]> jfr.im git - yt-dlp.git/blame - test/test_utils.py
projects / yt-dlp.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[20min] Fix embeds extraction
[yt-dlp.git] / test / test_utils.py
CommitLineData
e387eb5a 1#!/usr/bin/env python
9d4660ca 2# coding: utf-8
e387eb5a 3
4e408e47
PH		 4from __future__ import unicode_literals
5
44a5f171
PH		 6# Allow direct execution
7import os
dae7c920 8import sys
44fb3454 9import unittest
44a5f171 10sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
44fb3454 11
44a5f171
PH		 12
13# Various small unit tests
62e609ab 14import io
fac55558 15import json
44a5f171 16import xml.etree.ElementTree
dae7c920 17
a921f407 18from youtube_dl.utils import (
05900629 19 age_restricted,
cae97f65 20 args_to_str,
5eb6bdce 21 encode_base_n,
e4bdb37e 22 clean_html,
eb9c3edd 23 date_from_str,
a921f407 24 DateRange,
cae97f65 25 detect_exe_version,
5035536e 26 determine_ext,
cbecc9b9 27 dict_get,
6b77d52b 28 encode_compat_str,
29eb5174 29 encodeFilename,
cae97f65
PH		 30 escape_rfc3986,
31 escape_url,
8bb56eee 32 extract_attributes,
5379a2d4 33 ExtractorError,
a921f407 34 find_xpath_attr,
5aafe895 35 fix_xml_ampersands,
84c237fb 36 get_element_by_class,
2af12ad9
TC		 37 get_element_by_attribute,
38 get_elements_by_class,
39 get_elements_by_attribute,
9c44d242 40 InAdvancePagedList,
cae97f65 41 intlist_to_bytes,
61ca9a80 42 is_html,
cae97f65
PH		 43 js_to_json,
44 limit_length,
6562d34a 45 mimetype2ext,
f6717dec 46 month_by_name,
0c265486 47 multipart_encode,
5bc880b9 48 ohdave_rsa_encrypt,
cae97f65
PH		 49 OnDemandPagedList,
50 orderedSet,
a8795327 51 parse_age_limit,
608d11f5 52 parse_duration,
cae97f65 53 parse_filesize,
fb47597b 54 parse_count,
cae97f65 55 parse_iso8601,
f48409c7 56 pkcs1pad,
62e609ab 57 read_batch_urls,
29eb5174 58 sanitize_filename,
a2aaf4db 59 sanitize_path,
51098426 60 expand_path,
a4bcaad7 61 prepend_extension,
b3ed15b7 62 replace_extension,
46bc9b7d
S		 63 remove_start,
64 remove_end,
31b2051e 65 remove_quotes,
a6a173c2 66 shell_quote,
29eb5174 67 smuggle_url,
f53c966a 68 str_to_int,
cae97f65 69 strip_jsonp,
29eb5174
PH		 70 timeconvert,
71 unescapeHTML,
72 unified_strdate,
46f59e89 73 unified_timestamp,
29eb5174 74 unsmuggle_url,
cae97f65 75 uppercase_escape,
0fe2ff78 76 lowercase_escape,
29eb5174 77 url_basename,
02dc0a36 78 base_url,
e34c3361 79 urljoin,
b74fa8cd 80 urlencode_postdata,
1143535d 81 urshift,
fb640d0a 82 update_url_query,
5f9b8394 83 version_tuple,
cae97f65 84 xpath_with_ns,
87f70ab3 85 xpath_element,
5379a2d4 86 xpath_text,
87f70ab3 87 xpath_attr,
cfb56d1a 88 render_table,
347de493 89 match_str,
bf6427d2
YCH		 90 parse_dfxp_time_expr,
91 dfxp2srt,
f7126449
S		 92 cli_option,
93 cli_valueless_option,
94 cli_bool_option,
4f3c5e06 95 parse_codecs,
a921f407 96)
36e6f62c 97from youtube_dl.compat import (
8bb56eee 98 compat_chr,
36e6f62c 99 compat_etree_fromstring,
51098426 100 compat_getenv,
dee2ff1d 101 compat_os_name,
51098426 102 compat_setenv,
fb640d0a 103 compat_urlparse,
104 compat_parse_qs,
36e6f62c 105)
44fb3454 106
627dcfff 107
44fb3454 108class TestUtil(unittest.TestCase):
59ae15a5
PH		 109 def test_timeconvert(self):
110 self.assertTrue(timeconvert('') is None)
111 self.assertTrue(timeconvert('bougrg') is None)
112
113 def test_sanitize_filename(self):
114 self.assertEqual(sanitize_filename('abc'), 'abc')
115 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
116
117 self.assertEqual(sanitize_filename('123'), '123')
118
119 self.assertEqual('abc_de', sanitize_filename('abc/de'))
120 self.assertFalse('/' in sanitize_filename('abc/de///'))
121
122 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
123 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
124 self.assertEqual('yes no', sanitize_filename('yes? no'))
125 self.assertEqual('this - that', sanitize_filename('this: that'))
126
127 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
4e408e47 128 aumlaut = 'ä'
59ae15a5 129 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
4e408e47 130 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
59ae15a5
PH		 131 self.assertEqual(sanitize_filename(tests), tests)
132
2aeb06d6
PH		 133 self.assertEqual(
134 sanitize_filename('New World record at 0:12:34'),
135 'New World record at 0_12_34')
a7440261 136
5a42414b
PH		 137 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
138 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
a7440261
PH		 139 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
140 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
2aeb06d6 141
59ae15a5
PH		 142 forbidden = '"\0\\/'
143 for fc in forbidden:
144 for fbc in forbidden:
145 self.assertTrue(fbc not in sanitize_filename(fc))
146
147 def test_sanitize_filename_restricted(self):
148 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
149 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
150
151 self.assertEqual(sanitize_filename('123', restricted=True), '123')
152
153 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
154 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
155
156 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
157 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
158 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
159 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
160
79a2e94e
AT		 161 tests = 'aäb\u4e2d\u56fd\u7684c'
162 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
4e408e47 163 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
59ae15a5 164
627dcfff 165 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
59ae15a5
PH		 166 for fc in forbidden:
167 for fbc in forbidden:
168 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
169
170 # Handle a common case more neatly
4e408e47
PH		 171 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
172 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
59ae15a5
PH		 173 # .. but make sure the file name is never empty
174 self.assertTrue(sanitize_filename('-', restricted=True) != '')
175 self.assertTrue(sanitize_filename(':', restricted=True) != '')
176
79a2e94e 177 self.assertEqual(sanitize_filename(
b96f007e 178 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
179 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
79a2e94e 180
796173d0 181 def test_sanitize_ids(self):
314d506b
PH		 182 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
183 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
184 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
796173d0 185
a2aaf4db
S		 186 def test_sanitize_path(self):
