]> Raphaël G. Git Repositories - youtubedl/blob - test/test_utils.py
3cdb21d409b97c4dec4a1c205960ae133642c620
[youtubedl] / test / test_utils.py
1 #!/usr/bin/env python
2 # coding: utf-8
3
4 from __future__ import unicode_literals
5
6 # Allow direct execution
7 import os
8 import sys
9 import unittest
10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
11
12
13 # Various small unit tests
14 import io
15 import json
16 import xml.etree.ElementTree
17
18 from youtube_dl.utils import (
19 age_restricted,
20 args_to_str,
21 encode_base_n,
22 clean_html,
23 date_from_str,
24 DateRange,
25 detect_exe_version,
26 determine_ext,
27 dict_get,
28 encode_compat_str,
29 encodeFilename,
30 escape_rfc3986,
31 escape_url,
32 extract_attributes,
33 ExtractorError,
34 find_xpath_attr,
35 fix_xml_ampersands,
36 get_element_by_class,
37 get_element_by_attribute,
38 get_elements_by_class,
39 get_elements_by_attribute,
40 InAdvancePagedList,
41 intlist_to_bytes,
42 is_html,
43 js_to_json,
44 limit_length,
45 mimetype2ext,
46 month_by_name,
47 ohdave_rsa_encrypt,
48 OnDemandPagedList,
49 orderedSet,
50 parse_age_limit,
51 parse_duration,
52 parse_filesize,
53 parse_count,
54 parse_iso8601,
55 read_batch_urls,
56 sanitize_filename,
57 sanitize_path,
58 prepend_extension,
59 replace_extension,
60 remove_start,
61 remove_end,
62 remove_quotes,
63 shell_quote,
64 smuggle_url,
65 str_to_int,
66 strip_jsonp,
67 timeconvert,
68 unescapeHTML,
69 unified_strdate,
70 unified_timestamp,
71 unsmuggle_url,
72 uppercase_escape,
73 lowercase_escape,
74 url_basename,
75 base_url,
76 urljoin,
77 urlencode_postdata,
78 urshift,
79 update_url_query,
80 version_tuple,
81 xpath_with_ns,
82 xpath_element,
83 xpath_text,
84 xpath_attr,
85 render_table,
86 match_str,
87 parse_dfxp_time_expr,
88 dfxp2srt,
89 cli_option,
90 cli_valueless_option,
91 cli_bool_option,
92 parse_codecs,
93 )
94 from youtube_dl.compat import (
95 compat_chr,
96 compat_etree_fromstring,
97 compat_urlparse,
98 compat_parse_qs,
99 )
100
101
102 class TestUtil(unittest.TestCase):
103 def test_timeconvert(self):
104 self.assertTrue(timeconvert('') is None)
105 self.assertTrue(timeconvert('bougrg') is None)
106
107 def test_sanitize_filename(self):
108 self.assertEqual(sanitize_filename('abc'), 'abc')
109 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
110
111 self.assertEqual(sanitize_filename('123'), '123')
112
113 self.assertEqual('abc_de', sanitize_filename('abc/de'))
114 self.assertFalse('/' in sanitize_filename('abc/de///'))
115
116 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
117 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
118 self.assertEqual('yes no', sanitize_filename('yes? no'))
119 self.assertEqual('this - that', sanitize_filename('this: that'))
120
121 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
122 aumlaut = 'ä'
123 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
124 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
125 self.assertEqual(sanitize_filename(tests), tests)
126
127 self.assertEqual(
128 sanitize_filename('New World record at 0:12:34'),
129 'New World record at 0_12_34')
130
131 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
132 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
133 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
134 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
135
136 forbidden = '"\0\\/'
137 for fc in forbidden:
138 for fbc in forbidden:
139 self.assertTrue(fbc not in sanitize_filename(fc))
140
141 def test_sanitize_filename_restricted(self):
142 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
143 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
144
145 self.assertEqual(sanitize_filename('123', restricted=True), '123')
146
147 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
148 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
149
150 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
151 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
152 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
153 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
154
155 tests = 'aäb\u4e2d\u56fd\u7684c'
156 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
157 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
158
159 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
160 for fc in forbidden:
161 for fbc in forbidden:
162 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
163
164 # Handle a common case more neatly
165 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
166 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
167 # .. but make sure the file name is never empty
168 self.assertTrue(sanitize_filename('-', restricted=True) != '')
169 self.assertTrue(sanitize_filename(':', restricted=True) != '')
170
171 self.assertEqual(sanitize_filename(
172 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
173 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy')
174
175 def test_sanitize_ids(self):
176 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
177 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
178 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
179
180 def test_sanitize_path(self):
181 if sys.platform != 'win32':
182 return
183
184 self.assertEqual(sanitize_path('abc'), 'abc')
185 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
186 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
187 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
188 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
189 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
190 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
