]> Raphaël G. Git Repositories - youtubedl/blob - test/test_utils.py
New upstream release 2017.09.24.
[youtubedl] / test / test_utils.py
1 #!/usr/bin/env python
2 # coding: utf-8
3
4 from __future__ import unicode_literals
5
6 # Allow direct execution
7 import os
8 import sys
9 import unittest
10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
11
12
13 # Various small unit tests
14 import io
15 import json
16 import xml.etree.ElementTree
17
18 from youtube_dl.utils import (
19 age_restricted,
20 args_to_str,
21 encode_base_n,
22 clean_html,
23 date_from_str,
24 DateRange,
25 detect_exe_version,
26 determine_ext,
27 dict_get,
28 encode_compat_str,
29 encodeFilename,
30 escape_rfc3986,
31 escape_url,
32 extract_attributes,
33 ExtractorError,
34 find_xpath_attr,
35 fix_xml_ampersands,
36 get_element_by_class,
37 get_element_by_attribute,
38 get_elements_by_class,
39 get_elements_by_attribute,
40 InAdvancePagedList,
41 intlist_to_bytes,
42 is_html,
43 js_to_json,
44 limit_length,
45 mimetype2ext,
46 month_by_name,
47 multipart_encode,
48 ohdave_rsa_encrypt,
49 OnDemandPagedList,
50 orderedSet,
51 parse_age_limit,
52 parse_duration,
53 parse_filesize,
54 parse_count,
55 parse_iso8601,
56 pkcs1pad,
57 read_batch_urls,
58 sanitize_filename,
59 sanitize_path,
60 expand_path,
61 prepend_extension,
62 replace_extension,
63 remove_start,
64 remove_end,
65 remove_quotes,
66 shell_quote,
67 smuggle_url,
68 str_to_int,
69 strip_jsonp,
70 timeconvert,
71 unescapeHTML,
72 unified_strdate,
73 unified_timestamp,
74 unsmuggle_url,
75 uppercase_escape,
76 lowercase_escape,
77 url_basename,
78 base_url,
79 urljoin,
80 urlencode_postdata,
81 urshift,
82 update_url_query,
83 version_tuple,
84 xpath_with_ns,
85 xpath_element,
86 xpath_text,
87 xpath_attr,
88 render_table,
89 match_str,
90 parse_dfxp_time_expr,
91 dfxp2srt,
92 cli_option,
93 cli_valueless_option,
94 cli_bool_option,
95 parse_codecs,
96 )
97 from youtube_dl.compat import (
98 compat_chr,
99 compat_etree_fromstring,
100 compat_getenv,
101 compat_os_name,
102 compat_setenv,
103 compat_urlparse,
104 compat_parse_qs,
105 )
106
107
108 class TestUtil(unittest.TestCase):
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')
128 aumlaut = 'ä'
129 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
130 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
131 self.assertEqual(sanitize_filename(tests), tests)
132
133 self.assertEqual(
134 sanitize_filename('New World record at 0:12:34'),
135 'New World record at 0_12_34')
136
137 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
138 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
139 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
140 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
141
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
161 tests = 'aäb\u4e2d\u56fd\u7684c'
162 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
163 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
164
165 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
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
171 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
172 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
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
177 self.assertEqual(sanitize_filename(
178 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
179 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
180
181 def test_sanitize_ids(self):
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')
185
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
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
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
222 def test_expand_path(self):
223 def env(var):
224 return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
225
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'))
229 self.assertEqual(expand_path('~'), compat_getenv('HOME'))
230 self.assertEqual(
231 expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
232 '%s/expanded' % compat_getenv('HOME'))
233
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
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
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
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
269 def test_ordered_set(self):
270 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
271 self.assertEqual(orderedSet([]), [])
272 self.assertEqual(orderedSet([1]), [1])
273 # keep the list ordered
274 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
275
276 def test_unescape_html(self):
277 self.assertEqual(unescapeHTML('%20;'), '%20;')
278 self.assertEqual(unescapeHTML('&#x2F;'), '/')
279 self.assertEqual(unescapeHTML('&#47;'), '/')
280 self.assertEqual(unescapeHTML('&eacute;'), 'é')
281 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
282 self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
283 # HTML5 entities
284 self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
285
286 def test_date_from_str(self):
287 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
288 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
289 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
290 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
291 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
292
293 def test_daterange(self):
294 _20century = DateRange("19000101", "20000101")
295 self.assertFalse("17890714" in _20century)
296 _ac = DateRange("00010101")
297 self.assertTrue("19690721" in _ac)