187 if sys.platform != 'win32':
188 return
189
190 self.assertEqual(sanitize_path('abc'), 'abc')
191 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
192 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
193 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
194 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
195 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
196 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
197
198 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
199 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
200
201 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
202 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
203 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
204 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
205
f18ef2d1
S		 206 self.assertEqual(
207 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
208 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
209
210 self.assertEqual(
211 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
212 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
213 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
214 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
215 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
216
2ebfeaca
S		 217 self.assertEqual(sanitize_path('../abc'), '..\\abc')
218 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
219 self.assertEqual(sanitize_path('./abc'), 'abc')
220 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
221
51098426 222 def test_expand_path(self):
41c5e60d
S		 223 def env(var):
224 return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
225
a426ef6d
S		 226 compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
227 self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
228 self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
51098426 229 self.assertEqual(expand_path('~'), compat_getenv('HOME'))
41c5e60d 230 self.assertEqual(
a426ef6d 231 expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
41c5e60d 232 '%s/expanded' % compat_getenv('HOME'))
51098426 233
a4bcaad7
S		 234 def test_prepend_extension(self):
235 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
236 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
237 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
238 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
239 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
240 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
241
b3ed15b7
S		 242 def test_replace_extension(self):
243 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
244 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
245 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
246 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
247 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
248 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
249
46bc9b7d
S		 250 def test_remove_start(self):
251 self.assertEqual(remove_start(None, 'A - '), None)
252 self.assertEqual(remove_start('A - B', 'A - '), 'B')
253 self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
254
255 def test_remove_end(self):
256 self.assertEqual(remove_end(None, ' - B'), None)
257 self.assertEqual(remove_end('A - B', ' - B'), 'A')
258 self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
259
31b2051e
S		 260 def test_remove_quotes(self):
261 self.assertEqual(remove_quotes(None), None)
262 self.assertEqual(remove_quotes('"'), '"')
263 self.assertEqual(remove_quotes("'"), "'")
264 self.assertEqual(remove_quotes(';'), ';')
265 self.assertEqual(remove_quotes('";'), '";')
266 self.assertEqual(remove_quotes('""'), '')
267 self.assertEqual(remove_quotes('";"'), ';')
268
59ae15a5 269 def test_ordered_set(self):
627dcfff 270 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
59ae15a5
PH		 271 self.assertEqual(orderedSet([]), [])
272 self.assertEqual(orderedSet([1]), [1])
5f6a1245 273 # keep the list ordered
627dcfff 274 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
59ae15a5
PH		 275
276 def test_unescape_html(self):
4e408e47 277 self.assertEqual(unescapeHTML('%20;'), '%20;')
91757b0f
NJ		 278 self.assertEqual(unescapeHTML('&#x2F;'), '/')
279 self.assertEqual(unescapeHTML('&#47;'), '/')
7aefc49c
S		 280 self.assertEqual(unescapeHTML('&eacute;'), 'é')
281 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
55b2f099
YCH		 282 # HTML5 entities
283 self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
5f6a1245 284
eb9c3edd
JMF		 285 def test_date_from_str(self):
286 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
287 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
288 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
289 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
290 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
291
bd558525 292 def test_daterange(self):
5f6a1245 293 _20century = DateRange("19000101", "20000101")
bd558525
JMF		 294 self.assertFalse("17890714" in _20century)
295 _ac = DateRange("00010101")
296 self.assertTrue("19690721" in _ac)
297 _firstmilenium = DateRange(end="10000101")
298 self.assertTrue("07110427" in _firstmilenium)
37254abc 299
bf50b038
JMF		 300 def test_unified_dates(self):
301 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
302 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
303 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
304 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
a69801e2 305 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
026fcc04 306 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
99b67fec 307 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
42bdd9d0
PH		 308 self.assertEqual(
309 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
310 '20141126')
9bb8e0a3
PH		 311 self.assertEqual(
312 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
313 '20150202')
f160785c 314 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
8cf70de4 315 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
46f59e89 316 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
6a750402 317 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
c6eed6b8 318 self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
cb655f34
S		 319 self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
320 self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
321 self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
dae7c920 322
46f59e89
S		 323 def test_unified_timestamps(self):
