_dict
)
497 downloaded
= ydl
.downloaded_info_dicts
[0]
498 self
.assertEqual(downloaded
['format_id'], 'B')
500 ydl
= YDL({'format': 'best [filesize = 1000] [width!=450]'}
)
501 ydl
.process_ie_result(info_dict
)
502 downloaded
= ydl
.downloaded_info_dicts
[0]
503 self
.assertEqual(downloaded
['format_id'], 'C')
505 ydl
= YDL({'format': '[filesize>?1]'}
)
506 ydl
.process_ie_result(info_dict
)
507 downloaded
= ydl
.downloaded_info_dicts
[0]
508 self
.assertEqual(downloaded
['format_id'], 'G')
510 ydl
= YDL({'format': '[filesize<1M]'}
)
511 ydl
.process_ie_result(info_dict
)
512 downloaded
= ydl
.downloaded_info_dicts
[0]
513 self
.assertEqual(downloaded
['format_id'], 'E')
515 ydl
= YDL({'format': '[filesize<1MiB]'}
)
516 ydl
.process_ie_result(info_dict
)
517 downloaded
= ydl
.downloaded_info_dicts
[0]
518 self
.assertEqual(downloaded
['format_id'], 'G')
520 ydl
= YDL({'format': 'all[width>=400][width<=600]'}
)
521 ydl
.process_ie_result(info_dict
)
522 downloaded_ids
= [info
['format_id'] for info
in ydl
.downloaded_info_dicts
]
523 self
.assertEqual(downloaded_ids
, ['B', 'C', 'D'])
525 ydl
= YDL({'format': 'best[height<40]'}
)
527 ydl
.process_ie_result(info_dict
)
528 except ExtractorError
:
530 self
.assertEqual(ydl
.downloaded_info_dicts
, [])
532 def test_default_format_spec(self
):
533 ydl
= YDL({'simulate': True}
)
534 self
.assertEqual(ydl
._default
_format
_spec
({}), 'bestvideo*+bestaudio/best')
537 self
.assertEqual(ydl
._default
_format
_spec
({'is_live': True}
), 'best/bestvideo+bestaudio')
539 ydl
= YDL({'simulate': True}
)
540 self
.assertEqual(ydl
._default
_format
_spec
({'is_live': True}
), 'bestvideo*+bestaudio/best')
542 ydl
= YDL({'outtmpl': '-'}
)
543 self
.assertEqual(ydl
._default
_format
_spec
({}), 'best/bestvideo+bestaudio')
546 self
.assertEqual(ydl
._default
_format
_spec
({}, download
=False), 'bestvideo*+bestaudio/best')
547 self
.assertEqual(ydl
._default
_format
_spec
({'is_live': True}
), 'best/bestvideo+bestaudio')
550 class TestYoutubeDL(unittest
.TestCase
):
551 def test_subtitles(self
):
552 def s_formats(lang
, autocaption
=False):
555 'url': 'http://localhost/video.%s.%s' % (lang
, ext
),
556 '_auto': autocaption
,
557 } for ext
in ['vtt', 'srt', 'ass']]
558 subtitles
= dict((l
, s_formats(l
)) for l
in ['en', 'fr', 'es'])
559 auto_captions
= dict((l
, s_formats(l
, True)) for l
in ['it', 'pt', 'es'])
563 'url': 'http://localhost/video.mp4',
564 'subtitles': subtitles
,
565 'automatic_captions': auto_captions
,
567 'webpage_url': 'http://example.com/watch?v=shenanigans',
570 def get_info(params
={}):
571 params
.setdefault('simulate', True)
573 ydl
.report_warning
= lambda *args
, **kargs
: None
574 return ydl
.process_video_result(info_dict
, download
=False)
577 self
.assertFalse(result
.get('requested_subtitles'))
578 self
.assertEqual(result
['subtitles'], subtitles
)
579 self
.assertEqual(result
['automatic_captions'], auto_captions
)
581 result
= get_info({'writesubtitles': True}
)
582 subs
= result
['requested_subtitles']
583 self
.assertTrue(subs
)
584 self
.assertEqual(set(subs
.keys()), set(['en']))
585 self
.assertTrue(subs
['en'].get('data') is None)
586 self
.assertEqual(subs
['en']['ext'], 'ass')
588 result