298 _firstmilenium = DateRange(end="10000101")
299 self.assertTrue("07110427" in _firstmilenium)
300
301 def test_unified_dates(self):
302 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
303 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
304 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
305 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
306 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
307 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
308 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
309 self.assertEqual(
310 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
311 '20141126')
312 self.assertEqual(
313 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
314 '20150202')
315 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
316 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
317 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
318 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
319 self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
320 self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
321 self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
322 self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
323
324 def test_unified_timestamps(self):
325 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
326 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
327 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
328 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
329 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
330 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
331 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
332 self.assertEqual(
333 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
334 1417001400)
335 self.assertEqual(
336 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
337 1422902860)
338 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
339 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
340 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
341 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
342 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
343 self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
344 self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
345 self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
346
347 def test_determine_ext(self):
348 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
349 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
350 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
351 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
352 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
353
354 def test_find_xpath_attr(self):
355 testxml = '''<root>
356 <node/>
357 <node x="a"/>
358 <node x="a" y="c" />
359 <node x="b" y="d" />
360 <node x="" />
361 </root>'''
362 doc = compat_etree_fromstring(testxml)
363
364 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
365 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
366 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
367 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
368 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
369 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
370 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
371 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
372 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
373 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
374 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
375
376 def test_xpath_with_ns(self):
377 testxml = '''<root xmlns:media="http://example.com/">
378 <media:song>
379 <media:author>The Author</media:author>
380 <url>http://server.com/download.mp3</url>
381 </media:song>
382 </root>'''
383 doc = compat_etree_fromstring(testxml)
384 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
385 self.assertTrue(find('media:song') is not None)
386 self.assertEqual(find('media:song/media:author').text, 'The Author')
387 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
388
389 def test_xpath_element(self):
390 doc = xml.etree.ElementTree.Element('root')
391 div = xml.etree.ElementTree.SubElement(doc, 'div')
392 p = xml.etree.ElementTree.SubElement(div, 'p')
393 p.text = 'Foo'
394 self.assertEqual(xpath_element(doc, 'div/p'), p)
395 self.assertEqual(xpath_element(doc, ['div/p']), p)
396 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
397 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
398 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
399 self.assertTrue(xpath_element(doc, 'div/bar') is None)
400 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
401 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
402 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
403 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
404 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
405
406 def test_xpath_text(self):
407 testxml = '''<root>
408 <div>
409 <p>Foo</p>
410 </div>
411 </root>'''
412 doc = compat_etree_fromstring(testxml)
413 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
414 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
415 self.assertTrue(xpath_text(doc, 'div/bar') is None)
416 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
417
418 def test_xpath_attr(self):
419 testxml = '''<root>
420 <div>
421 <p x="a">Foo</p>
422 </div>
423 </root>'''
424 doc = compat_etree_fromstring(testxml)
425 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
