]> jfr.im git - yt-dlp.git/blame - test/test_utils.py
[viewster->contv] remove viewster extractor and add support for contv.com
[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,
5e1271c5 36 float_or_none,
84c237fb 37 get_element_by_class,
2af12ad9
TC
38 get_element_by_attribute,
39 get_elements_by_class,
40 get_elements_by_attribute,
9c44d242 41 InAdvancePagedList,
5e1271c5 42 int_or_none,
cae97f65 43 intlist_to_bytes,
61ca9a80 44 is_html,
cae97f65
PH
45 js_to_json,
46 limit_length,
6cc62232 47 merge_dicts,
6562d34a 48 mimetype2ext,
f6717dec 49 month_by_name,
0c265486 50 multipart_encode,
5bc880b9 51 ohdave_rsa_encrypt,
cae97f65
PH
52 OnDemandPagedList,
53 orderedSet,
a8795327 54 parse_age_limit,
608d11f5 55 parse_duration,
cae97f65 56 parse_filesize,
fb47597b 57 parse_count,
cae97f65 58 parse_iso8601,
b871d7e9 59 parse_resolution,
0dc41787 60 parse_bitrate,
f48409c7 61 pkcs1pad,
62e609ab 62 read_batch_urls,
29eb5174 63 sanitize_filename,
a2aaf4db 64 sanitize_path,
befa4708 65 sanitize_url,
51098426 66 expand_path,
a4bcaad7 67 prepend_extension,
b3ed15b7 68 replace_extension,
46bc9b7d
S
69 remove_start,
70 remove_end,
31b2051e 71 remove_quotes,
a6a173c2 72 shell_quote,
29eb5174 73 smuggle_url,
f53c966a 74 str_to_int,
cae97f65 75 strip_jsonp,
53cd37ba 76 strip_or_none,
29eb5174
PH
77 timeconvert,
78 unescapeHTML,
79 unified_strdate,
46f59e89 80 unified_timestamp,
29eb5174 81 unsmuggle_url,
cae97f65 82 uppercase_escape,
0fe2ff78 83 lowercase_escape,
29eb5174 84 url_basename,
af03000a 85 url_or_none,
02dc0a36 86 base_url,
e34c3361 87 urljoin,
b74fa8cd 88 urlencode_postdata,
1143535d 89 urshift,
fb640d0a 90 update_url_query,
5f9b8394 91 version_tuple,
cae97f65 92 xpath_with_ns,
87f70ab3 93 xpath_element,
5379a2d4 94 xpath_text,
87f70ab3 95 xpath_attr,
cfb56d1a 96 render_table,
347de493 97 match_str,
bf6427d2
YCH
98 parse_dfxp_time_expr,
99 dfxp2srt,
f7126449
S
100 cli_option,
101 cli_valueless_option,
102 cli_bool_option,
4f3c5e06 103 parse_codecs,
a921f407 104)
36e6f62c 105from youtube_dl.compat import (
8bb56eee 106 compat_chr,
36e6f62c 107 compat_etree_fromstring,
51098426 108 compat_getenv,
dee2ff1d 109 compat_os_name,
51098426 110 compat_setenv,
fb640d0a 111 compat_urlparse,
112 compat_parse_qs,
36e6f62c 113)
44fb3454 114
627dcfff 115
44fb3454 116class TestUtil(unittest.TestCase):
59ae15a5
PH
117 def test_timeconvert(self):
118 self.assertTrue(timeconvert('') is None)
119 self.assertTrue(timeconvert('bougrg') is None)
120
121 def test_sanitize_filename(self):
122 self.assertEqual(sanitize_filename('abc'), 'abc')
123 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
124
125 self.assertEqual(sanitize_filename('123'), '123')
126
127 self.assertEqual('abc_de', sanitize_filename('abc/de'))
128 self.assertFalse('/' in sanitize_filename('abc/de///'))
129
130 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
131 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
132 self.assertEqual('yes no', sanitize_filename('yes? no'))
133 self.assertEqual('this - that', sanitize_filename('this: that'))
134
135 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
4e408e47 136 aumlaut = 'ä'
59ae15a5 137 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
4e408e47 138 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
59ae15a5
PH
139 self.assertEqual(sanitize_filename(tests), tests)
140
2aeb06d6
PH
141 self.assertEqual(
142 sanitize_filename('New World record at 0:12:34'),
143 'New World record at 0_12_34')
a7440261 144
5a42414b
PH
145 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
146 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
a7440261
PH
147 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
148 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
2aeb06d6 149
59ae15a5
PH
150 forbidden = '"\0\\/'
151 for fc in forbidden:
152 for fbc in forbidden:
153 self.assertTrue(fbc not in sanitize_filename(fc))
154
155 def test_sanitize_filename_restricted(self):
156 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
157 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
158
159 self.assertEqual(sanitize_filename('123', restricted=True), '123')
160
161 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
162 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
163
164 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
165 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
166 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
167 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
168
79a2e94e
AT
169 tests = 'aäb\u4e2d\u56fd\u7684c'
170 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
4e408e47 171 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
59ae15a5 172
627dcfff 173 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
59ae15a5
PH
174 for fc in forbidden:
175 for fbc in forbidden:
176 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
177
178 # Handle a common case more neatly
4e408e47
PH
179 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
180 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
59ae15a5
PH
181 # .. but make sure the file name is never empty
182 self.assertTrue(sanitize_filename('-', restricted=True) != '')
183 self.assertTrue(sanitize_filename(':', restricted=True) != '')
184
79a2e94e 185 self.assertEqual(sanitize_filename(
b96f007e 186 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
fd35d8cd 187 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
79a2e94e 188
796173d0 189 def test_sanitize_ids(self):
314d506b
PH
190 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
191 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
192 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
796173d0 193
a2aaf4db
S
194 def test_sanitize_path(self):
195 if sys.platform != 'win32':
196 return
197
198 self.assertEqual(sanitize_path('abc'), 'abc')
199 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
200 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
201 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
202 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
203 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
204 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
205
206 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
207 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
208
209 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
210 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
211 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
212 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
213
f18ef2d1
S
214 self.assertEqual(
215 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
216 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
217
218 self.assertEqual(
219 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
220 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
221 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
222 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
223 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
224
2ebfeaca
S
225 self.assertEqual(sanitize_path('../abc'), '..\\abc')
226 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
227 self.assertEqual(sanitize_path('./abc'), 'abc')
228 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
229
befa4708
S
230 def test_sanitize_url(self):
231 self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
232 self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
233 self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
234 self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
235
51098426 236 def test_expand_path(self):
41c5e60d
S
237 def env(var):
