Release 2023.12.30
[yt-dlp.git] / test / test_utils.py
blobc3e387cd0d6fb1088f07be08e49c8f21488d400e
1 #!/usr/bin/env python3
3 # Allow direct execution
4 import os
5 import re
6 import sys
7 import unittest
8 import warnings
10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
13 import contextlib
14 import io
15 import itertools
16 import json
17 import subprocess
18 import xml.etree.ElementTree
20 from yt_dlp.compat import (
21 compat_etree_fromstring,
22 compat_HTMLParseError,
23 compat_os_name,
25 from yt_dlp.utils import (
26 Config,
27 DateRange,
28 ExtractorError,
29 InAdvancePagedList,
30 LazyList,
31 OnDemandPagedList,
32 Popen,
33 age_restricted,
34 args_to_str,
35 base_url,
36 caesar,
37 clean_html,
38 clean_podcast_url,
39 cli_bool_option,
40 cli_option,
41 cli_valueless_option,
42 date_from_str,
43 datetime_from_str,
44 detect_exe_version,
45 determine_ext,
46 determine_file_encoding,
47 dfxp2srt,
48 dict_get,
49 encode_base_n,
50 encode_compat_str,
51 encodeFilename,
52 expand_path,
53 extract_attributes,
54 extract_basic_auth,
55 find_xpath_attr,
56 fix_xml_ampersands,
57 float_or_none,
58 format_bytes,
59 get_compatible_ext,
60 get_element_by_attribute,
61 get_element_by_class,
62 get_element_html_by_attribute,
63 get_element_html_by_class,
64 get_element_text_and_html_by_tag,
65 get_elements_by_attribute,
66 get_elements_by_class,
67 get_elements_html_by_attribute,
68 get_elements_html_by_class,
69 get_elements_text_and_html_by_attribute,
70 int_or_none,
71 intlist_to_bytes,
72 iri_to_uri,
73 is_html,
74 js_to_json,
75 limit_length,
76 locked_file,
77 lowercase_escape,
78 match_str,
79 merge_dicts,
80 mimetype2ext,
81 month_by_name,
82 multipart_encode,
83 ohdave_rsa_encrypt,
84 orderedSet,
85 parse_age_limit,
86 parse_bitrate,
87 parse_codecs,
88 parse_count,
89 parse_dfxp_time_expr,
90 parse_duration,
91 parse_filesize,
92 parse_iso8601,
93 parse_qs,
94 parse_resolution,
95 pkcs1pad,
96 prepend_extension,
97 read_batch_urls,
98 remove_end,
99 remove_quotes,
100 remove_start,
101 render_table,
102 replace_extension,
103 rot47,
104 sanitize_filename,
105 sanitize_path,
106 sanitize_url,
107 shell_quote,
108 smuggle_url,
109 str_or_none,
110 str_to_int,
111 strip_jsonp,
112 strip_or_none,
113 subtitles_filename,
114 timeconvert,
115 traverse_obj,
116 try_call,
117 unescapeHTML,
118 unified_strdate,
119 unified_timestamp,
120 unsmuggle_url,
121 update_url_query,
122 uppercase_escape,
123 url_basename,
124 url_or_none,
125 urlencode_postdata,
126 urljoin,
127 urshift,
128 variadic,
129 version_tuple,
130 xpath_attr,
131 xpath_element,
132 xpath_text,
133 xpath_with_ns,
135 from yt_dlp.utils.networking import (
136 HTTPHeaderDict,
137 escape_rfc3986,
138 normalize_url,
139 remove_dot_segments,
143 class TestUtil(unittest.TestCase):
144 def test_timeconvert(self):
145 self.assertTrue(timeconvert('') is None)
146 self.assertTrue(timeconvert('bougrg') is None)
148 def test_sanitize_filename(self):
149 self.assertEqual(sanitize_filename(''), '')
150 self.assertEqual(sanitize_filename('abc'), 'abc')
151 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
153 self.assertEqual(sanitize_filename('123'), '123')
155 self.assertEqual('abc⧸de', sanitize_filename('abc/de'))
156 self.assertFalse('/' in sanitize_filename('abc/de///'))
158 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', is_id=False))
159 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', is_id=False))
160 self.assertEqual('yes no', sanitize_filename('yes? no', is_id=False))
161 self.assertEqual('this - that', sanitize_filename('this: that', is_id=False))
163 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
164 aumlaut = 'ä'
165 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
166 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
167 self.assertEqual(sanitize_filename(tests), tests)
169 self.assertEqual(
170 sanitize_filename('New World record at 0:12:34'),
171 'New World record at 0_12_34')
173 self.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf')
174 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
175 self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf')
176 self.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf')
177 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
178 self.assertEqual(sanitize_filename('.gasdgf', is_id=False), 'gasdgf')
180 forbidden = '"\0\\/'
181 for fc in forbidden:
182 for fbc in forbidden:
183 self.assertTrue(fbc not in sanitize_filename(fc))
185 def test_sanitize_filename_restricted(self):
186 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
187 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
189 self.assertEqual(sanitize_filename('123', restricted=True), '123')
191 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
192 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
194 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
195 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
196 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
197 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
199 tests = 'aäb\u4e2d\u56fd\u7684c'
200 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
201 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
203 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
204 for fc in forbidden:
205 for fbc in forbidden:
206 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
208 # Handle a common case more neatly
209 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
210 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
211 # .. but make sure the file name is never empty
212 self.assertTrue(sanitize_filename('-', restricted=True) != '')
213 self.assertTrue(sanitize_filename(':', restricted=True) != '')
215 self.assertEqual(sanitize_filename(
216 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
217 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
219 def test_sanitize_ids(self):
220 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
221 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
222 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
224 def test_sanitize_path(self):
225 if sys.platform != 'win32':
226 return
228 self.assertEqual(sanitize_path('abc'), 'abc')
229 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
230 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
231 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
232 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
233 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
234 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
236 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
237 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
239 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
240 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
241 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
242 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
244 self.assertEqual(
245 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
246 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
248 self.assertEqual(
249 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
250 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
251 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
252 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
253 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
255 self.assertEqual(sanitize_path('../abc'), '..\\abc')
256 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
257 self.assertEqual(sanitize_path('./abc'), 'abc')
258 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
260 def test_sanitize_url(self):
261 self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
262 self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
263 self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
264 self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
265 self.assertEqual(sanitize_url('foo bar'), 'foo bar')
267 def test_expand_path(self):
268 def env(var):
269 return f'%{var}%' if sys.platform == 'win32' else f'${var}'
271 os.environ['yt_dlp_EXPATH_PATH'] = 'expanded'
272 self.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded')
274 old_home = os.environ.get('HOME')
275 test_str = R'C:\Documents and Settings\тест\Application Data'
276 try:
277 os.environ['HOME'] = test_str
278 self.assertEqual(expand_path(env('HOME')), os.getenv('HOME'))
279 self.assertEqual(expand_path('~'), os.getenv('HOME'))
280 self.assertEqual(
281 expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')),
282 '%s/expanded' % os.getenv('HOME'))
283 finally:
284 os.environ['HOME'] = old_home or ''
286 def test_prepend_extension(self):
287 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
288 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
289 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
290 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
291 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
292 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
294 def test_replace_extension(self):
295 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
296 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
297 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
298 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
299 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
300 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
302 def test_subtitles_filename(self):
303 self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
304 self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
305 self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
307 def test_remove_start(self):
308 self.assertEqual(remove_start(None, 'A - '), None)
309 self.assertEqual(remove_start('A - B', 'A - '), 'B')
310 self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
312 def test_remove_end(self):
313 self.assertEqual(remove_end(None, ' - B'), None)
314 self.assertEqual(remove_end('A - B', ' - B'), 'A')
315 self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
317 def test_remove_quotes(self):
318 self.assertEqual(remove_quotes(None), None)
319 self.assertEqual(remove_quotes('"'), '"')
320 self.assertEqual(remove_quotes("'"), "'")
321 self.assertEqual(remove_quotes(';'), ';')
322 self.assertEqual(remove_quotes('";'), '";')
323 self.assertEqual(remove_quotes('""'), '')
324 self.assertEqual(remove_quotes('";"'), ';')
326 def test_ordered_set(self):
327 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
328 self.assertEqual(orderedSet([]), [])
329 self.assertEqual(orderedSet([1]), [1])
330 # keep the list ordered
331 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
333 def test_unescape_html(self):
334 self.assertEqual(unescapeHTML('%20;'), '%20;')
335 self.assertEqual(unescapeHTML('&#x2F;'), '/')
336 self.assertEqual(unescapeHTML('&#47;'), '/')
337 self.assertEqual(unescapeHTML('&eacute;'), 'é')
338 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
339 self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
340 # HTML5 entities
341 self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
343 def test_date_from_str(self):
344 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
345 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
346 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
347 self.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year'))
348 self.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month'))
350 def test_datetime_from_str(self):
351 self.assertEqual(datetime_from_str('yesterday', precision='day'), datetime_from_str('now-1day', precision='auto'))
352 self.assertEqual(datetime_from_str('now+7day', precision='day'), datetime_from_str('now+1week', precision='auto'))
353 self.assertEqual(datetime_from_str('now+14day', precision='day'), datetime_from_str('now+2week', precision='auto'))
354 self.assertEqual(datetime_from_str('20200229+365day', precision='day'), datetime_from_str('20200229+1year', precision='auto'))
355 self.assertEqual(datetime_from_str('20210131+28day', precision='day'), datetime_from_str('20210131+1month', precision='auto'))
356 self.assertEqual(datetime_from_str('20210131+59day', precision='day'), datetime_from_str('20210131+2month', precision='auto'))
357 self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto'))
358 self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto'))
360 def test_daterange(self):
361 _20century = DateRange("19000101", "20000101")
362 self.assertFalse("17890714" in _20century)
363 _ac = DateRange("00010101")
364 self.assertTrue("19690721" in _ac)
365 _firstmilenium = DateRange(end="10000101")
366 self.assertTrue("07110427" in _firstmilenium)
368 def test_unified_dates(self):
369 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
370 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
371 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
