]> gitweb @ CieloNegro.org - youtube-dl.git/blob - test/test_utils.py
Merge pull request #12861 from Tithen-Firion/cbsinteractive-fix
[youtube-dl.git] / 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_setenv,
102     compat_urlparse,
103     compat_parse_qs,
104 )
105
106
107 class TestUtil(unittest.TestCase):
108     def test_timeconvert(self):
109         self.assertTrue(timeconvert('') is None)
110         self.assertTrue(timeconvert('bougrg') is None)
111
112     def test_sanitize_filename(self):
113         self.assertEqual(sanitize_filename('abc'), 'abc')
114         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
115
116         self.assertEqual(sanitize_filename('123'), '123')
117
118         self.assertEqual('abc_de', sanitize_filename('abc/de'))
119         self.assertFalse('/' in sanitize_filename('abc/de///'))
120
121         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
122         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
123         self.assertEqual('yes no', sanitize_filename('yes? no'))
124         self.assertEqual('this - that', sanitize_filename('this: that'))
125
126         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
127         aumlaut = 'ä'
128         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
129         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
130         self.assertEqual(sanitize_filename(tests), tests)
131
132         self.assertEqual(
133             sanitize_filename('New World record at 0:12:34'),
134             'New World record at 0_12_34')
135
136         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
137         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
138         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
139         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
140
141         forbidden = '"\0\\/'
142         for fc in forbidden:
143             for fbc in forbidden:
144                 self.assertTrue(fbc not in sanitize_filename(fc))
145
146     def test_sanitize_filename_restricted(self):
147         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
148         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
149
150         self.assertEqual(sanitize_filename('123', restricted=True), '123')
151
152         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
153         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
154
155         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
156         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
157         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
158         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
159
160         tests = 'aäb\u4e2d\u56fd\u7684c'
161         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
162         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
163
164         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
165         for fc in forbidden:
166             for fbc in forbidden:
167                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
168
169         # Handle a common case more neatly
170         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
171         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
172         # .. but make sure the file name is never empty
173         self.assertTrue(sanitize_filename('-', restricted=True) != '')
174         self.assertTrue(sanitize_filename(':', restricted=True) != '')
175
176         self.assertEqual(sanitize_filename(
177             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
178             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
179
180     def test_sanitize_ids(self):
181         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
182         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
183         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
184
185     def test_sanitize_path(self):
186         if sys.platform != 'win32':
187             return
188
189         self.assertEqual(sanitize_path('abc'), 'abc')
190         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
191         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
192         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
193         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
194         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
195         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
196
197         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
198         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
199
200         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
201         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
202         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
203         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
204
205         self.assertEqual(
206             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
207             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
208
209         self.assertEqual(
210             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
211             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
212         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
213         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
214         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
215
216         self.assertEqual(sanitize_path('../abc'), '..\\abc')
217         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
218         self.assertEqual(sanitize_path('./abc'), 'abc')
219         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
220
221     def test_expand_path(self):
222         def env(var):
223             return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
224
225         compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
226         self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
227         self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
228         self.assertEqual(expand_path('~'), compat_getenv('HOME'))
229         self.assertEqual(
230             expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
231             '%s/expanded' % compat_getenv('HOME'))
232
233     def test_prepend_extension(self):
234         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
235         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
236         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
237         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
238         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
239         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
240
241     def test_replace_extension(self):
242         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
