]> jfr.im git - yt-dlp.git/blame_incremental - test/test_utils.py
[utils] Improve month_by_name and add tests
[yt-dlp.git] / test / test_utils.py
... / ...
CommitLineData
1#!/usr/bin/env python
2# coding: utf-8
3
4from __future__ import unicode_literals
5
6# Allow direct execution
7import os
8import sys
9import unittest
10sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
11
12
13# Various small unit tests
14import io
15import json
16import xml.etree.ElementTree
17
18from 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 InAdvancePagedList,
38 intlist_to_bytes,
39 is_html,
40 js_to_json,
41 limit_length,
42 mimetype2ext,
43 month_by_name,
44 ohdave_rsa_encrypt,
45 OnDemandPagedList,
46 orderedSet,
47 parse_age_limit,
48 parse_duration,
49 parse_filesize,
50 parse_count,
51 parse_iso8601,
52 read_batch_urls,
53 sanitize_filename,
54 sanitize_path,
55 prepend_extension,
56 replace_extension,
57 remove_start,
58 remove_end,
59 remove_quotes,
60 shell_quote,
61 smuggle_url,
62 str_to_int,
63 strip_jsonp,
64 timeconvert,
65 unescapeHTML,
66 unified_strdate,
67 unified_timestamp,
68 unsmuggle_url,
69 uppercase_escape,
70 lowercase_escape,
71 url_basename,
72 urlencode_postdata,
73 urshift,
74 update_url_query,
75 version_tuple,
76 xpath_with_ns,
77 xpath_element,
78 xpath_text,
79 xpath_attr,
80 render_table,
81 match_str,
82 parse_dfxp_time_expr,
83 dfxp2srt,
84 cli_option,
85 cli_valueless_option,
86 cli_bool_option,
87 parse_codecs,
88)
89from youtube_dl.compat import (
90 compat_chr,
91 compat_etree_fromstring,
92 compat_urlparse,
93 compat_parse_qs,
94)
95
96
97class TestUtil(unittest.TestCase):
98 def test_timeconvert(self):
99 self.assertTrue(timeconvert('') is None)
100 self.assertTrue(timeconvert('bougrg') is None)
101
102 def test_sanitize_filename(self):
103 self.assertEqual(sanitize_filename('abc'), 'abc')
104 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
105
106 self.assertEqual(sanitize_filename('123'), '123')
107
108 self.assertEqual('abc_de', sanitize_filename('abc/de'))
109 self.assertFalse('/' in sanitize_filename('abc/de///'))
110
111 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
112 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
113 self.assertEqual('yes no', sanitize_filename('yes? no'))
114 self.assertEqual('this - that', sanitize_filename('this: that'))
115
116 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
117 aumlaut = 'ä'
118 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
119 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
120 self.assertEqual(sanitize_filename(tests), tests)
121
122 self.assertEqual(
123 sanitize_filename('New World record at 0:12:34'),
124 'New World record at 0_12_34')
125
126 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
127 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
128 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
129 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
130
131 forbidden = '"\0\\/'
132 for fc in forbidden:
133 for fbc in forbidden:
134 self.assertTrue(fbc not in sanitize_filename(fc))
135
136 def test_sanitize_filename_restricted(self):
137 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
138 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
139
140 self.assertEqual(sanitize_filename('123', restricted=True), '123')
141
142 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
143 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
144
145 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
146 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
147 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
148 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
149
150 tests = 'aäb\u4e2d\u56fd\u7684c'
151 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
152 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
153
154 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
155 for fc in forbidden:
156 for fbc in forbidden:
157 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
158
159 # Handle a common case more neatly
160 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
161 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
162 # .. but make sure the file name is never empty
163 self.assertTrue(sanitize_filename('-', restricted=True) != '')
164 self.assertTrue(sanitize_filename(':', restricted=True) != '')
165
166 self.assertEqual(sanitize_filename(
167 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
168 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
169
170 def test_sanitize_ids(self):
171 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
172 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
173 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
174
175 def test_sanitize_path(self):
176 if sys.platform != 'win32':
177 return
178
179 self.assertEqual(sanitize_path('abc'), 'abc')
180 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
181 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
182 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
183 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