372 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
373 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
374 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
375 self.assertEqual(unified_strdate('31-07-2022 20:00'), '20220731')
376 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
377 self.assertEqual(
378 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
379 '20141126')
380 self.assertEqual(
381 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
382 '20150202')
383 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
384 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
385 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
386 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
387 self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
388 self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
389 self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
390 self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
391 self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
392 self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
394 def test_unified_timestamps(self):
395 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
396 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
397 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
398 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
399 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
400 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
401 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
402 self.assertEqual(
403 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
404 1417001400)
405 self.assertEqual(
406 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
407 1422902860)
408 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
409 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
410 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
411 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
412 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
413 self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
414 self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
415 self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
416 self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
417 self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
419 self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1)
420 self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86)
421 self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78)
423 def test_determine_ext(self):
424 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
425 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
426 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
427 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
428 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
429 self.assertEqual(determine_ext('foobar', None), None)
431 def test_find_xpath_attr(self):
432 testxml = '''<root>
433 <node/>
434 <node x="a"/>
435 <node x="a" y="c" />
436 <node x="b" y="d" />
437 <node x="" />
438 </root>'''
439 doc = compat_etree_fromstring(testxml)
441 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
442 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
443 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
444 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
445 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
446 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
447 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
448 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
449 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
450 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
451 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
453 def test_xpath_with_ns(self):
454 testxml = '''<root xmlns:media="http://example.com/">
455 <media:song>
456 <media:author>The Author</media:author>
457 <url>http://server.com/download.mp3</url>
458 </media:song>
459 </root>'''
460 doc = compat_etree_fromstring(testxml)
461 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
462 self.assertTrue(find('media:song') is not None)
463 self.assertEqual(find('media:song/media:author').text, 'The Author')
464 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
466 def test_xpath_element(self):
467 doc = xml.etree.ElementTree.Element('root')
468 div = xml.etree.ElementTree.SubElement(doc, 'div')
469 p = xml.etree.ElementTree.SubElement(div, 'p')
470 p.text = 'Foo'
471 self.assertEqual(xpath_element(doc, 'div/p'), p)
472 self.assertEqual(xpath_element(doc, ['div/p']), p)
473 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
474 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
475 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
476 self.assertTrue(xpath_element(doc, 'div/bar') is None)
477 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
478 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
479 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
480 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
481 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
483 def test_xpath_text(self):
484 testxml = '''<root>
485 <div>
486 <p>Foo</p>
487 </div>
488 </root>'''
489 doc = compat_etree_fromstring(testxml)
490 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
491 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
492 self.assertTrue(xpath_text(doc, 'div/bar') is None)
493 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
495 def test_xpath_attr(self):
496 testxml = '''<root>
497 <div>
498 <p x="a">Foo</p>
499 </div>
500 </root>'''
501 doc = compat_etree_fromstring(testxml)
502 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
503 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
504 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
505 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
506 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
507 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
508 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
510 def test_smuggle_url(self):
511 data = {"ö": "ö", "abc": [3]}
512 url = 'https://foo.bar/baz?x=y#a'
513 smug_url = smuggle_url(url, data)
514 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
515 self.assertEqual(url, unsmug_url)
516 self.assertEqual(data, unsmug_data)
518 res_url, res_data = unsmuggle_url(url)
519 self.assertEqual(res_url, url)
520 self.assertEqual(res_data, None)
522 smug_url = smuggle_url(url, {'a': 'b'})
523 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
524 res_url, res_data = unsmuggle_url(smug_smug_url)
525 self.assertEqual(res_url, url)
526 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
528 def test_shell_quote(self):
529 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
530 self.assertEqual(
531 shell_quote(args),
532 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
534 def test_float_or_none(self):
535 self.assertEqual(float_or_none('42.42'), 42.42)
536 self.assertEqual(float_or_none('42'), 42.0)
537 self.assertEqual(float_or_none(''), None)
538 self.assertEqual(float_or_none(None), None)
539 self.assertEqual(float_or_none([]), None)
540 self.assertEqual(float_or_none(set()), None)
542 def test_int_or_none(self):
543 self.assertEqual(int_or_none('42'), 42)
544 self.assertEqual(int_or_none(''), None)
545 self.assertEqual(int_or_none(None), None)
546 self.assertEqual(int_or_none([]), None)
547 self.assertEqual(int_or_none(set()), None)
549 def test_str_to_int(self):
550 self.assertEqual(str_to_int('123,456'), 123456)
551 self.assertEqual(str_to_int('123.456'), 123456)
552 self.assertEqual(str_to_int(523), 523)
553 self.assertEqual(str_to_int('noninteger'), None)
554 self.assertEqual(str_to_int([]), None)
556 def test_url_basename(self):
557 self.assertEqual(url_basename('http://foo.de/'), '')
558 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
559 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
560 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
561 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
562 self.assertEqual(
563 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
564 'trailer.mp4')
566 def test_base_url(self):
567 self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
568 self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
569 self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
570 self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
571 self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
572 self.assertEqual(base_url('http://foo.de/bar/baz&x=z&w=y/x/c'), 'http://foo.de/bar/baz&x=z&w=y/x/')
574 def test_urljoin(self):
575 self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
576 self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
577 self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
578 self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
579 self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
580 self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
581 self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
582 self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
583 self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
584 self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
585 self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
586 self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
587 self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
588 self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
589 self.assertEqual(urljoin('http://foo.de/', None), None)
590 self.assertEqual(urljoin('http://foo.de/', ''), None)
591 self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
592 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
593 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
594 self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
596 def test_url_or_none(self):
597 self.assertEqual(url_or_none(None), None)
598 self.assertEqual(url_or_none(''), None)
599 self.assertEqual(url_or_none('foo'), None)
600 self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
601 self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
602 self.assertEqual(url_or_none('http$://foo.de'), None)
603 self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
604 self.assertEqual(url_or_none('//foo.de'), '//foo.de')
605 self.assertEqual(url_or_none('s3://foo.de'), None)
606 self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
607 self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
608 self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
609 self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
611 def test_parse_age_limit(self):
612 self.assertEqual(parse_age_limit(None), None)
613 self.assertEqual(parse_age_limit(False), None)
614 self.assertEqual(parse_age_limit('invalid'), None)
615 self.assertEqual(parse_age_limit(0), 0)
616 self.assertEqual(parse_age_limit(18), 18)
617 self.assertEqual(parse_age_limit(21), 21)
618 self.assertEqual(parse_age_limit(22), None)
619 self.assertEqual(parse_age_limit('18'), 18)
620 self.assertEqual(parse_age_limit('18+'), 18)
621 self.assertEqual(parse_age_limit('PG-13'), 13)
622 self.assertEqual(parse_age_limit('TV-14'), 14)
623 self.assertEqual(parse_age_limit('TV-MA'), 17)
624 self.assertEqual(parse_age_limit('TV14'), 14)
625 self.assertEqual(parse_age_limit('TV_G'), 0)
627 def test_parse_duration(self):
628 self.assertEqual(parse_duration(None), None)
629 self.assertEqual(parse_duration(False), None)
630 self.assertEqual(parse_duration('invalid'), None)
631 self.assertEqual(parse_duration('1'), 1)
632 self.assertEqual(parse_duration('1337:12'), 80232)
633 self.assertEqual(parse_duration('9:12:43'), 33163)
634 self.assertEqual(parse_duration('12:00'), 720)
635 self.assertEqual(parse_duration('00:01:01'), 61)
636 self.assertEqual(parse_duration('x:y'), None)
637 self.assertEqual(parse_duration('3h11m53s'), 11513)
638 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
639 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
640 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
641 self.assertEqual(parse_duration('3 hours, 11 minutes, 53 seconds'), 11513)
642 self.assertEqual(parse_duration('3 hours, 11 mins, 53 secs'), 11513)
643 self.assertEqual(parse_duration('62m45s'), 3765)
644 self.assertEqual(parse_duration('6m59s'), 419)
645 self.assertEqual(parse_duration('49s'), 49)
646 self.assertEqual(parse_duration('0h0m0s'), 0)
647 self.assertEqual(parse_duration('0m0s'), 0)
648 self.assertEqual(parse_duration('0s'), 0)
649 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
650 self.assertEqual(parse_duration('T30M38S'), 1838)
651 self.assertEqual(parse_duration('5 s'), 5)
652 self.assertEqual(parse_duration('3 min'), 180)
653 self.assertEqual(parse_duration('2.5 hours'), 9000)
654 self.assertEqual(parse_duration('02:03:04'), 7384)