243         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
244         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
245         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
246         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
247         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
248
249     def test_remove_start(self):
250         self.assertEqual(remove_start(None, 'A - '), None)
251         self.assertEqual(remove_start('A - B', 'A - '), 'B')
252         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
253
254     def test_remove_end(self):
255         self.assertEqual(remove_end(None, ' - B'), None)
256         self.assertEqual(remove_end('A - B', ' - B'), 'A')
257         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
258
259     def test_remove_quotes(self):
260         self.assertEqual(remove_quotes(None), None)
261         self.assertEqual(remove_quotes('"'), '"')
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
268     def test_ordered_set(self):
269         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
270         self.assertEqual(orderedSet([]), [])
271         self.assertEqual(orderedSet([1]), [1])
272         # keep the list ordered
273         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
274
275     def test_unescape_html(self):
276         self.assertEqual(unescapeHTML('%20;'), '%20;')
277         self.assertEqual(unescapeHTML('&#x2F;'), '/')
278         self.assertEqual(unescapeHTML('&#47;'), '/')
279         self.assertEqual(unescapeHTML('&eacute;'), 'é')
280         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
281         # HTML5 entities
282         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
283
284     def test_date_from_str(self):
285         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
286         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
287         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
288         self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
289         self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
290
291     def test_daterange(self):
292         _20century = DateRange("19000101", "20000101")
293         self.assertFalse("17890714" in _20century)
294         _ac = DateRange("00010101")
295         self.assertTrue("19690721" in _ac)
296         _firstmilenium = DateRange(end="10000101")
297         self.assertTrue("07110427" in _firstmilenium)
298
299     def test_unified_dates(self):
300         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
301         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
302         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
303         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
304         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
305         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
306         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
307         self.assertEqual(
308             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
309             '20141126')
310         self.assertEqual(
311             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
312             '20150202')
313         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
314         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
315         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
316         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
317         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
318         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
319         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
320         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
321
322     def test_unified_timestamps(self):
323         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
324         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
325         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
326         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
327         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
328         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
329         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
330         self.assertEqual(
331             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
332             1417001400)
333         self.assertEqual(
334             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
335             1422902860)
336         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
337         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
338         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
339         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
340         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
341         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
342         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
343
344     def test_determine_ext(self):
345         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
346         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
347         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
348         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
349         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
350
351     def test_find_xpath_attr(self):
352         testxml = '''<root>
353             <node/>
354             <node x="a"/>
355             <node x="a" y="c" />
356             <node x="b" y="d" />
357             <node x="" />
358         </root>'''
359         doc = compat_etree_fromstring(testxml)
360
361         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
362         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
363         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
364         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
365         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
366         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
367         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
368         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
369         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
370         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
371         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
372
373     def test_xpath_with_ns(self):
374         testxml = '''<root xmlns:media="http://example.com/">
375             <media:song>
376                 <media:author>The Author</media:author>
377                 <url>http://server.com/download.mp3</url>
378             </media:song>
379         </root>'''
380         doc = compat_etree_fromstring(testxml)
381         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
382         self.assertTrue(find('media:song') is not None)