238 return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
239
a426ef6d
S
240 compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
241 self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
242 self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
51098426 243 self.assertEqual(expand_path('~'), compat_getenv('HOME'))
41c5e60d 244 self.assertEqual(
a426ef6d 245 expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
41c5e60d 246 '%s/expanded' % compat_getenv('HOME'))
51098426 247
a4bcaad7
S
248 def test_prepend_extension(self):
249 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
250 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
251 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
252 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
253 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
254 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
255
b3ed15b7
S
256 def test_replace_extension(self):
257 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
258 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
259 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
260 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
261 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
262 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
263
46bc9b7d
S
264 def test_remove_start(self):
265 self.assertEqual(remove_start(None, 'A - '), None)
266 self.assertEqual(remove_start('A - B', 'A - '), 'B')
267 self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
268
269 def test_remove_end(self):
270 self.assertEqual(remove_end(None, ' - B'), None)
271 self.assertEqual(remove_end('A - B', ' - B'), 'A')
272 self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
273
31b2051e
S
274 def test_remove_quotes(self):
275 self.assertEqual(remove_quotes(None), None)
276 self.assertEqual(remove_quotes('"'), '"')
277 self.assertEqual(remove_quotes("'"), "'")
278 self.assertEqual(remove_quotes(';'), ';')
279 self.assertEqual(remove_quotes('";'), '";')
280 self.assertEqual(remove_quotes('""'), '')
281 self.assertEqual(remove_quotes('";"'), ';')
282
59ae15a5 283 def test_ordered_set(self):
627dcfff 284 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
59ae15a5
PH
285 self.assertEqual(orderedSet([]), [])
286 self.assertEqual(orderedSet([1]), [1])
5f6a1245 287 # keep the list ordered
627dcfff 288 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
59ae15a5
PH
289
290 def test_unescape_html(self):
4e408e47 291 self.assertEqual(unescapeHTML('%20;'), '%20;')
91757b0f
NJ
292 self.assertEqual(unescapeHTML('&#x2F;'), '/')
293 self.assertEqual(unescapeHTML('&#47;'), '/')
7aefc49c
S
294 self.assertEqual(unescapeHTML('&eacute;'), 'é')
295 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
95f3f7c2 296 self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
55b2f099
YCH
297 # HTML5 entities
298 self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
5f6a1245 299
eb9c3edd
JMF
300 def test_date_from_str(self):
301 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
302 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
303 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
304 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
305 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
306
bd558525 307 def test_daterange(self):
5f6a1245 308 _20century = DateRange("19000101", "20000101")
bd558525
JMF
309 self.assertFalse("17890714" in _20century)
310 _ac = DateRange("00010101")
311 self.assertTrue("19690721" in _ac)
312 _firstmilenium = DateRange(end="10000101")
313 self.assertTrue("07110427" in _firstmilenium)
37254abc 314
bf50b038
JMF
315 def test_unified_dates(self):
316 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
317 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
318 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
319 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
a69801e2 320 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
026fcc04 321 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
99b67fec 322 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
42bdd9d0
PH
323 self.assertEqual(
324 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
325 '20141126')
9bb8e0a3
PH
326 self.assertEqual(
327 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
328 '20150202')
f160785c 329 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
8cf70de4 330 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
46f59e89 331 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
6a750402 332 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
c6eed6b8 333 self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
cb655f34
S
334 self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
335 self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
336 self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
dae7c920 337
46f59e89
S
338 def test_unified_timestamps(self):
339 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
340 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
341 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
342 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
343 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
344 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
345 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
346 self.assertEqual(
347 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
348 1417001400)
349 self.assertEqual(
350 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
351 1422902860)
352 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
353 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
354 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
355 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
7dc2a74e 356 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
c6eed6b8 357 self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
deef3195 358 self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
2ae2ffda 359 self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
b555ae9b 360 self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
f226880c 361 self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
46f59e89 362
5035536e
S
363 def test_determine_ext(self):
364 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
365 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
9cb9a5df
S
366 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
367 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
368 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
85750f89 369 self.assertEqual(determine_ext('foobar', None), None)
5035536e 370
59ae56fa 371 def test_find_xpath_attr(self):
4e408e47 372 testxml = '''<root>
59ae56fa
PH
373 <node/>
374 <node x="a"/>
375 <node x="a" y="c" />
376 <node x="b" y="d" />
ee114368 377 <node x="" />
59ae56fa 378 </root>'''
36e6f62c 379 doc = compat_etree_fromstring(testxml)
59ae56fa 380
ee114368 381 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
59ae56fa 382 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
ee114368
S
383 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
384 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
385 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
59ae56fa 386 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
ee114368
S
387 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
388 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
59ae56fa 389 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
ee114368
S
390 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
391 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
59ae56fa 392
d7e66d39 393 def test_xpath_with_ns(self):