191
192 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
193 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
194
195 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
196 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
197 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
198 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
199
200 self.assertEqual(
201 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
202 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
203
204 self.assertEqual(
205 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
206 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
207 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
208 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
209 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
210
211 self.assertEqual(sanitize_path('../abc'), '..\\abc')
212 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
213 self.assertEqual(sanitize_path('./abc'), 'abc')
214 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
215
216 def test_prepend_extension(self):
217 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
218 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
219 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
220 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
221 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
222 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
223
224 def test_replace_extension(self):
225 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
226 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
227 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
228 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
229 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
230 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
231
232 def test_remove_start(self):
233 self.assertEqual(remove_start(None, 'A - '), None)
234 self.assertEqual(remove_start('A - B', 'A - '), 'B')
235 self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
236
237 def test_remove_end(self):
238 self.assertEqual(remove_end(None, ' - B'), None)
239 self.assertEqual(remove_end('A - B', ' - B'), 'A')
240 self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
241
242 def test_remove_quotes(self):
243 self.assertEqual(remove_quotes(None), None)
244 self.assertEqual(remove_quotes('"'), '"')
245 self.assertEqual(remove_quotes("'"), "'")
246 self.assertEqual(remove_quotes(';'), ';')
247 self.assertEqual(remove_quotes('";'), '";')
248 self.assertEqual(remove_quotes('""'), '')
249 self.assertEqual(remove_quotes('";"'), ';')
250
251 def test_ordered_set(self):
252 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
253 self.assertEqual(orderedSet([]), [])
254 self.assertEqual(orderedSet([1]), [1])
255 # keep the list ordered
256 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
257
258 def test_unescape_html(self):
259 self.assertEqual(unescapeHTML('%20;'), '%20;')
260 self.assertEqual(unescapeHTML('&#x2F;'), '/')
261 self.assertEqual(unescapeHTML('&#47;'), '/')
262 self.assertEqual(unescapeHTML('&eacute;'), 'é')
263 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
264 # HTML5 entities
265 self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
266
267 def test_date_from_str(self):
268 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
269 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
270 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
271 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
272 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
273
274 def test_daterange(self):
275 _20century = DateRange("19000101", "20000101")
276 self.assertFalse("17890714" in _20century)
277 _ac = DateRange("00010101")
278 self.assertTrue("19690721" in _ac)
279 _firstmilenium = DateRange(end="10000101")
280 self.assertTrue("07110427" in _firstmilenium)
281
282 def test_unified_dates(self):
283 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
284 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
285 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
286 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
287 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
288 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
289 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
290 self.assertEqual(
291 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
292 '20141126')
293 self.assertEqual(
294 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
295 '20150202')
296 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
297 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
298 self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
299 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
300 self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
301 self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
302 self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
303 self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
304
305 def test_unified_timestamps(self):
306 self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
307 self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
308 self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
309 self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
310 self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
311 self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
312 self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
313 self.assertEqual(
314 unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
315 1417001400)
316 self.assertEqual(
