0857c0fc0cef84b7e09a09ef77dc815d4a508cef
[youtube-dl] / 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         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
347
348     def test_determine_ext(self):
349         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
350         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
351         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
352         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
353         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
354
355     def test_find_xpath_attr(self):
356         testxml = '''<root>
357             <node/>
358             <node x="a"/>
359             <node x="a" y="c" />
360             <node x="b" y="d" />
361             <node x="" />
362         </root>'''
363         doc = compat_etree_fromstring(testxml)
364
365         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
366         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
367         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
368         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
369         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
370         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
371         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
372         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
373         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
374         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
375         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
376
377     def test_xpath_with_ns(self):
378         testxml = '''<root xmlns:media="http://example.com/">
379             <media:song>
380                 <media:author>The Author</media:author>
381                 <url>http://server.com/download.mp3</url>
382             </media:song>
383         </root>'''
384         doc = compat_etree_fromstring(testxml)
385         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
386         self.assertTrue(find('media:song') is not None)
387         self.assertEqual(find('media:song/media:author').text, 'The Author')
388         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
389
390     def test_xpath_element(self):
391         doc = xml.etree.ElementTree.Element('root')
392         div = xml.etree.ElementTree.SubElement(doc, 'div')
393         p = xml.etree.ElementTree.SubElement(div, 'p')
394         p.text = 'Foo'
395         self.assertEqual(xpath_element(doc, 'div/p'), p)
396         self.assertEqual(xpath_element(doc, ['div/p']), p)
397         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
398         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
399         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
400         self.assertTrue(xpath_element(doc, 'div/bar') is None)
401         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
402         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
403         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
404         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
405         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
406
407     def test_xpath_text(self):
408         testxml = '''<root>
409             <div>
410                 <p>Foo</p>
411             </div>
412         </root>'''
413         doc = compat_etree_fromstring(testxml)
414         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
415         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
416         self.assertTrue(xpath_text(doc, 'div/bar') is None)
417         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
418
419     def test_xpath_attr(self):
420         testxml = '''<root>
421             <div>
422                 <p x="a">Foo</p>
423             </div>
424         </root>'''
425         doc = compat_etree_fromstring(testxml)
426         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
427         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
428         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
429         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
430         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
431         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
432         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
433
434     def test_smuggle_url(self):
435         data = {"ö": "ö", "abc": [3]}
436         url = 'https://foo.bar/baz?x=y#a'
437         smug_url = smuggle_url(url, data)
438         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
439         self.assertEqual(url, unsmug_url)
440         self.assertEqual(data, unsmug_data)
441
442         res_url, res_data = unsmuggle_url(url)
443         self.assertEqual(res_url, url)
444         self.assertEqual(res_data, None)
445
446         smug_url = smuggle_url(url, {'a': 'b'})
447         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
448         res_url, res_data = unsmuggle_url(smug_smug_url)
449         self.assertEqual(res_url, url)
450         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
451
452     def test_shell_quote(self):
453         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
454         self.assertEqual(
455             shell_quote(args),
456             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
457
458     def test_str_to_int(self):
459         self.assertEqual(str_to_int('123,456'), 123456)
460         self.assertEqual(str_to_int('123.456'), 123456)
461
462     def test_url_basename(self):
463         self.assertEqual(url_basename('http://foo.de/'), '')
464         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
465         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
466         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
467         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
468         self.assertEqual(
469             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
470             'trailer.mp4')
471
472     def test_base_url(self):
473         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
474         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
475         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
476         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
477         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
478
479     def test_urljoin(self):
480         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
481         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
482         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
483         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
484         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//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', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