184 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
185 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
186
187 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
188 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
189
190 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
191 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
192 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
193 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
194
195 self.assertEqual(
196 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
197 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
198
199 self.assertEqual(
200 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
201 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
202 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
203 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
204 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
205
206 self.assertEqual(sanitize_path('../abc'), '..\\abc')
207 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
208 self.assertEqual(sanitize_path('./abc'), 'abc')
209 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
210
211 def test_prepend_extension(self):
212 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
213 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
214 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
215 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
216 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
217 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
218
219 def test_replace_extension(self):
220 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
221 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
222 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
223 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
224 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
225 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
226
227 def test_remove_start(self):
228 self.assertEqual(remove_start(None, 'A - '), None)
229 self.assertEqual(remove_start('A - B', 'A - '), 'B')
230 self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
231
232 def test_remove_end(self):
233 self.assertEqual(remove_end(None, ' - B'), None)
234 self.assertEqual(remove_end('A - B', ' - B'), 'A')
235 self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
236
237 def test_remove_quotes(self):
238 self.assertEqual(remove_quotes(None), None)
239 self.assertEqual(remove_quotes('"'), '"')
240 self.assertEqual(remove_quotes("'"), "'")
241 self.assertEqual(remove_quotes(';'), ';')
242 self.assertEqual(remove_quotes('";'), '";')
243 self.assertEqual(remove_quotes('""'), '')
244 self.assertEqual(remove_quotes('";"'), ';')
245
246 def test_ordered_set(self):
247 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
248 self.assertEqual(orderedSet([]), [])
249 self.assertEqual(orderedSet([1]), [1])
250 # keep the list ordered
251 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
252
253 def test_unescape_html(self):
254 self.assertEqual(unescapeHTML('%20;'), '%20;')
255 self.assertEqual(unescapeHTML('&#x2F;'), '/')
256 self.assertEqual(unescapeHTML('&#47;'), '/')
257 self.assertEqual(unescapeHTML('&eacute;'), 'é')
258 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
259 # HTML5 entities
260 self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
261
262 def test_date_from_str(self):
263 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
264 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
265 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
266 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
267 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
268
269 def test_daterange(self):
270 _20century = DateRange("19000101", "20000101")
271 self.assertFalse("17890714" in _20century)
272 _ac = DateRange("00010101")
273 self.assertTrue("19690721" in _ac)
274 _firstmilenium = DateRange(end="10000101")
275 self.assertTrue("07110427" in _firstmilenium)
276
277 def test_unified_dates(self):
278 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
279 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
280 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
281 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
282 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
283 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
284 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
285 self.assertEqual(
286 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
287 '20141126')
288 self.assertEqual(
289 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
290 '20150202')
291 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
292 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
293 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
294 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
295
296 def test_unified_timestamps(self):
297 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
298 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
299 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