383         self.assertEqual(find('media:song/media:author').text, 'The Author')
384         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
385
386     def test_xpath_element(self):
387         doc = xml.etree.ElementTree.Element('root')
388         div = xml.etree.ElementTree.SubElement(doc, 'div')
389         p = xml.etree.ElementTree.SubElement(div, 'p')
390         p.text = 'Foo'
391         self.assertEqual(xpath_element(doc, 'div/p'), p)
392         self.assertEqual(xpath_element(doc, ['div/p']), p)
393         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
394         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
395         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
396         self.assertTrue(xpath_element(doc, 'div/bar') is None)
397         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
398         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
399         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
400         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
401         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
402
403     def test_xpath_text(self):
404         testxml = '''<root>
405             <div>
406                 <p>Foo</p>
407             </div>
408         </root>'''
409         doc = compat_etree_fromstring(testxml)
410         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
411         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
412         self.assertTrue(xpath_text(doc, 'div/bar') is None)
413         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
414
415     def test_xpath_attr(self):
416         testxml = '''<root>
417             <div>
418                 <p x="a">Foo</p>
419             </div>
420         </root>'''
421         doc = compat_etree_fromstring(testxml)
422         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
423         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
424         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
425         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
426         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
427         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
428         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
429
430     def test_smuggle_url(self):
431         data = {"ö": "ö", "abc": [3]}
432         url = 'https://foo.bar/baz?x=y#a'
433         smug_url = smuggle_url(url, data)
434         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
435         self.assertEqual(url, unsmug_url)
436         self.assertEqual(data, unsmug_data)
437
438         res_url, res_data = unsmuggle_url(url)
439         self.assertEqual(res_url, url)
440         self.assertEqual(res_data, None)
441
442         smug_url = smuggle_url(url, {'a': 'b'})
443         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
444         res_url, res_data = unsmuggle_url(smug_smug_url)
445         self.assertEqual(res_url, url)
446         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
447
448     def test_shell_quote(self):
449         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
450         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
451
452     def test_str_to_int(self):
453         self.assertEqual(str_to_int('123,456'), 123456)
454         self.assertEqual(str_to_int('123.456'), 123456)
455
456     def test_url_basename(self):
457         self.assertEqual(url_basename('http://foo.de/'), '')
458         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
459         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
460         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
461         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
462         self.assertEqual(
463             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
464             'trailer.mp4')
465
466     def test_base_url(self):
467         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
468         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
469         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
470         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
471         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
472
473     def test_urljoin(self):
474         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
475         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
476         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
477         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
478         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
479         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
480         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
481         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
482         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
483         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
484         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
485         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//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(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
488         self.assertEqual(urljoin('http://foo.de/', None), None)
489         self.assertEqual(urljoin('http://foo.de/', ''), None)
490         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
491         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
492
493     def test_parse_age_limit(self):
494         self.assertEqual(parse_age_limit(None), None)
495         self.assertEqual(parse_age_limit(False), None)
496         self.assertEqual(parse_age_limit('invalid'), None)
497         self.assertEqual(parse_age_limit(0), 0)
498         self.assertEqual(parse_age_limit(18), 18)
499         self.assertEqual(parse_age_limit(21), 21)
500         self.assertEqual(parse_age_limit(22), None)
501         self.assertEqual(parse_age_limit('18'), 18)
502         self.assertEqual(parse_age_limit('18+'), 18)
503         self.assertEqual(parse_age_limit('PG-13'), 13)
504         self.assertEqual(parse_age_limit('TV-14'), 14)
505         self.assertEqual(parse_age_limit('TV-MA'), 17)
506
507     def test_parse_duration(self):
508         self.assertEqual(parse_duration(None), None)
509         self.assertEqual(parse_duration(False), None)
510         self.assertEqual(parse_duration('invalid'), None)
511         self.assertEqual(parse_duration('1'), 1)
512         self.assertEqual(parse_duration('1337:12'), 80232)
513         self.assertEqual(parse_duration('9:12:43'), 33163)
514         self.assertEqual(parse_duration('12:00'), 720)
515         self.assertEqual(parse_duration('00:01:01'), 61)
516         self.assertEqual(parse_duration('x:y'), None)
517         self.assertEqual(parse_duration('3h11m53s'), 11513)