= get_info({'writesubtitles': True, 'subtitlesformat': 'foo/srt'}
)
589 subs
= result
['requested_subtitles']
590 self
.assertEqual(subs
['en']['ext'], 'srt')
592 result
= get_info({'writesubtitles': True, 'subtitleslangs': ['es', 'fr', 'it']}
)
593 subs
= result
['requested_subtitles']
594 self
.assertTrue(subs
)
595 self
.assertEqual(set(subs
.keys()), set(['es', 'fr']))
597 result
= get_info({'writesubtitles': True, 'subtitleslangs': ['all', '-en']}
)
598 subs
= result
['requested_subtitles']
599 self
.assertTrue(subs
)
600 self
.assertEqual(set(subs
.keys()), set(['es', 'fr']))
602 result
= get_info({'writesubtitles': True, 'subtitleslangs': ['en', 'fr', '-en']}
)
603 subs
= result
['requested_subtitles']
604 self
.assertTrue(subs
)
605 self
.assertEqual(set(subs
.keys()), set(['fr']))
607 result
= get_info({'writesubtitles': True, 'subtitleslangs': ['-en', 'en']}
)
608 subs
= result
['requested_subtitles']
609 self
.assertTrue(subs
)
610 self
.assertEqual(set(subs
.keys()), set(['en']))
612 result
= get_info({'writesubtitles': True, 'subtitleslangs': ['e.+']}
)
613 subs
= result
['requested_subtitles']
614 self
.assertTrue(subs
)
615 self
.assertEqual(set(subs
.keys()), set(['es', 'en']))
617 result
= get_info({'writesubtitles': True, 'writeautomaticsub': True, 'subtitleslangs': ['es', 'pt']}
)
618 subs
= result
['requested_subtitles']
619 self
.assertTrue(subs
)
620 self
.assertEqual(set(subs
.keys()), set(['es', 'pt']))
621 self
.assertFalse(subs
['es']['_auto'])
622 self
.assertTrue(subs
['pt']['_auto'])
624 result
= get_info({'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
.assertTrue(subs
['es']['_auto'])
629 self
.assertTrue(subs
['pt']['_auto'])
631 def test_add_extra_info(self
):
637 'playlist': 'funny videos',
639 YDL
.add_extra_info(test_dict
, extra_info
)
640 self
.assertEqual(test_dict
['extractor'], 'Foo')
641 self
.assertEqual(test_dict
['playlist'], 'funny videos')
650 'title3': 'foo/bar\\test',
651 'title4': 'foo "bar" test',
653 'timestamp': 1618488000,
656 'playlist_autonumber': 2,
657 '_last_playlist_index': 100,
659 'formats': [{'id': 'id1'}
, {'id': 'id2'}
, {'id': 'id3'}
]
662 def test_prepare_outtmpl_and_filename(self
):
663 def test(tmpl
, expected
, *, info
=None, **params
):
664 params
['outtmpl'] = tmpl
665 ydl
= YoutubeDL(params
)
666 ydl
._num
_downloads
= 1
667 self
.assertEqual(ydl
.validate_outtmpl(tmpl
), None)
669 outtmpl
, tmpl_dict
= ydl
.prepare_outtmpl(tmpl
, info
or self
.outtmpl_info
)
670 out
= ydl
.escape_outtmpl(outtmpl
) % tmpl_dict
671 fname
= ydl
.prepare_filename(info
or self
.outtmpl_info
)
673 if not isinstance(expected
, (list, tuple)):
674 expected
= (expected
, expected
)
675 for (name
, got
), expect
in zip((('outtmpl', out
), ('filename', fname
)), expected
):
677 self
.assertTrue(expect(got
), f
'Wrong {name} from {tmpl}')
679 self
.assertEqual(got
, expect
, f
'Wrong {name} from {tmpl}')
682 original_infodict
= dict(self
.outtmpl_info
)
683 test('foo.bar', 'foo.bar')
684 original_infodict
['epoch'] = self
.outtmpl_info
.get('epoch')
685 self
.assertTrue(isinstance(original_infodict
['epoch'], int))
686 test('%(epoch)d', int_or_none
)
687 self
.assertEqual(original_infodict
, self
.outtmpl_info
)
689 # Auto-generated fields