655 self.assertEqual(parse_duration('01:02:03:04'), 93784)
656 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
657 self.assertEqual(parse_duration('87 Min.'), 5220)
658 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
659 self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
660 self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
661 self.assertEqual(parse_duration('01:02:03:050'), 3723.05)
662 self.assertEqual(parse_duration('103:050'), 103.05)
663 self.assertEqual(parse_duration('1HR 3MIN'), 3780)
664 self.assertEqual(parse_duration('2hrs 3mins'), 7380)
666 def test_fix_xml_ampersands(self):
667 self.assertEqual(
668 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
669 self.assertEqual(
670 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
671 '"&amp;x=y&amp;wrong;&amp;z=a')
672 self.assertEqual(
673 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
674 '&amp;&apos;&gt;&lt;&quot;')
675 self.assertEqual(
676 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
677 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
679 def test_paged_list(self):
680 def testPL(size, pagesize, sliceargs, expected):
681 def get_page(pagenum):
682 firstid = pagenum * pagesize
683 upto = min(size, pagenum * pagesize + pagesize)
684 yield from range(firstid, upto)
686 pl = OnDemandPagedList(get_page, pagesize)
687 got = pl.getslice(*sliceargs)
688 self.assertEqual(got, expected)
690 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
691 got = iapl.getslice(*sliceargs)
692 self.assertEqual(got, expected)
694 testPL(5, 2, (), [0, 1, 2, 3, 4])
695 testPL(5, 2, (1,), [1, 2, 3, 4])
696 testPL(5, 2, (2,), [2, 3, 4])
697 testPL(5, 2, (4,), [4])
698 testPL(5, 2, (0, 3), [0, 1, 2])
699 testPL(5, 2, (1, 4), [1, 2, 3])
700 testPL(5, 2, (2, 99), [2, 3, 4])
701 testPL(5, 2, (20, 99), [])
703 def test_read_batch_urls(self):
704 f = io.StringIO('''\xef\xbb\xbf foo
705 bar\r
707 # More after this line\r
708 ; or after this
709 bam''')
710 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
712 def test_urlencode_postdata(self):
713 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
714 self.assertTrue(isinstance(data, bytes))
716 def test_update_url_query(self):
717 self.assertEqual(parse_qs(update_url_query(
718 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
719 parse_qs('http://example.com/path?quality=HD&format=mp4'))
720 self.assertEqual(parse_qs(update_url_query(
721 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
722 parse_qs('http://example.com/path?system=LINUX&system=WINDOWS'))
723 self.assertEqual(parse_qs(update_url_query(
724 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
725 parse_qs('http://example.com/path?fields=id,formats,subtitles'))
726 self.assertEqual(parse_qs(update_url_query(
727 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
728 parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
729 self.assertEqual(parse_qs(update_url_query(
730 'http://example.com/path?manifest=f4m', {'manifest': []})),
731 parse_qs('http://example.com/path'))
732 self.assertEqual(parse_qs(update_url_query(
733 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
734 parse_qs('http://example.com/path?system=LINUX'))
735 self.assertEqual(parse_qs(update_url_query(
736 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
737 parse_qs('http://example.com/path?fields=id,formats,subtitles'))
738 self.assertEqual(parse_qs(update_url_query(
739 'http://example.com/path', {'width': 1080, 'height': 720})),
740 parse_qs('http://example.com/path?width=1080&height=720'))
741 self.assertEqual(parse_qs(update_url_query(
742 'http://example.com/path', {'bitrate': 5020.43})),
743 parse_qs('http://example.com/path?bitrate=5020.43'))
744 self.assertEqual(parse_qs(update_url_query(
745 'http://example.com/path', {'test': '第二行тест'})),
746 parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
748 def test_multipart_encode(self):
749 self.assertEqual(
750 multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
751 b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
752 self.assertEqual(
753 multipart_encode({'欄位'.encode(): '值'.encode()}, boundary='AAAAAA')[0],
754 b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n')
755 self.assertRaises(
756 ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
758 def test_dict_get(self):
759 FALSE_VALUES = {
760 'none': None,
761 'false': False,
762 'zero': 0,
763 'empty_string': '',
764 'empty_list': [],
766 d = FALSE_VALUES.copy()
767 d['a'] = 42
768 self.assertEqual(dict_get(d, 'a'), 42)
769 self.assertEqual(dict_get(d, 'b'), None)
770 self.assertEqual(dict_get(d, 'b', 42), 42)
771 self.assertEqual(dict_get(d, ('a', )), 42)
772 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
773 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
774 self.assertEqual(dict_get(d, ('b', 'c', )), None)
775 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
776 for key, false_value in FALSE_VALUES.items():
777 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
778 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
780 def test_merge_dicts(self):
781 self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
782 self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
783 self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
784 self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
785 self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
786 self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
787 self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
788 self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
789 self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
791 def test_encode_compat_str(self):
792 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
793 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
795 def test_parse_iso8601(self):
796 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
797 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
798 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
799 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
800 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
801 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
803 def test_strip_jsonp(self):
804 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
805 d = json.loads(stripped)
806 self.assertEqual(d, [{"id": "532cb", "x": 3}])
808 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
809 d = json.loads(stripped)
810 self.assertEqual(d, {'STATUS': 'OK'})
812 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
813 d = json.loads(stripped)
814 self.assertEqual(d, {'status': 'success'})
816 stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
817 d = json.loads(stripped)
818 self.assertEqual(d, {'status': 'success'})
820 stripped = strip_jsonp('window.cb && cb({"status": "success"});')
821 d = json.loads(stripped)
822 self.assertEqual(d, {'status': 'success'})
824 stripped = strip_jsonp('({"status": "success"});')
825 d = json.loads(stripped)
826 self.assertEqual(d, {'status': 'success'})
828 def test_strip_or_none(self):
829 self.assertEqual(strip_or_none(' abc'), 'abc')
830 self.assertEqual(strip_or_none('abc '), 'abc')
831 self.assertEqual(strip_or_none(' abc '), 'abc')
832 self.assertEqual(strip_or_none('\tabc\t'), 'abc')
833 self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
834 self.assertEqual(strip_or_none('abc'), 'abc')
835 self.assertEqual(strip_or_none(''), '')
836 self.assertEqual(strip_or_none(None), None)
837 self.assertEqual(strip_or_none(42), None)
838 self.assertEqual(strip_or_none([]), None)
840 def test_uppercase_escape(self):
841 self.assertEqual(uppercase_escape('aä'), 'aä')
842 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
844 def test_lowercase_escape(self):
845 self.assertEqual(lowercase_escape('aä'), 'aä')
846 self.assertEqual(lowercase_escape('\\u0026'), '&')
848 def test_limit_length(self):
849 self.assertEqual(limit_length(None, 12), None)
850 self.assertEqual(limit_length('foo', 12), 'foo')
851 self.assertTrue(
852 limit_length('foo bar baz asd', 12).startswith('foo bar'))
853 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
855 def test_mimetype2ext(self):
856 self.assertEqual(mimetype2ext(None), None)
857 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
858 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
859 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
860 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
861 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
862 self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
863 self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
865 def test_month_by_name(self):
866 self.assertEqual(month_by_name(None), None)
867 self.assertEqual(month_by_name('December', 'en'), 12)
868 self.assertEqual(month_by_name('décembre', 'fr'), 12)
869 self.assertEqual(month_by_name('December'), 12)
870 self.assertEqual(month_by_name('décembre'), None)
871 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
873 def test_parse_codecs(self):
874 self.assertEqual(parse_codecs(''), {})
875 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
876 'vcodec': 'avc1.77.30',
877 'acodec': 'mp4a.40.2',
878 'dynamic_range': None,
880 self.assertEqual(parse_codecs('mp4a.40.2'), {
881 'vcodec': 'none',
882 'acodec': 'mp4a.40.2',
883 'dynamic_range': None,
885 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
886 'vcodec': 'avc1.42001e',
887 'acodec': 'mp4a.40.5',
888 'dynamic_range': None,
890 self.assertEqual(parse_codecs('avc3.640028'), {
891 'vcodec': 'avc3.640028',
892 'acodec': 'none',
893 'dynamic_range': None,
895 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
896 'vcodec': 'h264',
897 'acodec': 'aac',
898 'dynamic_range': None,
900 self.assertEqual(parse_codecs('av01.0.05M.08'), {
901 'vcodec': 'av01.0.05M.08',
902 'acodec': 'none',
903 'dynamic_range': None,
905 self.assertEqual(parse_codecs('vp9.2'), {
906 'vcodec': 'vp9.2',
907 'acodec': 'none',
908 'dynamic_range': 'HDR10',
910 self.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), {
911 'vcodec': 'av01.0.12M.10.0.110.09.16.09.0',
912 'acodec': 'none',
913 'dynamic_range': 'HDR10',
915 self.assertEqual(parse_codecs('dvhe'), {
916 'vcodec': 'dvhe',
917 'acodec': 'none',
918 'dynamic_range': 'DV',
920 self.assertEqual(parse_codecs('theora, vorbis'), {
921 'vcodec': 'theora',
922 'acodec': 'vorbis',
923 'dynamic_range': None,
925 self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
926 'vcodec': 'unknownvcodec',
927 'acodec': 'unknownacodec',
929 self.assertEqual(parse_codecs('unknown'), {})
931 def test_escape_rfc3986(self):
932 reserved = "!*'();:@&=+$,/?#[]"
933 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
934 self.assertEqual(escape_rfc3986(reserved), reserved)
935 self.assertEqual(escape_rfc3986(unreserved), unreserved)
936 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
937 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
938 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
939 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
941 def test_normalize_url(self):
942 self.assertEqual(
943 normalize_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
944 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
946 self.assertEqual(
947 normalize_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
948 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
950 self.assertEqual(
951 normalize_url('http://тест.рф/фрагмент'),
952 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
954 self.assertEqual(
955 normalize_url('http://тест.рф/абв?абв=абв#абв'),
956 'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
958 self.assertEqual(normalize_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
960 self.assertEqual(normalize_url('http://www.example.com/../a/b/../c/./d.html'), 'http://www.example.com/a/c/d.html')
962 def test_remove_dot_segments(self):
963 self.assertEqual(remove_dot_segments('/a/b/c/./../../g'), '/a/g')
964 self.assertEqual(remove_dot_segments('mid/content=5/../6'), 'mid/6')
965 self.assertEqual(remove_dot_segments('/ad/../cd'), '/cd')
966 self.assertEqual(remove_dot_segments('/ad/../cd/'), '/cd/')
967 self.assertEqual(remove_dot_segments('/..'), '/')
968 self.assertEqual(remove_dot_segments('/./'), '/')
969 self.assertEqual(remove_dot_segments('/./a'), '/a')
970 self.assertEqual(remove_dot_segments('/abc/./.././d/././e/.././f/./../../ghi'), '/ghi')
971 self.assertEqual(remove_dot_segments('/'), '/')
972 self.assertEqual(remove_dot_segments('/t'), '/t')
973 self.assertEqual(remove_dot_segments('t'), 't')
974 self.assertEqual(remove_dot_segments(''), '')
975 self.assertEqual(remove_dot_segments('/../a/b/c'), '/a/b/c')
976 self.assertEqual(remove_dot_segments('../a'), 'a')
977 self.assertEqual(remove_dot_segments('./a'), 'a')
978 self.assertEqual(remove_dot_segments('.'), '')