300 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
301 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
302 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
303 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
304 self.assertEqual(
305 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
306 1417001400)
307 self.assertEqual(
308 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
309 1422902860)
310 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
311 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
312 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
313 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
314 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
315
316 def test_determine_ext(self):
317 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
318 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
319 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
320 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
321 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
322
323 def test_find_xpath_attr(self):
324 testxml = '''<root>
325 <node/>
326 <node x="a"/>
327 <node x="a" y="c" />
328 <node x="b" y="d" />
329 <node x="" />
330 </root>'''
331 doc = compat_etree_fromstring(testxml)
332
333 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
334 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
335 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
336 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
337 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
338 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
339 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
340 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
341 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
342 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
343 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
344
345 def test_xpath_with_ns(self):
346 testxml = '''<root xmlns:media="http://example.com/">
347 <media:song>
348 <media:author>The Author</media:author>
349 <url>http://server.com/download.mp3</url>
350 </media:song>
351 </root>'''
352 doc = compat_etree_fromstring(testxml)
353 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
354 self.assertTrue(find('media:song') is not None)
355 self.assertEqual(find('media:song/media:author').text, 'The Author')
356 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
357
358 def test_xpath_element(self):
359 doc = xml.etree.ElementTree.Element('root')
360 div = xml.etree.ElementTree.SubElement(doc, 'div')
361 p = xml.etree.ElementTree.SubElement(div, 'p')
362 p.text = 'Foo'
363 self.assertEqual(xpath_element(doc, 'div/p'), p)
364 self.assertEqual(xpath_element(doc, ['div/p']), p)
365 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
366 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
367 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
368 self.assertTrue(xpath_element(doc, 'div/bar') is None)
369 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
370 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
371 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
372 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
373 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
374
375 def test_xpath_text(self):
376 testxml = '''<root>
377 <div>
378 <p>Foo</p>
379 </div>
380 </root>'''
381 doc = compat_etree_fromstring(testxml)
382 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
383 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
384 self.assertTrue(xpath_text(doc, 'div/bar') is None)
385 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
386
387 def test_xpath_attr(self):
388 testxml = '''<root>
389 <div>
390 <p x="a">Foo</p>
391 </div>
392 </root>'''
393 doc = compat_etree_fromstring(testxml)
394 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
395 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
396 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
397 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
398 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
399 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
400 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
401
402 def test_smuggle_url(self):
403 data = {"ö": "ö", "abc": [3]}
404 url = 'https://foo.bar/baz?x=y#a'
405 smug_url = smuggle_url(url, data)
406 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
407 self.assertEqual(url, unsmug_url)
408 self.assertEqual(data, unsmug_data)
409
410 res_url, res_data = unsmuggle_url(url)
411 self.assertEqual(res_url, url)
412 self.assertEqual(res_data, None)
413
414 smug_url = smuggle_url(url, {'a': 'b'})
415 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
416 res_url, res_data = unsmuggle_url(smug_smug_url)
417 self.assertEqual(res_url, url)
418 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
419
420 def test_shell_quote(self):
421 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
422 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
423
424 def test_str_to_int(self):
425 self.assertEqual(str_to_int('123,456'), 123456)