426 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
427 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
428 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
429 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
430 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
431 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
432
433 def test_smuggle_url(self):
434 data = {"ö": "ö", "abc": [3]}
435 url = 'https://foo.bar/baz?x=y#a'
436 smug_url = smuggle_url(url, data)
437 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
438 self.assertEqual(url, unsmug_url)
439 self.assertEqual(data, unsmug_data)
440
441 res_url, res_data = unsmuggle_url(url)
442 self.assertEqual(res_url, url)
443 self.assertEqual(res_data, None)
444
445 smug_url = smuggle_url(url, {'a': 'b'})
446 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
447 res_url, res_data = unsmuggle_url(smug_smug_url)
448 self.assertEqual(res_url, url)
449 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
450
451 def test_shell_quote(self):
452 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
453 self.assertEqual(
454 shell_quote(args),
455 """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
456
457 def test_str_to_int(self):
458 self.assertEqual(str_to_int('123,456'), 123456)
459 self.assertEqual(str_to_int('123.456'), 123456)
460
461 def test_url_basename(self):
462 self.assertEqual(url_basename('http://foo.de/'), '')
463 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
464 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
465 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
466 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
467 self.assertEqual(
468 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
469 'trailer.mp4')
470
471 def test_base_url(self):
472 self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
473 self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
474 self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
475 self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
476 self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
477
478 def test_urljoin(self):
479 self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
480 self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
481 self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
482 self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
483 self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//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', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
487 self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
488 self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
489 self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
490 self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
491 self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
492 self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
493 self.assertEqual(urljoin('http://foo.de/', None), None)
494 self.assertEqual(urljoin('http://foo.de/', ''), None)
495 self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
496 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
497
498 def test_parse_age_limit(self):
499 self.assertEqual(parse_age_limit(None), None)
500 self.assertEqual(parse_age_limit(False), None)
501 self.assertEqual(parse_age_limit('invalid'), None)
502 self.assertEqual(parse_age_limit(0), 0)
503 self.assertEqual(parse_age_limit(18), 18)
504 self.assertEqual(parse_age_limit(21), 21)
505 self.assertEqual(parse_age_limit(22), None)
506 self.assertEqual(parse_age_limit('18'), 18)
507 self.assertEqual(parse_age_limit('18+'), 18)
508 self.assertEqual(parse_age_limit('PG-13'), 13)
509 self.assertEqual(parse_age_limit('TV-14'), 14)
510 self.assertEqual(parse_age_limit('TV-MA'), 17)
511
512 def test_parse_duration(self):
513 self.assertEqual(parse_duration(None), None)
514 self.assertEqual(parse_duration(False), None)
515 self.assertEqual(parse_duration('invalid'), None)
516 self.assertEqual(parse_duration('1'), 1)
517 self.assertEqual(parse_duration('1337:12'), 80232)
518 self.assertEqual(parse_duration('9:12:43'), 33163)
519 self.assertEqual(parse_duration('12:00'), 720)
520 self.assertEqual(parse_duration('00:01:01'), 61)
521 self.assertEqual(parse_duration('x:y'), None)
522 self.assertEqual(parse_duration('3h11m53s'), 11513)
523 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
524 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
525 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
526 self.assertEqual(parse_duration('62m45s'), 3765)
527 self.assertEqual(parse_duration('6m59s'), 419)
528 self.assertEqual(parse_duration('49s'), 49)
529 self.assertEqual(parse_duration('0h0m0s'), 0)
530 self.assertEqual(parse_duration('0m0s'), 0)
531 self.assertEqual(parse_duration('0s'), 0)
532 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
533 self.assertEqual(parse_duration('T30M38S'), 1838)
534 self.assertEqual(parse_duration('5 s'), 5)
535 self.assertEqual(parse_duration('3 min'), 180)
536 self.assertEqual(parse_duration('2.5 hours'), 9000)
537 self.assertEqual(parse_duration('02:03:04'), 7384)
538 self.assertEqual(parse_duration('01:02:03:04'), 93784)
539 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
540 self.assertEqual(parse_duration('87 Min.'), 5220)
541 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
542 self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
543
544 def test_fix_xml_ampersands(self):
545 self.assertEqual(