979 self.assertEqual(remove_dot_segments('////'), '////')
981 def test_js_to_json_vars_strings(self):
982 self.assertDictEqual(
983 json.loads(js_to_json(
984 '''{
985 'null': a,
986 'nullStr': b,
987 'true': c,
988 'trueStr': d,
989 'false': e,
990 'falseStr': f,
991 'unresolvedVar': g,
992 }''',
994 'a': 'null',
995 'b': '"null"',
996 'c': 'true',
997 'd': '"true"',
998 'e': 'false',
999 'f': '"false"',
1000 'g': 'var',
1004 'null': None,
1005 'nullStr': 'null',
1006 'true': True,
1007 'trueStr': 'true',
1008 'false': False,
1009 'falseStr': 'false',
1010 'unresolvedVar': 'var'
1014 self.assertDictEqual(
1015 json.loads(js_to_json(
1016 '''{
1017 'int': a,
1018 'intStr': b,
1019 'float': c,
1020 'floatStr': d,
1021 }''',
1023 'a': '123',
1024 'b': '"123"',
1025 'c': '1.23',
1026 'd': '"1.23"',
1030 'int': 123,
1031 'intStr': '123',
1032 'float': 1.23,
1033 'floatStr': '1.23',
1037 self.assertDictEqual(
1038 json.loads(js_to_json(
1039 '''{
1040 'object': a,
1041 'objectStr': b,
1042 'array': c,
1043 'arrayStr': d,
1044 }''',
1046 'a': '{}',
1047 'b': '"{}"',
1048 'c': '[]',
1049 'd': '"[]"',
1053 'object': {},
1054 'objectStr': '{}',
1055 'array': [],
1056 'arrayStr': '[]',
1060 def test_js_to_json_realworld(self):
1061 inp = '''{
1062 'clip':{'provider':'pseudo'}
1063 }'''
1064 self.assertEqual(js_to_json(inp), '''{
1065 "clip":{"provider":"pseudo"}
1066 }''')
1067 json.loads(js_to_json(inp))
1069 inp = '''{
1070 'playlist':[{'controls':{'all':null}}]
1071 }'''
1072 self.assertEqual(js_to_json(inp), '''{
1073 "playlist":[{"controls":{"all":null}}]
1074 }''')
1076 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
1077 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
1079 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
1080 json_code = js_to_json(inp)
1081 self.assertEqual(json.loads(json_code), json.loads(inp))
1083 inp = '''{
1084 0:{src:'skipped', type: 'application/dash+xml'},
1085 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
1086 }'''
1087 self.assertEqual(js_to_json(inp), '''{
1088 "0":{"src":"skipped", "type": "application/dash+xml"},
1089 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
1090 }''')
1092 inp = '''{"foo":101}'''
1093 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
1095 inp = '''{"duration": "00:01:07"}'''
1096 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
1098 inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
1099 self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
1101 def test_js_to_json_edgecases(self):
1102 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
1103 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
1105 on = js_to_json('{"abc": true}')
1106 self.assertEqual(json.loads(on), {'abc': True})
1108 # Ignore JavaScript code as well
1109 on = js_to_json('''{
1110 "x": 1,
1111 y: "a",
1112 z: some.code
1113 }''')
1114 d = json.loads(on)
1115 self.assertEqual(d['x'], 1)
1116 self.assertEqual(d['y'], 'a')
1118 # Just drop ! prefix for now though this results in a wrong value
1119 on = js_to_json('''{
1120 a: !0,
1121 b: !1,
1122 c: !!0,
1123 d: !!42.42,
1124 e: !!![],
1125 f: !"abc",
1126 g: !"",
1127 !42: 42
1128 }''')
1129 self.assertEqual(json.loads(on), {
1130 'a': 0,
1131 'b': 1,
1132 'c': 0,
1133 'd': 42.42,
1134 'e': [],
1135 'f': "abc",
1136 'g': "",
1137 '42': 42
1140 on = js_to_json('["abc", "def",]')
1141 self.assertEqual(json.loads(on), ['abc', 'def'])
1143 on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
1144 self.assertEqual(json.loads(on), ['abc', 'def'])
1146 on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
1147 self.assertEqual(json.loads(on), ['abc', 'def'])
1149 on = js_to_json('{"abc": "def",}')
1150 self.assertEqual(json.loads(on), {'abc': 'def'})
1152 on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
1153 self.assertEqual(json.loads(on), {'abc': 'def'})
1155 on = js_to_json('{ 0: /* " \n */ ",]" , }')
1156 self.assertEqual(json.loads(on), {'0': ',]'})
1158 on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
1159 self.assertEqual(json.loads(on), {'0': ',]'})
1161 on = js_to_json('{ 0: // comment\n1 }')
1162 self.assertEqual(json.loads(on), {'0': 1})
1164 on = js_to_json(r'["<p>x<\/p>"]')
1165 self.assertEqual(json.loads(on), ['<p>x</p>'])
1167 on = js_to_json(r'["\xaa"]')
1168 self.assertEqual(json.loads(on), ['\u00aa'])
1170 on = js_to_json("['a\\\nb']")
1171 self.assertEqual(json.loads(on), ['ab'])
1173 on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
1174 self.assertEqual(json.loads(on), ['ab'])
1176 on = js_to_json('{0xff:0xff}')
1177 self.assertEqual(json.loads(on), {'255': 255})
1179 on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
1180 self.assertEqual(json.loads(on), {'255': 255})
1182 on = js_to_json('{077:077}')
1183 self.assertEqual(json.loads(on), {'63': 63})
1185 on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
1186 self.assertEqual(json.loads(on), {'63': 63})
1188 on = js_to_json('{42:42}')
1189 self.assertEqual(json.loads(on), {'42': 42})
1191 on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
1192 self.assertEqual(json.loads(on), {'42': 42})
1194 on = js_to_json('{42:4.2e1}')
1195 self.assertEqual(json.loads(on), {'42': 42.0})
1197 on = js_to_json('{ "0x40": "0x40" }')
1198 self.assertEqual(json.loads(on), {'0x40': '0x40'})
1200 on = js_to_json('{ "040": "040" }')
1201 self.assertEqual(json.loads(on), {'040': '040'})
1203 on = js_to_json('[1,//{},\n2]')
1204 self.assertEqual(json.loads(on), [1, 2])
1206 on = js_to_json(R'"\^\$\#"')
1207 self.assertEqual(json.loads(on), R'^$#', msg='Unnecessary escapes should be stripped')
1209 on = js_to_json('\'"\\""\'')
1210 self.assertEqual(json.loads(on), '"""', msg='Unnecessary quote escape should be escaped')
1212 on = js_to_json('[new Date("spam"), \'("eggs")\']')
1213 self.assertEqual(json.loads(on), ['spam', '("eggs")'], msg='Date regex should match a single string')
1215 def test_js_to_json_malformed(self):
1216 self.assertEqual(js_to_json('42a1'), '42"a1"')
1217 self.assertEqual(js_to_json('42a-1'), '42"a"-1')
1219 def test_js_to_json_template_literal(self):
1220 self.assertEqual(js_to_json('`Hello ${name}`', {'name': '"world"'}), '"Hello world"')
1221 self.assertEqual(js_to_json('`${name}${name}`', {'name': '"X"'}), '"XX"')
1222 self.assertEqual(js_to_json('`${name}${name}`', {'name': '5'}), '"55"')
1223 self.assertEqual(js_to_json('`${name}"${name}"`', {'name': '5'}), '"5\\"5\\""')
1224 self.assertEqual(js_to_json('`${name}`', {}), '"name"')
1226 def test_js_to_json_common_constructors(self):
1227 self.assertEqual(json.loads(js_to_json('new Map([["a", 5]])')), {'a': 5})
1228 self.assertEqual(json.loads(js_to_json('Array(5, 10)')), [5, 10])
1229 self.assertEqual(json.loads(js_to_json('new Array(15,5)')), [15, 5])
1230 self.assertEqual(json.loads(js_to_json('new Map([Array(5, 10),new Array(15,5)])')), {'5': 10, '15': 5})
1231 self.assertEqual(json.loads(js_to_json('new Date("123")')), "123")
1232 self.assertEqual(json.loads(js_to_json('new Date(\'2023-10-19\')')), "2023-10-19")
1234 def test_extract_attributes(self):
1235 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
1236 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
1237 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
1238 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
1239 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
1240 self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
1241 self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
1242 self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'}) # XML
1243 self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
1244 self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
1245 self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
1246 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
1247 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
1248 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
1249 self.assertEqual(extract_attributes('<e x >'), {'x': None})
1250 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1251 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1252 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1253 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1254 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1255 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1256 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
1257 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
1258 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1259 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1260 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1261 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1262 self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1263 # "Narrow" Python builds don't support unicode code points outside BMP.
1264 try:
1265 chr(0x10000)
1266 supports_outside_bmp = True
1267 except ValueError:
1268 supports_outside_bmp = False
1269 if supports_outside_bmp:
1270 self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
1271 # Malformed HTML should not break attributes extraction on older Python
1272 self.assertEqual(extract_attributes('<mal"formed/>'), {})
1274 def test_clean_html(self):
1275 self.assertEqual(clean_html('a:\nb'), 'a: b')
1276 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
1277 self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
1279 def test_intlist_to_bytes(self):
1280 self.assertEqual(
1281 intlist_to_bytes([0, 1, 127, 128, 255]),
1282 b'\x00\x01\x7f\x80\xff')
1284 def test_args_to_str(self):
1285 self.assertEqual(
1286 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
1287 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
1290 def test_parse_filesize(self):
1291 self.assertEqual(parse_filesize(None), None)
1292 self.assertEqual(parse_filesize(''), None)
1293 self.assertEqual(parse_filesize('91 B'), 91)
1294 self.assertEqual(parse_filesize('foobar'), None)
1295 self.assertEqual(parse_filesize('2 MiB'), 2097152)
1296 self.assertEqual(parse_filesize('5 GB'), 5000000000)
1297 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
1298 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
1299 self.assertEqual(parse_filesize('1,24 KB'), 1240)
1300 self.assertEqual(parse_filesize('1,24 kb'), 1240)
1301 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
1303 def test_parse_count(self):
1304 self.assertEqual(parse_count(None), None)
1305 self.assertEqual(parse_count(''), None)
1306 self.assertEqual(parse_count('0'), 0)
1307 self.assertEqual(parse_count('1000'), 1000)
1308 self.assertEqual(parse_count('1.000'), 1000)
1309 self.assertEqual(parse_count('1.1k'), 1100)
1310 self.assertEqual(parse_count('1.1 k'), 1100)
1311 self.assertEqual(parse_count('1,1 k'), 1100)
1312 self.assertEqual(parse_count('1.1kk'), 1100000)
1313 self.assertEqual(parse_count('1.1kk '), 1100000)
1314 self.assertEqual(parse_count('1,1kk'), 1100000)
1315 self.assertEqual(parse_count('100 views'), 100)
1316 self.assertEqual(parse_count('1,100 views'), 1100)
1317 self.assertEqual(parse_count('1.1kk views'), 1100000)
1318 self.assertEqual(parse_count('10M views'), 10000000)
1319 self.assertEqual(parse_count('has 10M views'), 10000000)
1321 def test_parse_resolution(self):
1322 self.assertEqual(parse_resolution(None), {})
1323 self.assertEqual(parse_resolution(''), {})
1324 self.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080})
1325 self.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080})
1326 self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1327 self.assertEqual(parse_resolution('720p'), {'height': 720})
1328 self.assertEqual(parse_resolution('4k'), {'height': 2160})
1329 self.assertEqual(parse_resolution('8K'), {'height': 4320})
1330 self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080})
1331 self.assertEqual(parse_resolution('ep1x2'), {})
1332 self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080})
1334 def test_parse_bitrate(self):
1335 self.assertEqual(parse_bitrate(None), None)
1336 self.assertEqual(parse_bitrate(''), None)
1337 self.assertEqual(parse_bitrate('300kbps'), 300)
1338 self.assertEqual(parse_bitrate('1500kbps'), 1500)
1339 self.assertEqual(parse_bitrate('300 kbps'), 300)
1341 def test_version_tuple(self):
1342 self.assertEqual(version_tuple('1'), (1,))
1343 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
1344 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
1346 def test_detect_exe_version(self):
1347 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1348 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1349 configuration: --prefix=/usr --extra-'''), '1.2.1')
1350 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1351 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1352 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1353 Trying to open render node...