324 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
325 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
326 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
327 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
328 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
329 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
330 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
331 self.assertEqual(
332 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
333 1417001400)
334 self.assertEqual(
335 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
336 1422902860)
337 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
338 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
339 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
340 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
7dc2a74e 341 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
c6eed6b8 342 self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
deef3195 343 self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
2ae2ffda 344 self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
46f59e89 345
5035536e
S		 346 def test_determine_ext(self):
347 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
348 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
9cb9a5df
S		 349 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
350 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
351 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
5035536e 352
59ae56fa 353 def test_find_xpath_attr(self):
4e408e47 354 testxml = '''<root>
59ae56fa
PH		 355 <node/>
356 <node x="a"/>
357 <node x="a" y="c" />
358 <node x="b" y="d" />
ee114368 359 <node x="" />
59ae56fa 360 </root>'''
36e6f62c 361 doc = compat_etree_fromstring(testxml)
59ae56fa 362
ee114368 363 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
59ae56fa 364 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
ee114368
S		 365 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
366 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
367 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
59ae56fa 368 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
ee114368
S		 369 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
370 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
59ae56fa 371 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
ee114368
S		 372 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
373 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
59ae56fa 374
d7e66d39 375 def test_xpath_with_ns(self):
4e408e47 376 testxml = '''<root xmlns:media="http://example.com/">
d7e66d39
JMF		 377 <media:song>
378 <media:author>The Author</media:author>
379 <url>http://server.com/download.mp3</url>
380 </media:song>
381 </root>'''
36e6f62c 382 doc = compat_etree_fromstring(testxml)
d7e66d39
JMF		 383 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
384 self.assertTrue(find('media:song') is not None)
4e408e47
PH		 385 self.assertEqual(find('media:song/media:author').text, 'The Author')
386 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
d7e66d39 387
87f70ab3
S		 388 def test_xpath_element(self):
389 doc = xml.etree.ElementTree.Element('root')
390 div = xml.etree.ElementTree.SubElement(doc, 'div')
391 p = xml.etree.ElementTree.SubElement(div, 'p')
392 p.text = 'Foo'
393 self.assertEqual(xpath_element(doc, 'div/p'), p)
578c0745
S		 394 self.assertEqual(xpath_element(doc, ['div/p']), p)
395 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
87f70ab3 396 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
578c0745 397 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
87f70ab3 398 self.assertTrue(xpath_element(doc, 'div/bar') is None)
578c0745
S		 399 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
400 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
87f70ab3 401 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
578c0745
S		 402 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
403 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
87f70ab3 404
5379a2d4
JMF		 405 def test_xpath_text(self):
406 testxml = '''<root>
407 <div>
408 <p>Foo</p>
409 </div>
410 </root>'''
36e6f62c 411 doc = compat_etree_fromstring(testxml)
5379a2d4 412 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
87f70ab3 413 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
5379a2d4
JMF		 414 self.assertTrue(xpath_text(doc, 'div/bar') is None)
415 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
416
87f70ab3
S		 417 def test_xpath_attr(self):
418 testxml = '''<root>
419 <div>
420 <p x="a">Foo</p>
421 </div>
422 </root>'''
36e6f62c 423 doc = compat_etree_fromstring(testxml)
87f70ab3
S		 424 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
425 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
426 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
427 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
428 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
429 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
430 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
431
9d4660ca 432 def test_smuggle_url(self):
e075a44a 433 data = {"ö": "ö", "abc": [3]}
9d4660ca
PH		 434 url = 'https://foo.bar/baz?x=y#a'
435 smug_url = smuggle_url(url, data)
436 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
437 self.assertEqual(url, unsmug_url)
438 self.assertEqual(data, unsmug_data)
439
440 res_url, res_data = unsmuggle_url(url)
441 self.assertEqual(res_url, url)
442 self.assertEqual(res_data, None)
443
dfaa86b7
RA		 444 smug_url = smuggle_url(url, {'a': 'b'})
445 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
446 res_url, res_data = unsmuggle_url(smug_smug_url)
447 self.assertEqual(res_url, url)
448 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
449
a6a173c2 450 def test_shell_quote(self):
4e408e47 451 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
dee2ff1d
S		 452 self.assertEqual(
453 shell_quote(args),
454 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
a6a173c2 455
f53c966a
JMF		 456 def test_str_to_int(self):
457 self.assertEqual(str_to_int('123,456'), 123456)
458 self.assertEqual(str_to_int('123.456'), 123456)
459
29eb5174 460 def test_url_basename(self):
4e408e47
PH		 461 self.assertEqual(url_basename('http://foo.de/'), '')
462 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
463 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
464 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
465 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