518         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
519         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
520         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
521         self.assertEqual(parse_duration('62m45s'), 3765)
522         self.assertEqual(parse_duration('6m59s'), 419)
523         self.assertEqual(parse_duration('49s'), 49)
524         self.assertEqual(parse_duration('0h0m0s'), 0)
525         self.assertEqual(parse_duration('0m0s'), 0)
526         self.assertEqual(parse_duration('0s'), 0)
527         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
528         self.assertEqual(parse_duration('T30M38S'), 1838)
529         self.assertEqual(parse_duration('5 s'), 5)
530         self.assertEqual(parse_duration('3 min'), 180)
531         self.assertEqual(parse_duration('2.5 hours'), 9000)
532         self.assertEqual(parse_duration('02:03:04'), 7384)
533         self.assertEqual(parse_duration('01:02:03:04'), 93784)
534         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
535         self.assertEqual(parse_duration('87 Min.'), 5220)
536         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
537         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
538
539     def test_fix_xml_ampersands(self):
540         self.assertEqual(
541             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
542         self.assertEqual(
543             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
544             '"&amp;x=y&amp;wrong;&amp;z=a')
545         self.assertEqual(
546             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
547             '&amp;&apos;&gt;&lt;&quot;')
548         self.assertEqual(
549             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
550         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
551
552     def test_paged_list(self):
553         def testPL(size, pagesize, sliceargs, expected):
554             def get_page(pagenum):
555                 firstid = pagenum * pagesize
556                 upto = min(size, pagenum * pagesize + pagesize)
557                 for i in range(firstid, upto):
558                     yield i
559
560             pl = OnDemandPagedList(get_page, pagesize)
561             got = pl.getslice(*sliceargs)
562             self.assertEqual(got, expected)
563
564             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
565             got = iapl.getslice(*sliceargs)
566             self.assertEqual(got, expected)
567
568         testPL(5, 2, (), [0, 1, 2, 3, 4])
569         testPL(5, 2, (1,), [1, 2, 3, 4])
570         testPL(5, 2, (2,), [2, 3, 4])
571         testPL(5, 2, (4,), [4])
572         testPL(5, 2, (0, 3), [0, 1, 2])
573         testPL(5, 2, (1, 4), [1, 2, 3])
574         testPL(5, 2, (2, 99), [2, 3, 4])
575         testPL(5, 2, (20, 99), [])
576
577     def test_read_batch_urls(self):
578         f = io.StringIO('''\xef\xbb\xbf foo
579             bar\r
580             baz
581             # More after this line\r
582             ; or after this
583             bam''')
584         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
585
586     def test_urlencode_postdata(self):
587         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
588         self.assertTrue(isinstance(data, bytes))
589
590     def test_update_url_query(self):
591         def query_dict(url):
592             return compat_parse_qs(compat_urlparse.urlparse(url).query)
593         self.assertEqual(query_dict(update_url_query(
594             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
595             query_dict('http://example.com/path?quality=HD&format=mp4'))
596         self.assertEqual(query_dict(update_url_query(
597             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
598             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
599         self.assertEqual(query_dict(update_url_query(
600             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
601             query_dict('http://example.com/path?fields=id,formats,subtitles'))
602         self.assertEqual(query_dict(update_url_query(
603             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
604             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
605         self.assertEqual(query_dict(update_url_query(
606             'http://example.com/path?manifest=f4m', {'manifest': []})),
607             query_dict('http://example.com/path'))
608         self.assertEqual(query_dict(update_url_query(
609             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
610             query_dict('http://example.com/path?system=LINUX'))
611         self.assertEqual(query_dict(update_url_query(
612             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
613             query_dict('http://example.com/path?fields=id,formats,subtitles'))
614         self.assertEqual(query_dict(update_url_query(
615             'http://example.com/path', {'width': 1080, 'height': 720})),
616             query_dict('http://example.com/path?width=1080&height=720'))
617         self.assertEqual(query_dict(update_url_query(
618             'http://example.com/path', {'bitrate': 5020.43})),
619             query_dict('http://example.com/path?bitrate=5020.43'))
620         self.assertEqual(query_dict(update_url_query(
621             'http://example.com/path', {'test': '第二行тест'})),
622             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
623
624     def test_multipart_encode(self):
625         self.assertEqual(
626             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
627             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
628         self.assertEqual(
629             multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
630             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')
631         self.assertRaises(
632             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
633
634     def test_dict_get(self):
635         FALSE_VALUES = {
636             'none': None,
637             'false': False,
638             'zero': 0,
639             'empty_string': '',
640             'empty_list': [],
641         }
642         d = FALSE_VALUES.copy()
643         d['a'] = 42
644         self.assertEqual(dict_get(d, 'a'), 42)
645         self.assertEqual(dict_get(d, 'b'), None)
646         self.assertEqual(dict_get(d, 'b', 42), 42)
647         self.assertEqual(dict_get(d, ('a', )), 42)
648         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
649         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
650         self.assertEqual(dict_get(d, ('b', 'c', )), None)
651         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
652         for key, false_value in FALSE_VALUES.items():
653             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