1354 Success at /dev/dri/renderD128.
1355 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1357 def test_age_restricted(self):
1358 self.assertFalse(age_restricted(None, 10)) # unrestricted content
1359 self.assertFalse(age_restricted(1, None)) # unrestricted policy
1360 self.assertFalse(age_restricted(8, 10))
1361 self.assertTrue(age_restricted(18, 14))
1362 self.assertFalse(age_restricted(18, 18))
1364 def test_is_html(self):
1365 self.assertFalse(is_html(b'\x49\x44\x43<html'))
1366 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1367 self.assertTrue(is_html( # UTF-8 with BOM
1368 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1369 self.assertTrue(is_html( # UTF-16-LE
1370 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1372 self.assertTrue(is_html( # UTF-16-BE
1373 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1375 self.assertTrue(is_html( # UTF-32-BE
1376 b'\x00\x00\xFE\xFF\x00\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4'))
1377 self.assertTrue(is_html( # UTF-32-LE
1378 b'\xFF\xFE\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4\x00\x00\x00'))
1380 def test_render_table(self):
1381 self.assertEqual(
1382 render_table(
1383 ['a', 'empty', 'bcd'],
1384 [[123, '', 4], [9999, '', 51]]),
1385 'a empty bcd\n'
1386 '123 4\n'
1387 '9999 51')
1389 self.assertEqual(
1390 render_table(
1391 ['a', 'empty', 'bcd'],
1392 [[123, '', 4], [9999, '', 51]],
1393 hide_empty=True),
1394 'a bcd\n'
1395 '123 4\n'
1396 '9999 51')
1398 self.assertEqual(
1399 render_table(
1400 ['\ta', 'bcd'],
1401 [['1\t23', 4], ['\t9999', 51]]),
1402 ' a bcd\n'
1403 '1 23 4\n'
1404 '9999 51')
1406 self.assertEqual(
1407 render_table(
1408 ['a', 'bcd'],
1409 [[123, 4], [9999, 51]],
1410 delim='-'),
1411 'a bcd\n'
1412 '--------\n'
1413 '123 4\n'
1414 '9999 51')
1416 self.assertEqual(
1417 render_table(
1418 ['a', 'bcd'],
1419 [[123, 4], [9999, 51]],
1420 delim='-', extra_gap=2),
1421 'a bcd\n'
1422 '----------\n'
1423 '123 4\n'
1424 '9999 51')
1426 def test_match_str(self):
1427 # Unary
1428 self.assertFalse(match_str('xy', {'x': 1200}))
1429 self.assertTrue(match_str('!xy', {'x': 1200}))
1430 self.assertTrue(match_str('x', {'x': 1200}))
1431 self.assertFalse(match_str('!x', {'x': 1200}))
1432 self.assertTrue(match_str('x', {'x': 0}))
1433 self.assertTrue(match_str('is_live', {'is_live': True}))
1434 self.assertFalse(match_str('is_live', {'is_live': False}))
1435 self.assertFalse(match_str('is_live', {'is_live': None}))
1436 self.assertFalse(match_str('is_live', {}))
1437 self.assertFalse(match_str('!is_live', {'is_live': True}))
1438 self.assertTrue(match_str('!is_live', {'is_live': False}))
1439 self.assertTrue(match_str('!is_live', {'is_live': None}))
1440 self.assertTrue(match_str('!is_live', {}))
1441 self.assertTrue(match_str('title', {'title': 'abc'}))
1442 self.assertTrue(match_str('title', {'title': ''}))
1443 self.assertFalse(match_str('!title', {'title': 'abc'}))
1444 self.assertFalse(match_str('!title', {'title': ''}))
1446 # Numeric
1447 self.assertFalse(match_str('x>0', {'x': 0}))
1448 self.assertFalse(match_str('x>0', {}))
1449 self.assertTrue(match_str('x>?0', {}))
1450 self.assertTrue(match_str('x>1K', {'x': 1200}))
1451 self.assertFalse(match_str('x>2K', {'x': 1200}))
1452 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1453 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1454 self.assertTrue(match_str('x > 1:0:0', {'x': 3700}))
1456 # String
1457 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1458 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1459 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1460 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1461 self.assertTrue(match_str('y^=foo', {'y': 'foobar42'}))
1462 self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'}))
1463 self.assertFalse(match_str('y^=bar', {'y': 'foobar42'}))
1464 self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'}))
1465 self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42})
1466 self.assertTrue(match_str('y*=bar', {'y': 'foobar42'}))
1467 self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'}))
1468 self.assertFalse(match_str('y*=baz', {'y': 'foobar42'}))
1469 self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'}))
1470 self.assertTrue(match_str('y$=42', {'y': 'foobar42'}))
1471 self.assertFalse(match_str('y$=43', {'y': 'foobar42'}))
1473 # And
1474 self.assertFalse(match_str(
1475 'like_count > 100 & dislike_count <? 50 & description',
1476 {'like_count': 90, 'description': 'foo'}))
1477 self.assertTrue(match_str(
1478 'like_count > 100 & dislike_count <? 50 & description',
1479 {'like_count': 190, 'description': 'foo'}))
1480 self.assertFalse(match_str(
1481 'like_count > 100 & dislike_count <? 50 & description',
1482 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1483 self.assertFalse(match_str(
1484 'like_count > 100 & dislike_count <? 50 & description',
1485 {'like_count': 190, 'dislike_count': 10}))
1487 # Regex
1488 self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'}))
1489 self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'}))
1490 self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'}))
1491 self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'}))
1493 # Quotes
1494 self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'}))
1495 self.assertFalse(match_str(r'x^="foo "', {'x': 'foo "bar"'}))
1496 self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'}))
1497 self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'}))
1499 # Escaping &
1500 self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'}))
1501 self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'}))
1502 self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'}))
1503 self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'}))
1505 # Example from docs
1506 self.assertTrue(match_str(
1507 r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'",
1508 {'description': 'Raining Cats & Dogs'}))
1510 # Incomplete
1511 self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True))
1512 self.assertTrue(match_str('x', {'id': 'foo'}, True))
1513 self.assertTrue(match_str('!x', {'id': 'foo'}, True))
1514 self.assertFalse(match_str('x', {'id': 'foo'}, False))
1516 def test_parse_dfxp_time_expr(self):
1517 self.assertEqual(parse_dfxp_time_expr(None), None)
1518 self.assertEqual(parse_dfxp_time_expr(''), None)
1519 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1520 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1521 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1522 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1523 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1525 def test_dfxp2srt(self):
1526 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1527 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1528 <body>
1529 <div xml:lang="en">
1530 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1531 <p begin="1" end="2">第二行<br/>♪♪</p>
1532 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1533 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1534 <p begin="-1" end="-1">Ignore, two</p>
1535 <p begin="3" dur="-1">Ignored, three</p>
1536 </div>
1537 </body>
1538 </tt>'''.encode()
1539 srt_data = '''1
1540 00:00:00,000 --> 00:00:01,000
1541 The following line contains Chinese characters and special symbols
1544 00:00:01,000 --> 00:00:02,000
1545 第二行
1546 ♪♪
1549 00:00:02,000 --> 00:00:03,000
1550 Third
1551 Line
1554 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1556 dfxp_data_no_default_namespace = b'''<?xml version="1.0" encoding="UTF-8"?>
1557 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1558 <body>
1559 <div xml:lang="en">
1560 <p begin="0" end="1">The first line</p>
1561 </div>
1562 </body>
1563 </tt>'''
1564 srt_data = '''1
1565 00:00:00,000 --> 00:00:01,000
1566 The first line
1569 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1571 dfxp_data_with_style = b'''<?xml version="1.0" encoding="utf-8"?>
1572 <tt xmlns="http://www.w3.org/2006/10/ttaf1" xmlns:ttp="http://www.w3.org/2006/10/ttaf1#parameter" ttp:timeBase="media" xmlns:tts="http://www.w3.org/2006/10/ttaf1#style" xml:lang="en" xmlns:ttm="http://www.w3.org/2006/10/ttaf1#metadata">
1573 <head>
1574 <styling>
1575 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1576 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1577 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1578 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1579 </styling>
1580 </head>
1581 <body tts:textAlign="center" style="s0">
1582 <div>
1583 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1584 <p style="s2" begin="00:00:02.08" id="p0" end="00:00:05.84"><span tts:color="lime">part 1<br /></span><span tts:color="cyan">part 2</span></p>
1585 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1586 <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p>
1587 </div>
1588 </body>
1589 </tt>'''
1590 srt_data = '''1
1591 00:00:02,080 --> 00:00:05,840
1592 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1595 00:00:02,080 --> 00:00:05,840
1596 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1597 </font>part 2</font></b>
1600 00:00:05,840 --> 00:00:09,560
1601 <u><font color="lime">line 3
1602 part 3</font></u>
1605 00:00:09,560 --> 00:00:12,360
1606 <i><u><font color="yellow"><font color="lime">inner
1607 </font>style</font></u></i>
1610 self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1612 dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1613 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1614 <body>
1615 <div xml:lang="en">
1616 <p begin="0" end="1">Line 1</p>
1617 <p begin="1" end="2">第二行</p>
1618 </div>
1619 </body>
1620 </tt>'''.encode('utf-16')
1621 srt_data = '''1
1622 00:00:00,000 --> 00:00:01,000
1623 Line 1
1626 00:00:01,000 --> 00:00:02,000
1627 第二行
1630 self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1632 def test_cli_option(self):
1633 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1634 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1635 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1636 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1638 def test_cli_valueless_option(self):
1639 self.assertEqual(cli_valueless_option(
1640 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1641 self.assertEqual(cli_valueless_option(
1642 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1643 self.assertEqual(cli_valueless_option(
1644 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1645 self.assertEqual(cli_valueless_option(
1646 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1647 self.assertEqual(cli_valueless_option(
1648 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1649 self.assertEqual(cli_valueless_option(
1650 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1652 def test_cli_bool_option(self):
1653 self.assertEqual(
1654 cli_bool_option(
1655 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1656 ['--no-check-certificate', 'true'])
1657 self.assertEqual(
1658 cli_bool_option(
1659 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1660 ['--no-check-certificate=true'])
1661 self.assertEqual(
1662 cli_bool_option(
1663 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1664 ['--check-certificate', 'false'])
1665 self.assertEqual(
1666 cli_bool_option(
1667 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1668 ['--check-certificate=false'])
1669 self.assertEqual(
1670 cli_bool_option(
1671 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1672 ['--check-certificate', 'true'])
1673 self.assertEqual(
1674 cli_bool_option(
1675 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1676 ['--check-certificate=true'])
1677 self.assertEqual(
1678 cli_bool_option(
1679 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1682 def test_ohdave_rsa_encrypt(self):
1683 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1684 e = 65537
1686 self.assertEqual(
1687 ohdave_rsa_encrypt(b'aa111222', e, N),
1688 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1690 def test_pkcs1pad(self):
1691 data = [1, 2, 3]
1692 padded_data = pkcs1pad(data, 32)
1693 self.assertEqual(padded_data[:2], [0, 2])
1694 self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1696 self.assertRaises(ValueError, pkcs1pad, data, 8)
1698 def test_encode_base_n(self):
1699 self.assertEqual(encode_base_n(0, 30), '0')
1700 self.assertEqual(encode_base_n(80, 30), '2k')
1702 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1703 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1704 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1706 self.assertRaises(ValueError, encode_base_n, 0, 70)
1707 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1709 def test_caesar(self):
1710 self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
1711 self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
1712 self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
1713 self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
1714 self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
1715 self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
1716 self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
1717 self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
1719 def test_rot47(self):
1720 self.assertEqual(rot47('yt-dlp'), r'JE\5=A')
1721 self.assertEqual(rot47('YT-DLP'), r'*%\s{!')