426 self.assertEqual(str_to_int('123.456'), 123456)
427
428 def test_url_basename(self):
429 self.assertEqual(url_basename('http://foo.de/'), '')
430 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
431 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
432 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
433 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
434 self.assertEqual(
435 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
436 'trailer.mp4')
437
438 def test_parse_age_limit(self):
439 self.assertEqual(parse_age_limit(None), None)
440 self.assertEqual(parse_age_limit(False), None)
441 self.assertEqual(parse_age_limit('invalid'), None)
442 self.assertEqual(parse_age_limit(0), 0)
443 self.assertEqual(parse_age_limit(18), 18)
444 self.assertEqual(parse_age_limit(21), 21)
445 self.assertEqual(parse_age_limit(22), None)
446 self.assertEqual(parse_age_limit('18'), 18)
447 self.assertEqual(parse_age_limit('18+'), 18)
448 self.assertEqual(parse_age_limit('PG-13'), 13)
449 self.assertEqual(parse_age_limit('TV-14'), 14)
450 self.assertEqual(parse_age_limit('TV-MA'), 17)
451
452 def test_parse_duration(self):
453 self.assertEqual(parse_duration(None), None)
454 self.assertEqual(parse_duration(False), None)
455 self.assertEqual(parse_duration('invalid'), None)
456 self.assertEqual(parse_duration('1'), 1)
457 self.assertEqual(parse_duration('1337:12'), 80232)
458 self.assertEqual(parse_duration('9:12:43'), 33163)
459 self.assertEqual(parse_duration('12:00'), 720)
460 self.assertEqual(parse_duration('00:01:01'), 61)
461 self.assertEqual(parse_duration('x:y'), None)
462 self.assertEqual(parse_duration('3h11m53s'), 11513)
463 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
464 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
465 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
466 self.assertEqual(parse_duration('62m45s'), 3765)
467 self.assertEqual(parse_duration('6m59s'), 419)
468 self.assertEqual(parse_duration('49s'), 49)
469 self.assertEqual(parse_duration('0h0m0s'), 0)
470 self.assertEqual(parse_duration('0m0s'), 0)
471 self.assertEqual(parse_duration('0s'), 0)
472 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
473 self.assertEqual(parse_duration('T30M38S'), 1838)
474 self.assertEqual(parse_duration('5 s'), 5)
475 self.assertEqual(parse_duration('3 min'), 180)
476 self.assertEqual(parse_duration('2.5 hours'), 9000)
477 self.assertEqual(parse_duration('02:03:04'), 7384)
478 self.assertEqual(parse_duration('01:02:03:04'), 93784)
479 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
480 self.assertEqual(parse_duration('87 Min.'), 5220)
481 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
482
483 def test_fix_xml_ampersands(self):
484 self.assertEqual(
485 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
486 self.assertEqual(
487 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
488 '"&amp;x=y&amp;wrong;&amp;z=a')
489 self.assertEqual(
490 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
491 '&amp;&apos;&gt;&lt;&quot;')
492 self.assertEqual(
493 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
494 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
495
496 def test_paged_list(self):
497 def testPL(size, pagesize, sliceargs, expected):
498 def get_page(pagenum):
499 firstid = pagenum * pagesize
500 upto = min(size, pagenum * pagesize + pagesize)
501 for i in range(firstid, upto):
502 yield i
503
504 pl = OnDemandPagedList(get_page, pagesize)
505 got = pl.getslice(*sliceargs)
506 self.assertEqual(got, expected)
507
508 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
509 got = iapl.getslice(*sliceargs)
510 self.assertEqual(got, expected)
511
512 testPL(5, 2, (), [0, 1, 2, 3, 4])
513 testPL(5, 2, (1,), [1, 2, 3, 4])
514 testPL(5, 2, (2,), [2, 3, 4])
515 testPL(5, 2, (4,), [4])
516 testPL(5, 2, (0, 3), [0, 1, 2])
517 testPL(5, 2, (1, 4), [1, 2, 3])
518 testPL(5, 2, (2, 99), [2, 3, 4])
519 testPL(5, 2, (20, 99), [])
520
521 def test_read_batch_urls(self):
522 f = io.StringIO('''\xef\xbb\xbf foo
523 bar\r
524 baz
525 # More after this line\r
526 ; or after this
527 bam''')
528 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
529
530 def test_urlencode_postdata(self):
531 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
532 self.assertTrue(isinstance(data, bytes))
533
534 def test_update_url_query(self):
535 def query_dict(url):
536 return compat_parse_qs(compat_urlparse.urlparse(url).query)
537 self.assertEqual(query_dict(update_url_query(
538 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
539 query_dict('http://example.com/path?quality=HD&format=mp4'))
540 self.assertEqual(query_dict(update_url_query(
541 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
542 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
543 self.assertEqual(query_dict(update_url_query(
544 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
545 query_dict('http://example.com/path?fields=id,formats,subtitles'))
546 self.assertEqual(query_dict(update_url_query(
547 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