488         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
489         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
490         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
491         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
492         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
493         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
494         self.assertEqual(urljoin('http://foo.de/', None), None)
495         self.assertEqual(urljoin('http://foo.de/', ''), None)
496         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
497         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
498
499     def test_parse_age_limit(self):
500         self.assertEqual(parse_age_limit(None), None)
501         self.assertEqual(parse_age_limit(False), None)
502         self.assertEqual(parse_age_limit('invalid'), None)
503         self.assertEqual(parse_age_limit(0), 0)
504         self.assertEqual(parse_age_limit(18), 18)
505         self.assertEqual(parse_age_limit(21), 21)
506         self.assertEqual(parse_age_limit(22), None)
507         self.assertEqual(parse_age_limit('18'), 18)
508         self.assertEqual(parse_age_limit('18+'), 18)
509         self.assertEqual(parse_age_limit('PG-13'), 13)
510         self.assertEqual(parse_age_limit('TV-14'), 14)
511         self.assertEqual(parse_age_limit('TV-MA'), 17)
512
513     def test_parse_duration(self):
514         self.assertEqual(parse_duration(None), None)
515         self.assertEqual(parse_duration(False), None)
516         self.assertEqual(parse_duration('invalid'), None)
517         self.assertEqual(parse_duration('1'), 1)
518         self.assertEqual(parse_duration('1337:12'), 80232)
519         self.assertEqual(parse_duration('9:12:43'), 33163)
520         self.assertEqual(parse_duration('12:00'), 720)
521         self.assertEqual(parse_duration('00:01:01'), 61)
522         self.assertEqual(parse_duration('x:y'), None)
523         self.assertEqual(parse_duration('3h11m53s'), 11513)
524         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
525         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
526         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
527         self.assertEqual(parse_duration('62m45s'), 3765)
528         self.assertEqual(parse_duration('6m59s'), 419)
529         self.assertEqual(parse_duration('49s'), 49)
530         self.assertEqual(parse_duration('0h0m0s'), 0)
531         self.assertEqual(parse_duration('0m0s'), 0)
532         self.assertEqual(parse_duration('0s'), 0)
533         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
534         self.assertEqual(parse_duration('T30M38S'), 1838)
535         self.assertEqual(parse_duration('5 s'), 5)
536         self.assertEqual(parse_duration('3 min'), 180)
537         self.assertEqual(parse_duration('2.5 hours'), 9000)
538         self.assertEqual(parse_duration('02:03:04'), 7384)
539         self.assertEqual(parse_duration('01:02:03:04'), 93784)
540         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
541         self.assertEqual(parse_duration('87 Min.'), 5220)
542         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
543         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
544         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
545
546     def test_fix_xml_ampersands(self):
547         self.assertEqual(
548             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
549         self.assertEqual(
550             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
551             '"&amp;x=y&amp;wrong;&amp;z=a')
552         self.assertEqual(
553             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
554             '&amp;&apos;&gt;&lt;&quot;')
555         self.assertEqual(
556             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
557         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
558
559     def test_paged_list(self):
560         def testPL(size, pagesize, sliceargs, expected):
561             def get_page(pagenum):
562                 firstid = pagenum * pagesize
563                 upto = min(size, pagenum * pagesize + pagesize)
564                 for i in range(firstid, upto):
565                     yield i
566
567             pl = OnDemandPagedList(get_page, pagesize)
568             got = pl.getslice(*sliceargs)
569             self.assertEqual(got, expected)
570
571             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
572             got = iapl.getslice(*sliceargs)
573             self.assertEqual(got, expected)
574
575         testPL(5, 2, (), [0, 1, 2, 3, 4])
576         testPL(5, 2, (1,), [1, 2, 3, 4])
577         testPL(5, 2, (2,), [2, 3, 4])
578         testPL(5, 2, (4,), [4])
579         testPL(5, 2, (0, 3), [0, 1, 2])
580         testPL(5, 2, (1, 4), [1, 2, 3])
581         testPL(5, 2, (2, 99), [2, 3, 4])
582         testPL(5, 2, (20, 99), [])
583
584     def test_read_batch_urls(self):
585         f = io.StringIO('''\xef\xbb\xbf foo
586             bar\r
587             baz
588             # More after this line\r
589             ; or after this
590             bam''')
591         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
592
593     def test_urlencode_postdata(self):
594         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
595         self.assertTrue(isinstance(data, bytes))
596
597     def test_update_url_query(self):
598         def query_dict(url):
599             return compat_parse_qs(compat_urlparse.urlparse(url).query)
600         self.assertEqual(query_dict(update_url_query(
601             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
602             query_dict('http://example.com/path?quality=HD&format=mp4'))
603         self.assertEqual(query_dict(update_url_query(
604             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
605             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
606         self.assertEqual(query_dict(update_url_query(
607             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
608             query_dict('http://example.com/path?fields=id,formats,subtitles'))
609         self.assertEqual(query_dict(update_url_query(
610             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
611             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
612         self.assertEqual(query_dict(update_url_query(
613             'http://example.com/path?manifest=f4m', {'manifest': []})),
614             query_dict('http://example.com/path'))
615         self.assertEqual(query_dict(update_url_query(
616             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
617             query_dict('http://example.com/path?system=LINUX'))