690 test('%(id)s.%(ext)s', '1234.mp4')
691 test('%(duration_string)s', ('27:46:40', '27-46-40'))
692 test('%(resolution)s', '1080p')
693 test('%(playlist_index)s', '001')
694 test('%(playlist_autonumber)s', '02')
695 test('%(autonumber)s', '00001')
696 test('%(autonumber+2)03d', '005', autonumber_start
=3)
697 test('%(autonumber)s', '001', autonumber_size
=3)
706 test('%abc%', '%abc%')
707 test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
708 test('%%%(height)s', '%1080')
709 test('%(width)06d.%(ext)s', 'NA.mp4')
710 test('%(width)06d.%%(ext)s', 'NA.%(ext)s')
711 test('%%(width)06d.%(ext)s', '%(width)06d.mp4')
714 test('%(id)s', '_abcd', info
={'id': '_abcd'}
)
715 test('%(some_id)s', '_abcd', info
={'some_id': '_abcd'}
)
716 test('%(formats.0.id)s', '_abcd', info
={'formats': [{'id': '_abcd'}
]})
717 test('%(id)s', '-abcd', info
={'id': '-abcd'}
)
718 test('%(id)s', '.abcd', info
={'id': '.abcd'}
)
719 test('%(id)s', 'ab__cd', info
={'id': 'ab__cd'}
)
720 test('%(id)s', ('ab:cd', 'ab -cd'), info
={'id': 'ab:cd'}
)
723 self
.assertTrue(isinstance(YoutubeDL
.validate_outtmpl('%(title)'), ValueError))
724 test('%(invalid@tmpl|def)s', 'none', outtmpl_na_placeholder
='none')
728 def expect_same_infodict(out
):
729 got_dict
= json
.loads(out
)
730 for info_field
, expected
in self
.outtmpl_info
.items():
731 self
.assertEqual(got_dict
.get(info_field
), expected
, info_field
)
734 test('%()j', (expect_same_infodict
, str))
737 NA_TEST_OUTTMPL
= '%(uploader_date)s-%(width)d-%(x|def)s-%(id)s.%(ext)s'
738 test(NA_TEST_OUTTMPL
, 'NA-NA-def-1234.mp4')
739 test(NA_TEST_OUTTMPL
, 'none-none-def-1234.mp4', outtmpl_na_placeholder
='none')
740 test(NA_TEST_OUTTMPL
, '--def-1234.mp4', outtmpl_na_placeholder
='')
743 FMT_TEST_OUTTMPL
= '%%(height)%s.%%(ext)s'
744 test(FMT_TEST_OUTTMPL
% 's', '1080.mp4')
745 test(FMT_TEST_OUTTMPL
% 'd', '1080.mp4')
746 test(FMT_TEST_OUTTMPL
% '6d', ' 1080.mp4')
747 test(FMT_TEST_OUTTMPL
% '-6d', '1080 .mp4')
748 test(FMT_TEST_OUTTMPL
% '06d', '001080.mp4')
749 test(FMT_TEST_OUTTMPL
% ' 06d', ' 01080.mp4')
750 test(FMT_TEST_OUTTMPL
% ' 06d', ' 01080.mp4')
751 test(FMT_TEST_OUTTMPL
% '0 6d', ' 01080.mp4')
752 test(FMT_TEST_OUTTMPL
% '0 6d', ' 01080.mp4')
753 test(FMT_TEST_OUTTMPL
% ' 0 6d', ' 01080.mp4')
756 test('%(id)d', '1234')
757 test('%(height)c', '1')
759 test('%(id)d %(id)r', "1234 '1234'")
760 test('%(id)r %(height)r', "'1234' 1080")
761 test('%(ext)s-%(ext|def)d', 'mp4-def')
762 test('%(width|0)04d', '0000')
763 test('a%(width|)d', 'a', outtmpl_na_placeholder
='none')
765 FORMATS
= self
.outtmpl_info
['formats']
766 sanitize
= lambda x
: x
.replace(':', ' -').replace('"', "'")
768 # Custom type casting
769 test('%(formats.:.id)l', 'id1, id2, id3')
770 test('%(formats.:.id)#l', ('id1\nid2\nid3', 'id1 id2 id3'))
771 test('%(ext)l', 'mp4')
772 test('%(formats.:.id) 15l', ' id1, id2, id3')
773 test('%(formats)j', (json
.dumps(FORMATS
), sanitize(json
.dumps(FORMATS
))))
774 test('%(title5).3B', 'á')
775 test('%(title5)U', 'áéí 𝐀')
776 test('%(title5)#U', 'a\u0301e\u0301i\u0301 𝐀')
777 test('%(title5)+U', 'áéí A')
778 test('%(title5)+#U', 'a\u0301e\u0301i\u0301 A')
779 if compat_os_name
== 'nt':
780 test('%(title4)q', ('"foo \\"bar\\" test"', "'foo _'bar_' test'"))