1723 def test_urshift(self):
1724 self.assertEqual(urshift(3, 1), 1)
1725 self.assertEqual(urshift(-3, 1), 2147483646)
1727 GET_ELEMENT_BY_CLASS_TEST_STRING = '''
1728 <span class="foo bar">nice</span>
1731 def test_get_element_by_class(self):
1732 html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1734 self.assertEqual(get_element_by_class('foo', html), 'nice')
1735 self.assertEqual(get_element_by_class('no-such-class', html), None)
1737 def test_get_element_html_by_class(self):
1738 html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1740 self.assertEqual(get_element_html_by_class('foo', html), html.strip())
1741 self.assertEqual(get_element_by_class('no-such-class', html), None)
1743 GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = '''
1744 <div itemprop="author" itemscope>foo</div>
1747 def test_get_element_by_attribute(self):
1748 html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1750 self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1751 self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1752 self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1754 html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1756 self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1758 def test_get_element_html_by_attribute(self):
1759 html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
1761 self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip())
1762 self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None)
1763 self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None)
1765 html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
1767 self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip())
1769 GET_ELEMENTS_BY_CLASS_TEST_STRING = '''
1770 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1772 GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
1774 def test_get_elements_by_class(self):
1775 html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1777 self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1778 self.assertEqual(get_elements_by_class('no-such-class', html), [])
1780 def test_get_elements_html_by_class(self):
1781 html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1783 self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES)
1784 self.assertEqual(get_elements_html_by_class('no-such-class', html), [])
1786 def test_get_elements_by_attribute(self):
1787 html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1789 self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1790 self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1791 self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1793 def test_get_elements_html_by_attribute(self):
1794 html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1796 self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES)
1797 self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), [])
1798 self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), [])
1800 def test_get_elements_text_and_html_by_attribute(self):
1801 html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
1803 self.assertEqual(
1804 list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)),
1805 list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES)))
1806 self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), [])
1807 self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), [])
1809 self.assertEqual(list(get_elements_text_and_html_by_attribute(
1810 'class', 'foo', '<a class="foo">nice</a><span class="foo">nice</span>', tag='a')), [('nice', '<a class="foo">nice</a>')])
1812 GET_ELEMENT_BY_TAG_TEST_STRING = '''
1813 random text lorem ipsum</p>
1814 <div>
1815 this should be returned
1816 <span>this should also be returned</span>
1817 <div>
1818 this should also be returned
1819 </div>
1820 closing tag above should not trick, so this should also be returned
1821 </div>
1822 but this text should not be returned
1824 GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276]
1825 GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6]
1826 GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119]
1827 GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7]
1829 def test_get_element_text_and_html_by_tag(self):
1830 html = self.GET_ELEMENT_BY_TAG_TEST_STRING
1832 self.assertEqual(
1833 get_element_text_and_html_by_tag('div', html),
1834 (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML))
1835 self.assertEqual(
1836 get_element_text_and_html_by_tag('span', html),
1837 (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML))
1838 self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html)
1840 def test_iri_to_uri(self):
1841 self.assertEqual(
1842 iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
1843 'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b') # Same
1844 self.assertEqual(
1845 iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'), # German for cheese sauce stirring spoon
1846 'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel')
1847 self.assertEqual(
1848 iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'),
1849 'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#')
1850 self.assertEqual(
1851 iri_to_uri('http://правозащита38.рф/category/news/'),
1852 'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1853 self.assertEqual(
1854 iri_to_uri('http://www.правозащита38.рф/category/news/'),
1855 'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/')
1856 self.assertEqual(
1857 iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'),
1858 'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA')
1859 self.assertEqual(
1860 iri_to_uri('http://日本語.jp/'),
1861 'http://xn--wgv71a119e.jp/')
1862 self.assertEqual(
1863 iri_to_uri('http://导航.中国/'),
1864 'http://xn--fet810g.xn--fiqs8s/')
1866 def test_clean_podcast_url(self):
1867 self.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3')
1868 self.assertEqual(clean_podcast_url('https://play.podtrac.com/npr-344098539/edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3'), 'https://edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3')
1869 self.assertEqual(clean_podcast_url('https://pdst.fm/e/2.gum.fm/chtbl.com/track/chrt.fm/track/34D33/pscrb.fm/rss/p/traffic.megaphone.fm/ITLLC7765286967.mp3?updated=1687282661'), 'https://traffic.megaphone.fm/ITLLC7765286967.mp3?updated=1687282661')
1870 self.assertEqual(clean_podcast_url('https://pdst.fm/e/https://mgln.ai/e/441/www.buzzsprout.com/1121972/13019085-ep-252-the-deep-life-stack.mp3'), 'https://www.buzzsprout.com/1121972/13019085-ep-252-the-deep-life-stack.mp3')
1872 def test_LazyList(self):
1873 it = list(range(10))
1875 self.assertEqual(list(LazyList(it)), it)
1876 self.assertEqual(LazyList(it).exhaust(), it)
1877 self.assertEqual(LazyList(it)[5], it[5])
1879 self.assertEqual(LazyList(it)[5:], it[5:])
1880 self.assertEqual(LazyList(it)[:5], it[:5])
1881 self.assertEqual(LazyList(it)[::2], it[::2])
1882 self.assertEqual(LazyList(it)[1::2], it[1::2])
1883 self.assertEqual(LazyList(it)[5::-1], it[5::-1])
1884 self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2])
1885 self.assertEqual(LazyList(it)[::-1], it[::-1])
1887 self.assertTrue(LazyList(it))
1888 self.assertFalse(LazyList(range(0)))
1889 self.assertEqual(len(LazyList(it)), len(it))
1890 self.assertEqual(repr(LazyList(it)), repr(it))
1891 self.assertEqual(str(LazyList(it)), str(it))
1893 self.assertEqual(list(LazyList(it, reverse=True)), it[::-1])
1894 self.assertEqual(list(reversed(LazyList(it))[::-1]), it)
1895 self.assertEqual(list(reversed(LazyList(it))[1:3:7]), it[::-1][1:3:7])
1897 def test_LazyList_laziness(self):
1899 def test(ll, idx, val, cache):
1900 self.assertEqual(ll[idx], val)
1901 self.assertEqual(ll._cache, list(cache))
1903 ll = LazyList(range(10))
1904 test(ll, 0, 0, range(1))
1905 test(ll, 5, 5, range(6))
1906 test(ll, -3, 7, range(10))
1908 ll = LazyList(range(10), reverse=True)
1909 test(ll, -1, 0, range(1))
1910 test(ll, 3, 6, range(10))
1912 ll = LazyList(itertools.count())
1913 test(ll, 10, 10, range(11))
1914 ll = reversed(ll)
1915 test(ll, -15, 14, range(15))
1917 def test_format_bytes(self):
1918 self.assertEqual(format_bytes(0), '0.00B')
1919 self.assertEqual(format_bytes(1000), '1000.00B')
1920 self.assertEqual(format_bytes(1024), '1.00KiB')
1921 self.assertEqual(format_bytes(1024**2), '1.00MiB')
1922 self.assertEqual(format_bytes(1024**3), '1.00GiB')
1923 self.assertEqual(format_bytes(1024**4), '1.00TiB')
1924 self.assertEqual(format_bytes(1024**5), '1.00PiB')
1925 self.assertEqual(format_bytes(1024**6), '1.00EiB')
1926 self.assertEqual(format_bytes(1024**7), '1.00ZiB')
1927 self.assertEqual(format_bytes(1024**8), '1.00YiB')
1928 self.assertEqual(format_bytes(1024**9), '1024.00YiB')
1930 def test_hide_login_info(self):
1931 self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']),
1932 ['-u', 'PRIVATE', '-p', 'PRIVATE'])
1933 self.assertEqual(Config.hide_login_info(['-u']), ['-u'])
1934 self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']),
1935 ['-u', 'PRIVATE', '-u', 'PRIVATE'])
1936 self.assertEqual(Config.hide_login_info(['--username=foo']),
1937 ['--username=PRIVATE'])
1939 def test_locked_file(self):
1940 TEXT = 'test_locked_file\n'
1941 FILE = 'test_locked_file.ytdl'
1942 MODES = 'war' # Order is important
1944 try:
1945 for lock_mode in MODES:
1946 with locked_file(FILE, lock_mode, False) as f:
1947 if lock_mode == 'r':
1948 self.assertEqual(f.read(), TEXT * 2, 'Wrong file content')
1949 else:
1950 f.write(TEXT)
1951 for test_mode in MODES:
1952 testing_write = test_mode != 'r'
1953 try:
1954 with locked_file(FILE, test_mode, False):
1955 pass
1956 except (BlockingIOError, PermissionError):
1957 if not testing_write: # FIXME
1958 print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})')
1959 continue
1960 self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}')
1961 else:
1962 self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}')
1963 finally:
1964 with contextlib.suppress(OSError):
1965 os.remove(FILE)
1967 def test_determine_file_encoding(self):
1968 self.assertEqual(determine_file_encoding(b''), (None, 0))
1969 self.assertEqual(determine_file_encoding(b'--verbose -x --audio-format mkv\n'), (None, 0))
1971 self.assertEqual(determine_file_encoding(b'\xef\xbb\xbf'), ('utf-8', 3))
1972 self.assertEqual(determine_file_encoding(b'\x00\x00\xfe\xff'), ('utf-32-be', 4))
1973 self.assertEqual(determine_file_encoding(b'\xff\xfe'), ('utf-16-le', 2))
1975 self.assertEqual(determine_file_encoding(b'\xff\xfe# coding: utf-8\n--verbose'), ('utf-16-le', 2))
1977 self.assertEqual(determine_file_encoding(b'# coding: utf-8\n--verbose'), ('utf-8', 0))
1978 self.assertEqual(determine_file_encoding(b'# coding: someencodinghere-12345\n--verbose'), ('someencodinghere-12345', 0))
1980 self.assertEqual(determine_file_encoding(b'#coding:utf-8\n--verbose'), ('utf-8', 0))
1981 self.assertEqual(determine_file_encoding(b'# coding: utf-8 \r\n--verbose'), ('utf-8', 0))
1983 self.assertEqual(determine_file_encoding('# coding: utf-32-be'.encode('utf-32-be')), ('utf-32-be', 0))
1984 self.assertEqual(determine_file_encoding('# coding: utf-16-le'.encode('utf-16-le')), ('utf-16-le', 0))
1986 def test_get_compatible_ext(self):
1987 self.assertEqual(get_compatible_ext(