618         self.assertEqual(query_dict(update_url_query(
619             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
620             query_dict('http://example.com/path?fields=id,formats,subtitles'))
621         self.assertEqual(query_dict(update_url_query(
622             'http://example.com/path', {'width': 1080, 'height': 720})),
623             query_dict('http://example.com/path?width=1080&height=720'))
624         self.assertEqual(query_dict(update_url_query(
625             'http://example.com/path', {'bitrate': 5020.43})),
626             query_dict('http://example.com/path?bitrate=5020.43'))
627         self.assertEqual(query_dict(update_url_query(
628             'http://example.com/path', {'test': '第二行тест'})),
629             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
630
631     def test_multipart_encode(self):
632         self.assertEqual(
633             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
634             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
635         self.assertEqual(
636             multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
637             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')
638         self.assertRaises(
639             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
640
641     def test_dict_get(self):
642         FALSE_VALUES = {
643             'none': None,
644             'false': False,
645             'zero': 0,
646             'empty_string': '',
647             'empty_list': [],
648         }
649         d = FALSE_VALUES.copy()
650         d['a'] = 42
651         self.assertEqual(dict_get(d, 'a'), 42)
652         self.assertEqual(dict_get(d, 'b'), None)
653         self.assertEqual(dict_get(d, 'b', 42), 42)
654         self.assertEqual(dict_get(d, ('a', )), 42)
655         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
656         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
657         self.assertEqual(dict_get(d, ('b', 'c', )), None)
658         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
659         for key, false_value in FALSE_VALUES.items():
660             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
661             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
662
663     def test_encode_compat_str(self):
664         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
665         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
666
667     def test_parse_iso8601(self):
668         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
669         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
670         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
671         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
672         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
673         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
674
675     def test_strip_jsonp(self):
676         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
677         d = json.loads(stripped)
678         self.assertEqual(d, [{"id": "532cb", "x": 3}])
679
680         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
681         d = json.loads(stripped)
682         self.assertEqual(d, {'STATUS': 'OK'})
683
684         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
685         d = json.loads(stripped)
686         self.assertEqual(d, {'status': 'success'})
687
688         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
689         d = json.loads(stripped)
690         self.assertEqual(d, {'status': 'success'})
691
692         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
693         d = json.loads(stripped)
694         self.assertEqual(d, {'status': 'success'})
695
696     def test_uppercase_escape(self):
697         self.assertEqual(uppercase_escape('aä'), 'aä')
698         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
699
700     def test_lowercase_escape(self):
701         self.assertEqual(lowercase_escape('aä'), 'aä')
702         self.assertEqual(lowercase_escape('\\u0026'), '&')
703
704     def test_limit_length(self):
705         self.assertEqual(limit_length(None, 12), None)
706         self.assertEqual(limit_length('foo', 12), 'foo')
707         self.assertTrue(
708             limit_length('foo bar baz asd', 12).startswith('foo bar'))
709         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
710
711     def test_mimetype2ext(self):
712         self.assertEqual(mimetype2ext(None), None)
713         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
714         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
715         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
716         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
717         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
718
719     def test_month_by_name(self):
720         self.assertEqual(month_by_name(None), None)
721         self.assertEqual(month_by_name('December', 'en'), 12)
722         self.assertEqual(month_by_name('décembre', 'fr'), 12)
723         self.assertEqual(month_by_name('December'), 12)
724         self.assertEqual(month_by_name('décembre'), None)
725         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
726
727     def test_parse_codecs(self):
728         self.assertEqual(parse_codecs(''), {})
729         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
730             'vcodec': 'avc1.77.30',
731             'acodec': 'mp4a.40.2',
732         })
733         self.assertEqual(parse_codecs('mp4a.40.2'), {
734             'vcodec': 'none',
735             'acodec': 'mp4a.40.2',
736         })
737         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
738             'vcodec': 'avc1.42001e',
739             'acodec': 'mp4a.40.5',
740         })
741         self.assertEqual(parse_codecs('avc3.640028'), {
742             'vcodec': 'avc3.640028',
743             'acodec': 'none',
744         })
745         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
746             'vcodec': 'h264',
747             'acodec': 'aac',
748         })
749
750     def test_escape_rfc3986(self):
751         reserved = "!*'();:@&=+$,/?#[]"
752         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
753         self.assertEqual(escape_rfc3986(reserved), reserved)
754         self.assertEqual(escape_rfc3986(unreserved), unreserved)
755         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
756         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
757         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
758         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
759
760     def test_escape_url(self):
761         self.assertEqual(
762             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