782 test('%(title4)q', ('\'foo "bar" test\'', "'foo 'bar' test'"))
784 # Internal formatting
785 test('%(timestamp-1000>%H-%M-%S)s', '11-43-20')
786 test('%(title|%)s %(title|%%)s', '% %%')
787 test('%(id+1-height+3)05d', '00158')
788 test('%(width+100)05d', 'NA')
789 test('%(formats.0) 15s', ('% 15s' % FORMATS
[0], '% 15s' % sanitize(str(FORMATS
[0]))))
790 test('%(formats.0)r', (repr(FORMATS
[0]), sanitize(repr(FORMATS
[0]))))
791 test('%(height.0)03d', '001')
792 test('%(-height.0)04d', '-001')
793 test('%(formats.-1.id)s', FORMATS
[-1]['id'])
794 test('%(formats.0.id.-1)d', FORMATS
[0]['id'][-1])
795 test('%(formats.3)s', 'NA')
796 test('%(formats.:2:-1)r', repr(FORMATS
[:2:-1]))
797 test('%(formats.0.id.-1+id)f', '1235.000000')
798 test('%(formats.0.id.-1+formats.1.id.-1)d', '3')
801 test('%(title,id)s', '1234')
802 test('%(width-100,height+20|def)d', '1100')
803 test('%(width-100,height+width|def)s', 'def')
804 test('%(timestamp-x>%H\\,%M\\,%S,timestamp>%H\\,%M\\,%S)s', '12,00,00')
809 raise self
.assertTrue(False, 'LazyList should not be evaluated till here')
810 test('%(key.4)s', '4', info
={'key': LazyList(gen())}
)
813 test('%(foo|)s-%(bar|)s.%(ext)s', '-.mp4')
814 # test('%(foo|)s.%(ext)s', ('.mp4', '_.mp4')) # fixme
815 # test('%(foo|)s', ('', '_')) # fixme
817 # Environment variable expansion for prepare_filename
818 compat_setenv('__yt_dlp_var', 'expanded')
819 envvar
= '%__yt_dlp_var%' if compat_os_name
== 'nt' else '$__yt_dlp_var'
820 test(envvar
, (envvar
, 'expanded'))
822 # Path expansion and escaping
823 test('Hello %(title1)s', 'Hello $PATH')
824 test('Hello %(title2)s', 'Hello %PATH%')
825 test('%(title3)s', ('foo/bar\\test', 'foo_bar_test'))
826 test('folder/%(title3)s', ('folder/foo/bar\\test', 'folder%sfoo_bar_test' % os
.path
.sep
))
828 def test_format_note(self
):
830 self
.assertEqual(ydl
._format
_note
({}), '')
831 assertRegexpMatches(self
, ydl
._format
_note
({
834 assertRegexpMatches(self
, ydl
._format
_note
({
838 def test_postprocessors(self
):
839 filename
= 'post-processor-testfile.mp4'
840 audiofile
= filename
+ '.mp3'
842 class SimplePP(PostProcessor
):
844 with open(audiofile
, 'wt') as f
:
846 return [info
['filepath']], info
848 def run_pp(params
, PP
):
849 with open(filename
, 'wt') as f
:
851 ydl
= YoutubeDL(params
)
852 ydl
.add_post_processor(PP())
853 ydl
.post_process(filename
, {'filepath': filename}
)
855 run_pp({'keepvideo': True}
, SimplePP
)
856 self
.assertTrue(os
.path
.exists(filename
), '%s doesn\'t exist' % filename
)
857 self
.assertTrue(os
.path
.exists(audiofile
), '%s doesn\'t exist' % audiofile
)
861 run_pp({'keepvideo': False}
, SimplePP
)
862 self
.assertFalse(os
.path
.exists(filename
), '%s exists' % filename
)
863 self
.assertTrue(os
.path
.exists(audiofile
), '%s doesn\'t exist' % audiofile
)
866 class ModifierPP(PostProcessor
):
868 with open(info
['filepath'], 'wt') as f
:
872 run_pp({'keepvideo': False}
, ModifierPP
)
873 self
.assertTrue(os
.path
.exists(filename
), '%s doesn\'t exist' % filename
)
876 def test_match_filter(self
):
877 class FilterYDL(YDL
):
878 def __init__(self
, *args
, **kwargs
):
879 super(FilterYDL
, self
).__init
__(*args
, **kwargs
)
880 self