548 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
549 self.assertEqual(query_dict(update_url_query(
550 'http://example.com/path?manifest=f4m', {'manifest': []})),
551 query_dict('http://example.com/path'))
552 self.assertEqual(query_dict(update_url_query(
553 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
554 query_dict('http://example.com/path?system=LINUX'))
555 self.assertEqual(query_dict(update_url_query(
556 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
557 query_dict('http://example.com/path?fields=id,formats,subtitles'))
558 self.assertEqual(query_dict(update_url_query(
559 'http://example.com/path', {'width': 1080, 'height': 720})),
560 query_dict('http://example.com/path?width=1080&height=720'))
561 self.assertEqual(query_dict(update_url_query(
562 'http://example.com/path', {'bitrate': 5020.43})),
563 query_dict('http://example.com/path?bitrate=5020.43'))
564 self.assertEqual(query_dict(update_url_query(
565 'http://example.com/path', {'test': '第二行тест'})),
566 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
567
568 def test_dict_get(self):
569 FALSE_VALUES = {
570 'none': None,
571 'false': False,
572 'zero': 0,
573 'empty_string': '',
574 'empty_list': [],
575 }
576 d = FALSE_VALUES.copy()
577 d['a'] = 42
578 self.assertEqual(dict_get(d, 'a'), 42)
579 self.assertEqual(dict_get(d, 'b'), None)
580 self.assertEqual(dict_get(d, 'b', 42), 42)
581 self.assertEqual(dict_get(d, ('a', )), 42)
582 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
583 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
584 self.assertEqual(dict_get(d, ('b', 'c', )), None)
585 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
586 for key, false_value in FALSE_VALUES.items():
587 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
588 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
589
590 def test_encode_compat_str(self):
591 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
592 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
593
594 def test_parse_iso8601(self):
595 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
596 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
597 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
598 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
599 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
600 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
601
602 def test_strip_jsonp(self):
603 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
604 d = json.loads(stripped)
605 self.assertEqual(d, [{"id": "532cb", "x": 3}])
606
607 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
608 d = json.loads(stripped)
609 self.assertEqual(d, {'STATUS': 'OK'})
610
611 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
612 d = json.loads(stripped)
613 self.assertEqual(d, {'status': 'success'})
614
615 def test_uppercase_escape(self):
616 self.assertEqual(uppercase_escape('aä'), 'aä')
617 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
618
619 def test_lowercase_escape(self):
620 self.assertEqual(lowercase_escape('aä'), 'aä')
621 self.assertEqual(lowercase_escape('\\u0026'), '&')
622
623 def test_limit_length(self):
624 self.assertEqual(limit_length(None, 12), None)
625 self.assertEqual(limit_length('foo', 12), 'foo')
626 self.assertTrue(
627 limit_length('foo bar baz asd', 12).startswith('foo bar'))
628 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
629
630 def test_mimetype2ext(self):
631 self.assertEqual(mimetype2ext(None), None)
632 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
633 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
634 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
635 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
636 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
637
638 def test_month_by_name(self):
639 self.assertEqual(month_by_name(None), None)
640 self.assertEqual(month_by_name('December', 'en'), 12)
641 self.assertEqual(month_by_name('decembre', 'fr'), 12)
642 self.assertEqual(month_by_name('December'), 12)
643 self.assertEqual(month_by_name('decembre'), None)
644 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
645
646 def test_m
647
648 def test_parse_codecs(self):
649 self.assertEqual(parse_codecs(''), {})
650 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
651 'vcodec': 'avc1.77.30',
652 'acodec': 'mp4a.40.2',
653 })
654 self.assertEqual(parse_codecs('mp4a.40.2'), {
655 'vcodec': 'none',
656 'acodec': 'mp4a.40.2',
657 })
658 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
659 'vcodec': 'avc1.42001e',
660 'acodec': 'mp4a.40.5',
661 })
662 self.assertEqual(parse_codecs('avc3.640028'), {
663 'vcodec': 'avc3.640028',
664 'acodec': 'none',
665 })
666 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
667 'vcodec': 'h264',
668 'acodec': 'aac',
669 })
670
671 def test_escape_rfc3986(self):
672 reserved = "!*'();:@&=+$,/?#[]"
673 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
674 self.assertEqual(escape_rfc3986(reserved), reserved)
675 self.assertEqual(escape_rfc3986(unreserved), unreserved)