317 unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
318 1422902860)
319 self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
320 self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
321 self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
322 self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
323 self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
324 self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
325
326 def test_determine_ext(self):
327 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
328 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
329 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
330 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
331 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
332
333 def test_find_xpath_attr(self):
334 testxml = '''<root>
335 <node/>
336 <node x="a"/>
337 <node x="a" y="c" />
338 <node x="b" y="d" />
339 <node x="" />
340 </root>'''
341 doc = compat_etree_fromstring(testxml)
342
343 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
344 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
345 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
346 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
347 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
348 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
349 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
350 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
351 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
352 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
353 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
354
355 def test_xpath_with_ns(self):
356 testxml = '''<root xmlns:media="http://example.com/">
357 <media:song>
358 <media:author>The Author</media:author>
359 <url>http://server.com/download.mp3</url>
360 </media:song>
361 </root>'''
362 doc = compat_etree_fromstring(testxml)
363 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
364 self.assertTrue(find('media:song') is not None)
365 self.assertEqual(find('media:song/media:author').text, 'The Author')
366 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
367
368 def test_xpath_element(self):
369 doc = xml.etree.ElementTree.Element('root')
370 div = xml.etree.ElementTree.SubElement(doc, 'div')
371 p = xml.etree.ElementTree.SubElement(div, 'p')
372 p.text = 'Foo'
373 self.assertEqual(xpath_element(doc, 'div/p'), p)
374 self.assertEqual(xpath_element(doc, ['div/p']), p)
375 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
376 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
377 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
378 self.assertTrue(xpath_element(doc, 'div/bar') is None)
379 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
380 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
381 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
382 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
383 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
384
385 def test_xpath_text(self):
386 testxml = '''<root>
387 <div>
388 <p>Foo</p>
389 </div>
390 </root>'''
391 doc = compat_etree_fromstring(testxml)
392 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
393 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
394 self.assertTrue(xpath_text(doc, 'div/bar') is None)
395 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
396
397 def test_xpath_attr(self):
398 testxml = '''<root>
399 <div>
400 <p x="a">Foo</p>
401 </div>
402 </root>'''
403 doc = compat_etree_fromstring(testxml)
404 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
405 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
406 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
407 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
408 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
409 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
410 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
411
412 def test_smuggle_url(self):
413 data = {"ö": "ö", "abc": [3]}
414 url = 'https://foo.bar/baz?x=y#a'
415 smug_url = smuggle_url(url, data)
416 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
417 self.assertEqual(url, unsmug_url)
418 self.assertEqual(data, unsmug_data)
419
420 res_url, res_data = unsmuggle_url(url)
421 self.assertEqual(res_url, url)
422 self.assertEqual(res_data, None)
423
424 smug_url = smuggle_url(url, {'a': 'b'})
425 smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
426 res_url, res_data = unsmuggle_url(smug_smug_url)
427 self.assertEqual(res_url, url)
428 self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
429
430 def test_shell_quote(self):
431 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
432 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
433
434 def test_str_to_int(self):
435 self.assertEqual(str_to_int('123,456'), 123456)
436 self.assertEqual(str_to_int('123.456'), 123456)
437
438 def test_url_basename(self):
439 self.assertEqual(url_basename('http://foo.de/'), '')
440 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
441 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
442 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
443 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
444 self.assertEqual(
445 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
446 'trailer.mp4')
447
448 def test_base_url(self):
449 self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
450 self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
451 self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
452 self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
453 self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