d6c7a367 466 self.assertEqual(
4e408e47
PH		 467 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
468 'trailer.mp4')
9d4660ca 469
02dc0a36
S		 470 def test_base_url(self):
471 self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
472 self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
473 self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
474 self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
475 self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
476
e34c3361
S		 477 def test_urljoin(self):
478 self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
4b5de77b
S		 479 self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
480 self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
481 self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
b0c65c67 482 self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
e34c3361
S		 483 self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
484 self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
485 self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
486 self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
b0c65c67 487 self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
e34c3361 488 self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
b0c65c67 489 self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
e34c3361
S		 490 self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
491 self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
492 self.assertEqual(urljoin('http://foo.de/', None), None)
493 self.assertEqual(urljoin('http://foo.de/', ''), None)
494 self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
7fe15920 495 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
e34c3361 496
a8795327
S		 497 def test_parse_age_limit(self):
498 self.assertEqual(parse_age_limit(None), None)
499 self.assertEqual(parse_age_limit(False), None)
500 self.assertEqual(parse_age_limit('invalid'), None)
501 self.assertEqual(parse_age_limit(0), 0)
502 self.assertEqual(parse_age_limit(18), 18)
503 self.assertEqual(parse_age_limit(21), 21)
504 self.assertEqual(parse_age_limit(22), None)
505 self.assertEqual(parse_age_limit('18'), 18)
506 self.assertEqual(parse_age_limit('18+'), 18)
507 self.assertEqual(parse_age_limit('PG-13'), 13)
508 self.assertEqual(parse_age_limit('TV-14'), 14)
509 self.assertEqual(parse_age_limit('TV-MA'), 17)
510
608d11f5
PH		 511 def test_parse_duration(self):
512 self.assertEqual(parse_duration(None), None)
a5fb718c
S		 513 self.assertEqual(parse_duration(False), None)
514 self.assertEqual(parse_duration('invalid'), None)
608d11f5
PH		 515 self.assertEqual(parse_duration('1'), 1)
516 self.assertEqual(parse_duration('1337:12'), 80232)
517 self.assertEqual(parse_duration('9:12:43'), 33163)
2db806b4
S		 518 self.assertEqual(parse_duration('12:00'), 720)
519 self.assertEqual(parse_duration('00:01:01'), 61)
608d11f5 520 self.assertEqual(parse_duration('x:y'), None)
2db806b4 521 self.assertEqual(parse_duration('3h11m53s'), 11513)
ca7b3246
S		 522 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
523 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
524 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
2db806b4
S		 525 self.assertEqual(parse_duration('62m45s'), 3765)
526 self.assertEqual(parse_duration('6m59s'), 419)
527 self.assertEqual(parse_duration('49s'), 49)
528 self.assertEqual(parse_duration('0h0m0s'), 0)
529 self.assertEqual(parse_duration('0m0s'), 0)
530 self.assertEqual(parse_duration('0s'), 0)
7adcbe75 531 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
6a68bb57 532 self.assertEqual(parse_duration('T30M38S'), 1838)
e8df5cee
PH		 533 self.assertEqual(parse_duration('5 s'), 5)
534 self.assertEqual(parse_duration('3 min'), 180)
535 self.assertEqual(parse_duration('2.5 hours'), 9000)
8f4b58d7
PH		 536 self.assertEqual(parse_duration('02:03:04'), 7384)
537 self.assertEqual(parse_duration('01:02:03:04'), 93784)
3e675fab 538 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
9c29bc69 539 self.assertEqual(parse_duration('87 Min.'), 5220)
acaff495 540 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
15846398 541 self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
608d11f5 542
5aafe895
PH		 543 def test_fix_xml_ampersands(self):
544 self.assertEqual(
545 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
546 self.assertEqual(
547 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
548 '"&amp;x=y&amp;wrong;&amp;z=a')
549 self.assertEqual(
550 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
551 '&amp;&apos;&gt;&lt;&quot;')
552 self.assertEqual(
553 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
554 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
555
b7ab0590
PH		 556 def test_paged_list(self):
557 def testPL(size, pagesize, sliceargs, expected):
558 def get_page(pagenum):
559 firstid = pagenum * pagesize
560 upto = min(size, pagenum * pagesize + pagesize)
561 for i in range(firstid, upto):
562 yield i
563
9c44d242 564 pl = OnDemandPagedList(get_page, pagesize)
b7ab0590
PH		 565 got = pl.getslice(*sliceargs)
566 self.assertEqual(got, expected)
567
9c44d242
PH		 568 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
569 got = iapl.getslice(*sliceargs)
570 self.assertEqual(got, expected)
571
b7ab0590
PH		 572 testPL(5, 2, (), [0, 1, 2, 3, 4])
573 testPL(5, 2, (1,), [1, 2, 3, 4])
574 testPL(5, 2, (2,), [2, 3, 4])
575 testPL(5, 2, (4,), [4])
576 testPL(5, 2, (0, 3), [0, 1, 2])
577 testPL(5, 2, (1, 4), [1, 2, 3])
578 testPL(5, 2, (2, 99), [2, 3, 4])
579 testPL(5, 2, (20, 99), [])
580
62e609ab 581 def test_read_batch_urls(self):
4e408e47 582 f = io.StringIO('''\xef\xbb\xbf foo
62e609ab
PH		 583 bar\r
584 baz
585 # More after this line\r
586 ; or after this
587 bam''')
4e408e47 588 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
62e609ab 589
b74fa8cd
JMF		 590 def test_urlencode_postdata(self):
591 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
592 self.assertTrue(isinstance(data, bytes))
593
fb640d0a 594 def test_update_url_query(self):
595 def query_dict(url):
596 return compat_parse_qs(compat_urlparse.urlparse(url).query)
597 self.assertEqual(query_dict(update_url_query(
598 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
599 query_dict('http://example.com/path?quality=HD&format=mp4'))
600 self.assertEqual(query_dict(update_url_query(
601 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
602 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