676 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
677 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
678 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
679 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
680
681 def test_escape_url(self):
682 self.assertEqual(
683 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
684 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
685 )
686 self.assertEqual(
687 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
688 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
689 )
690 self.assertEqual(
691 escape_url('http://тест.рф/фрагмент'),
692 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
693 )
694 self.assertEqual(
695 escape_url('http://тест.рф/абв?абв=абв#абв'),
696 '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'
697 )
698 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
699
700 def test_js_to_json_realworld(self):
701 inp = '''{
702 'clip':{'provider':'pseudo'}
703 }'''
704 self.assertEqual(js_to_json(inp), '''{
705 "clip":{"provider":"pseudo"}
706 }''')
707 json.loads(js_to_json(inp))
708
709 inp = '''{
710 'playlist':[{'controls':{'all':null}}]
711 }'''
712 self.assertEqual(js_to_json(inp), '''{
713 "playlist":[{"controls":{"all":null}}]
714 }''')
715
716 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
717 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
718
719 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
720 json_code = js_to_json(inp)
721 self.assertEqual(json.loads(json_code), json.loads(inp))
722
723 inp = '''{
724 0:{src:'skipped', type: 'application/dash+xml'},
725 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
726 }'''
727 self.assertEqual(js_to_json(inp), '''{
728 "0":{"src":"skipped", "type": "application/dash+xml"},
729 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
730 }''')
731
732 inp = '''{"foo":101}'''
733 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
734
735 inp = '''{"duration": "00:01:07"}'''
736 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
737
738 def test_js_to_json_edgecases(self):
739 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
740 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
741
742 on = js_to_json('{"abc": true}')
743 self.assertEqual(json.loads(on), {'abc': True})
744
745 # Ignore JavaScript code as well
746 on = js_to_json('''{
747 "x": 1,
748 y: "a",
749 z: some.code
750 }''')
751 d = json.loads(on)
752 self.assertEqual(d['x'], 1)
753 self.assertEqual(d['y'], 'a')
754
755 on = js_to_json('["abc", "def",]')
756 self.assertEqual(json.loads(on), ['abc', 'def'])
757
758 on = js_to_json('{"abc": "def",}')
759 self.assertEqual(json.loads(on), {'abc': 'def'})
760
761 on = js_to_json('{ 0: /* " \n */ ",]" , }')
762 self.assertEqual(json.loads(on), {'0': ',]'})
763
764 on = js_to_json(r'["<p>x<\/p>"]')
765 self.assertEqual(json.loads(on), ['<p>x</p>'])
766
767 on = js_to_json(r'["\xaa"]')
768 self.assertEqual(json.loads(on), ['\u00aa'])
769
770 on = js_to_json("['a\\\nb']")
771 self.assertEqual(json.loads(on), ['ab'])
772
773 on = js_to_json('{0xff:0xff}')
774 self.assertEqual(json.loads(on), {'255': 255})
775
776 on = js_to_json('{077:077}')
777 self.assertEqual(json.loads(on), {'63': 63})
778
779 on = js_to_json('{42:42}')
780 self.assertEqual(json.loads(on), {'42': 42})
781
782 def test_extract_attributes(self):
783 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
784 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
785 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
786 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
787 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
788 self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
789 self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
790 self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'}) # XML
791 self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
792 self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
793 self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
794 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
795 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
796 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
797 self.assertEqual(extract_attributes('<e x >'), {'x': None})
798 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
799 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
800 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
801 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
802 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
803 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
804 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
805 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
806 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
807 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
808 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
809 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
810 self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
811 # "Narrow" Python builds don't support unicode code points outside BMP.