454
455 def test_urljoin(self):
456 self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
457 self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
458 self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
459 self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
460 self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
461 self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
462 self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
463 self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
464 self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
465 self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
466 self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
467 self.assertEqual(urljoin('http://foo.de/', None), None)
468 self.assertEqual(urljoin('http://foo.de/', ''), None)
469 self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
470 self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
471
472 def test_parse_age_limit(self):
473 self.assertEqual(parse_age_limit(None), None)
474 self.assertEqual(parse_age_limit(False), None)
475 self.assertEqual(parse_age_limit('invalid'), None)
476 self.assertEqual(parse_age_limit(0), 0)
477 self.assertEqual(parse_age_limit(18), 18)
478 self.assertEqual(parse_age_limit(21), 21)
479 self.assertEqual(parse_age_limit(22), None)
480 self.assertEqual(parse_age_limit('18'), 18)
481 self.assertEqual(parse_age_limit('18+'), 18)
482 self.assertEqual(parse_age_limit('PG-13'), 13)
483 self.assertEqual(parse_age_limit('TV-14'), 14)
484 self.assertEqual(parse_age_limit('TV-MA'), 17)
485
486 def test_parse_duration(self):
487 self.assertEqual(parse_duration(None), None)
488 self.assertEqual(parse_duration(False), None)
489 self.assertEqual(parse_duration('invalid'), None)
490 self.assertEqual(parse_duration('1'), 1)
491 self.assertEqual(parse_duration('1337:12'), 80232)
492 self.assertEqual(parse_duration('9:12:43'), 33163)
493 self.assertEqual(parse_duration('12:00'), 720)
494 self.assertEqual(parse_duration('00:01:01'), 61)
495 self.assertEqual(parse_duration('x:y'), None)
496 self.assertEqual(parse_duration('3h11m53s'), 11513)
497 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
498 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
499 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
500 self.assertEqual(parse_duration('62m45s'), 3765)
501 self.assertEqual(parse_duration('6m59s'), 419)
502 self.assertEqual(parse_duration('49s'), 49)
503 self.assertEqual(parse_duration('0h0m0s'), 0)
504 self.assertEqual(parse_duration('0m0s'), 0)
505 self.assertEqual(parse_duration('0s'), 0)
506 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
507 self.assertEqual(parse_duration('T30M38S'), 1838)
508 self.assertEqual(parse_duration('5 s'), 5)
509 self.assertEqual(parse_duration('3 min'), 180)
510 self.assertEqual(parse_duration('2.5 hours'), 9000)
511 self.assertEqual(parse_duration('02:03:04'), 7384)
512 self.assertEqual(parse_duration('01:02:03:04'), 93784)
513 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
514 self.assertEqual(parse_duration('87 Min.'), 5220)
515 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
516 self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
517
518 def test_fix_xml_ampersands(self):
519 self.assertEqual(
520 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
521 self.assertEqual(
522 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
523 '"&amp;x=y&amp;wrong;&amp;z=a')
524 self.assertEqual(
525 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
526 '&amp;&apos;&gt;&lt;&quot;')
527 self.assertEqual(
528 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
529 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
530
531 def test_paged_list(self):
532 def testPL(size, pagesize, sliceargs, expected):
533 def get_page(pagenum):
534 firstid = pagenum * pagesize
535 upto = min(size, pagenum * pagesize + pagesize)
536 for i in range(firstid, upto):
537 yield i
538
539 pl = OnDemandPagedList(get_page, pagesize)
540 got = pl.getslice(*sliceargs)
541 self.assertEqual(got, expected)
542
543 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
544 got = iapl.getslice(*sliceargs)
545 self.assertEqual(got, expected)
546
547 testPL(5, 2, (), [0, 1, 2, 3, 4])
548 testPL(5, 2, (1,), [1, 2, 3, 4])
549 testPL(5, 2, (2,), [2, 3, 4])
550 testPL(5, 2, (4,), [4])
551 testPL(5, 2, (0, 3), [0, 1, 2])
552 testPL(5, 2, (1, 4), [1, 2, 3])
553 testPL(5, 2, (2, 99), [2, 3, 4])
554 testPL(5, 2, (20, 99), [])
555
556 def test_read_batch_urls(self):
557 f = io.StringIO('''\xef\xbb\xbf foo
558 bar\r
559 baz
560 # More after this line\r
561 ; or after this
562 bam''')
563 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
564
565 def test_urlencode_postdata(self):
566 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
567 self.assertTrue(isinstance(data, bytes))
568
569 def test_update_url_query(self):
570 def query_dict(url):
571 return compat_parse_qs(compat_urlparse.urlparse(url).query)
572 self.assertEqual(query_dict(update_url_query(
573 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
574 query_dict('http://example.com/path?quality=HD&format=mp4'))
575 self.assertEqual(query_dict(update_url_query(
576 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
577 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
578 self.assertEqual(query_dict(update_url_query(
579 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
580 query_dict('http://example.com/path?fields=id,formats,subtitles'))
581 self.assertEqual(query_dict(update_url_query(
582 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