4e408e47 394 testxml = '''<root xmlns:media="http://example.com/">
d7e66d39
JMF
395 <media:song>
396 <media:author>The Author</media:author>
397 <url>http://server.com/download.mp3</url>
398 </media:song>
399 </root>'''
36e6f62c 400 doc = compat_etree_fromstring(testxml)
d7e66d39
JMF
401 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
402 self.assertTrue(find('media:song') is not None)
4e408e47
PH
403 self.assertEqual(find('media:song/media:author').text, 'The Author')
404 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
d7e66d39 405
87f70ab3
S
406 def test_xpath_element(self):
407 doc = xml.etree.ElementTree.Element('root')
408 div = xml.etree.ElementTree.SubElement(doc, 'div')
409 p = xml.etree.ElementTree.SubElement(div, 'p')
410 p.text = 'Foo'
411 self.assertEqual(xpath_element(doc, 'div/p'), p)
578c0745
S
412 self.assertEqual(xpath_element(doc, ['div/p']), p)
413 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
87f70ab3 414 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
578c0745 415 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
87f70ab3 416 self.assertTrue(xpath_element(doc, 'div/bar') is None)
578c0745
S
417 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
418 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
87f70ab3 419 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
578c0745
S
420 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
421 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
87f70ab3 422
5379a2d4
JMF
423 def test_xpath_text(self):
424 testxml = '''<root>
425 <div>
426 <p>Foo</p>
427 </div>
428 </root>'''
36e6f62c 429 doc = compat_etree_fromstring(testxml)
5379a2d4 430 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
87f70ab3 431 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
5379a2d4
JMF
432 self.assertTrue(xpath_text(doc, 'div/bar') is None)
433 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
434
87f70ab3
S
435 def test_xpath_attr(self):
436 testxml = '''<root>
437 <div>
438 <p x="a">Foo</p>
439 </div>
440 </root>'''
36e6f62c 441 doc = compat_etree_fromstring(testxml)
87f70ab3
S
442 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
443 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
444 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
445 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
446 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
447 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
448 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
449
9d4660ca 450 def test_smuggle_url(self):
e075a44a 451 data = {"ö": "ö", "abc": [3]}
9d4660ca
PH
452 url = 'https://foo.bar/baz?x=y#a'
453 smug_url = smuggle_url(url, data)
454 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
455 self.assertEqual(url, unsmug_url)
456 self.assertEqual(data, unsmug_data)
457
458 res_url, res_data = unsmuggle_url(url)
459 self.assertEqual(res_url, url)
460 self.assertEqual(res_data, None)
461
dfaa86b7
RA
462 smug_url = smuggle_url(url, {'a': 'b'})
463 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
464 res_url, res_data = unsmuggle_url(smug_smug_url)
465 self.assertEqual(res_url, url)
466 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
467
a6a173c2 468 def test_shell_quote(self):
4e408e47 469 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
dee2ff1d
S
470 self.assertEqual(
471 shell_quote(args),
472 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
a6a173c2 473
5e1271c5
S
474 def test_float_or_none(self):
475 self.assertEqual(float_or_none('42.42'), 42.42)
476 self.assertEqual(float_or_none('42'), 42.0)
477 self.assertEqual(float_or_none(''), None)
478 self.assertEqual(float_or_none(None), None)
479 self.assertEqual(float_or_none([]), None)
480 self.assertEqual(float_or_none(set()), None)
481
482 def test_int_or_none(self):
483 self.assertEqual(int_or_none('42'), 42)
484 self.assertEqual(int_or_none(''), None)
485 self.assertEqual(int_or_none(None), None)
486 self.assertEqual(int_or_none([]), None)
487 self.assertEqual(int_or_none(set()), None)
488
f53c966a
JMF
489 def test_str_to_int(self):
490 self.assertEqual(str_to_int('123,456'), 123456)
491 self.assertEqual(str_to_int('123.456'), 123456)
492
29eb5174 493 def test_url_basename(self):
4e408e47
PH
494 self.assertEqual(url_basename('http://foo.de/'), '')
495 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
496 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
497 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
498 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
d6c7a367 499 self.assertEqual(
4e408e47
PH
500 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
501 'trailer.mp4')
9d4660ca 502
02dc0a36
S
503 def test_base_url(self):
504 self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
505 self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
506 self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
507 self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
508 self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
509
e34c3361
S
510 def test_urljoin(self):
511 self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
4b5de77b
S
512 self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
513 self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
514 self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
b0c65c67 515 self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
e34c3361
S
516 self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
517 self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
518 self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
519 self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
b0c65c67 520 self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
e34c3361 521 self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
b0c65c67 522 self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
e34c3361
S
523 self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
524 self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
525 self.assertEqual(urljoin('http://foo.de/', None), None)
526 self.assertEqual(urljoin('http://foo.de/', ''), None)
527 self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
7fe15920 528 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
fad4ceb5
S
529 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
530 self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
e34c3361 531
af03000a
S
532 def test_url_or_none(self):
533 self.assertEqual(url_or_none(None), None)
534 self.assertEqual(url_or_none(''), None)
535 self.assertEqual(url_or_none('foo'), None)
536 self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
537 self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
538 self.assertEqual(url_or_none('http$://foo.de'), None)
539 self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
540 self.assertEqual(url_or_none('//foo.de'), '//foo.de')
541
a8795327
S
542 def test_parse_age_limit(self):
543 self.assertEqual(parse_age_limit(None), None)
544 self.assertEqual(parse_age_limit(False), None)
545 self.assertEqual(parse_age_limit('invalid'), None)
546 self.assertEqual(parse_age_limit(0), 0)
547 self.assertEqual(parse_age_limit(18), 18)
548 self.assertEqual(parse_age_limit(21), 21)
549 self.assertEqual(parse_age_limit(22), None)
550 self.assertEqual(parse_age_limit('18'), 18)
551 self.assertEqual(parse_age_limit('18+'), 18)
552 self.assertEqual(parse_age_limit('PG-13'), 13)
553 self.assertEqual(parse_age_limit('TV-14'), 14)
554 self.assertEqual(parse_age_limit('TV-MA'), 17)
3bb3ff38
RA
555 self.assertEqual(parse_age_limit('TV14'), 14)
556 self.assertEqual(parse_age_limit('TV_G'), 0)
a8795327 557
