]>
Commit | Line | Data |
---|---|---|
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 | merge_dicts, | |
46 | mimetype2ext, | |
47 | month_by_name, | |
48 | multipart_encode, | |
49 | ohdave_rsa_encrypt, | |
50 | OnDemandPagedList, | |
51 | orderedSet, | |
52 | parse_age_limit, | |
53 | parse_duration, | |
54 | parse_filesize, | |
55 | parse_count, | |
56 | parse_iso8601, | |
57 | parse_resolution, | |
58 | pkcs1pad, | |
59 | read_batch_urls, | |
60 | sanitize_filename, | |
61 | sanitize_path, | |
62 | sanitize_url, | |
63 | expand_path, | |
64 | prepend_extension, | |
65 | replace_extension, | |
66 | remove_start, | |
67 | remove_end, | |
68 | remove_quotes, | |
69 | shell_quote, | |
70 | smuggle_url, | |
71 | str_to_int, | |
72 | strip_jsonp, | |
73 | timeconvert, | |
74 | unescapeHTML, | |
75 | unified_strdate, | |
76 | unified_timestamp, | |
77 | unsmuggle_url, | |
78 | uppercase_escape, | |
79 | lowercase_escape, | |
80 | url_basename, | |
81 | url_or_none, | |
82 | base_url, | |
83 | urljoin, | |
84 | urlencode_postdata, | |
85 | urshift, | |
86 | update_url_query, | |
87 | version_tuple, | |
88 | xpath_with_ns, | |
89 | xpath_element, | |
90 | xpath_text, | |
91 | xpath_attr, | |
92 | render_table, | |
93 | match_str, | |
94 | parse_dfxp_time_expr, | |
95 | dfxp2srt, | |
96 | cli_option, | |
97 | cli_valueless_option, | |
98 | cli_bool_option, | |
99 | parse_codecs, | |
100 | ) | |
101 | from youtube_dl.compat import ( | |
102 | compat_chr, | |
103 | compat_etree_fromstring, | |
104 | compat_getenv, | |
105 | compat_os_name, | |
106 | compat_setenv, | |
107 | compat_urlparse, | |
108 | compat_parse_qs, | |
109 | ) | |
110 | ||
111 | ||
112 | class TestUtil(unittest.TestCase): | |
113 | def test_timeconvert(self): | |
114 | self.assertTrue(timeconvert('') is None) | |
115 | self.assertTrue(timeconvert('bougrg') is None) | |
116 | ||
117 | def test_sanitize_filename(self): | |
118 | self.assertEqual(sanitize_filename('abc'), 'abc') | |
119 | self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e') | |
120 | ||
121 | self.assertEqual(sanitize_filename('123'), '123') | |
122 | ||
123 | self.assertEqual('abc_de', sanitize_filename('abc/de')) | |
124 | self.assertFalse('/' in sanitize_filename('abc/de///')) | |
125 | ||
126 | self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de')) | |
127 | self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|')) | |
128 | self.assertEqual('yes no', sanitize_filename('yes? no')) | |
129 | self.assertEqual('this - that', sanitize_filename('this: that')) | |
130 | ||
131 | self.assertEqual(sanitize_filename('AT&T'), 'AT&T') | |
132 | aumlaut = 'ä' | |
133 | self.assertEqual(sanitize_filename(aumlaut), aumlaut) | |
134 | tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430' | |
135 | self.assertEqual(sanitize_filename(tests), tests) | |
136 | ||
137 | self.assertEqual( | |
138 | sanitize_filename('New World record at 0:12:34'), | |
139 | 'New World record at 0_12_34') | |
140 | ||
141 | self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf') | |
142 | self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf') | |
143 | self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf') | |
144 | self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf') | |
145 | ||
146 | forbidden = '"\0\\/' | |
147 | for fc in forbidden: | |
148 | for fbc in forbidden: | |
149 | self.assertTrue(fbc not in sanitize_filename(fc)) | |
150 | ||
151 | def test_sanitize_filename_restricted(self): | |
152 | self.assertEqual(sanitize_filename('abc', restricted=True), 'abc') | |
153 | self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e') | |
154 | ||
155 | self.assertEqual(sanitize_filename('123', restricted=True), '123') | |
156 | ||
157 | self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True)) | |
158 | self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True)) | |
159 | ||
160 | self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True)) | |
161 | self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True)) | |
162 | self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True)) | |
163 | self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True)) | |
164 | ||
165 | tests = 'aäb\u4e2d\u56fd\u7684c' | |
166 | self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c') | |
167 | self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename | |
168 | ||
169 | forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#' | |
170 | for fc in forbidden: | |
171 | for fbc in forbidden: | |
172 | self.assertTrue(fbc not in sanitize_filename(fc, restricted=True)) | |
173 | ||
174 | # Handle a common case more neatly | |
175 | self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song') | |
176 | self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech') | |
177 | # .. but make sure the file name is never empty | |
178 | self.assertTrue(sanitize_filename('-', restricted=True) != '') | |