603 self.assertEqual(query_dict(update_url_query(
604 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
605 query_dict('http://example.com/path?fields=id,formats,subtitles'))
606 self.assertEqual(query_dict(update_url_query(
607 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
608 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
609 self.assertEqual(query_dict(update_url_query(
610 'http://example.com/path?manifest=f4m', {'manifest': []})),
611 query_dict('http://example.com/path'))
612 self.assertEqual(query_dict(update_url_query(
613 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
614 query_dict('http://example.com/path?system=LINUX'))
615 self.assertEqual(query_dict(update_url_query(
616 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
617 query_dict('http://example.com/path?fields=id,formats,subtitles'))
3201a67f 618 self.assertEqual(query_dict(update_url_query(
619 'http://example.com/path', {'width': 1080, 'height': 720})),
620 query_dict('http://example.com/path?width=1080&height=720'))
621 self.assertEqual(query_dict(update_url_query(
622 'http://example.com/path', {'bitrate': 5020.43})),
623 query_dict('http://example.com/path?bitrate=5020.43'))
624 self.assertEqual(query_dict(update_url_query(
625 'http://example.com/path', {'test': '第二行тест'})),
626 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
fb640d0a 627
0c265486
YCH		 628 def test_multipart_encode(self):
629 self.assertEqual(
630 multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
631 b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
632 self.assertEqual(
633 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
634 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')
635 self.assertRaises(
636 ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
637
cbecc9b9 638 def test_dict_get(self):
86296ad2
S		 639 FALSE_VALUES = {
640 'none': None,
641 'false': False,
642 'zero': 0,
643 'empty_string': '',
644 'empty_list': [],
cbecc9b9 645 }
86296ad2
S		 646 d = FALSE_VALUES.copy()
647 d['a'] = 42
cbecc9b9
S		 648 self.assertEqual(dict_get(d, 'a'), 42)
649 self.assertEqual(dict_get(d, 'b'), None)
650 self.assertEqual(dict_get(d, 'b', 42), 42)
651 self.assertEqual(dict_get(d, ('a', )), 42)
652 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
653 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
654 self.assertEqual(dict_get(d, ('b', 'c', )), None)
655 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
86296ad2
S		 656 for key, false_value in FALSE_VALUES.items():
657 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
658 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
cbecc9b9 659
6b77d52b
S		 660 def test_encode_compat_str(self):
661 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
662 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
663
912b38b4
PH		 664 def test_parse_iso8601(self):
665 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
666 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
667 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
6ad4013d 668 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
52c3a6e4
S		 669 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
670 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
912b38b4 671
fac55558
PH		 672 def test_strip_jsonp(self):
673 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
674 d = json.loads(stripped)
675 self.assertEqual(d, [{"id": "532cb", "x": 3}])
676
609a61e3
PH		 677 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
678 d = json.loads(stripped)
679 self.assertEqual(d, {'STATUS': 'OK'})
680
8411229b
S		 681 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
682 d = json.loads(stripped)
683 self.assertEqual(d, {'status': 'success'})
684
5552c9eb
YCH		 685 stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
686 d = json.loads(stripped)
687 self.assertEqual(d, {'status': 'success'})
688
689 stripped = strip_jsonp('window.cb && cb({"status": "success"});')
690 d = json.loads(stripped)
691 self.assertEqual(d, {'status': 'success'})
692
173a7026 693 def test_uppercase_escape(self):
4e408e47
PH		 694 self.assertEqual(uppercase_escape('aä'), 'aä')
695 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
fac55558 696
0fe2ff78
YCH		 697 def test_lowercase_escape(self):
698 self.assertEqual(lowercase_escape('aä'), 'aä')
699 self.assertEqual(lowercase_escape('\\u0026'), '&')
700
a020a0dc
PH		 701 def test_limit_length(self):
702 self.assertEqual(limit_length(None, 12), None)
703 self.assertEqual(limit_length('foo', 12), 'foo')
704 self.assertTrue(
705 limit_length('foo bar baz asd', 12).startswith('foo bar'))
706 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
707
6562d34a
S		 708 def test_mimetype2ext(self):
709 self.assertEqual(mimetype2ext(None), None)
710 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
711 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
712 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
713 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
714 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
715
f6717dec
S		 716 def test_month_by_name(self):
717 self.assertEqual(month_by_name(None), None)
718 self.assertEqual(month_by_name('December', 'en'), 12)
3e4185c3 719 self.assertEqual(month_by_name('décembre', 'fr'), 12)
f6717dec 720 self.assertEqual(month_by_name('December'), 12)
3e4185c3 721 self.assertEqual(month_by_name('décembre'), None)
f6717dec
S		 722 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
723
4f3c5e06 724 def test_parse_codecs(self):
725 self.assertEqual(parse_codecs(''), {})
726 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
727 'vcodec': 'avc1.77.30',
728 'acodec': 'mp4a.40.2',
729 })
730 self.assertEqual(parse_codecs('mp4a.40.2'), {
731 'vcodec': 'none',
732 'acodec': 'mp4a.40.2',
733 })
734 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
735 'vcodec': 'avc1.42001e',
736 'acodec': 'mp4a.40.5',
737 })
738 self.assertEqual(parse_codecs('avc3.640028'), {
739 'vcodec': 'avc3.640028',
740 'acodec': 'none',
741 })
742 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
743 'vcodec': 'h264',
744 'acodec': 'aac',
745 })
746
d05cfe06
S		 747 def test_escape_rfc3986(self):
748 reserved = "!*'();:@&=+$,/?#[]"
749 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
750 self.assertEqual(escape_rfc3986(reserved), reserved)
751 self.assertEqual(escape_rfc3986(unreserved), unreserved)
752 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
753 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