546 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
547 self.assertEqual(
548 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
549 '"&amp;x=y&amp;wrong;&amp;z=a')
550 self.assertEqual(
551 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
552 '&amp;&apos;&gt;&lt;&quot;')
553 self.assertEqual(
554 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
555 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
556
557 def test_paged_list(self):
558 def testPL(size, pagesize, sliceargs, expected):
559 def get_page(pagenum):
560 firstid = pagenum * pagesize
561 upto = min(size, pagenum * pagesize + pagesize)
562 for i in range(firstid, upto):
563 yield i
564
565 pl = OnDemandPagedList(get_page, pagesize)
566 got = pl.getslice(*sliceargs)
567 self.assertEqual(got, expected)
568
569 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
570 got = iapl.getslice(*sliceargs)
571 self.assertEqual(got, expected)
572
573 testPL(5, 2, (), [0, 1, 2, 3, 4])
574 testPL(5, 2, (1,), [1, 2, 3, 4])
575 testPL(5, 2, (2,), [2, 3, 4])
576 testPL(5, 2, (4,), [4])
577 testPL(5, 2, (0, 3), [0, 1, 2])
578 testPL(5, 2, (1, 4), [1, 2, 3])
579 testPL(5, 2, (2, 99), [2, 3, 4])
580 testPL(5, 2, (20, 99), [])
581
582 def test_read_batch_urls(self):
583 f = io.StringIO('''\xef\xbb\xbf foo
584 bar\r
585 baz
586 # More after this line\r
587 ; or after this
588 bam''')
589 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
590
591 def test_urlencode_postdata(self):
592 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
593 self.assertTrue(isinstance(data, bytes))
594
595 def test_update_url_query(self):
596 def query_dict(url):
597 return compat_parse_qs(compat_urlparse.urlparse(url).query)
598 self.assertEqual(query_dict(update_url_query(
599 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
600 query_dict('http://example.com/path?quality=HD&format=mp4'))
601 self.assertEqual(query_dict(update_url_query(
602 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
603 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
604 self.assertEqual(query_dict(update_url_query(
605 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
606 query_dict('http://example.com/path?fields=id,formats,subtitles'))
607 self.assertEqual(query_dict(update_url_query(
608 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
609 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
610 self.assertEqual(query_dict(update_url_query(
611 'http://example.com/path?manifest=f4m', {'manifest': []})),
612 query_dict('http://example.com/path'))
613 self.assertEqual(query_dict(update_url_query(
614 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
615 query_dict('http://example.com/path?system=LINUX'))
616 self.assertEqual(query_dict(update_url_query(
617 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
618 query_dict('http://example.com/path?fields=id,formats,subtitles'))
619 self.assertEqual(query_dict(update_url_query(
620 'http://example.com/path', {'width': 1080, 'height': 720})),
621 query_dict('http://example.com/path?width=1080&height=720'))
622 self.assertEqual(query_dict(update_url_query(
623 'http://example.com/path', {'bitrate': 5020.43})),
624 query_dict('http://example.com/path?bitrate=5020.43'))
625 self.assertEqual(query_dict(update_url_query(
626 'http://example.com/path', {'test': '第二行тест'})),
627 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
628
629 def test_multipart_encode(self):
630 self.assertEqual(
631 multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
632 b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
633 self.assertEqual(
634 multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
635 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')
636 self.assertRaises(
637 ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
638
639 def test_dict_get(self):
640 FALSE_VALUES = {
641 'none': None,
642 'false': False,
643 'zero': 0,
644 'empty_string': '',
645 'empty_list': [],
646 }
647 d = FALSE_VALUES.copy()
648 d['a'] = 42
649 self.assertEqual(dict_get(d, 'a'), 42)
650 self.assertEqual(dict_get(d, 'b'), None)
651 self.assertEqual(dict_get(d, 'b', 42), 42)
652 self.assertEqual(dict_get(d, ('a', )), 42)
653 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
654 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
655 self.assertEqual(dict_get(d, ('b', 'c', )), None)
656 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
657 for key, false_value in FALSE_VALUES.items():
658 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
659 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
660
661 def test_encode_compat_str(self):
662 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
663 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
664
665 def test_parse_iso8601(self):
666 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
667 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
668 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
669 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
670 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
671 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
672
673 def test_strip_jsonp(self):
674 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
675 d = json.loads(stripped)
676 self.assertEqual(d, [{"id": "532cb", "x": 3}])
677
678 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
679 d = json.loads(stripped)
680 self.assertEqual(d, {'STATUS': 'OK'})
681
682 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