608d11f5
PH
558 def test_parse_duration(self):
559 self.assertEqual(parse_duration(None), None)
a5fb718c
S
560 self.assertEqual(parse_duration(False), None)
561 self.assertEqual(parse_duration('invalid'), None)
608d11f5
PH
562 self.assertEqual(parse_duration('1'), 1)
563 self.assertEqual(parse_duration('1337:12'), 80232)
564 self.assertEqual(parse_duration('9:12:43'), 33163)
2db806b4
S
565 self.assertEqual(parse_duration('12:00'), 720)
566 self.assertEqual(parse_duration('00:01:01'), 61)
608d11f5 567 self.assertEqual(parse_duration('x:y'), None)
2db806b4 568 self.assertEqual(parse_duration('3h11m53s'), 11513)
ca7b3246
S
569 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
570 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
571 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
2db806b4
S
572 self.assertEqual(parse_duration('62m45s'), 3765)
573 self.assertEqual(parse_duration('6m59s'), 419)
574 self.assertEqual(parse_duration('49s'), 49)
575 self.assertEqual(parse_duration('0h0m0s'), 0)
576 self.assertEqual(parse_duration('0m0s'), 0)
577 self.assertEqual(parse_duration('0s'), 0)
7adcbe75 578 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
6a68bb57 579 self.assertEqual(parse_duration('T30M38S'), 1838)
e8df5cee
PH
580 self.assertEqual(parse_duration('5 s'), 5)
581 self.assertEqual(parse_duration('3 min'), 180)
582 self.assertEqual(parse_duration('2.5 hours'), 9000)
8f4b58d7
PH
583 self.assertEqual(parse_duration('02:03:04'), 7384)
584 self.assertEqual(parse_duration('01:02:03:04'), 93784)
3e675fab 585 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
9c29bc69 586 self.assertEqual(parse_duration('87 Min.'), 5220)
acaff495 587 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
15846398 588 self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
056653bb 589 self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
608d11f5 590
5aafe895
PH
591 def test_fix_xml_ampersands(self):
592 self.assertEqual(
593 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
594 self.assertEqual(
595 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
596 '"&amp;x=y&amp;wrong;&amp;z=a')
597 self.assertEqual(
598 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
599 '&amp;&apos;&gt;&lt;&quot;')
600 self.assertEqual(
601 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
602 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
603
b7ab0590
PH
604 def test_paged_list(self):
605 def testPL(size, pagesize, sliceargs, expected):
606 def get_page(pagenum):
607 firstid = pagenum * pagesize
608 upto = min(size, pagenum * pagesize + pagesize)
609 for i in range(firstid, upto):
610 yield i
611
9c44d242 612 pl = OnDemandPagedList(get_page, pagesize)
b7ab0590
PH
613 got = pl.getslice(*sliceargs)
614 self.assertEqual(got, expected)
615
9c44d242
PH
616 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
617 got = iapl.getslice(*sliceargs)
618 self.assertEqual(got, expected)
619
b7ab0590
PH
620 testPL(5, 2, (), [0, 1, 2, 3, 4])
621 testPL(5, 2, (1,), [1, 2, 3, 4])
622 testPL(5, 2, (2,), [2, 3, 4])
623 testPL(5, 2, (4,), [4])
624 testPL(5, 2, (0, 3), [0, 1, 2])
625 testPL(5, 2, (1, 4), [1, 2, 3])
626 testPL(5, 2, (2, 99), [2, 3, 4])
627 testPL(5, 2, (20, 99), [])
628
62e609ab 629 def test_read_batch_urls(self):
4e408e47 630 f = io.StringIO('''\xef\xbb\xbf foo
62e609ab
PH
631 bar\r
632 baz
633 # More after this line\r
634 ; or after this
635 bam''')
4e408e47 636 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
62e609ab 637
b74fa8cd
JMF
638 def test_urlencode_postdata(self):
639 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
640 self.assertTrue(isinstance(data, bytes))
641
fb640d0a 642 def test_update_url_query(self):
643 def query_dict(url):
644 return compat_parse_qs(compat_urlparse.urlparse(url).query)
645 self.assertEqual(query_dict(update_url_query(
646 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
647 query_dict('http://example.com/path?quality=HD&format=mp4'))
648 self.assertEqual(query_dict(update_url_query(
649 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
650 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
651 self.assertEqual(query_dict(update_url_query(
652 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
653 query_dict('http://example.com/path?fields=id,formats,subtitles'))
654 self.assertEqual(query_dict(update_url_query(
655 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
656 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
657 self.assertEqual(query_dict(update_url_query(
658 'http://example.com/path?manifest=f4m', {'manifest': []})),
659 query_dict('http://example.com/path'))
660 self.assertEqual(query_dict(update_url_query(
661 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
662 query_dict('http://example.com/path?system=LINUX'))
663 self.assertEqual(query_dict(update_url_query(
664 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
665 query_dict('http://example.com/path?fields=id,formats,subtitles'))
3201a67f 666 self.assertEqual(query_dict(update_url_query(
667 'http://example.com/path', {'width': 1080, 'height': 720})),
668 query_dict('http://example.com/path?width=1080&height=720'))
669 self.assertEqual(query_dict(update_url_query(
670 'http://example.com/path', {'bitrate': 5020.43})),
671 query_dict('http://example.com/path?bitrate=5020.43'))
672 self.assertEqual(query_dict(update_url_query(
673 'http://example.com/path', {'test': '第二行тест'})),
674 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 675
0c265486
YCH
676 def test_multipart_encode(self):
677 self.assertEqual(
678 multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
679 b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
680 self.assertEqual(
681 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
682 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')
683 self.assertRaises(
684 ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
685
cbecc9b9 686 def test_dict_get(self):
86296ad2
S
687 FALSE_VALUES = {
688 'none': None,
689 'false': False,
690 'zero': 0,
691 'empty_string': '',
692 'empty_list': [],
cbecc9b9 693 }
86296ad2
S
694 d = FALSE_VALUES.copy()
695 d['a'] = 42
cbecc9b9
S
696 self.assertEqual(dict_get(d, 'a'), 42)
697 self.assertEqual(dict_get(d, 'b'), None)
698 self.assertEqual(dict_get(d, 'b', 42), 42)
699 self.assertEqual(dict_get(d, ('a', )), 42)
700 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
701 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
702 self.assertEqual(dict_get(d, ('b', 'c', )), None)
703 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
86296ad2
S
704 for key, false_value in FALSE_VALUES.items():
705 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
706 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
cbecc9b9 707
6cc62232
S
708 def test_merge_dicts(self):
709 self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
710 self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
711 self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
712 self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
713 self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
714 self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
715 self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
716 self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
717 self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
718
6b77d52b
S
719 def test_encode_compat_str(self):
720 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
721 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
722
912b38b4
PH
723 def test_parse_iso8601(self):
724 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