754 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
755 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
756
757 def test_escape_url(self):
758 self.assertEqual(
759 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
760 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
761 )
762 self.assertEqual(
763 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
764 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
765 )
766 self.assertEqual(
767 escape_url('http://тест.рф/фрагмент'),
2d60465e 768 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
d05cfe06
S		 769 )
770 self.assertEqual(
771 escape_url('http://тест.рф/абв?абв=абв#абв'),
81f36eba 772 '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'
d05cfe06
S		 773 )
774 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
775
e7b6d122 776 def test_js_to_json_realworld(self):
410f3e73 777 inp = '''{
e7b6d122 778 'clip':{'provider':'pseudo'}
410f3e73
PH		 779 }'''
780 self.assertEqual(js_to_json(inp), '''{
e7b6d122 781 "clip":{"provider":"pseudo"}
410f3e73
PH		 782 }''')
783 json.loads(js_to_json(inp))
784
e7b6d122
PH		 785 inp = '''{
786 'playlist':[{'controls':{'all':null}}]
787 }'''
788 self.assertEqual(js_to_json(inp), '''{
789 "playlist":[{"controls":{"all":null}}]
790 }''')
791
d01949dc
S		 792 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
793 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
794
d305dd73
PH		 795 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
796 json_code = js_to_json(inp)
797 self.assertEqual(json.loads(json_code), json.loads(inp))
798
89ac4a19
S		 799 inp = '''{
800 0:{src:'skipped', type: 'application/dash+xml'},
801 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
802 }'''
803 self.assertEqual(js_to_json(inp), '''{
804 "0":{"src":"skipped", "type": "application/dash+xml"},
805 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
806 }''')
807
47212f7b
YCH		 808 inp = '''{"foo":101}'''
809 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
810
e4659b45
YCH		 811 inp = '''{"duration": "00:01:07"}'''
812 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
813
e7b6d122
PH		 814 def test_js_to_json_edgecases(self):
815 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
816 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
817
818 on = js_to_json('{"abc": true}')
819 self.assertEqual(json.loads(on), {'abc': True})
820
8f4b58d7
PH		 821 # Ignore JavaScript code as well
822 on = js_to_json('''{
823 "x": 1,
824 y: "a",
825 z: some.code
826 }''')
827 d = json.loads(on)
828 self.assertEqual(d['x'], 1)
829 self.assertEqual(d['y'], 'a')
830
ba9e68f4
S		 831 on = js_to_json('["abc", "def",]')
832 self.assertEqual(json.loads(on), ['abc', 'def'])
833
4195096e
S		 834 on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
835 self.assertEqual(json.loads(on), ['abc', 'def'])
836
837 on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
838 self.assertEqual(json.loads(on), ['abc', 'def'])
839
ba9e68f4
S		 840 on = js_to_json('{"abc": "def",}')
841 self.assertEqual(json.loads(on), {'abc': 'def'})
842
4195096e
S		 843 on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
844 self.assertEqual(json.loads(on), {'abc': 'def'})
845
bd1e4844 846 on = js_to_json('{ 0: /* " \n */ ",]" , }')
847 self.assertEqual(json.loads(on), {'0': ',]'})
848
4195096e
S		 849 on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
850 self.assertEqual(json.loads(on), {'0': ',]'})
851
b3ee552e
MČ		 852 on = js_to_json('{ 0: // comment\n1 }')
853 self.assertEqual(json.loads(on), {'0': 1})
854
bd1e4844 855 on = js_to_json(r'["<p>x<\/p>"]')
856 self.assertEqual(json.loads(on), ['<p>x</p>'])
857
858 on = js_to_json(r'["\xaa"]')
859 self.assertEqual(json.loads(on), ['\u00aa'])
860
861 on = js_to_json("['a\\\nb']")
862 self.assertEqual(json.loads(on), ['ab'])
863
4195096e
S		 864 on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
865 self.assertEqual(json.loads(on), ['ab'])
866
89ac4a19
S		 867 on = js_to_json('{0xff:0xff}')
868 self.assertEqual(json.loads(on), {'255': 255})
869
4195096e
S		 870 on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
871 self.assertEqual(json.loads(on), {'255': 255})
872
89ac4a19
S		 873 on = js_to_json('{077:077}')
874 self.assertEqual(json.loads(on), {'63': 63})
875
4195096e
S		 876 on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
877 self.assertEqual(json.loads(on), {'63': 63})
878
89ac4a19
S		 879 on = js_to_json('{42:42}')
880 self.assertEqual(json.loads(on), {'42': 42})
881
4195096e
S		 882 on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
883 self.assertEqual(json.loads(on), {'42': 42})
884
8bb56eee
BF		 885 def test_extract_attributes(self):
886 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
887 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
888 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
889 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
890 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
891 self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
892 self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
893 self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'}) # XML
894 self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
c5229f39
S		 895 self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
896 self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
8bb56eee
BF		 897 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
898 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
899 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
900 self.assertEqual(extract_attributes('<e x >'), {'x': None})
901 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
902 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
903 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
904 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
905 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
906 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
907 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
c5229f39 908 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
8bb56eee
BF		 909 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
910 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
911 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
912 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
913 self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
914 # "Narrow" Python builds don't support unicode code points outside BMP.