654             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
655
656     def test_encode_compat_str(self):
657         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
658         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
659
660     def test_parse_iso8601(self):
661         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
662         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
663         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
664         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
665         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
666         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
667
668     def test_strip_jsonp(self):
669         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
670         d = json.loads(stripped)
671         self.assertEqual(d, [{"id": "532cb", "x": 3}])
672
673         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
674         d = json.loads(stripped)
675         self.assertEqual(d, {'STATUS': 'OK'})
676
677         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
678         d = json.loads(stripped)
679         self.assertEqual(d, {'status': 'success'})
680
681     def test_uppercase_escape(self):
682         self.assertEqual(uppercase_escape('aä'), 'aä')
683         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
684
685     def test_lowercase_escape(self):
686         self.assertEqual(lowercase_escape('aä'), 'aä')
687         self.assertEqual(lowercase_escape('\\u0026'), '&')
688
689     def test_limit_length(self):
690         self.assertEqual(limit_length(None, 12), None)
691         self.assertEqual(limit_length('foo', 12), 'foo')
692         self.assertTrue(
693             limit_length('foo bar baz asd', 12).startswith('foo bar'))
694         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
695
696     def test_mimetype2ext(self):
697         self.assertEqual(mimetype2ext(None), None)
698         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
699         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
700         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
701         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
702         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
703
704     def test_month_by_name(self):
705         self.assertEqual(month_by_name(None), None)
706         self.assertEqual(month_by_name('December', 'en'), 12)
707         self.assertEqual(month_by_name('décembre', 'fr'), 12)
708         self.assertEqual(month_by_name('December'), 12)
709         self.assertEqual(month_by_name('décembre'), None)
710         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
711
712     def test_parse_codecs(self):
713         self.assertEqual(parse_codecs(''), {})
714         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
715             'vcodec': 'avc1.77.30',
716             'acodec': 'mp4a.40.2',
717         })
718         self.assertEqual(parse_codecs('mp4a.40.2'), {
719             'vcodec': 'none',
720             'acodec': 'mp4a.40.2',
721         })
722         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
723             'vcodec': 'avc1.42001e',
724             'acodec': 'mp4a.40.5',
725         })
726         self.assertEqual(parse_codecs('avc3.640028'), {
727             'vcodec': 'avc3.640028',
728             'acodec': 'none',
729         })
730         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
731             'vcodec': 'h264',
732             'acodec': 'aac',
733         })
734
735     def test_escape_rfc3986(self):
736         reserved = "!*'();:@&=+$,/?#[]"
737         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
738         self.assertEqual(escape_rfc3986(reserved), reserved)
739         self.assertEqual(escape_rfc3986(unreserved), unreserved)
740         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
741         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
742         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
743         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
744
745     def test_escape_url(self):
746         self.assertEqual(
747             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
748             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
749         )
750         self.assertEqual(
751             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
752             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
753         )
754         self.assertEqual(
755             escape_url('http://тест.рф/фрагмент'),
756             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
757         )
758         self.assertEqual(
759             escape_url('http://тест.рф/абв?абв=абв#абв'),
760             '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'
761         )
762         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
763
764     def test_js_to_json_realworld(self):
765         inp = '''{
766             'clip':{'provider':'pseudo'}
767         }'''
768         self.assertEqual(js_to_json(inp), '''{
769             "clip":{"provider":"pseudo"}
770         }''')
771         json.loads(js_to_json(inp))
772
773         inp = '''{
774             'playlist':[{'controls':{'all':null}}]
775         }'''
776         self.assertEqual(js_to_json(inp), '''{
777             "playlist":[{"controls":{"all":null}}]
778         }''')
779
780         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
781         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
782
783         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
784         json_code = js_to_json(inp)
785         self.assertEqual(json.loads(json_code), json.loads(inp))
786
787         inp = '''{
788             0:{src:'skipped', type: 'application/dash+xml'},
789             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
790         }'''
791         self.assertEqual(js_to_json(inp), '''{
792             "0":{"src":"skipped", "type": "application/dash+xml"},
793             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
794         }''')
795
796         inp = '''{"foo":101}'''
797         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
798
799         inp = '''{"duration": "00:01:07"}'''
800         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
801
802     def test_js_to_json_edgecases(self):
803         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
804         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
805
806         on = js_to_json('{"abc": true}')
807         self.assertEqual(json.loads(on), {'abc': True})
808
809         # Ignore JavaScript code as well
810         on = js_to_json('''{
811             "x": 1,
812             y: "a",