725 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
726 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
6ad4013d 727 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
52c3a6e4
S
728 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
729 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
912b38b4 730
fac55558
PH
731 def test_strip_jsonp(self):
732 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
733 d = json.loads(stripped)
734 self.assertEqual(d, [{"id": "532cb", "x": 3}])
735
609a61e3
PH
736 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
737 d = json.loads(stripped)
738 self.assertEqual(d, {'STATUS': 'OK'})
739
8411229b
S
740 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
741 d = json.loads(stripped)
742 self.assertEqual(d, {'status': 'success'})
743
5552c9eb
YCH
744 stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
745 d = json.loads(stripped)
746 self.assertEqual(d, {'status': 'success'})
747
748 stripped = strip_jsonp('window.cb && cb({"status": "success"});')
749 d = json.loads(stripped)
750 self.assertEqual(d, {'status': 'success'})
e9c671d5
S
751
752 stripped = strip_jsonp('({"status": "success"});')
753 d = json.loads(stripped)
754 self.assertEqual(d, {'status': 'success'})
5552c9eb 755
53cd37ba
S
756 def test_strip_or_none(self):
757 self.assertEqual(strip_or_none(' abc'), 'abc')
758 self.assertEqual(strip_or_none('abc '), 'abc')
759 self.assertEqual(strip_or_none(' abc '), 'abc')
760 self.assertEqual(strip_or_none('\tabc\t'), 'abc')
761 self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
762 self.assertEqual(strip_or_none('abc'), 'abc')
763 self.assertEqual(strip_or_none(''), '')
764 self.assertEqual(strip_or_none(None), None)
765 self.assertEqual(strip_or_none(42), None)
766 self.assertEqual(strip_or_none([]), None)
767
173a7026 768 def test_uppercase_escape(self):
4e408e47
PH
769 self.assertEqual(uppercase_escape('aä'), 'aä')
770 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
fac55558 771
0fe2ff78
YCH
772 def test_lowercase_escape(self):
773 self.assertEqual(lowercase_escape('aä'), 'aä')
774 self.assertEqual(lowercase_escape('\\u0026'), '&')
775
a020a0dc
PH
776 def test_limit_length(self):
777 self.assertEqual(limit_length(None, 12), None)
778 self.assertEqual(limit_length('foo', 12), 'foo')
779 self.assertTrue(
780 limit_length('foo bar baz asd', 12).startswith('foo bar'))
781 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
782
6562d34a
S
783 def test_mimetype2ext(self):
784 self.assertEqual(mimetype2ext(None), None)
785 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
786 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
787 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
788 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
789 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
790
f6717dec
S
791 def test_month_by_name(self):
792 self.assertEqual(month_by_name(None), None)
793 self.assertEqual(month_by_name('December', 'en'), 12)
3e4185c3 794 self.assertEqual(month_by_name('décembre', 'fr'), 12)
f6717dec 795 self.assertEqual(month_by_name('December'), 12)
3e4185c3 796 self.assertEqual(month_by_name('décembre'), None)
f6717dec
S
797 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
798
4f3c5e06 799 def test_parse_codecs(self):
800 self.assertEqual(parse_codecs(''), {})
801 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
802 'vcodec': 'avc1.77.30',
803 'acodec': 'mp4a.40.2',
804 })
805 self.assertEqual(parse_codecs('mp4a.40.2'), {
806 'vcodec': 'none',
807 'acodec': 'mp4a.40.2',
808 })
809 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
810 'vcodec': 'avc1.42001e',
811 'acodec': 'mp4a.40.5',
812 })
813 self.assertEqual(parse_codecs('avc3.640028'), {
814 'vcodec': 'avc3.640028',
815 'acodec': 'none',
816 })
817 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
818 'vcodec': 'h264',
819 'acodec': 'aac',
820 })
25d110be
S
821 self.assertEqual(parse_codecs('av01.0.05M.08'), {
822 'vcodec': 'av01.0.05M.08',
823 'acodec': 'none',
824 })
28cc2241
S
825 self.assertEqual(parse_codecs('theora, vorbis'), {
826 'vcodec': 'theora',
827 'acodec': 'vorbis',
828 })
829 self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
830 'vcodec': 'unknownvcodec',
831 'acodec': 'unknownacodec',
832 })
833 self.assertEqual(parse_codecs('unknown'), {})
4f3c5e06 834
d05cfe06
S
835 def test_escape_rfc3986(self):
836 reserved = "!*'();:@&=+$,/?#[]"
837 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
838 self.assertEqual(escape_rfc3986(reserved), reserved)
839 self.assertEqual(escape_rfc3986(unreserved), unreserved)
840 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
841 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
842 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
843 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
844
845 def test_escape_url(self):
846 self.assertEqual(
847 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
848 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
849 )
850 self.assertEqual(
851 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
852 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
853 )
854 self.assertEqual(
855 escape_url('http://тест.рф/фрагмент'),
2d60465e 856 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
d05cfe06
S
857 )
858 self.assertEqual(
859 escape_url('http://тест.рф/абв?абв=абв#абв'),
81f36eba 860 '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
861 )
862 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
863
e7b6d122 864 def test_js_to_json_realworld(self):
410f3e73 865 inp = '''{
e7b6d122 866 'clip':{'provider':'pseudo'}
410f3e73
PH
867 }'''
868 self.assertEqual(js_to_json(inp), '''{
e7b6d122 869 "clip":{"provider":"pseudo"}
410f3e73
PH
870 }''')
871 json.loads(js_to_json(inp))
872
e7b6d122
PH
873 inp = '''{
874 'playlist':[{'controls':{'all':null}}]
875 }'''
876 self.assertEqual(js_to_json(inp), '''{
877 "playlist":[{"controls":{"all":null}}]
878 }''')
879
d01949dc
S
880 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
881 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
882
d305dd73
PH
883 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
884 json_code = js_to_json(inp)
885 self.assertEqual(json.loads(json_code), json.loads(inp))
886
89ac4a19
S
887 inp = '''{
888 0:{src:'skipped', type: 'application/dash+xml'},
889 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
890 }'''
891 self.assertEqual(js_to_json(inp), '''{
892 "0":{"src":"skipped", "type": "application/dash+xml"},
893 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
894 }''')
895
47212f7b
YCH
896 inp = '''{"foo":101}'''
897 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
898
e4659b45
YCH
899 inp = '''{"duration": "00:01:07"}'''
900 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
901
c384d537
MF
902 inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
903 self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
904
e7b6d122
PH
905 def test_js_to_json_edgecases(self):
906 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
907 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
908
909 on = js_to_json('{"abc": true}')
910 self.assertEqual(json.loads(on), {'abc': True})
911
8f4b58d7
PH
912 # Ignore JavaScript code as well
913 on = js_to_json('''{
914 "x": 1,
915 y: "a",
916 z: some.code
917 }''')
918 d = json.loads(on)
919 self.assertEqual(d['x'], 1)
920 self.assertEqual(d['y'], 'a')
921
ba9e68f4
S
922 on = js_to_json('["abc", "def",]')
923 self.assertEqual(json.loads(on), ['abc', 'def'])
924
4195096e
S
925 on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
926 self.assertEqual(json.loads(on), ['abc', 'def'])