915 try:
916 compat_chr(0x10000)
917 supports_outside_bmp = True
918 except ValueError:
919 supports_outside_bmp = False
920 if supports_outside_bmp:
921 self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
b4a3d461
S		 922 # Malformed HTML should not break attributes extraction on older Python
923 self.assertEqual(extract_attributes('<mal"formed/>'), {})
8bb56eee 924
e4bdb37e
PH		 925 def test_clean_html(self):
926 self.assertEqual(clean_html('a:\nb'), 'a: b')
927 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
9222d945 928 self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
e4bdb37e 929
4c0924bb
PH		 930 def test_intlist_to_bytes(self):
931 self.assertEqual(
932 intlist_to_bytes([0, 1, 127, 128, 255]),
933 b'\x00\x01\x7f\x80\xff')
934
7d4111ed
PH		 935 def test_args_to_str(self):
936 self.assertEqual(
937 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
dee2ff1d 938 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
7d4111ed
PH		 939 )
940
be64b5b0
PH		 941 def test_parse_filesize(self):
942 self.assertEqual(parse_filesize(None), None)
943 self.assertEqual(parse_filesize(''), None)
944 self.assertEqual(parse_filesize('91 B'), 91)
945 self.assertEqual(parse_filesize('foobar'), None)
946 self.assertEqual(parse_filesize('2 MiB'), 2097152)
947 self.assertEqual(parse_filesize('5 GB'), 5000000000)
948 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
13585d76 949 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
4349c07d 950 self.assertEqual(parse_filesize('1,24 KB'), 1240)
13585d76 951 self.assertEqual(parse_filesize('1,24 kb'), 1240)
70852b47 952 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
be64b5b0 953
fb47597b
S		 954 def test_parse_count(self):
955 self.assertEqual(parse_count(None), None)
956 self.assertEqual(parse_count(''), None)
957 self.assertEqual(parse_count('0'), 0)
958 self.assertEqual(parse_count('1000'), 1000)
959 self.assertEqual(parse_count('1.000'), 1000)
960 self.assertEqual(parse_count('1.1k'), 1100)
961 self.assertEqual(parse_count('1.1kk'), 1100000)
782b1b5b
JMF		 962 self.assertEqual(parse_count('1.1kk '), 1100000)
963 self.assertEqual(parse_count('1.1kk views'), 1100000)
fb47597b 964
5f9b8394
PH		 965 def test_version_tuple(self):
966 self.assertEqual(version_tuple('1'), (1,))
967 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
47d7c642 968 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
5f9b8394 969
cae97f65
PH		 970 def test_detect_exe_version(self):
971 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
972built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
973configuration: --prefix=/usr --extra-'''), '1.2.1')
974 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
975built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
976 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
977Trying to open render node...
978Success at /dev/dri/renderD128.
979ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
980
05900629
PH		 981 def test_age_restricted(self):
982 self.assertFalse(age_restricted(None, 10)) # unrestricted content
983 self.assertFalse(age_restricted(1, None)) # unrestricted policy
984 self.assertFalse(age_restricted(8, 10))
985 self.assertTrue(age_restricted(18, 14))
986 self.assertFalse(age_restricted(18, 18))
987
61ca9a80
PH		 988 def test_is_html(self):
989 self.assertFalse(is_html(b'\x49\x44\x43<html'))
990 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
991 self.assertTrue(is_html( # UTF-8 with BOM
992 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
993 self.assertTrue(is_html( # UTF-16-LE
994 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
995 ))
996 self.assertTrue(is_html( # UTF-16-BE
997 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
998 ))
999 self.assertTrue(is_html( # UTF-32-BE
1000 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'))
1001 self.assertTrue(is_html( # UTF-32-LE
1002 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'))
1003
cfb56d1a
PH		 1004 def test_render_table(self):
1005 self.assertEqual(
1006 render_table(
1007 ['a', 'bcd'],
1008 [[123, 4], [9999, 51]]),
1009 'a bcd\n'
1010 '123 4\n'
1011 '9999 51')
1012
347de493
PH		 1013 def test_match_str(self):
1014 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1015 self.assertFalse(match_str('xy', {'x': 1200}))
1016 self.assertTrue(match_str('!xy', {'x': 1200}))
1017 self.assertTrue(match_str('x', {'x': 1200}))
1018 self.assertFalse(match_str('!x', {'x': 1200}))
1019 self.assertTrue(match_str('x', {'x': 0}))
1020 self.assertFalse(match_str('x>0', {'x': 0}))
1021 self.assertFalse(match_str('x>0', {}))
1022 self.assertTrue(match_str('x>?0', {}))
1023 self.assertTrue(match_str('x>1K', {'x': 1200}))
1024 self.assertFalse(match_str('x>2K', {'x': 1200}))
1025 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1026 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1027 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1028 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1029 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1030 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1031 self.assertFalse(match_str(
1032 'like_count > 100 & dislike_count <? 50 & description',
1033 {'like_count': 90, 'description': 'foo'}))
1034 self.assertTrue(match_str(
1035 'like_count > 100 & dislike_count <? 50 & description',
1036 {'like_count': 190, 'description': 'foo'}))
1037 self.assertFalse(match_str(
1038 'like_count > 100 & dislike_count <? 50 & description',
1039 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1040 self.assertFalse(match_str(
1041 'like_count > 100 & dislike_count <? 50 & description',
1042 {'like_count': 190, 'dislike_count': 10}))
1043
bf6427d2 1044 def test_parse_dfxp_time_expr(self):
d631d5f9
YCH		 1045 self.assertEqual(parse_dfxp_time_expr(None), None)
1046 self.assertEqual(parse_dfxp_time_expr(''), None)
bf6427d2
YCH		 1047 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1048 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1049 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1050 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
db2fe38b 1051 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
bf6427d2
YCH		 1052
1053 def test_dfxp2srt(self):
1054 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1055 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1056 <body>
1057 <div xml:lang="en">
1058 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1059 <p begin="1" end="2">第二行<br/>♪♪</p>
7dff0363 1060 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
d631d5f9
YCH		 1061 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1062 <p begin="-1" end="-1">Ignore, two</p>