.params
['simulate'] = True
882 def process_info(self
, info_dict
):
883 super(YDL
, self
).process_info(info_dict
)
885 def _match_entry(self
, info_dict
, incomplete
=False):
886 res
= super(FilterYDL
, self
)._match
_entry
(info_dict
, incomplete
)
888 self
.downloaded_info_dicts
.append(info_dict
)
897 'filesize': 10 * 1024,
899 'uploader': "變態妍字幕版 太妍 тест",
900 'creator': "тест ' 123 ' тест--",
901 'webpage_url': 'http://example.com/watch?v=shenanigans',
909 'description': 'foo',
910 'filesize': 5 * 1024,
912 'uploader': "тест 123",
913 'webpage_url': 'http://example.com/watch?v=SHENANIGANS',
915 videos
= [first
, second
]
917 def get_videos(filter_
=None):
918 ydl
= FilterYDL({'match_filter': filter_}
)
920 ydl
.process_ie_result(v
, download
=True)
921 return [v
['id'] for v
in ydl
.downloaded_info_dicts
]
924 self
.assertEqual(res
, ['1', '2'])
930 return 'Video id is not 1'
932 self
.assertEqual(res
, ['1'])
934 f
= match_filter_func('duration < 30')
936 self
.assertEqual(res
, ['2'])
938 f
= match_filter_func('description = foo')
940 self
.assertEqual(res
, ['2'])
942 f
= match_filter_func('description =? foo')
944 self
.assertEqual(res
, ['1', '2'])
946 f
= match_filter_func('filesize > 5KiB')
948 self
.assertEqual(res
, ['1'])
950 f
= match_filter_func('playlist_id = 42')
952 self
.assertEqual(res
, ['1'])
954 f
= match_filter_func('uploader = "變態妍字幕版 太妍 тест"')
956 self
.assertEqual(res
, ['1'])
958 f
= match_filter_func('uploader != "變態妍字幕版 太妍 тест"')
960 self
.assertEqual(res
, ['2'])
962 f
= match_filter_func('creator = "тест \' 123 \' тест--"')
964 self
.assertEqual(res
, ['1'])
966 f
= match_filter_func("creator = 'тест \\' 123 \\' тест--'")
968 self
.assertEqual(res
, ['1'])
970 f
= match_filter_func(r
"creator = 'тест \' 123 \' тест--' & duration > 30")
972 self
.assertEqual(res
, [])
974 def test_playlist_items_selection(self
):
977 'title': compat_str(i
),
979 } for i
in range(1, 5)]
984 'extractor': 'test:playlist',
985 'extractor_key': 'test:playlist',
986 'webpage_url': 'http://example.com',
989 def get_downloaded_info_dicts(params
):
991 # make a deep copy because the dictionary and nested entries
993 ydl
.process_ie_result(copy
.deepcopy(playlist
))
994 return ydl
.downloaded_info_dicts
996 def test_selection(params
, expected_ids
):
998 (v
['playlist_autonumber'] - 1, (int(v
['id']), v
['playlist_index']))
999 for v
in get_downloaded_info_dicts(params
)]
1000 self
.assertEqual(results
, list(enumerate(zip(expected_ids
, expected_ids
))))
1002 test_selection({}, [1, 2, 3, 4])
1003 test_selection({'playlistend': 10}
, [1, 2, 3, 4])
1004 test_selection({'playlistend': 2}
, [1, 2])
1005 test_selection({'playliststart': 10}
, [])
1006 test_selection({'playliststart': 2}
, [2, 3, 4])
1007 test_selection({'playlist_items': '2-4'}
, [2, 3, 4])
1008 test_selection({'playlist_items': '2,4'}
, [2, 4])
1009 test_selection({'playlist_items': '10'}
, [])
1010 test_selection({'playlist_items': '0'}
, [])
1012 # Tests for https://github.com/ytdl-org/youtube-dl/issues/10591
1013 test_selection({'playlist_items': '2-4,3-4,3'}
, [2, 3, 4])
1014 test_selection({'playlist_items': '4,2'}
, [4, 2])
1016 # Tests for https://github.com/yt-dlp/yt-dlp/issues/720