927
928 on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
929 self.assertEqual(json.loads(on), ['abc', 'def'])
930
ba9e68f4
S
931 on = js_to_json('{"abc": "def",}')
932 self.assertEqual(json.loads(on), {'abc': 'def'})
933
4195096e
S
934 on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
935 self.assertEqual(json.loads(on), {'abc': 'def'})
936
bd1e4844 937 on = js_to_json('{ 0: /* " \n */ ",]" , }')
938 self.assertEqual(json.loads(on), {'0': ',]'})
939
4195096e
S
940 on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
941 self.assertEqual(json.loads(on), {'0': ',]'})
942
b3ee552e
943 on = js_to_json('{ 0: // comment\n1 }')
944 self.assertEqual(json.loads(on), {'0': 1})
945
bd1e4844 946 on = js_to_json(r'["<p>x<\/p>"]')
947 self.assertEqual(json.loads(on), ['<p>x</p>'])
948
949 on = js_to_json(r'["\xaa"]')
950 self.assertEqual(json.loads(on), ['\u00aa'])
951
952 on = js_to_json("['a\\\nb']")
953 self.assertEqual(json.loads(on), ['ab'])
954
4195096e
S
955 on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
956 self.assertEqual(json.loads(on), ['ab'])
957
89ac4a19
S
958 on = js_to_json('{0xff:0xff}')
959 self.assertEqual(json.loads(on), {'255': 255})
960
4195096e
S
961 on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
962 self.assertEqual(json.loads(on), {'255': 255})
963
89ac4a19
S
964 on = js_to_json('{077:077}')
965 self.assertEqual(json.loads(on), {'63': 63})
966
4195096e
S
967 on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
968 self.assertEqual(json.loads(on), {'63': 63})
969
89ac4a19
S
970 on = js_to_json('{42:42}')
971 self.assertEqual(json.loads(on), {'42': 42})
972
4195096e
S
973 on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
974 self.assertEqual(json.loads(on), {'42': 42})
975
c384d537
MF
976 on = js_to_json('{42:4.2e1}')
977 self.assertEqual(json.loads(on), {'42': 42.0})
978
c707b1d8
S
979 def test_js_to_json_malformed(self):
980 self.assertEqual(js_to_json('42a1'), '42"a1"')
981 self.assertEqual(js_to_json('42a-1'), '42"a"-1')
982
8bb56eee
BF
983 def test_extract_attributes(self):
984 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
985 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
986 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
987 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
988 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
989 self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
990 self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
991 self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'}) # XML
992 self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
c5229f39
S
993 self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
994 self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
8bb56eee
BF
995 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
996 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
997 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
998 self.assertEqual(extract_attributes('<e x >'), {'x': None})
999 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
1000 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
1001 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
1002 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
1003 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
1004 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
1005 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
c5229f39 1006 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
8bb56eee
BF
1007 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
1008 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
1009 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
1010 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
1011 self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
1012 # "Narrow" Python builds don't support unicode code points outside BMP.
1013 try:
1014 compat_chr(0x10000)
1015 supports_outside_bmp = True
1016 except ValueError:
1017 supports_outside_bmp = False
1018 if supports_outside_bmp:
1019 self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
b4a3d461
S
1020 # Malformed HTML should not break attributes extraction on older Python
1021 self.assertEqual(extract_attributes('<mal"formed/>'), {})
8bb56eee 1022
e4bdb37e
PH
1023 def test_clean_html(self):
1024 self.assertEqual(clean_html('a:\nb'), 'a: b')
1025 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
9222d945 1026 self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
e4bdb37e 1027
4c0924bb
PH
1028 def test_intlist_to_bytes(self):
1029 self.assertEqual(
1030 intlist_to_bytes([0, 1, 127, 128, 255]),
1031 b'\x00\x01\x7f\x80\xff')
1032
7d4111ed
PH
1033 def test_args_to_str(self):
1034 self.assertEqual(
1035 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
dee2ff1d 1036 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
7d4111ed
PH
1037 )
1038
be64b5b0
PH
1039 def test_parse_filesize(self):
1040 self.assertEqual(parse_filesize(None), None)
1041 self.assertEqual(parse_filesize(''), None)
1042 self.assertEqual(parse_filesize('91 B'), 91)
1043 self.assertEqual(parse_filesize('foobar'), None)
1044 self.assertEqual(parse_filesize('2 MiB'), 2097152)
1045 self.assertEqual(parse_filesize('5 GB'), 5000000000)
1046 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
13585d76 1047 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
4349c07d 1048 self.assertEqual(parse_filesize('1,24 KB'), 1240)
13585d76 1049 self.assertEqual(parse_filesize('1,24 kb'), 1240)
70852b47 1050 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
be64b5b0 1051
fb47597b
S
1052 def test_parse_count(self):
1053 self.assertEqual(parse_count(None), None)
1054 self.assertEqual(parse_count(''), None)
1055 self.assertEqual(parse_count('0'), 0)
1056 self.assertEqual(parse_count('1000'), 1000)
1057 self.assertEqual(parse_count('1.000'), 1000)
1058 self.assertEqual(parse_count('1.1k'), 1100)
1059 self.assertEqual(parse_count('1.1kk'), 1100000)
782b1b5b
JMF
1060 self.assertEqual(parse_count('1.1kk '), 1100000)
1061 self.assertEqual(parse_count('1.1kk views'), 1100000)
fb47597b 1062
b871d7e9
S
1063 def test_parse_resolution(self):
1064 self.assertEqual(parse_resolution(None), {})
1065 self.assertEqual(parse_resolution(''), {})
1066 self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080})
1067 self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080})
1068 self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
1069 self.assertEqual(parse_resolution('720p'), {'height': 720})
1070 self.assertEqual(parse_resolution('4k'), {'height': 2160})
1071 self.assertEqual(parse_resolution('8K'), {'height': 4320})
1072
0dc41787
S
1073 def test_parse_bitrate(self):
1074 self.assertEqual(parse_bitrate(None), None)
1075 self.assertEqual(parse_bitrate(''), None)
1076 self.assertEqual(parse_bitrate('300kbps'), 300)
1077 self.assertEqual(parse_bitrate('1500kbps'), 1500)
1078 self.assertEqual(parse_bitrate('300 kbps'), 300)
1079
5f9b8394
PH
1080 def test_version_tuple(self):
1081 self.assertEqual(version_tuple('1'), (1,))
1082 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
47d7c642 1083 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
5f9b8394 1084
cae97f65
PH
1085 def test_detect_exe_version(self):
1086 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
1087built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
1088configuration: --prefix=/usr --extra-'''), '1.2.1')
1089 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
1090built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
1091 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
1092Trying to open render node...