683 d = json.loads(stripped)
684 self.assertEqual(d, {'status': 'success'})
685
686 stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
687 d = json.loads(stripped)
688 self.assertEqual(d, {'status': 'success'})
689
690 stripped = strip_jsonp('window.cb && cb({"status": "success"});')
691 d = json.loads(stripped)
692 self.assertEqual(d, {'status': 'success'})
693
694 def test_uppercase_escape(self):
695 self.assertEqual(uppercase_escape('aä'), 'aä')
696 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
697
698 def test_lowercase_escape(self):
699 self.assertEqual(lowercase_escape('aä'), 'aä')
700 self.assertEqual(lowercase_escape('\\u0026'), '&')
701
702 def test_limit_length(self):
703 self.assertEqual(limit_length(None, 12), None)
704 self.assertEqual(limit_length('foo', 12), 'foo')
705 self.assertTrue(
706 limit_length('foo bar baz asd', 12).startswith('foo bar'))
707 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
708
709 def test_mimetype2ext(self):
710 self.assertEqual(mimetype2ext(None), None)
711 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
712 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
713 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
714 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
715 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
716
717 def test_month_by_name(self):
718 self.assertEqual(month_by_name(None), None)
719 self.assertEqual(month_by_name('December', 'en'), 12)
720 self.assertEqual(month_by_name('décembre', 'fr'), 12)
721 self.assertEqual(month_by_name('December'), 12)
722 self.assertEqual(month_by_name('décembre'), None)
723 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
724
725 def test_parse_codecs(self):
726 self.assertEqual(parse_codecs(''), {})
727 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
728 'vcodec': 'avc1.77.30',
729 'acodec': 'mp4a.40.2',
730 })
731 self.assertEqual(parse_codecs('mp4a.40.2'), {
732 'vcodec': 'none',
733 'acodec': 'mp4a.40.2',
734 })
735 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
736 'vcodec': 'avc1.42001e',
737 'acodec': 'mp4a.40.5',
738 })
739 self.assertEqual(parse_codecs('avc3.640028'), {
740 'vcodec': 'avc3.640028',
741 'acodec': 'none',
742 })
743 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
744 'vcodec': 'h264',
745 'acodec': 'aac',
746 })
747
748 def test_escape_rfc3986(self):
749 reserved = "!*'();:@&=+$,/?#[]"
750 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
751 self.assertEqual(escape_rfc3986(reserved), reserved)
752 self.assertEqual(escape_rfc3986(unreserved), unreserved)
753 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
754 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
755 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
756 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
757
758 def test_escape_url(self):
759 self.assertEqual(
760 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
761 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
762 )
763 self.assertEqual(
764 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
765 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
766 )
767 self.assertEqual(
768 escape_url('http://тест.рф/фрагмент'),
769 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
770 )
771 self.assertEqual(
772 escape_url('http://тест.рф/абв?абв=абв#абв'),
773 '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'
774 )
775 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
776
777 def test_js_to_json_realworld(self):
778 inp = '''{
779 'clip':{'provider':'pseudo'}
780 }'''
781 self.assertEqual(js_to_json(inp), '''{
782 "clip":{"provider":"pseudo"}
783 }''')
784 json.loads(js_to_json(inp))
785
786 inp = '''{
787 'playlist':[{'controls':{'all':null}}]
788 }'''
789 self.assertEqual(js_to_json(inp), '''{
790 "playlist":[{"controls":{"all":null}}]
791 }''')
792
793 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
794 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
795
796 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
797 json_code = js_to_json(inp)
798 self.assertEqual(json.loads(json_code), json.loads(inp))
799
800 inp = '''{
801 0:{src:'skipped', type: 'application/dash+xml'},
802 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
803 }'''
804 self.assertEqual(js_to_json(inp), '''{
805 "0":{"src":"skipped", "type": "application/dash+xml"},
806 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
807 }''')
808
809 inp = '''{"foo":101}'''
810 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
811
812 inp = '''{"duration": "00:01:07"}'''
813 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
814
815 def test_js_to_json_edgecases(self):
816 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
817 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
818
819 on = js_to_json('{"abc": true}')
820 self.assertEqual(json.loads(on), {'abc': True})
821
822 # Ignore JavaScript code as well
823 on = js_to_json('''{
824 "x": 1,
825 y: "a",
826 z: some.code
827 }''')
828 d = json.loads(on)
829 self.assertEqual(d['x'], 1)
830 self.assertEqual(d['y'], 'a')
831
832 on = js_to_json('["abc", "def",]')
833 self.assertEqual(json.loads(on), ['abc', 'def'])
834
835 on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
836 self.assertEqual(json.loads(on), ['abc', 'def'])
837
838 on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
839 self.assertEqual(json.loads(on), ['abc', 'def'])
840
841 on = js_to_json('{"abc": "def",}')
842 self.assertEqual(json.loads(on), {'abc': 'def'})