813             z: some.code
814         }''')
815         d = json.loads(on)
816         self.assertEqual(d['x'], 1)
817         self.assertEqual(d['y'], 'a')
818
819         on = js_to_json('["abc", "def",]')
820         self.assertEqual(json.loads(on), ['abc', 'def'])
821
822         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
823         self.assertEqual(json.loads(on), ['abc', 'def'])
824
825         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
826         self.assertEqual(json.loads(on), ['abc', 'def'])
827
828         on = js_to_json('{"abc": "def",}')
829         self.assertEqual(json.loads(on), {'abc': 'def'})
830
831         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
832         self.assertEqual(json.loads(on), {'abc': 'def'})
833
834         on = js_to_json('{ 0: /* " \n */ ",]" , }')
835         self.assertEqual(json.loads(on), {'0': ',]'})
836
837         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
838         self.assertEqual(json.loads(on), {'0': ',]'})
839
840         on = js_to_json('{ 0: // comment\n1 }')
841         self.assertEqual(json.loads(on), {'0': 1})
842
843         on = js_to_json(r'["<p>x<\/p>"]')
844         self.assertEqual(json.loads(on), ['<p>x</p>'])
845
846         on = js_to_json(r'["\xaa"]')
847         self.assertEqual(json.loads(on), ['\u00aa'])
848
849         on = js_to_json("['a\\\nb']")
850         self.assertEqual(json.loads(on), ['ab'])
851
852         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
853         self.assertEqual(json.loads(on), ['ab'])
854
855         on = js_to_json('{0xff:0xff}')
856         self.assertEqual(json.loads(on), {'255': 255})
857
858         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
859         self.assertEqual(json.loads(on), {'255': 255})
860
861         on = js_to_json('{077:077}')
862         self.assertEqual(json.loads(on), {'63': 63})
863
864         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
865         self.assertEqual(json.loads(on), {'63': 63})
866
867         on = js_to_json('{42:42}')
868         self.assertEqual(json.loads(on), {'42': 42})
869
870         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
871         self.assertEqual(json.loads(on), {'42': 42})
872
873     def test_extract_attributes(self):
874         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
875         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
876         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
877         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
878         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
879         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
880         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
881         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
882         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
883         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
884         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
885         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
886         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
887         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
888         self.assertEqual(extract_attributes('<e x >'), {'x': None})
889         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
890         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
891         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
892         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
893         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
894         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
895         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
896         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
897         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
898         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
899         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
900         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
901         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
902         # "Narrow" Python builds don't support unicode code points outside BMP.
903         try:
904             compat_chr(0x10000)
905             supports_outside_bmp = True
906         except ValueError:
907             supports_outside_bmp = False
908         if supports_outside_bmp:
909             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
910
911     def test_clean_html(self):
912         self.assertEqual(clean_html('a:\nb'), 'a: b')
913         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
914         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
915
916     def test_intlist_to_bytes(self):
917         self.assertEqual(
918             intlist_to_bytes([0, 1, 127, 128, 255]),
919             b'\x00\x01\x7f\x80\xff')
920
921     def test_args_to_str(self):
922         self.assertEqual(
923             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
924             'foo ba/r -baz \'2 be\' \'\''
925         )
926
927     def test_parse_filesize(self):
928         self.assertEqual(parse_filesize(None), None)
929         self.assertEqual(parse_filesize(''), None)
930         self.assertEqual(parse_filesize('91 B'), 91)
931         self.assertEqual(parse_filesize('foobar'), None)
932         self.assertEqual(parse_filesize('2 MiB'), 2097152)
933         self.assertEqual(parse_filesize('5 GB'), 5000000000)
934         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
935         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
936         self.assertEqual(parse_filesize('1,24 KB'), 1240)
937         self.assertEqual(parse_filesize('1,24 kb'), 1240)
938         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
939
940     def test_parse_count(self):
941         self.assertEqual(parse_count(None), None)
942         self.assertEqual(parse_count(''), None)
943         self.assertEqual(parse_count('0'), 0)
944         self.assertEqual(parse_count('1000'), 1000)
945         self.assertEqual(parse_count('1.000'), 1000)
946         self.assertEqual(parse_count('1.1k'), 1100)
947         self.assertEqual(parse_count('1.1kk'), 1100000)
948         self.assertEqual(parse_count('1.1kk '), 1100000)
949         self.assertEqual(parse_count('1.1kk views'), 1100000)
950
951     def test_version_tuple(self):
952         self.assertEqual(version_tuple('1'), (1,))
953         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
954         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
955
956     def test_detect_exe_version(self):
957         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
958 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
959 configuration: --prefix=/usr --extra-'''), '1.2.1')