812 try:
813 compat_chr(0x10000)
814 supports_outside_bmp = True
815 except ValueError:
816 supports_outside_bmp = False
817 if supports_outside_bmp:
818 self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
819
820 def test_clean_html(self):
821 self.assertEqual(clean_html('a:\nb'), 'a: b')
822 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
823
824 def test_intlist_to_bytes(self):
825 self.assertEqual(
826 intlist_to_bytes([0, 1, 127, 128, 255]),
827 b'\x00\x01\x7f\x80\xff')
828
829 def test_args_to_str(self):
830 self.assertEqual(
831 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
832 'foo ba/r -baz \'2 be\' \'\''
833 )
834
835 def test_parse_filesize(self):
836 self.assertEqual(parse_filesize(None), None)
837 self.assertEqual(parse_filesize(''), None)
838 self.assertEqual(parse_filesize('91 B'), 91)
839 self.assertEqual(parse_filesize('foobar'), None)
840 self.assertEqual(parse_filesize('2 MiB'), 2097152)
841 self.assertEqual(parse_filesize('5 GB'), 5000000000)
842 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
843 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
844 self.assertEqual(parse_filesize('1,24 KB'), 1240)
845 self.assertEqual(parse_filesize('1,24 kb'), 1240)
846 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
847
848 def test_parse_count(self):
849 self.assertEqual(parse_count(None), None)
850 self.assertEqual(parse_count(''), None)
851 self.assertEqual(parse_count('0'), 0)
852 self.assertEqual(parse_count('1000'), 1000)
853 self.assertEqual(parse_count('1.000'), 1000)
854 self.assertEqual(parse_count('1.1k'), 1100)
855 self.assertEqual(parse_count('1.1kk'), 1100000)
856 self.assertEqual(parse_count('1.1kk '), 1100000)
857 self.assertEqual(parse_count('1.1kk views'), 1100000)
858
859 def test_version_tuple(self):
860 self.assertEqual(version_tuple('1'), (1,))
861 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
862 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
863
864 def test_detect_exe_version(self):
865 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
866built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
867configuration: --prefix=/usr --extra-'''), '1.2.1')
868 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
869built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
870 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
871Trying to open render node...
872Success at /dev/dri/renderD128.
873ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
874
875 def test_age_restricted(self):
876 self.assertFalse(age_restricted(None, 10)) # unrestricted content
877 self.assertFalse(age_restricted(1, None)) # unrestricted policy
878 self.assertFalse(age_restricted(8, 10))
879 self.assertTrue(age_restricted(18, 14))
880 self.assertFalse(age_restricted(18, 18))
881
882 def test_is_html(self):
883 self.assertFalse(is_html(b'\x49\x44\x43<html'))
884 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
885 self.assertTrue(is_html( # UTF-8 with BOM
886 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
887 self.assertTrue(is_html( # UTF-16-LE
888 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
889 ))
890 self.assertTrue(is_html( # UTF-16-BE
891 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
892 ))
893 self.assertTrue(is_html( # UTF-32-BE
894 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'))
895 self.assertTrue(is_html( # UTF-32-LE
896 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'))
897
898 def test_render_table(self):
899 self.assertEqual(
900 render_table(
901 ['a', 'bcd'],
902 [[123, 4], [9999, 51]]),
903 'a bcd\n'
904 '123 4\n'
905 '9999 51')
906
907 def test_match_str(self):
908 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
909 self.assertFalse(match_str('xy', {'x': 1200}))
910 self.assertTrue(match_str('!xy', {'x': 1200}))
911 self.assertTrue(match_str('x', {'x': 1200}))
912 self.assertFalse(match_str('!x', {'x': 1200}))
913 self.assertTrue(match_str('x', {'x': 0}))
914 self.assertFalse(match_str('x>0', {'x': 0}))
915 self.assertFalse(match_str('x>0', {}))
916 self.assertTrue(match_str('x>?0', {}))
917 self.assertTrue(match_str('x>1K', {'x': 1200}))
918 self.assertFalse(match_str('x>2K', {'x': 1200}))
919 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
920 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
921 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
922 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
923 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
924 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
925 self.assertFalse(match_str(
926 'like_count > 100 & dislike_count <? 50 & description',
927 {'like_count': 90, 'description': 'foo'}))
928 self.assertTrue(match_str(
929 'like_count > 100 & dislike_count <? 50 & description',
930 {'like_count': 190, 'description': 'foo'}))
931 self.assertFalse(match_str(