843
844 on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
845 self.assertEqual(json.loads(on), {'abc': 'def'})
846
847 on = js_to_json('{ 0: /* " \n */ ",]" , }')
848 self.assertEqual(json.loads(on), {'0': ',]'})
849
850 on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
851 self.assertEqual(json.loads(on), {'0': ',]'})
852
853 on = js_to_json('{ 0: // comment\n1 }')
854 self.assertEqual(json.loads(on), {'0': 1})
855
856 on = js_to_json(r'["<p>x<\/p>"]')
857 self.assertEqual(json.loads(on), ['<p>x</p>'])
858
859 on = js_to_json(r'["\xaa"]')
860 self.assertEqual(json.loads(on), ['\u00aa'])
861
862 on = js_to_json("['a\\\nb']")
863 self.assertEqual(json.loads(on), ['ab'])
864
865 on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
866 self.assertEqual(json.loads(on), ['ab'])
867
868 on = js_to_json('{0xff:0xff}')
869 self.assertEqual(json.loads(on), {'255': 255})
870
871 on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
872 self.assertEqual(json.loads(on), {'255': 255})
873
874 on = js_to_json('{077:077}')
875 self.assertEqual(json.loads(on), {'63': 63})
876
877 on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
878 self.assertEqual(json.loads(on), {'63': 63})
879
880 on = js_to_json('{42:42}')
881 self.assertEqual(json.loads(on), {'42': 42})
882
883 on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
884 self.assertEqual(json.loads(on), {'42': 42})
885
886 def test_extract_attributes(self):
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=y>'), {'x': 'y'})
890 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
891 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
892 self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
893 self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
894 self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'}) # XML
895 self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
896 self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
897 self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
898 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
899 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
900 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
901 self.assertEqual(extract_attributes('<e x >'), {'x': None})
902 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
903 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
904 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
905 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
906 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
907 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
908 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
909 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
910 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
911 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
912 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
913 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
914 self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
915 # "Narrow" Python builds don't support unicode code points outside BMP.
916 try:
917 compat_chr(0x10000)
918 supports_outside_bmp = True
919 except ValueError:
920 supports_outside_bmp = False
921 if supports_outside_bmp:
922 self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
923 # Malformed HTML should not break attributes extraction on older Python
924 self.assertEqual(extract_attributes('<mal"formed/>'), {})
925
926 def test_clean_html(self):
927 self.assertEqual(clean_html('a:\nb'), 'a: b')
928 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
929 self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
930
931 def test_intlist_to_bytes(self):
932 self.assertEqual(
933 intlist_to_bytes([0, 1, 127, 128, 255]),
934 b'\x00\x01\x7f\x80\xff')
935
936 def test_args_to_str(self):
937 self.assertEqual(
938 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
939 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
940 )
941
942 def test_parse_filesize(self):
943 self.assertEqual(parse_filesize(None), None)
944 self.assertEqual(parse_filesize(''), None)
945 self.assertEqual(parse_filesize('91 B'), 91)
946 self.assertEqual(parse_filesize('foobar'), None)
947 self.assertEqual(parse_filesize('2 MiB'), 2097152)
948 self.assertEqual(parse_filesize('5 GB'), 5000000000)
949 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
950 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
951 self.assertEqual(parse_filesize('1,24 KB'), 1240)
952 self.assertEqual(parse_filesize('1,24 kb'), 1240)
953 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
954
955 def test_parse_count(self):
956 self.assertEqual(parse_count(None), None)
957 self.assertEqual(parse_count(''), None)
958 self.assertEqual(parse_count('0'), 0)
959 self.assertEqual(parse_count('1000'), 1000)
960 self.assertEqual(parse_count('1.000'), 1000)
961 self.assertEqual(parse_count('1.1k'), 1100)
962 self.assertEqual(parse_count('1.1kk'), 1100000)
963 self.assertEqual(parse_count('1.1kk '), 1100000)
964 self.assertEqual(parse_count('1.1kk views'), 1100000)
965
966 def test_version_tuple(self):
967 self.assertEqual(version_tuple('1'), (1,))
968 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
969 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
970
971 def test_detect_exe_version(self):
972 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
973 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
974 configuration: --prefix=/usr --extra-'''), '1.2.1')