583 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
584 self.assertEqual(query_dict(update_url_query(
585 'http://example.com/path?manifest=f4m', {'manifest': []})),
586 query_dict('http://example.com/path'))
587 self.assertEqual(query_dict(update_url_query(
588 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
589 query_dict('http://example.com/path?system=LINUX'))
590 self.assertEqual(query_dict(update_url_query(
591 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
592 query_dict('http://example.com/path?fields=id,formats,subtitles'))
593 self.assertEqual(query_dict(update_url_query(
594 'http://example.com/path', {'width': 1080, 'height': 720})),
595 query_dict('http://example.com/path?width=1080&height=720'))
596 self.assertEqual(query_dict(update_url_query(
597 'http://example.com/path', {'bitrate': 5020.43})),
598 query_dict('http://example.com/path?bitrate=5020.43'))
599 self.assertEqual(query_dict(update_url_query(
600 'http://example.com/path', {'test': '第二行тест'})),
601 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
602
603 def test_dict_get(self):
604 FALSE_VALUES = {
605 'none': None,
606 'false': False,
607 'zero': 0,
608 'empty_string': '',
609 'empty_list': [],
610 }
611 d = FALSE_VALUES.copy()
612 d['a'] = 42
613 self.assertEqual(dict_get(d, 'a'), 42)
614 self.assertEqual(dict_get(d, 'b'), None)
615 self.assertEqual(dict_get(d, 'b', 42), 42)
616 self.assertEqual(dict_get(d, ('a', )), 42)
617 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
618 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
619 self.assertEqual(dict_get(d, ('b', 'c', )), None)
620 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
621 for key, false_value in FALSE_VALUES.items():
622 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
623 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
624
625 def test_encode_compat_str(self):
626 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
627 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
628
629 def test_parse_iso8601(self):
630 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
631 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
632 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
633 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
634 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
635 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
636
637 def test_strip_jsonp(self):
638 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
639 d = json.loads(stripped)
640 self.assertEqual(d, [{"id": "532cb", "x": 3}])
641
642 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
643 d = json.loads(stripped)
644 self.assertEqual(d, {'STATUS': 'OK'})
645
646 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
647 d = json.loads(stripped)
648 self.assertEqual(d, {'status': 'success'})
649
650 def test_uppercase_escape(self):
651 self.assertEqual(uppercase_escape('aä'), 'aä')
652 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
653
654 def test_lowercase_escape(self):
655 self.assertEqual(lowercase_escape('aä'), 'aä')
656 self.assertEqual(lowercase_escape('\\u0026'), '&')
657
658 def test_limit_length(self):
659 self.assertEqual(limit_length(None, 12), None)
660 self.assertEqual(limit_length('foo', 12), 'foo')
661 self.assertTrue(
662 limit_length('foo bar baz asd', 12).startswith('foo bar'))
663 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
664
665 def test_mimetype2ext(self):
666 self.assertEqual(mimetype2ext(None), None)
667 self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
668 self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
669 self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
670 self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
671 self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
672
673 def test_month_by_name(self):
674 self.assertEqual(month_by_name(None), None)
675 self.assertEqual(month_by_name('December', 'en'), 12)
676 self.assertEqual(month_by_name('décembre', 'fr'), 12)
677 self.assertEqual(month_by_name('December'), 12)
678 self.assertEqual(month_by_name('décembre'), None)
679 self.assertEqual(month_by_name('Unknown', 'unknown'), None)
680
681 def test_parse_codecs(self):
682 self.assertEqual(parse_codecs(''), {})
683 self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
684 'vcodec': 'avc1.77.30',
685 'acodec': 'mp4a.40.2',
686 })
687 self.assertEqual(parse_codecs('mp4a.40.2'), {
688 'vcodec': 'none',
689 'acodec': 'mp4a.40.2',
690 })
691 self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
692 'vcodec': 'avc1.42001e',
693 'acodec': 'mp4a.40.5',
694 })
695 self.assertEqual(parse_codecs('avc3.640028'), {
696 'vcodec': 'avc3.640028',
697 'acodec': 'none',
698 })
699 self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
700 'vcodec': 'h264',
701 'acodec': 'aac',
702 })
703
704 def test_escape_rfc3986(self):
705 reserved = "!*'();:@&=+$,/?#[]"
706 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
707 self.assertEqual(escape_rfc3986(reserved), reserved)
708 self.assertEqual(escape_rfc3986(unreserved), unreserved)
709 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
710 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
711 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
712 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
713
714 def test_escape_url(self):
715 self.assertEqual(
716 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
717 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
718 )
719 self.assertEqual(
720 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
721 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