763             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
764         )
765         self.assertEqual(
766             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
767             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
768         )
769         self.assertEqual(
770             escape_url('http://тест.рф/фрагмент'),
771             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
772         )
773         self.assertEqual(
774             escape_url('http://тест.рф/абв?абв=абв#абв'),
775             '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'
776         )
777         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
778
779     def test_js_to_json_realworld(self):
780         inp = '''{
781             'clip':{'provider':'pseudo'}
782         }'''
783         self.assertEqual(js_to_json(inp), '''{
784             "clip":{"provider":"pseudo"}
785         }''')
786         json.loads(js_to_json(inp))
787
788         inp = '''{
789             'playlist':[{'controls':{'all':null}}]
790         }'''
791         self.assertEqual(js_to_json(inp), '''{
792             "playlist":[{"controls":{"all":null}}]
793         }''')
794
795         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
796         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
797
798         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
799         json_code = js_to_json(inp)
800         self.assertEqual(json.loads(json_code), json.loads(inp))
801
802         inp = '''{
803             0:{src:'skipped', type: 'application/dash+xml'},
804             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
805         }'''
806         self.assertEqual(js_to_json(inp), '''{
807             "0":{"src":"skipped", "type": "application/dash+xml"},
808             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
809         }''')
810
811         inp = '''{"foo":101}'''
812         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
813
814         inp = '''{"duration": "00:01:07"}'''
815         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
816
817     def test_js_to_json_edgecases(self):
818         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
819         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
820
821         on = js_to_json('{"abc": true}')
822         self.assertEqual(json.loads(on), {'abc': True})
823
824         # Ignore JavaScript code as well
825         on = js_to_json('''{
826             "x": 1,
827             y: "a",
828             z: some.code
829         }''')
830         d = json.loads(on)
831         self.assertEqual(d['x'], 1)
832         self.assertEqual(d['y'], 'a')
833
834         on = js_to_json('["abc", "def",]')
835         self.assertEqual(json.loads(on), ['abc', 'def'])
836
837         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
838         self.assertEqual(json.loads(on), ['abc', 'def'])
839
840         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
841         self.assertEqual(json.loads(on), ['abc', 'def'])
842
843         on = js_to_json('{"abc": "def",}')
844         self.assertEqual(json.loads(on), {'abc': 'def'})
845
846         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
847         self.assertEqual(json.loads(on), {'abc': 'def'})
848
849         on = js_to_json('{ 0: /* " \n */ ",]" , }')
850         self.assertEqual(json.loads(on), {'0': ',]'})
851
852         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
853         self.assertEqual(json.loads(on), {'0': ',]'})
854
855         on = js_to_json('{ 0: // comment\n1 }')
856         self.assertEqual(json.loads(on), {'0': 1})
857
858         on = js_to_json(r'["<p>x<\/p>"]')
859         self.assertEqual(json.loads(on), ['<p>x</p>'])
860
861         on = js_to_json(r'["\xaa"]')
862         self.assertEqual(json.loads(on), ['\u00aa'])
863
864         on = js_to_json("['a\\\nb']")
865         self.assertEqual(json.loads(on), ['ab'])
866
867         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
868         self.assertEqual(json.loads(on), ['ab'])
869
870         on = js_to_json('{0xff:0xff}')
871         self.assertEqual(json.loads(on), {'255': 255})
872
873         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
874         self.assertEqual(json.loads(on), {'255': 255})
875
876         on = js_to_json('{077:077}')
877         self.assertEqual(json.loads(on), {'63': 63})
878
879         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
880         self.assertEqual(json.loads(on), {'63': 63})
881
882         on = js_to_json('{42:42}')
883         self.assertEqual(json.loads(on), {'42': 42})
884
885         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
886         self.assertEqual(json.loads(on), {'42': 42})
887
888     def test_extract_attributes(self):
889         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
890         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
891         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
892         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
893         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
894         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
895         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
896         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
897         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
898         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
899         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
900         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
901         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
902         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
903         self.assertEqual(extract_attributes('<e x >'), {'x': None})
904         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
905         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
906         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
907         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
908         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
909         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
910         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
911         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
912         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
913         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
914         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
915         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
916         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
917         # "Narrow" Python builds don't support unicode code points outside BMP.