1988 vcodecs=[None], acodecs=[None, None], vexts=['mp4'], aexts=['m4a', 'm4a']), 'mkv')
1989 self.assertEqual(get_compatible_ext(
1990 vcodecs=[None], acodecs=[None], vexts=['flv'], aexts=['flv']), 'flv')
1992 self.assertEqual(get_compatible_ext(
1993 vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['m4a']), 'mp4')
1994 self.assertEqual(get_compatible_ext(
1995 vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['webm']), 'mkv')
1996 self.assertEqual(get_compatible_ext(
1997 vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['m4a']), 'mkv')
1998 self.assertEqual(get_compatible_ext(
1999 vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['webm']), 'webm')
2000 self.assertEqual(get_compatible_ext(
2001 vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['weba']), 'webm')
2003 self.assertEqual(get_compatible_ext(
2004 vcodecs=['h264'], acodecs=['mp4a'], vexts=['mov'], aexts=['m4a']), 'mp4')
2005 self.assertEqual(get_compatible_ext(
2006 vcodecs=['av01.0.12M.08'], acodecs=['opus'], vexts=['mp4'], aexts=['webm']), 'webm')
2008 self.assertEqual(get_compatible_ext(
2009 vcodecs=['vp9'], acodecs=['opus'], vexts=['webm'], aexts=['webm'], preferences=['flv', 'mp4']), 'mp4')
2010 self.assertEqual(get_compatible_ext(
2011 vcodecs=['av1'], acodecs=['mp4a'], vexts=['webm'], aexts=['m4a'], preferences=('webm', 'mkv')), 'mkv')
2013 def test_try_call(self):
2014 def total(*x, **kwargs):
2015 return sum(x) + sum(kwargs.values())
2017 self.assertEqual(try_call(None), None,
2018 msg='not a fn should give None')
2019 self.assertEqual(try_call(lambda: 1), 1,
2020 msg='int fn with no expected_type should give int')
2021 self.assertEqual(try_call(lambda: 1, expected_type=int), 1,
2022 msg='int fn with expected_type int should give int')
2023 self.assertEqual(try_call(lambda: 1, expected_type=dict), None,
2024 msg='int fn with wrong expected_type should give None')
2025 self.assertEqual(try_call(total, args=(0, 1, 0, ), expected_type=int), 1,
2026 msg='fn should accept arglist')
2027 self.assertEqual(try_call(total, kwargs={'a': 0, 'b': 1, 'c': 0}, expected_type=int), 1,
2028 msg='fn should accept kwargs')
2029 self.assertEqual(try_call(lambda: 1, expected_type=dict), None,
2030 msg='int fn with no expected_type should give None')
2031 self.assertEqual(try_call(lambda x: {}, total, args=(42, ), expected_type=int), 42,
2032 msg='expect first int result with expected_type int')
2034 def test_variadic(self):
2035 self.assertEqual(variadic(None), (None, ))
2036 self.assertEqual(variadic('spam'), ('spam', ))
2037 self.assertEqual(variadic('spam', allowed_types=dict), 'spam')
2038 with warnings.catch_warnings():
2039 warnings.simplefilter('ignore')
2040 self.assertEqual(variadic('spam', allowed_types=[dict]), 'spam')
2042 def test_traverse_obj(self):
2043 _TEST_DATA = {
2044 100: 100,
2045 1.2: 1.2,
2046 'str': 'str',
2047 'None': None,
2048 '...': ...,
2049 'urls': [
2050 {'index': 0, 'url': 'https://www.example.com/0'},
2051 {'index': 1, 'url': 'https://www.example.com/1'},
2053 'data': (
2054 {'index': 2},
2055 {'index': 3},
2057 'dict': {},
2060 # Test base functionality
2061 self.assertEqual(traverse_obj(_TEST_DATA, ('str',)), 'str',
2062 msg='allow tuple path')
2063 self.assertEqual(traverse_obj(_TEST_DATA, ['str']), 'str',
2064 msg='allow list path')
2065 self.assertEqual(traverse_obj(_TEST_DATA, (value for value in ("str",))), 'str',
2066 msg='allow iterable path')
2067 self.assertEqual(traverse_obj(_TEST_DATA, 'str'), 'str',
2068 msg='single items should be treated as a path')
2069 self.assertEqual(traverse_obj(_TEST_DATA, None), _TEST_DATA)
2070 self.assertEqual(traverse_obj(_TEST_DATA, 100), 100)
2071 self.assertEqual(traverse_obj(_TEST_DATA, 1.2), 1.2)
2073 # Test Ellipsis behavior
2074 self.assertCountEqual(traverse_obj(_TEST_DATA, ...),
2075 (item for item in _TEST_DATA.values() if item not in (None, {})),
2076 msg='`...` should give all non discarded values')
2077 self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, ...)), _TEST_DATA['urls'][0].values(),
2078 msg='`...` selection for dicts should select all values')
2079 self.assertEqual(traverse_obj(_TEST_DATA, (..., ..., 'url')),
2080 ['https://www.example.com/0', 'https://www.example.com/1'],
2081 msg='nested `...` queries should work')
2082 self.assertCountEqual(traverse_obj(_TEST_DATA, (..., ..., 'index')), range(4),
2083 msg='`...` query result should be flattened')
2084 self.assertEqual(traverse_obj(iter(range(4)), ...), list(range(4)),
2085 msg='`...` should accept iterables')
2087 # Test function as key
2088 self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)),
2089 [_TEST_DATA['urls']],
2090 msg='function as query key should perform a filter based on (key, value)')
2091 self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'},
2092 msg='exceptions in the query function should be catched')
2093 self.assertEqual(traverse_obj(iter(range(4)), lambda _, x: x % 2 == 0), [0, 2],
2094 msg='function key should accept iterables')
2095 if __debug__:
2096 with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
2097 traverse_obj(_TEST_DATA, lambda a: ...)
2098 with self.assertRaises(Exception, msg='Wrong function signature should raise in debug'):
2099 traverse_obj(_TEST_DATA, lambda a, b, c: ...)
2101 # Test set as key (transformation/type, like `expected_type`)
2102 self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper}, )), ['STR'],
2103 msg='Function in set should be a transformation')
2104 self.assertEqual(traverse_obj(_TEST_DATA, (..., {str})), ['str'],
2105 msg='Type in set should be a type filter')
2106 self.assertEqual(traverse_obj(_TEST_DATA, {dict}), _TEST_DATA,
2107 msg='A single set should be wrapped into a path')
2108 self.assertEqual(traverse_obj(_TEST_DATA, (..., {str.upper})), ['STR'],
2109 msg='Transformation function should not raise')
2110 self.assertEqual(traverse_obj(_TEST_DATA, (..., {str_or_none})),
2111 [item for item in map(str_or_none, _TEST_DATA.values()) if item is not None],
2112 msg='Function in set should be a transformation')
2113 self.assertEqual(traverse_obj(_TEST_DATA, ('fail', {lambda _: 'const'})), 'const',
2114 msg='Function in set should always be called')
2115 if __debug__:
2116 with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
2117 traverse_obj(_TEST_DATA, set())
2118 with self.assertRaises(Exception, msg='Sets with length != 1 should raise in debug'):
2119 traverse_obj(_TEST_DATA, {str.upper, str})
2121 # Test `slice` as a key
2122 _SLICE_DATA = [0, 1, 2, 3, 4]
2123 self.assertEqual(traverse_obj(_TEST_DATA, ('dict', slice(1))), None,
2124 msg='slice on a dictionary should not throw')
2125 self.assertEqual(traverse_obj(_SLICE_DATA, slice(1)), _SLICE_DATA[:1],
2126 msg='slice key should apply slice to sequence')
2127 self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 2)), _SLICE_DATA[1:2],
2128 msg='slice key should apply slice to sequence')
2129 self.assertEqual(traverse_obj(_SLICE_DATA, slice(1, 4, 2)), _SLICE_DATA[1:4:2],
2130 msg='slice key should apply slice to sequence')
2132 # Test alternative paths
2133 self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str',
2134 msg='multiple `paths` should be treated as alternative paths')
2135 self.assertEqual(traverse_obj(_TEST_DATA, 'str', 100), 'str',
2136 msg='alternatives should exit early')
2137 self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'fail'), None,
2138 msg='alternatives should return `default` if exhausted')
2139 self.assertEqual(traverse_obj(_TEST_DATA, (..., 'fail'), 100), 100,
2140 msg='alternatives should track their own branching return')
2141 self.assertEqual(traverse_obj(_TEST_DATA, ('dict', ...), ('data', ...)), list(_TEST_DATA['data']),
2142 msg='alternatives on empty objects should search further')
2144 # Test branch and path nesting
2145 self.assertEqual(traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')), ['https://www.example.com/0'],
2146 msg='tuple as key should be treated as branches')
2147 self.assertEqual(traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')), ['https://www.example.com/0'],
2148 msg='list as key should be treated as branches')
2149 self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))), ['https://www.example.com/0'],
2150 msg='double nesting in path should be treated as paths')
2151 self.assertEqual(traverse_obj(['0', [1, 2]], [(0, 1), 0]), [1],
2152 msg='do not fail early on branching')
2153 self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', ((1, ('fail', 'url')), (0, 'url')))),
2154 ['https://www.example.com/0', 'https://www.example.com/1'],
2155 msg='tripple nesting in path should be treated as branches')
2156 self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))),
2157 ['https://www.example.com/0', 'https://www.example.com/1'],
2158 msg='ellipsis as branch path start gets flattened')
2160 # Test dictionary as key
2161 self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}), {0: 100, 1: 1.2},
2162 msg='dict key should result in a dict with the same keys')
2163 self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}),
2164 {0: 'https://www.example.com/0'},
2165 msg='dict key should allow paths')
2166 self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}),
2167 {0: ['https://www.example.com/0']},
2168 msg='tuple in dict path should be treated as branches')
2169 self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}),
2170 {0: ['https://www.example.com/0']},
2171 msg='double nesting in dict path should be treated as paths')
2172 self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}),
2173 {0: ['https://www.example.com/1', 'https://www.example.com/0']},
2174 msg='tripple nesting in dict path should be treated as branches')
2175 self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}), {},
2176 msg='remove `None` values when top level dict key fails')
2177 self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}, default=...), {0: ...},
2178 msg='use `default` if key fails and `default`')
2179 self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {},
2180 msg='remove empty values when dict key')
2181 self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=...), {0: ...},
2182 msg='use `default` when dict key and `default`')
2183 self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}), {},
2184 msg='remove empty values when nested dict key fails')
2185 self.assertEqual(traverse_obj(None, {0: 'fail'}), {},
2186 msg='default to dict if pruned')
2187 self.assertEqual(traverse_obj(None, {0: 'fail'}, default=...), {0: ...},
2188 msg='default to dict if pruned and default is given')
2189 self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 'fail'}}, default=...), {0: {0: ...}},
2190 msg='use nested `default` when nested dict key fails and `default`')
2191 self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', ...)}), {},
2192 msg='remove key if branch in dict key not successful')
2194 # Testing default parameter behavior
2195 _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []}
2196 self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail'), None,
2197 msg='default value should be `None`')
2198 self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...), ...,
2199 msg='chained fails should result in default')
2200 self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', 'int'), 0,