932 'like_count > 100 & dislike_count <? 50 & description',
933 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
934 self.assertFalse(match_str(
935 'like_count > 100 & dislike_count <? 50 & description',
936 {'like_count': 190, 'dislike_count': 10}))
937
938 def test_parse_dfxp_time_expr(self):
939 self.assertEqual(parse_dfxp_time_expr(None), None)
940 self.assertEqual(parse_dfxp_time_expr(''), None)
941 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
942 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
943 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
944 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
945 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
946
947 def test_dfxp2srt(self):
948 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
949 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
950 <body>
951 <div xml:lang="en">
952 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
953 <p begin="1" end="2">第二行<br/>♪♪</p>
954 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
955 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
956 <p begin="-1" end="-1">Ignore, two</p>
957 <p begin="3" dur="-1">Ignored, three</p>
958 </div>
959 </body>
960 </tt>'''
961 srt_data = '''1
96200:00:00,000 --> 00:00:01,000
963The following line contains Chinese characters and special symbols
964
9652
96600:00:01,000 --> 00:00:02,000
967第二行
968♪♪
969
9703
97100:00:02,000 --> 00:00:03,000
972Third
973Line
974
975'''
976 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
977
978 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
979 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
980 <body>
981 <div xml:lang="en">
982 <p begin="0" end="1">The first line</p>
983 </div>
984 </body>
985 </tt>'''
986 srt_data = '''1
98700:00:00,000 --> 00:00:01,000
988The first line
989
990'''
991 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
992
993 def test_cli_option(self):
994 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
995 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
996 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
997 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
998
999 def test_cli_valueless_option(self):
1000 self.assertEqual(cli_valueless_option(
1001 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1002 self.assertEqual(cli_valueless_option(
1003 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1004 self.assertEqual(cli_valueless_option(
1005 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1006 self.assertEqual(cli_valueless_option(
1007 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1008 self.assertEqual(cli_valueless_option(
1009 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1010 self.assertEqual(cli_valueless_option(
1011 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1012
1013 def test_cli_bool_option(self):
1014 self.assertEqual(
1015 cli_bool_option(
1016 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1017 ['--no-check-certificate', 'true'])
1018 self.assertEqual(
1019 cli_bool_option(
1020 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1021 ['--no-check-certificate=true'])
1022 self.assertEqual(
1023 cli_bool_option(
1024 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1025 ['--check-certificate', 'false'])
1026 self.assertEqual(
1027 cli_bool_option(
1028 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1029 ['--check-certificate=false'])
1030 self.assertEqual(
1031 cli_bool_option(
1032 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1033 ['--check-certificate', 'true'])
1034 self.assertEqual(
1035 cli_bool_option(
1036 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1037 ['--check-certificate=true'])
1038
1039 def test_ohdave_rsa_encrypt(self):
1040 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1041 e = 65537
1042
1043 self.assertEqual(
1044 ohdave_rsa_encrypt(b'aa111222', e, N),
1045 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1046
1047 def test_encode_base_n(self):
1048 self.assertEqual(encode_base_n(0, 30), '0')
1049 self.assertEqual(encode_base_n(80, 30), '2k')
1050
1051 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1052 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1053 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1054
1055 self.assertRaises(ValueError, encode_base_n, 0, 70)
1056 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1057
1058 def test_urshift(self):
1059 self.assertEqual(urshift(3, 1), 1)
1060 self.assertEqual(urshift(-3, 1), 2147483646)
1061
1062 def test_get_element_by_class(self):
1063 html = '''
1064 <span class="foo bar">nice</span>
1065 '''
1066
1067 self.assertEqual(get_element_by_class('foo', html), 'nice')
1068 self.assertEqual(get_element_by_class('no-such-class', html), None)
1069
1070if __name__ == '__main__':
1071 unittest.main()