1093Success at /dev/dri/renderD128.
1094ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
1095
05900629
PH
1096 def test_age_restricted(self):
1097 self.assertFalse(age_restricted(None, 10)) # unrestricted content
1098 self.assertFalse(age_restricted(1, None)) # unrestricted policy
1099 self.assertFalse(age_restricted(8, 10))
1100 self.assertTrue(age_restricted(18, 14))
1101 self.assertFalse(age_restricted(18, 18))
1102
61ca9a80
PH
1103 def test_is_html(self):
1104 self.assertFalse(is_html(b'\x49\x44\x43<html'))
1105 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
1106 self.assertTrue(is_html( # UTF-8 with BOM
1107 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
1108 self.assertTrue(is_html( # UTF-16-LE
1109 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
1110 ))
1111 self.assertTrue(is_html( # UTF-16-BE
1112 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1113 ))
1114 self.assertTrue(is_html( # UTF-32-BE
1115 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'))
1116 self.assertTrue(is_html( # UTF-32-LE
1117 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'))
1118
cfb56d1a
PH
1119 def test_render_table(self):
1120 self.assertEqual(
1121 render_table(
1122 ['a', 'bcd'],
1123 [[123, 4], [9999, 51]]),
1124 'a bcd\n'
1125 '123 4\n'
1126 '9999 51')
1127
347de493
PH
1128 def test_match_str(self):
1129 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1130 self.assertFalse(match_str('xy', {'x': 1200}))
1131 self.assertTrue(match_str('!xy', {'x': 1200}))
1132 self.assertTrue(match_str('x', {'x': 1200}))
1133 self.assertFalse(match_str('!x', {'x': 1200}))
1134 self.assertTrue(match_str('x', {'x': 0}))
1135 self.assertFalse(match_str('x>0', {'x': 0}))
1136 self.assertFalse(match_str('x>0', {}))
1137 self.assertTrue(match_str('x>?0', {}))
1138 self.assertTrue(match_str('x>1K', {'x': 1200}))
1139 self.assertFalse(match_str('x>2K', {'x': 1200}))
1140 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1141 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1142 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1143 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1144 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1145 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1146 self.assertFalse(match_str(
1147 'like_count > 100 & dislike_count <? 50 & description',
1148 {'like_count': 90, 'description': 'foo'}))
1149 self.assertTrue(match_str(
1150 'like_count > 100 & dislike_count <? 50 & description',
1151 {'like_count': 190, 'description': 'foo'}))
1152 self.assertFalse(match_str(
1153 'like_count > 100 & dislike_count <? 50 & description',
1154 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1155 self.assertFalse(match_str(
1156 'like_count > 100 & dislike_count <? 50 & description',
1157 {'like_count': 190, 'dislike_count': 10}))
1cc47c66
S
1158 self.assertTrue(match_str('is_live', {'is_live': True}))
1159 self.assertFalse(match_str('is_live', {'is_live': False}))
1160 self.assertFalse(match_str('is_live', {'is_live': None}))
1161 self.assertFalse(match_str('is_live', {}))
1162 self.assertFalse(match_str('!is_live', {'is_live': True}))
1163 self.assertTrue(match_str('!is_live', {'is_live': False}))
1164 self.assertTrue(match_str('!is_live', {'is_live': None}))
1165 self.assertTrue(match_str('!is_live', {}))
1166 self.assertTrue(match_str('title', {'title': 'abc'}))
1167 self.assertTrue(match_str('title', {'title': ''}))
1168 self.assertFalse(match_str('!title', {'title': 'abc'}))
1169 self.assertFalse(match_str('!title', {'title': ''}))
347de493 1170
bf6427d2 1171 def test_parse_dfxp_time_expr(self):
d631d5f9
YCH
1172 self.assertEqual(parse_dfxp_time_expr(None), None)
1173 self.assertEqual(parse_dfxp_time_expr(''), None)
bf6427d2
YCH
1174 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1175 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1176 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1177 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
db2fe38b 1178 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
bf6427d2
YCH
1179
1180 def test_dfxp2srt(self):
1181 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1182 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1183 <body>
1184 <div xml:lang="en">
1185 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1186 <p begin="1" end="2">第二行<br/>♪♪</p>
7dff0363 1187 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
d631d5f9
YCH
1188 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1189 <p begin="-1" end="-1">Ignore, two</p>
1190 <p begin="3" dur="-1">Ignored, three</p>
bf6427d2
YCH
1191 </div>
1192 </body>
3869028f 1193 </tt>'''.encode('utf-8')
bf6427d2
YCH
1194 srt_data = '''1
119500:00:00,000 --> 00:00:01,000
1196The following line contains Chinese characters and special symbols
1197
11982
119900:00:01,000 --> 00:00:02,000
1200第二行
1201♪♪
1202
12033
120400:00:02,000 --> 00:00:03,000
1205Third
1206Line
1207
1208'''
1209 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1210
1b0427e6
YCH
1211 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1212 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1213 <body>
1214 <div xml:lang="en">
1215 <p begin="0" end="1">The first line</p>
1216 </div>
1217 </body>
3869028f 1218 </tt>'''.encode('utf-8')
1b0427e6
YCH
1219 srt_data = '''1
122000:00:00,000 --> 00:00:01,000
1221The first line
1222
1223'''
1224 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1225
5b995f71
RA
1226 dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1227<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">
1228 <head>
1229 <styling>
1230 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1231 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1232 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1233 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1234 </styling>
1235 </head>
1236 <body tts:textAlign="center" style="s0">
1237 <div>