2201 msg='should not short cirquit on `None`')
2202 self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', default=1), 1,
2203 msg='invalid dict key should result in `default`')
2204 self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', default=1), 1,
2205 msg='`None` is a deliberate sentinel and should become `default`')
2206 self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', 10)), None,
2207 msg='`IndexError` should result in `default`')
2208 self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1), 1,
2209 msg='if branched but not successful return `default` if defined, not `[]`')
2210 self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=None), None,
2211 msg='if branched but not successful return `default` even if `default` is `None`')
2212 self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail')), [],
2213 msg='if branched but not successful return `[]`, not `default`')
2214 self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', ...)), [],
2215 msg='if branched but object is empty return `[]`, not `default`')
2216 self.assertEqual(traverse_obj(None, ...), [],
2217 msg='if branched but object is `None` return `[]`, not `default`')
2218 self.assertEqual(traverse_obj({0: None}, (0, ...)), [],
2219 msg='if branched but state is `None` return `[]`, not `default`')
2221 branching_paths = [
2222 ('fail', ...),
2223 (..., 'fail'),
2224 100 * ('fail',) + (...,),
2225 (...,) + 100 * ('fail',),
2227 for branching_path in branching_paths:
2228 self.assertEqual(traverse_obj({}, branching_path), [],
2229 msg='if branched but state is `None`, return `[]` (not `default`)')
2230 self.assertEqual(traverse_obj({}, 'fail', branching_path), [],
2231 msg='if branching in last alternative and previous did not match, return `[]` (not `default`)')
2232 self.assertEqual(traverse_obj({0: 'x'}, 0, branching_path), 'x',
2233 msg='if branching in last alternative and previous did match, return single value')
2234 self.assertEqual(traverse_obj({0: 'x'}, branching_path, 0), 'x',
2235 msg='if branching in first alternative and non-branching path does match, return single value')
2236 self.assertEqual(traverse_obj({}, branching_path, 'fail'), None,
2237 msg='if branching in first alternative and non-branching path does not match, return `default`')
2239 # Testing expected_type behavior
2240 _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0}
2241 self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str),
2242 'str', msg='accept matching `expected_type` type')
2243 self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int),
2244 None, msg='reject non matching `expected_type` type')
2245 self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)),
2246 '0', msg='transform type using type function')
2247 self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=lambda _: 1 / 0),
2248 None, msg='wrap expected_type fuction in try_call')
2249 self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str),
2250 ['str'], msg='eliminate items that expected_type fails on')
2251 self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}, expected_type=int),
2252 {0: 100}, msg='type as expected_type should filter dict values')
2253 self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2, 2: 'None'}, expected_type=str_or_none),
2254 {0: '100', 1: '1.2'}, msg='function as expected_type should transform dict values')
2255 self.assertEqual(traverse_obj(_TEST_DATA, ({0: 1.2}, 0, {int_or_none}), expected_type=int),
2256 1, msg='expected_type should not filter non final dict values')
2257 self.assertEqual(traverse_obj(_TEST_DATA, {0: {0: 100, 1: 'str'}}, expected_type=int),
2258 {0: {0: 100}}, msg='expected_type should transform deep dict values')
2259 self.assertEqual(traverse_obj(_TEST_DATA, [({0: '...'}, {0: '...'})], expected_type=type(...)),
2260 [{0: ...}, {0: ...}], msg='expected_type should transform branched dict values')
2261 self.assertEqual(traverse_obj({1: {3: 4}}, [(1, 2), 3], expected_type=int),
2262 [4], msg='expected_type regression for type matching in tuple branching')
2263 self.assertEqual(traverse_obj(_TEST_DATA, ['data', ...], expected_type=int),
2264 [], msg='expected_type regression for type matching in dict result')
2266 # Test get_all behavior
2267 _GET_ALL_DATA = {'key': [0, 1, 2]}
2268 self.assertEqual(traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False), 0,
2269 msg='if not `get_all`, return only first matching value')
2270 self.assertEqual(traverse_obj(_GET_ALL_DATA, ..., get_all=False), [0, 1, 2],
2271 msg='do not overflatten if not `get_all`')
2273 # Test casesense behavior
2274 _CASESENSE_DATA = {
2275 'KeY': 'value0',
2276 0: {
2277 'KeY': 'value1',
2278 0: {'KeY': 'value2'},
2281 self.assertEqual(traverse_obj(_CASESENSE_DATA, 'key'), None,
2282 msg='dict keys should be case sensitive unless `casesense`')
2283 self.assertEqual(traverse_obj(_CASESENSE_DATA, 'keY',
2284 casesense=False), 'value0',
2285 msg='allow non matching key case if `casesense`')
2286 self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ('keY',)),
2287 casesense=False), ['value1'],
2288 msg='allow non matching key case in branch if `casesense`')
2289 self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ((0, 'keY'),)),
2290 casesense=False), ['value2'],
2291 msg='allow non matching key case in branch path if `casesense`')
2293 # Test traverse_string behavior
2294 _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2}
2295 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)), None,
2296 msg='do not traverse into string if not `traverse_string`')
2297 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0),
2298 traverse_string=True), 's',
2299 msg='traverse into string if `traverse_string`')
2300 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1),
2301 traverse_string=True), '.',
2302 msg='traverse into converted data if `traverse_string`')
2303 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...),
2304 traverse_string=True), 'str',
2305 msg='`...` should result in string (same value) if `traverse_string`')
2306 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', slice(0, None, 2)),
2307 traverse_string=True), 'sr',
2308 msg='`slice` should result in string if `traverse_string`')
2309 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda i, v: i or v == "s"),
2310 traverse_string=True), 'str',
2311 msg='function should result in string if `traverse_string`')
2312 self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)),
2313 traverse_string=True), ['s', 'r'],
2314 msg='branching should result in list if `traverse_string`')
2315 self.assertEqual(traverse_obj({}, (0, ...), traverse_string=True), [],
2316 msg='branching should result in list if `traverse_string`')
2317 self.assertEqual(traverse_obj({}, (0, lambda x, y: True), traverse_string=True), [],
2318 msg='branching should result in list if `traverse_string`')
2319 self.assertEqual(traverse_obj({}, (0, slice(1)), traverse_string=True), [],
2320 msg='branching should result in list if `traverse_string`')
2322 # Test re.Match as input obj
2323 mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123')
2324 self.assertEqual(traverse_obj(mobj, ...), [x for x in mobj.groups() if x is not None],
2325 msg='`...` on a `re.Match` should give its `groups()`')
2326 self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 2)), ['0123', '3'],
2327 msg='function on a `re.Match` should give groupno, value starting at 0')
2328 self.assertEqual(traverse_obj(mobj, 'group'), '3',
2329 msg='str key on a `re.Match` should give group with that name')
2330 self.assertEqual(traverse_obj(mobj, 2), '3',
2331 msg='int key on a `re.Match` should give group with that name')
2332 self.assertEqual(traverse_obj(mobj, 'gRoUp', casesense=False), '3',
2333 msg='str key on a `re.Match` should respect casesense')
2334 self.assertEqual(traverse_obj(mobj, 'fail'), None,
2335 msg='failing str key on a `re.Match` should return `default`')
2336 self.assertEqual(traverse_obj(mobj, 'gRoUpS', casesense=False), None,
2337 msg='failing str key on a `re.Match` should return `default`')
2338 self.assertEqual(traverse_obj(mobj, 8), None,
2339 msg='failing int key on a `re.Match` should return `default`')
2340 self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 'group')), ['0123', '3'],
2341 msg='function on a `re.Match` should give group name as well')
2343 def test_http_header_dict(self):
2344 headers = HTTPHeaderDict()
2345 headers['ytdl-test'] = b'0'
2346 self.assertEqual(list(headers.items()), [('Ytdl-Test', '0')])
2347 headers['ytdl-test'] = 1
2348 self.assertEqual(list(headers.items()), [('Ytdl-Test', '1')])
2349 headers['Ytdl-test'] = '2'
2350 self.assertEqual(list(headers.items()), [('Ytdl-Test', '2')])
2351 self.assertTrue('ytDl-Test' in headers)
2352 self.assertEqual(str(headers), str(dict(headers)))
2353 self.assertEqual(repr(headers), str(dict(headers)))
2355 headers.update({'X-dlp': 'data'})
2356 self.assertEqual(set(headers.items()), {('Ytdl-Test', '2'), ('X-Dlp', 'data')})
2357 self.assertEqual(dict(headers), {'Ytdl-Test': '2', 'X-Dlp': 'data'})
2358 self.assertEqual(len(headers), 2)
2359 self.assertEqual(headers.copy(), headers)
2360 headers2 = HTTPHeaderDict({'X-dlp': 'data3'}, **headers, **{'X-dlp': 'data2'})
2361 self.assertEqual(set(headers2.items()), {('Ytdl-Test', '2'), ('X-Dlp', 'data2')})
2362 self.assertEqual(len(headers2), 2)
2363 headers2.clear()
2364 self.assertEqual(len(headers2), 0)
2366 # ensure we prefer latter headers
2367 headers3 = HTTPHeaderDict({'Ytdl-TeSt': 1}, {'Ytdl-test': 2})
2368 self.assertEqual(set(headers3.items()), {('Ytdl-Test', '2')})
2369 del headers3['ytdl-tesT']
2370 self.assertEqual(dict(headers3), {})
2372 headers4 = HTTPHeaderDict({'ytdl-test': 'data;'})
2373 self.assertEqual(set(headers4.items()), {('Ytdl-Test', 'data;')})
2375 # common mistake: strip whitespace from values
2376 # https://github.com/yt-dlp/yt-dlp/issues/8729
2377 headers5 = HTTPHeaderDict({'ytdl-test': ' data; '})
2378 self.assertEqual(set(headers5.items()), {('Ytdl-Test', 'data;')})
2380 def test_extract_basic_auth(self):
2381 assert extract_basic_auth('http://:foo.bar') == ('http://:foo.bar', None)
2382 assert extract_basic_auth('http://foo.bar') == ('http://foo.bar', None)
2383 assert extract_basic_auth('http://@foo.bar') == ('http://foo.bar', 'Basic Og==')
2384 assert extract_basic_auth('http://:pass@foo.bar') == ('http://foo.bar', 'Basic OnBhc3M=')
2385 assert extract_basic_auth('http://user:@foo.bar') == ('http://foo.bar', 'Basic dXNlcjo=')
2386 assert extract_basic_auth('http://user:pass@foo.bar') == ('http://foo.bar', 'Basic dXNlcjpwYXNz')
2388 @unittest.skipUnless(compat_os_name == 'nt', 'Only relevant on Windows')
2389 def test_Popen_windows_escaping(self):
2390 def run_shell(args):
2391 stdout, stderr, error = Popen.run(
2392 args, text=True, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
2393 assert not stderr
2394 assert not error
2395 return stdout
2397 # Test escaping
2398 assert run_shell(['echo', 'test"&']) == '"test""&"\n'
2399 # Test if delayed expansion is disabled
2400 assert run_shell(['echo', '^!']) == '"^!"\n'
2401 assert run_shell('echo "^!"') == '"^!"\n'
2404 if __name__ == '__main__':
2405 unittest.main()