1017 # https://github.com/yt-dlp/yt-dlp/issues/302
1018 test_selection({'playlistreverse': True}
, [4, 3, 2, 1])
1019 test_selection({'playliststart': 2, 'playlistreverse': True}
, [4, 3, 2])
1020 test_selection({'playlist_items': '2,4', 'playlistreverse': True}
, [4, 2])
1021 test_selection({'playlist_items': '4,2'}
, [4, 2])
1023 def test_urlopen_no_file_protocol(self
):
1024 # see https://github.com/ytdl-org/youtube-dl/issues/8227
1026 self
.assertRaises(compat_urllib_error
.URLError
, ydl
.urlopen
, 'file:///etc/passwd')
1028 def test_do_not_override_ie_key_in_url_transparent(self
):
1031 class Foo1IE(InfoExtractor
):
1032 _VALID_URL
= r
'foo1:'
1034 def _real_extract(self
, url
):
1036 '_type': 'url_transparent',
1039 'title': 'foo1 title',
1043 class Foo2IE(InfoExtractor
):
1044 _VALID_URL
= r
'foo2:'
1046 def _real_extract(self
, url
):
1053 class Foo3IE(InfoExtractor
):
1054 _VALID_URL
= r
'foo3:'
1056 def _real_extract(self
, url
):
1057 return _make_result([{'url': TEST_URL}
], title
='foo3 title')
1059 ydl
.add_info_extractor(Foo1IE(ydl
))
1060 ydl
.add_info_extractor(Foo2IE(ydl
))
1061 ydl
.add_info_extractor(Foo3IE(ydl
))
1062 ydl
.extract_info('foo1:')
1063 downloaded
= ydl
.downloaded_info_dicts
[0]
1064 self
.assertEqual(downloaded
['url'], TEST_URL
)
1065 self
.assertEqual(downloaded
['title'], 'foo1 title')
1066 self
.assertEqual(downloaded
['id'], 'testid')
1067 self
.assertEqual(downloaded
['extractor'], 'testex')
1068 self
.assertEqual(downloaded
['extractor_key'], 'TestEx')
1070 # Test case for https://github.com/ytdl-org/youtube-dl/issues/27064
1071 def test_ignoreerrors_for_playlist_with_url_transparent_iterable_entries(self
):
1074 def __init__(self
, *args
, **kwargs
):
1075 super(_YDL
, self
).__init
__(*args
, **kwargs
)
1077 def trouble(self
, s
, tb
=None):
1082 'ignoreerrors': True,
1085 class VideoIE(InfoExtractor
):
1086 _VALID_URL
= r
'video:(?P<id>\d+)'
1088 def _real_extract(self
, url
):
1089 video_id
= self
._match
_id
(url
)
1091 'format_id': 'default',
1095 raise ExtractorError('foo')
1098 'format_id': 'extra',
1103 'title': 'Video %s' % video_id
,
1107 class PlaylistIE(InfoExtractor
):
1108 _VALID_URL
= r
'playlist:'
1112 video_id
= compat_str(n
)
1114 '_type': 'url_transparent',
1115 'ie_key': VideoIE
.ie_key(),
1117 'url': 'video:%s' % video_id
,
1118 'title': 'Video Transparent %s' % video_id
,
1121 def _real_extract(self
, url
):
1122 return self
.playlist_result(self
._entries
())
1124 ydl
.add_info_extractor(VideoIE(ydl
))
1125 ydl
.add_info_extractor(PlaylistIE(ydl
))
1126 info
= ydl
.extract_info('playlist:')
1127 entries
= info
['entries']
1128 self
.assertEqual(len(entries
), 3)
1129 self
.assertTrue(entries
[0] is None)
1130 self
.assertTrue(entries
[1] is None)
1131 self
.assertEqual(len(ydl
.downloaded_info_dicts
), 1)
1132 downloaded
= ydl
.downloaded_info_dicts
[0]
1133 self
.assertEqual(entries
[2], downloaded
)
1134 self
.assertEqual(downloaded
['url'], TEST_URL
)
1135 self
.assertEqual(downloaded
['title'], 'Video Transparent 2')
1136 self
.assertEqual(downloaded
['id'], '2')
1137 self
.assertEqual(downloaded
['extractor'], 'Video')
1138 self
.assertEqual(downloaded
['extractor_key'], 'Video')
1141 if __name__
== '__main__':