722 )
723 self.assertEqual(
724 escape_url('http://тест.рф/фрагмент'),
725 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
726 )
727 self.assertEqual(
728 escape_url('http://тест.рф/абв?абв=абв#абв'),
729 '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'
730 )
731 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
732
733 def test_js_to_json_realworld(self):
734 inp = '''{
735 'clip':{'provider':'pseudo'}
736 }'''
737 self.assertEqual(js_to_json(inp), '''{
738 "clip":{"provider":"pseudo"}
739 }''')
740 json.loads(js_to_json(inp))
741
742 inp = '''{
743 'playlist':[{'controls':{'all':null}}]
744 }'''
745 self.assertEqual(js_to_json(inp), '''{
746 "playlist":[{"controls":{"all":null}}]
747 }''')
748
749 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
750 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
751
752 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
753 json_code = js_to_json(inp)
754 self.assertEqual(json.loads(json_code), json.loads(inp))
755
756 inp = '''{
757 0:{src:'skipped', type: 'application/dash+xml'},
758 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
759 }'''
760 self.assertEqual(js_to_json(inp), '''{
761 "0":{"src":"skipped", "type": "application/dash+xml"},
762 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
763 }''')
764
765 inp = '''{"foo":101}'''
766 self.assertEqual(js_to_json(inp), '''{"foo":101}''')
767
768 inp = '''{"duration": "00:01:07"}'''
769 self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
770
771 def test_js_to_json_edgecases(self):
772 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
773 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
774
775 on = js_to_json('{"abc": true}')
776 self.assertEqual(json.loads(on), {'abc': True})
777
778 # Ignore JavaScript code as well
779 on = js_to_json('''{
780 "x": 1,
781 y: "a",
782 z: some.code
783 }''')
784 d = json.loads(on)
785 self.assertEqual(d['x'], 1)
786 self.assertEqual(d['y'], 'a')
787
788 on = js_to_json('["abc", "def",]')
789 self.assertEqual(json.loads(on), ['abc', 'def'])
790
791 on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
792 self.assertEqual(json.loads(on), ['abc', 'def'])
793
794 on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
795 self.assertEqual(json.loads(on), ['abc', 'def'])
796
797 on = js_to_json('{"abc": "def",}')
798 self.assertEqual(json.loads(on), {'abc': 'def'})
799
800 on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
801 self.assertEqual(json.loads(on), {'abc': 'def'})
802
803 on = js_to_json('{ 0: /* " \n */ ",]" , }')
804 self.assertEqual(json.loads(on), {'0': ',]'})
805
806 on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
807 self.assertEqual(json.loads(on), {'0': ',]'})
808
809 on = js_to_json('{ 0: // comment\n1 }')
810 self.assertEqual(json.loads(on), {'0': 1})
811
812 on = js_to_json(r'["<p>x<\/p>"]')
813 self.assertEqual(json.loads(on), ['<p>x</p>'])
814
815 on = js_to_json(r'["\xaa"]')
816 self.assertEqual(json.loads(on), ['\u00aa'])
817
818 on = js_to_json("['a\\\nb']")
819 self.assertEqual(json.loads(on), ['ab'])
820
821 on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
822 self.assertEqual(json.loads(on), ['ab'])
823
824 on = js_to_json('{0xff:0xff}')
825 self.assertEqual(json.loads(on), {'255': 255})
826
827 on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
828 self.assertEqual(json.loads(on), {'255': 255})
829
830 on = js_to_json('{077:077}')
831 self.assertEqual(json.loads(on), {'63': 63})
832
833 on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
834 self.assertEqual(json.loads(on), {'63': 63})
835
836 on = js_to_json('{42:42}')
837 self.assertEqual(json.loads(on), {'42': 42})
838
839 on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
840 self.assertEqual(json.loads(on), {'42': 42})
841
842 def test_extract_attributes(self):
843 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
844 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
845 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
846 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
847 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
848 self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
849 self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
850 self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'}) # XML
851 self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
852 self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'}) # HTML 3.2
853 self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'}) # HTML 4.0
854 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
855 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
856 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
857 self.assertEqual(extract_attributes('<e x >'), {'x': None})
858 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
859 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
860 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
861 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
862 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
863 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
864 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
865 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
866 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
867 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
868 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
869 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
870 self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
871 # "Narrow" Python builds don't support unicode code points outside BMP.