1063 <p begin="3" dur="-1">Ignored, three</p>
bf6427d2
YCH		 1064 </div>
1065 </body>
1066 </tt>'''
1067 srt_data = '''1
106800:00:00,000 --> 00:00:01,000
1069The following line contains Chinese characters and special symbols
1070
10712
107200:00:01,000 --> 00:00:02,000
1073第二行
1074♪♪
1075
10763
107700:00:02,000 --> 00:00:03,000
1078Third
1079Line
1080
1081'''
1082 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1083
1b0427e6
YCH		 1084 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1085 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1086 <body>
1087 <div xml:lang="en">
1088 <p begin="0" end="1">The first line</p>
1089 </div>
1090 </body>
1091 </tt>'''
1092 srt_data = '''1
109300:00:00,000 --> 00:00:01,000
1094The first line
1095
1096'''
1097 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1098
5b995f71
RA		 1099 dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1100<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">
1101 <head>
1102 <styling>
1103 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1104 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1105 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1106 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1107 </styling>
1108 </head>
1109 <body tts:textAlign="center" style="s0">
1110 <div>
1111 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1112 <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>
1113 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1114 <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>
1115 </div>
1116 </body>
1117</tt>'''
1118 srt_data = '''1
111900:00:02,080 --> 00:00:05,839
1120<font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1121
11222
112300:00:02,080 --> 00:00:05,839
1124<b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1125</font>part 2</font></b>
1126
11273
112800:00:05,839 --> 00:00:09,560
1129<u><font color="lime">line 3
1130part 3</font></u>
1131
11324
113300:00:09,560 --> 00:00:12,359
1134<i><u><font color="yellow"><font color="lime">inner
1135 </font>style</font></u></i>
1136
1137'''
1138 self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1139
f7126449
S		 1140 def test_cli_option(self):
1141 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1142 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1143 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
5f2c2b79 1144 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
f7126449
S		 1145
1146 def test_cli_valueless_option(self):
1147 self.assertEqual(cli_valueless_option(
1148 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1149 self.assertEqual(cli_valueless_option(
1150 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1151 self.assertEqual(cli_valueless_option(
1152 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1153 self.assertEqual(cli_valueless_option(
1154 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1155 self.assertEqual(cli_valueless_option(
1156 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1157 self.assertEqual(cli_valueless_option(
1158 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1159
1160 def test_cli_bool_option(self):
1161 self.assertEqual(
1162 cli_bool_option(
1163 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1164 ['--no-check-certificate', 'true'])
1165 self.assertEqual(
1166 cli_bool_option(
1167 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1168 ['--no-check-certificate=true'])
1169 self.assertEqual(
1170 cli_bool_option(
1171 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1172 ['--check-certificate', 'false'])
1173 self.assertEqual(
1174 cli_bool_option(
1175 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1176 ['--check-certificate=false'])
1177 self.assertEqual(
1178 cli_bool_option(
1179 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1180 ['--check-certificate', 'true'])
1181 self.assertEqual(
1182 cli_bool_option(
1183 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1184 ['--check-certificate=true'])
1185
5bc880b9
YCH		 1186 def test_ohdave_rsa_encrypt(self):
1187 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1188 e = 65537
1189
1190 self.assertEqual(
1191 ohdave_rsa_encrypt(b'aa111222', e, N),
1192 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
cfb56d1a 1193
f48409c7
YCH		 1194 def test_pkcs1pad(self):
1195 data = [1, 2, 3]
1196 padded_data = pkcs1pad(data, 32)
1197 self.assertEqual(padded_data[:2], [0, 2])
1198 self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1199
1200 self.assertRaises(ValueError, pkcs1pad, data, 8)
1201
5eb6bdce
YCH		 1202 def test_encode_base_n(self):
1203 self.assertEqual(encode_base_n(0, 30), '0')
1204 self.assertEqual(encode_base_n(80, 30), '2k')
1205
1206 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1207 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1208 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1209
1210 self.assertRaises(ValueError, encode_base_n, 0, 70)
1211 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1212
1143535d
YCH		 1213 def test_urshift(self):
1214 self.assertEqual(urshift(3, 1), 1)
1215 self.assertEqual(urshift(-3, 1), 2147483646)
1216
84c237fb
YCH		 1217 def test_get_element_by_class(self):
1218 html = '''
1219 <span class="foo bar">nice</span>
1220 '''
1221
1222 self.assertEqual(get_element_by_class('foo', html), 'nice')
1223 self.assertEqual(get_element_by_class('no-such-class', html), None)
1224
2af12ad9
TC		 1225 def test_get_element_by_attribute(self):
1226 html = '''
1227 <span class="foo bar">nice</span>
1228 '''
1229
1230 self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1231 self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1232 self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1233
609ff8ca
YCH		 1234 html = '''
1235 <div itemprop="author" itemscope>foo</div>
1236 '''
1237
1238 self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1239
2af12ad9
TC		 1240 def test_get_elements_by_class(self):
1241 html = '''
1242 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1243 '''
1244
1245 self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1246 self.assertEqual(get_elements_by_class('no-such-class', html), [])
1247
1248 def test_get_elements_by_attribute(self):
1249 html = '''
1250 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1251 '''
1252
1253 self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1254 self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1255 self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1256
582be358 1257
dae7c920 1258if __name__ == '__main__':
59ae15a5 1259 unittest.main()
fork of https://github.com/yt-dlp/yt-dlp/