918         try:
919             compat_chr(0x10000)
920             supports_outside_bmp = True
921         except ValueError:
922             supports_outside_bmp = False
923         if supports_outside_bmp:
924             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
925         # Malformed HTML should not break attributes extraction on older Python
926         self.assertEqual(extract_attributes('<mal"formed/>'), {})
927
928     def test_clean_html(self):
929         self.assertEqual(clean_html('a:\nb'), 'a: b')
930         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
931         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
932
933     def test_intlist_to_bytes(self):
934         self.assertEqual(
935             intlist_to_bytes([0, 1, 127, 128, 255]),
936             b'\x00\x01\x7f\x80\xff')
937
938     def test_args_to_str(self):
939         self.assertEqual(
940             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
941             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
942         )
943
944     def test_parse_filesize(self):
945         self.assertEqual(parse_filesize(None), None)
946         self.assertEqual(parse_filesize(''), None)
947         self.assertEqual(parse_filesize('91 B'), 91)
948         self.assertEqual(parse_filesize('foobar'), None)
949         self.assertEqual(parse_filesize('2 MiB'), 2097152)
950         self.assertEqual(parse_filesize('5 GB'), 5000000000)
951         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
952         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
953         self.assertEqual(parse_filesize('1,24 KB'), 1240)
954         self.assertEqual(parse_filesize('1,24 kb'), 1240)
955         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
956
957     def test_parse_count(self):
958         self.assertEqual(parse_count(None), None)
959         self.assertEqual(parse_count(''), None)
960         self.assertEqual(parse_count('0'), 0)
961         self.assertEqual(parse_count('1000'), 1000)
962         self.assertEqual(parse_count('1.000'), 1000)
963         self.assertEqual(parse_count('1.1k'), 1100)
964         self.assertEqual(parse_count('1.1kk'), 1100000)
965         self.assertEqual(parse_count('1.1kk '), 1100000)
966         self.assertEqual(parse_count('1.1kk views'), 1100000)
967
968     def test_version_tuple(self):
969         self.assertEqual(version_tuple('1'), (1,))
970         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
971         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
972
973     def test_detect_exe_version(self):
974         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
975 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
976 configuration: --prefix=/usr --extra-'''), '1.2.1')