975 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
976 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
977 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
978 Trying to open render node...
979 Success at /dev/dri/renderD128.
980 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
981
982 def test_age_restricted(self):
983 self.assertFalse(age_restricted(None, 10)) # unrestricted content
984 self.assertFalse(age_restricted(1, None)) # unrestricted policy
985 self.assertFalse(age_restricted(8, 10))
986 self.assertTrue(age_restricted(18, 14))
987 self.assertFalse(age_restricted(18, 18))
988
989 def test_is_html(self):
990 self.assertFalse(is_html(b'\x49\x44\x43<html'))
991 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
992 self.assertTrue(is_html( # UTF-8 with BOM
993 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
994 self.assertTrue(is_html( # UTF-16-LE
995 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
996 ))
997 self.assertTrue(is_html( # UTF-16-BE
998 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
999 ))
1000 self.assertTrue(is_html( # UTF-32-BE
1001 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'))
1002 self.assertTrue(is_html( # UTF-32-LE
1003 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'))
1004
1005 def test_render_table(self):
1006 self.assertEqual(
1007 render_table(
1008 ['a', 'bcd'],
1009 [[123, 4], [9999, 51]]),
1010 'a bcd\n'
1011 '123 4\n'
1012 '9999 51')
1013
1014 def test_match_str(self):
1015 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1016 self.assertFalse(match_str('xy', {'x': 1200}))
1017 self.assertTrue(match_str('!xy', {'x': 1200}))
1018 self.assertTrue(match_str('x', {'x': 1200}))
1019 self.assertFalse(match_str('!x', {'x': 1200}))
1020 self.assertTrue(match_str('x', {'x': 0}))
1021 self.assertFalse(match_str('x>0', {'x': 0}))
1022 self.assertFalse(match_str('x>0', {}))
1023 self.assertTrue(match_str('x>?0', {}))
1024 self.assertTrue(match_str('x>1K', {'x': 1200}))
1025 self.assertFalse(match_str('x>2K', {'x': 1200}))
1026 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1027 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1028 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1029 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1030 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1031 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1032 self.assertFalse(match_str(
1033 'like_count > 100 & dislike_count <? 50 & description',
1034 {'like_count': 90, 'description': 'foo'}))
1035 self.assertTrue(match_str(
1036 'like_count > 100 & dislike_count <? 50 & description',
1037 {'like_count': 190, 'description': 'foo'}))
1038 self.assertFalse(match_str(
1039 'like_count > 100 & dislike_count <? 50 & description',
1040 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1041 self.assertFalse(match_str(
1042 'like_count > 100 & dislike_count <? 50 & description',
1043 {'like_count': 190, 'dislike_count': 10}))
1044
1045 def test_parse_dfxp_time_expr(self):
1046 self.assertEqual(parse_dfxp_time_expr(None), None)
1047 self.assertEqual(parse_dfxp_time_expr(''), None)
1048 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1049 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1050 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1051 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1052 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1053
1054 def test_dfxp2srt(self):
1055 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1056 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1057 <body>
1058 <div xml:lang="en">
1059 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1060 <p begin="1" end="2">第二行<br/>♪♪</p>
1061 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1062 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1063 <p begin="-1" end="-1">Ignore, two</p>
1064 <p begin="3" dur="-1">Ignored, three</p>
1065 </div>
1066 </body>
1067 </tt>'''.encode('utf-8')
1068 srt_data = '''1
1069 00:00:00,000 --> 00:00:01,000
1070 The following line contains Chinese characters and special symbols
1071
1072 2
1073 00:00:01,000 --> 00:00:02,000
1074 第二行
1075 ♪♪
1076
1077 3
1078 00:00:02,000 --> 00:00:03,000
1079 Third
1080 Line
1081
1082 '''
1083 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1084
1085 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1086 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1087 <body>
1088 <div xml:lang="en">
1089 <p begin="0" end="1">The first line</p>
1090 </div>
1091 </body>
1092 </tt>'''.encode('utf-8')
1093 srt_data = '''1
1094 00:00:00,000 --> 00:00:01,000
1095 The first line
1096
1097 '''
1098 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1099
1100 dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1101 <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">
1102 <head>
1103 <styling>
1104 <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1105 <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1106 <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1107 <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1108 </styling>
1109 </head>
1110 <body tts:textAlign="center" style="s0">
1111 <div>
1112 <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1113 <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>
1114 <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1115 <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>
1116 </div>
1117 </body>