960         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
961 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
962         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
963 Trying to open render node...
964 Success at /dev/dri/renderD128.
965 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
966
967     def test_age_restricted(self):
968         self.assertFalse(age_restricted(None, 10))  # unrestricted content
969         self.assertFalse(age_restricted(1, None))  # unrestricted policy
970         self.assertFalse(age_restricted(8, 10))
971         self.assertTrue(age_restricted(18, 14))
972         self.assertFalse(age_restricted(18, 18))
973
974     def test_is_html(self):
975         self.assertFalse(is_html(b'\x49\x44\x43<html'))
976         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
977         self.assertTrue(is_html(  # UTF-8 with BOM
978             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
979         self.assertTrue(is_html(  # UTF-16-LE
980             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
981         ))
982         self.assertTrue(is_html(  # UTF-16-BE
983             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
984         ))
985         self.assertTrue(is_html(  # UTF-32-BE
986             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'))
987         self.assertTrue(is_html(  # UTF-32-LE
988             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'))
989
990     def test_render_table(self):
991         self.assertEqual(
992             render_table(
993                 ['a', 'bcd'],
994                 [[123, 4], [9999, 51]]),
995             'a    bcd\n'
996             '123  4\n'
997             '9999 51')
998
999     def test_match_str(self):
1000         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
1001         self.assertFalse(match_str('xy', {'x': 1200}))
1002         self.assertTrue(match_str('!xy', {'x': 1200}))
1003         self.assertTrue(match_str('x', {'x': 1200}))
1004         self.assertFalse(match_str('!x', {'x': 1200}))
1005         self.assertTrue(match_str('x', {'x': 0}))
1006         self.assertFalse(match_str('x>0', {'x': 0}))
1007         self.assertFalse(match_str('x>0', {}))
1008         self.assertTrue(match_str('x>?0', {}))
1009         self.assertTrue(match_str('x>1K', {'x': 1200}))
1010         self.assertFalse(match_str('x>2K', {'x': 1200}))
1011         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
1012         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
1013         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
1014         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
1015         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
1016         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
1017         self.assertFalse(match_str(
1018             'like_count > 100 & dislike_count <? 50 & description',
1019             {'like_count': 90, 'description': 'foo'}))
1020         self.assertTrue(match_str(
1021             'like_count > 100 & dislike_count <? 50 & description',
1022             {'like_count': 190, 'description': 'foo'}))
1023         self.assertFalse(match_str(
1024             'like_count > 100 & dislike_count <? 50 & description',
1025             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
1026         self.assertFalse(match_str(
1027             'like_count > 100 & dislike_count <? 50 & description',
1028             {'like_count': 190, 'dislike_count': 10}))
1029
1030     def test_parse_dfxp_time_expr(self):
1031         self.assertEqual(parse_dfxp_time_expr(None), None)
1032         self.assertEqual(parse_dfxp_time_expr(''), None)
1033         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1034         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1035         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1036         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1037         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1038
1039     def test_dfxp2srt(self):
1040         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1041             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1042             <body>
1043                 <div xml:lang="en">
1044                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1045                     <p begin="1" end="2">第二行<br/>♪♪</p>
1046                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1047                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1048                     <p begin="-1" end="-1">Ignore, two</p>
1049                     <p begin="3" dur="-1">Ignored, three</p>
1050                 </div>
1051             </body>
1052             </tt>'''
1053         srt_data = '''1
1054 00:00:00,000 --> 00:00:01,000
1055 The following line contains Chinese characters and special symbols
1056
1057 2
1058 00:00:01,000 --> 00:00:02,000
1059 第二行
1060 ♪♪
1061
1062 3
1063 00:00:02,000 --> 00:00:03,000
1064 Third
1065 Line
1066
1067 '''
1068         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1069
1070         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1071             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1072             <body>
1073                 <div xml:lang="en">
1074                     <p begin="0" end="1">The first line</p>
1075                 </div>
1076             </body>
1077             </tt>'''
1078         srt_data = '''1
1079 00:00:00,000 --> 00:00:01,000
1080 The first line
1081
1082 '''
1083         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1084
1085         dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
1086 <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">
1087   <head>
1088     <styling>
1089       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
1090       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
1091       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