977         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
978 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
979         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
980 Trying to open render node...
981 Success at /dev/dri/renderD128.
982 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
983
984     def test_age_restricted(self):
985         self.assertFalse(age_restricted(None, 10))  # unrestricted content
986         self.assertFalse(age_restricted(1, None))  # unrestricted policy
987         self.assertFalse(age_restricted(8, 10))
988         self.assertTrue(age_restricted(18, 14))
989         self.assertFalse(age_restricted(18, 18))
990
991     def test_is_html(self):
992         self.assertFalse(is_html(b'\x49\x44\x43<html'))
993         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
994         self.assertTrue(is_html(  # UTF-8 with BOM
995             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
996         self.assertTrue(is_html(  # UTF-16-LE
997             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
998         ))
999         self.assertTrue(is_html(  # UTF-16-BE
1000             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
1001         ))
1002         self.assertTrue(is_html(  # UTF-32-BE
1003             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'))
1004         self.assertTrue(is_html(  # UTF-32-LE
1005             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'))
1006
1007     def test_render_table(self):
1008         self.assertEqual(
1009             render_table(
1010                 ['a', 'bcd'],
1011                 [[123, 4], [9999, 51]]),
1012             'a    bcd\n'
1013             '123  4\n'
1014             '9999 51')
1015
1016     def test_match_str(self):
1017         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1018         self.assertFalse(match_str('xy', {'x': 1200}))
1019         self.assertTrue(match_str('!xy', {'x': 1200}))
1020         self.assertTrue(match_str('x', {'x': 1200}))
1021         self.assertFalse(match_str('!x', {'x': 1200}))
1022         self.assertTrue(match_str('x', {'x': 0}))
1023         self.assertFalse(match_str('x>0', {'x': 0}))
1024         self.assertFalse(match_str('x>0', {}))
1025         self.assertTrue(match_str('x>?0', {}))
1026         self.assertTrue(match_str('x>1K', {'x': 1200}))
1027         self.assertFalse(match_str('x>2K', {'x': 1200}))
1028         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1029         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1030         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1031         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1032         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1033         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1034         self.assertFalse(match_str(
1035             'like_count > 100 & dislike_count <? 50 & description',
1036             {'like_count': 90, 'description': 'foo'}))
1037         self.assertTrue(match_str(
1038             'like_count > 100 & dislike_count <? 50 & description',
1039             {'like_count': 190, 'description': 'foo'}))
1040         self.assertFalse(match_str(
1041             'like_count > 100 & dislike_count <? 50 & description',
1042             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1043         self.assertFalse(match_str(
1044             'like_count > 100 & dislike_count <? 50 & description',
1045             {'like_count': 190, 'dislike_count': 10}))
1046
1047     def test_parse_dfxp_time_expr(self):
1048         self.assertEqual(parse_dfxp_time_expr(None), None)
1049         self.assertEqual(parse_dfxp_time_expr(''), None)
1050         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1051         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1052         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1053         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1054         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1055
1056     def test_dfxp2srt(self):
1057         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1058             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1059             <body>
1060                 <div xml:lang="en">
1061                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1062                     <p begin="1" end="2">第二行<br/>♪♪</p>
1063                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1064                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1065                     <p begin="-1" end="-1">Ignore, two</p>
1066                     <p begin="3" dur="-1">Ignored, three</p>
1067                 </div>
1068             </body>
1069             </tt>'''.encode('utf-8')
1070         srt_data = '''1
1071 00:00:00,000 --> 00:00:01,000
1072 The following line contains Chinese characters and special symbols
1073
1074 2
1075 00:00:01,000 --> 00:00:02,000
1076 第二行
1077 ♪♪
1078
1079 3
1080 00:00:02,000 --> 00:00:03,000
1081 Third
1082 Line
1083
1084 '''
1085         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1086
1087         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1088             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1089             <body>
1090                 <div xml:lang="en">
1091                     <p begin="0" end="1">The first line</p>
1092                 </div>
1093             </body>
1094             </tt>'''.encode('utf-8')
1095         srt_data = '''1
1096 00:00:00,000 --> 00:00:01,000
1097 The first line
1098
1099 '''
1100         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1101
1102         dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1103 <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">
1104   <head>
1105     <styling>
1106       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1107       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1108       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1109       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1110     </styling>
1111   </head>
1112   <body tts:textAlign="center" style="s0">
1113     <div>
1114       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1115       <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>
1116       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1117       <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>
1118     </div>