1118 </tt>'''.encode('utf-8')
1119 srt_data = '''1
1120 00:00:02,080 --> 00:00:05,839
1121 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1122
1123 2
1124 00:00:02,080 --> 00:00:05,839
1125 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1126 </font>part 2</font></b>
1127
1128 3
1129 00:00:05,839 --> 00:00:09,560
1130 <u><font color="lime">line 3
1131 part 3</font></u>
1132
1133 4
1134 00:00:09,560 --> 00:00:12,359
1135 <i><u><font color="yellow"><font color="lime">inner
1136 </font>style</font></u></i>
1137
1138 '''
1139 self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1140
1141 dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1142 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1143 <body>
1144 <div xml:lang="en">
1145 <p begin="0" end="1">Line 1</p>
1146 <p begin="1" end="2">第二行</p>
1147 </div>
1148 </body>
1149 </tt>'''.encode('utf-16')
1150 srt_data = '''1
1151 00:00:00,000 --> 00:00:01,000
1152 Line 1
1153
1154 2
1155 00:00:01,000 --> 00:00:02,000
1156 第二行
1157
1158 '''
1159 self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1160
1161 def test_cli_option(self):
1162 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1163 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1164 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1165 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1166
1167 def test_cli_valueless_option(self):
1168 self.assertEqual(cli_valueless_option(
1169 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1170 self.assertEqual(cli_valueless_option(
1171 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1172 self.assertEqual(cli_valueless_option(
1173 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1174 self.assertEqual(cli_valueless_option(
1175 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1176 self.assertEqual(cli_valueless_option(
1177 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1178 self.assertEqual(cli_valueless_option(
1179 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1180
1181 def test_cli_bool_option(self):
1182 self.assertEqual(
1183 cli_bool_option(
1184 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1185 ['--no-check-certificate', 'true'])
1186 self.assertEqual(
1187 cli_bool_option(
1188 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1189 ['--no-check-certificate=true'])
1190 self.assertEqual(
1191 cli_bool_option(
1192 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1193 ['--check-certificate', 'false'])
1194 self.assertEqual(
1195 cli_bool_option(
1196 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1197 ['--check-certificate=false'])
1198 self.assertEqual(
1199 cli_bool_option(
1200 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1201 ['--check-certificate', 'true'])
1202 self.assertEqual(
1203 cli_bool_option(
1204 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1205 ['--check-certificate=true'])
1206 self.assertEqual(
1207 cli_bool_option(
1208 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1209 [])
1210
1211 def test_ohdave_rsa_encrypt(self):
1212 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1213 e = 65537
1214
1215 self.assertEqual(
1216 ohdave_rsa_encrypt(b'aa111222', e, N),
1217 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1218
1219 def test_pkcs1pad(self):
1220 data = [1, 2, 3]
1221 padded_data = pkcs1pad(data, 32)
1222 self.assertEqual(padded_data[:2], [0, 2])
1223 self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1224
1225 self.assertRaises(ValueError, pkcs1pad, data, 8)
1226
1227 def test_encode_base_n(self):
1228 self.assertEqual(encode_base_n(0, 30), '0')
1229 self.assertEqual(encode_base_n(80, 30), '2k')
1230
1231 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1232 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1233 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1234
1235 self.assertRaises(ValueError, encode_base_n, 0, 70)
1236 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1237
1238 def test_urshift(self):
1239 self.assertEqual(urshift(3, 1), 1)
1240 self.assertEqual(urshift(-3, 1), 2147483646)
1241
1242 def test_get_element_by_class(self):
1243 html = '''
1244 <span class="foo bar">nice</span>
1245 '''
1246
1247 self.assertEqual(get_element_by_class('foo', html), 'nice')
1248 self.assertEqual(get_element_by_class('no-such-class', html), None)
1249
1250 def test_get_element_by_attribute(self):
1251 html = '''
1252 <span class="foo bar">nice</span>
1253 '''
1254
1255 self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1256 self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1257 self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1258
1259 html = '''
1260 <div itemprop="author" itemscope>foo</div>
1261 '''
1262
1263 self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1264
1265 def test_get_elements_by_class(self):
1266 html = '''
1267 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1268 '''
1269
1270 self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1271 self.assertEqual(get_elements_by_class('no-such-class', html), [])
1272
1273 def test_get_elements_by_attribute(self):
1274 html = '''
1275 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1276 '''
1277
1278 self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1279 self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1280 self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1281
1282
1283 if __name__ == '__main__':
1284 unittest.main()