872 try:
873 compat_chr(0x10000)
874 supports_outside_bmp = True
875 except ValueError:
876 supports_outside_bmp = False
877 if supports_outside_bmp:
878 self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
879
880 def test_clean_html(self):
881 self.assertEqual(clean_html('a:\nb'), 'a: b')
882 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
883
884 def test_intlist_to_bytes(self):
885 self.assertEqual(
886 intlist_to_bytes([0, 1, 127, 128, 255]),
887 b'\x00\x01\x7f\x80\xff')
888
889 def test_args_to_str(self):
890 self.assertEqual(
891 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
892 'foo ba/r -baz \'2 be\' \'\''
893 )
894
895 def test_parse_filesize(self):
896 self.assertEqual(parse_filesize(None), None)
897 self.assertEqual(parse_filesize(''), None)
898 self.assertEqual(parse_filesize('91 B'), 91)
899 self.assertEqual(parse_filesize('foobar'), None)
900 self.assertEqual(parse_filesize('2 MiB'), 2097152)
901 self.assertEqual(parse_filesize('5 GB'), 5000000000)
902 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
903 self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
904 self.assertEqual(parse_filesize('1,24 KB'), 1240)
905 self.assertEqual(parse_filesize('1,24 kb'), 1240)
906 self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
907
908 def test_parse_count(self):
909 self.assertEqual(parse_count(None), None)
910 self.assertEqual(parse_count(''), None)
911 self.assertEqual(parse_count('0'), 0)
912 self.assertEqual(parse_count('1000'), 1000)
913 self.assertEqual(parse_count('1.000'), 1000)
914 self.assertEqual(parse_count('1.1k'), 1100)
915 self.assertEqual(parse_count('1.1kk'), 1100000)
916 self.assertEqual(parse_count('1.1kk '), 1100000)
917 self.assertEqual(parse_count('1.1kk views'), 1100000)
918
919 def test_version_tuple(self):
920 self.assertEqual(version_tuple('1'), (1,))
921 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
922 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
923
924 def test_detect_exe_version(self):
925 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
926 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
927 configuration: --prefix=/usr --extra-'''), '1.2.1')
928 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
929 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
930 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
931 Trying to open render node...
932 Success at /dev/dri/renderD128.
933 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
934
935 def test_age_restricted(self):
936 self.assertFalse(age_restricted(None, 10)) # unrestricted content
937 self.assertFalse(age_restricted(1, None)) # unrestricted policy
938 self.assertFalse(age_restricted(8, 10))
939 self.assertTrue(age_restricted(18, 14))
940 self.assertFalse(age_restricted(18, 18))
941
942 def test_is_html(self):
943 self.assertFalse(is_html(b'\x49\x44\x43<html'))
944 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
945 self.assertTrue(is_html( # UTF-8 with BOM
946 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
947 self.assertTrue(is_html( # UTF-16-LE
948 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
949 ))
950 self.assertTrue(is_html( # UTF-16-BE
951 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
952 ))
953 self.assertTrue(is_html( # UTF-32-BE
954 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'))
955 self.assertTrue(is_html( # UTF-32-LE
956 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'))
957
958 def test_render_table(self):
959 self.assertEqual(
960 render_table(
961 ['a', 'bcd'],
962 [[123, 4], [9999, 51]]),
963 'a bcd\n'
964 '123 4\n'
965 '9999 51')
966
967 def test_match_str(self):
968 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
969 self.assertFalse(match_str('xy', {'x': 1200}))
970 self.assertTrue(match_str('!xy', {'x': 1200}))
971 self.assertTrue(match_str('x', {'x': 1200}))
972 self.assertFalse(match_str('!x', {'x': 1200}))
973 self.assertTrue(match_str('x', {'x': 0}))
974 self.assertFalse(match_str('x>0', {'x': 0}))
975 self.assertFalse(match_str('x>0', {}))
976 self.assertTrue(match_str('x>?0', {}))
977 self.assertTrue(match_str('x>1K', {'x': 1200}))
978 self.assertFalse(match_str('x>2K', {'x': 1200}))
979 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
980 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
981 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
982 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
983 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
984 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
985 self.assertFalse(match_str(
986 'like_count > 100 & dislike_count <? 50 & description',
987 {'like_count': 90, 'description': 'foo'}))
988 self.assertTrue(match_str(
989 'like_count > 100 & dislike_count <? 50 & description',
990 {'like_count': 190, 'description': 'foo'}))
991 self.assertFalse(match_str(
992 'like_count > 100 & dislike_count <? 50 & description',
993 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
994 self.assertFalse(match_str(
995 'like_count > 100 & dislike_count <? 50 & description',
996 {'like_count': 190, 'dislike_count': 10}))
997
998 def test_parse_dfxp_time_expr(self):
999 self.assertEqual(parse_dfxp_time_expr(None), None)
1000 self.assertEqual(parse_dfxp_time_expr(''), None)
1001 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
1002 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