1092       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
1093     </styling>
1094   </head>
1095   <body tts:textAlign="center" style="s0">
1096     <div>
1097       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
1098       <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>
1099       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
1100       <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>
1101     </div>
1102   </body>
1103 </tt>'''
1104         srt_data = '''1
1105 00:00:02,080 --> 00:00:05,839
1106 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
1107
1108 2
1109 00:00:02,080 --> 00:00:05,839
1110 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
1111 </font>part 2</font></b>
1112
1113 3
1114 00:00:05,839 --> 00:00:09,560
1115 <u><font color="lime">line 3
1116 part 3</font></u>
1117
1118 4
1119 00:00:09,560 --> 00:00:12,359
1120 <i><u><font color="yellow"><font color="lime">inner
1121  </font>style</font></u></i>
1122
1123 '''
1124         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
1125
1126     def test_cli_option(self):
1127         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1128         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1129         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1130         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1131
1132     def test_cli_valueless_option(self):
1133         self.assertEqual(cli_valueless_option(
1134             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1135         self.assertEqual(cli_valueless_option(
1136             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1137         self.assertEqual(cli_valueless_option(
1138             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1139         self.assertEqual(cli_valueless_option(
1140             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1141         self.assertEqual(cli_valueless_option(
1142             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1143         self.assertEqual(cli_valueless_option(
1144             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1145
1146     def test_cli_bool_option(self):
1147         self.assertEqual(
1148             cli_bool_option(
1149                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1150             ['--no-check-certificate', 'true'])
1151         self.assertEqual(
1152             cli_bool_option(
1153                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1154             ['--no-check-certificate=true'])
1155         self.assertEqual(
1156             cli_bool_option(
1157                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1158             ['--check-certificate', 'false'])
1159         self.assertEqual(
1160             cli_bool_option(
1161                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1162             ['--check-certificate=false'])
1163         self.assertEqual(
1164             cli_bool_option(
1165                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1166             ['--check-certificate', 'true'])
1167         self.assertEqual(
1168             cli_bool_option(
1169                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1170             ['--check-certificate=true'])
1171
1172     def test_ohdave_rsa_encrypt(self):
1173         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1174         e = 65537
1175
1176         self.assertEqual(
1177             ohdave_rsa_encrypt(b'aa111222', e, N),
1178             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1179
1180     def test_pkcs1pad(self):
1181         data = [1, 2, 3]
1182         padded_data = pkcs1pad(data, 32)
1183         self.assertEqual(padded_data[:2], [0, 2])
1184         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
1185
1186         self.assertRaises(ValueError, pkcs1pad, data, 8)
1187
1188     def test_encode_base_n(self):
1189         self.assertEqual(encode_base_n(0, 30), '0')
1190         self.assertEqual(encode_base_n(80, 30), '2k')
1191
1192         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1193         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1194         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1195
1196         self.assertRaises(ValueError, encode_base_n, 0, 70)
1197         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1198
1199     def test_urshift(self):
1200         self.assertEqual(urshift(3, 1), 1)
1201         self.assertEqual(urshift(-3, 1), 2147483646)
1202
1203     def test_get_element_by_class(self):
1204         html = '''
1205             <span class="foo bar">nice</span>
1206         '''
1207
1208         self.assertEqual(get_element_by_class('foo', html), 'nice')
1209         self.assertEqual(get_element_by_class('no-such-class', html), None)
1210
1211     def test_get_element_by_attribute(self):
1212         html = '''
1213             <span class="foo bar">nice</span>
1214         '''
1215
1216         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1217         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1218         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1219
1220     def test_get_elements_by_class(self):
1221         html = '''
1222             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1223         '''
1224
1225         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1226         self.assertEqual(get_elements_by_class('no-such-class', html), [])
1227
1228     def test_get_elements_by_attribute(self):
1229         html = '''
1230             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1231         '''
1232
1233         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1234         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1235         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1236
1237
1238 if __name__ == '__main__':
1239     unittest.main()