1119   </body>
1120 </tt>'''.encode('utf-8')
1121         srt_data = '''1
1122 00:00:02,080 --> 00:00:05,839
1123 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1124
1125 2
1126 00:00:02,080 --> 00:00:05,839
1127 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1128 </font>part 2</font></b>
1129
1130 3
1131 00:00:05,839 --> 00:00:09,560
1132 <u><font color="lime">line 3
1133 part 3</font></u>
1134
1135 4
1136 00:00:09,560 --> 00:00:12,359
1137 <i><u><font color="yellow"><font color="lime">inner
1138  </font>style</font></u></i>
1139
1140 '''
1141         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1142
1143         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
1144             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1145             <body>
1146                 <div xml:lang="en">
1147                     <p begin="0" end="1">Line 1</p>
1148                     <p begin="1" end="2">第二行</p>
1149                 </div>
1150             </body>
1151             </tt>'''.encode('utf-16')
1152         srt_data = '''1
1153 00:00:00,000 --> 00:00:01,000
1154 Line 1
1155
1156 2
1157 00:00:01,000 --> 00:00:02,000
1158 第二行
1159
1160 '''
1161         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
1162
1163     def test_cli_option(self):
1164         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1165         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1166         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1167         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1168
1169     def test_cli_valueless_option(self):
1170         self.assertEqual(cli_valueless_option(
1171             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1172         self.assertEqual(cli_valueless_option(
1173             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1174         self.assertEqual(cli_valueless_option(
1175             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1176         self.assertEqual(cli_valueless_option(
1177             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1178         self.assertEqual(cli_valueless_option(
1179             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1180         self.assertEqual(cli_valueless_option(
1181             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1182
1183     def test_cli_bool_option(self):
1184         self.assertEqual(
1185             cli_bool_option(
1186                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1187             ['--no-check-certificate', 'true'])
1188         self.assertEqual(
1189             cli_bool_option(
1190                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1191             ['--no-check-certificate=true'])
1192         self.assertEqual(
1193             cli_bool_option(
1194                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1195             ['--check-certificate', 'false'])
1196         self.assertEqual(
1197             cli_bool_option(
1198                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1199             ['--check-certificate=false'])
1200         self.assertEqual(
1201             cli_bool_option(
1202                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1203             ['--check-certificate', 'true'])
1204         self.assertEqual(
1205             cli_bool_option(
1206                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1207             ['--check-certificate=true'])
1208         self.assertEqual(
1209             cli_bool_option(
1210                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1211             [])
1212
1213     def test_ohdave_rsa_encrypt(self):
1214         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1215         e = 65537
1216
1217         self.assertEqual(
1218             ohdave_rsa_encrypt(b'aa111222', e, N),
1219             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1220
1221     def test_pkcs1pad(self):
1222         data = [1, 2, 3]
1223         padded_data = pkcs1pad(data, 32)
1224         self.assertEqual(padded_data[:2], [0, 2])
1225         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1226
1227         self.assertRaises(ValueError, pkcs1pad, data, 8)
1228
1229     def test_encode_base_n(self):
1230         self.assertEqual(encode_base_n(0, 30), '0')
1231         self.assertEqual(encode_base_n(80, 30), '2k')
1232
1233         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1234         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1235         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1236
1237         self.assertRaises(ValueError, encode_base_n, 0, 70)
1238         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1239
1240     def test_urshift(self):
1241         self.assertEqual(urshift(3, 1), 1)
1242         self.assertEqual(urshift(-3, 1), 2147483646)
1243
1244     def test_get_element_by_class(self):
1245         html = '''
1246             <span class="foo bar">nice</span>
1247         '''
1248
1249         self.assertEqual(get_element_by_class('foo', html), 'nice')
1250         self.assertEqual(get_element_by_class('no-such-class', html), None)
1251
1252     def test_get_element_by_attribute(self):
1253         html = '''
1254             <span class="foo bar">nice</span>
1255         '''
1256
1257         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1258         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1259         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1260
1261         html = '''
1262             <div itemprop="author" itemscope>foo</div>
1263         '''
1264
1265         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
1266
1267     def test_get_elements_by_class(self):
1268         html = '''
1269             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1270         '''
1271
1272         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1273         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1274
1275     def test_get_elements_by_attribute(self):
1276         html = '''
1277             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1278         '''
1279
1280         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1281         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1282         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1283
1284
1285 if __name__ == '__main__':
1286     unittest.main()