1003 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
1004 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
1005 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
1006
1007 def test_dfxp2srt(self):
1008 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
1009 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1010 <body>
1011 <div xml:lang="en">
1012 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
1013 <p begin="1" end="2">第二行<br/>♪♪</p>
1014 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
1015 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
1016 <p begin="-1" end="-1">Ignore, two</p>
1017 <p begin="3" dur="-1">Ignored, three</p>
1018 </div>
1019 </body>
1020 </tt>'''
1021 srt_data = '''1
1022 00:00:00,000 --> 00:00:01,000
1023 The following line contains Chinese characters and special symbols
1024
1025 2
1026 00:00:01,000 --> 00:00:02,000
1027 第二行
1028 ♪♪
1029
1030 3
1031 00:00:02,000 --> 00:00:03,000
1032 Third
1033 Line
1034
1035 '''
1036 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
1037
1038 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
1039 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
1040 <body>
1041 <div xml:lang="en">
1042 <p begin="0" end="1">The first line</p>
1043 </div>
1044 </body>
1045 </tt>'''
1046 srt_data = '''1
1047 00:00:00,000 --> 00:00:01,000
1048 The first line
1049
1050 '''
1051 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
1052
1053 def test_cli_option(self):
1054 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
1055 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
1056 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
1057 self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
1058
1059 def test_cli_valueless_option(self):
1060 self.assertEqual(cli_valueless_option(
1061 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
1062 self.assertEqual(cli_valueless_option(
1063 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
1064 self.assertEqual(cli_valueless_option(
1065 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
1066 self.assertEqual(cli_valueless_option(
1067 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
1068 self.assertEqual(cli_valueless_option(
1069 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
1070 self.assertEqual(cli_valueless_option(
1071 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
1072
1073 def test_cli_bool_option(self):
1074 self.assertEqual(
1075 cli_bool_option(
1076 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
1077 ['--no-check-certificate', 'true'])
1078 self.assertEqual(
1079 cli_bool_option(
1080 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
1081 ['--no-check-certificate=true'])
1082 self.assertEqual(
1083 cli_bool_option(
1084 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1085 ['--check-certificate', 'false'])
1086 self.assertEqual(
1087 cli_bool_option(
1088 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1089 ['--check-certificate=false'])
1090 self.assertEqual(
1091 cli_bool_option(
1092 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
1093 ['--check-certificate', 'true'])
1094 self.assertEqual(
1095 cli_bool_option(
1096 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
1097 ['--check-certificate=true'])
1098
1099 def test_ohdave_rsa_encrypt(self):
1100 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
1101 e = 65537
1102
1103 self.assertEqual(
1104 ohdave_rsa_encrypt(b'aa111222', e, N),
1105 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
1106
1107 def test_encode_base_n(self):
1108 self.assertEqual(encode_base_n(0, 30), '0')
1109 self.assertEqual(encode_base_n(80, 30), '2k')
1110
1111 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
1112 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
1113 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
1114
1115 self.assertRaises(ValueError, encode_base_n, 0, 70)
1116 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
1117
1118 def test_urshift(self):
1119 self.assertEqual(urshift(3, 1), 1)
1120 self.assertEqual(urshift(-3, 1), 2147483646)
1121
1122 def test_get_element_by_class(self):
1123 html = '''
1124 <span class="foo bar">nice</span>
1125 '''
1126
1127 self.assertEqual(get_element_by_class('foo', html), 'nice')
1128 self.assertEqual(get_element_by_class('no-such-class', html), None)
1129
1130 def test_get_element_by_attribute(self):
1131 html = '''
1132 <span class="foo bar">nice</span>
1133 '''
1134
1135 self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
1136 self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
1137 self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
1138
1139 def test_get_elements_by_class(self):
1140 html = '''
1141 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1142 '''
1143
1144 self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
1145 self.assertEqual(get_elements_by_class('no-such-class', html), [])
1146
1147 def test_get_elements_by_attribute(self):
1148 html = '''
1149 <span class="foo bar">nice</span><span class="foo bar">also nice</span>
1150 '''
1151
1152 self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
1153 self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
1154 self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
1155
1156
1157 if __name__ == '__main__':
1158 unittest.main()