1238 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1239 <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>
1240 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1241 <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>
1242 </div>
1243 </body>
3869028f 1244</tt>'''.encode('utf-8')
5b995f71
RA
1245 srt_data = '''1
124600:00:02,080 --> 00:00:05,839
1247<font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1248
12492
125000:00:02,080 --> 00:00:05,839
1251<b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1252</font>part 2</font></b>
1253
12543
125500:00:05,839 --> 00:00:09,560
1256<u><font color="lime">line 3
1257part 3</font></u>
1258
12594
126000:00:09,560 --> 00:00:12,359
1261<i><u><font color="yellow"><font color="lime">inner
1262 </font>style</font></u></i>
1263
1264'''
1265 self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1266
3869028f
YCH
1267 dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1268 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1269 <body>
1270 <div xml:lang="en">
1271 <p begin="0" end="1">Line 1</p>
1272 <p begin="1" end="2">第二行</p>
1273 </div>
1274 </body>
1275 </tt>'''.encode('utf-16')
1276 srt_data = '''1
127700:00:00,000 --> 00:00:01,000
1278Line 1
1279
12802
128100:00:01,000 --> 00:00:02,000
1282第二行
1283
1284'''
1285 self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1286
f7126449
S
1287 def test_cli_option(self):
1288 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1289 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1290 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
5f2c2b79 1291 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
f7126449
S
1292
1293 def test_cli_valueless_option(self):
1294 self.assertEqual(cli_valueless_option(
1295 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1296 self.assertEqual(cli_valueless_option(
1297 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1298 self.assertEqual(cli_valueless_option(
1299 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1300 self.assertEqual(cli_valueless_option(
1301 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1302 self.assertEqual(cli_valueless_option(
1303 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1304 self.assertEqual(cli_valueless_option(
1305 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1306
1307 def test_cli_bool_option(self):
1308 self.assertEqual(
1309 cli_bool_option(
1310 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1311 ['--no-check-certificate', 'true'])
1312 self.assertEqual(
1313 cli_bool_option(
1314 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1315 ['--no-check-certificate=true'])
1316 self.assertEqual(
1317 cli_bool_option(
1318 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1319 ['--check-certificate', 'false'])
1320 self.assertEqual(
1321 cli_bool_option(
1322 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1323 ['--check-certificate=false'])
1324 self.assertEqual(
1325 cli_bool_option(
1326 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1327 ['--check-certificate', 'true'])
1328 self.assertEqual(
1329 cli_bool_option(
1330 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1331 ['--check-certificate=true'])
5b232f46
S
1332 self.assertEqual(
1333 cli_bool_option(
1334 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1335 [])
f7126449 1336
5bc880b9
YCH
1337 def test_ohdave_rsa_encrypt(self):
1338 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1339 e = 65537
1340
1341 self.assertEqual(
1342 ohdave_rsa_encrypt(b'aa111222', e, N),
1343 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
cfb56d1a 1344
f48409c7
YCH
1345 def test_pkcs1pad(self):
1346 data = [1, 2, 3]
1347 padded_data = pkcs1pad(data, 32)
1348 self.assertEqual(padded_data[:2], [0, 2])
1349 self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1350
1351 self.assertRaises(ValueError, pkcs1pad, data, 8)
1352
5eb6bdce
YCH
1353 def test_encode_base_n(self):
1354 self.assertEqual(encode_base_n(0, 30), '0')
1355 self.assertEqual(encode_base_n(80, 30), '2k')
1356
1357 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1358 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1359 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1360
1361 self.assertRaises(ValueError, encode_base_n, 0, 70)
1362 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1363
1143535d
YCH
1364 def test_urshift(self):
1365 self.assertEqual(urshift(3, 1), 1)
1366 self.assertEqual(urshift(-3, 1), 2147483646)
1367
84c237fb
YCH
1368 def test_get_element_by_class(self):
1369 html = '''
1370 <span class="foo bar">nice</span>
1371 '''
1372
1373 self.assertEqual(get_element_by_class('foo', html), 'nice')
1374 self.assertEqual(get_element_by_class('no-such-class', html), None)
1375
2af12ad9
TC
1376 def test_get_element_by_attribute(self):
1377 html = '''
1378 <span class="foo bar">nice</span>
1379 '''
1380
1381 self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1382 self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1383 self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1384
609ff8ca
YCH
1385 html = '''
1386 <div itemprop="author" itemscope>foo</div>
1387 '''
1388
1389 self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1390
2af12ad9
TC
1391 def test_get_elements_by_class(self):
1392 html = '''
1393 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1394 '''
1395
1396 self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1397 self.assertEqual(get_elements_by_class('no-such-class', html), [])
1398
1399 def test_get_elements_by_attribute(self):
1400 html = '''
1401 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1402 '''
1403
1404 self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1405 self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1406 self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1407
582be358 1408
dae7c920 1409if __name__ == '__main__':
59ae15a5 1410 unittest.main()