179 | self.assertTrue(sanitize_filename(':', restricted=True) != '') | |
180 | ||
181 | self.assertEqual(sanitize_filename( | |
182 | 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True), | |
183 | 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYPssaaaaaaaeceeeeiiiionooooooooeuuuuuypy') | |
184 | ||
185 | def test_sanitize_ids(self): | |
186 | self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw') | |
187 | self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw') | |
188 | self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI') | |
189 | ||
190 | def test_sanitize_path(self): | |
191 | if sys.platform != 'win32': | |
192 | return | |
193 | ||
194 | self.assertEqual(sanitize_path('abc'), 'abc') | |
195 | self.assertEqual(sanitize_path('abc/def'), 'abc\\def') | |
196 | self.assertEqual(sanitize_path('abc\\def'), 'abc\\def') | |
197 | self.assertEqual(sanitize_path('abc|def'), 'abc#def') | |
198 | self.assertEqual(sanitize_path('<>:"|?*'), '#######') | |
199 | self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def') | |
200 | self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def') | |
201 | ||
202 | self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc') | |
203 | self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc') | |
204 | ||
205 | self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') | |
206 | self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc') | |
207 | self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f') | |
208 | self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') | |
209 | ||
210 | self.assertEqual( | |
211 | sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'), | |
212 | 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s') | |
213 | ||
214 | self.assertEqual( | |
215 | sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'), | |
216 | 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part') | |
217 | self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#') | |
218 | self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def') | |
219 | self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#') | |
220 | ||
221 | self.assertEqual(sanitize_path('../abc'), '..\\abc') | |
222 | self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc') | |
223 | self.assertEqual(sanitize_path('./abc'), 'abc') | |
224 | self.assertEqual(sanitize_path('./../abc'), '..\\abc') | |
225 | ||
226 | def test_sanitize_url(self): | |
227 | self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar') | |
228 | self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar') | |
229 | self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar') | |
230 | self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar') | |
231 | ||
232 | def test_expand_path(self): | |
233 | def env(var): | |
234 | return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var) | |
235 | ||
236 | compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded') | |
237 | self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded') | |
238 | self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME')) | |
239 | self.assertEqual(expand_path('~'), compat_getenv('HOME')) | |
240 | self.assertEqual( | |
241 | expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')), | |
242 | '%s/expanded' % compat_getenv('HOME')) | |
243 | ||
244 | def test_prepend_extension(self): | |
245 | self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext') | |
246 | self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext') | |
247 | self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp') | |
248 | self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp') | |
249 | self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp') | |
250 | self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext') | |
251 | ||
252 | def test_replace_extension(self): | |
253 | self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp') | |
254 | self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp') | |
255 | self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp') | |
256 | self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp') | |
257 | self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp') | |
258 | self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp') | |
259 | ||
260 | def test_remove_start(self): | |
261 | self.assertEqual(remove_start(None, 'A - '), None) | |
262 | self.assertEqual(remove_start('A - B', 'A - '), 'B') | |
263 | self.assertEqual(remove_start('B - A', 'A - '), 'B - A') | |
264 | ||
265 | def test_remove_end(self): | |
266 | self.assertEqual(remove_end(None, ' - B'), None) | |
267 | self.assertEqual(remove_end('A - B', ' - B'), 'A') | |
268 | self.assertEqual(remove_end('B - A', ' - B'), 'B - A') | |
269 | ||
270 | def test_remove_quotes(self): | |
271 | self.assertEqual(remove_quotes(None), None) | |
272 | self.assertEqual(remove_quotes('"'), '"') | |
273 | self.assertEqual(remove_quotes("'"), "'") | |
274 | self.assertEqual(remove_quotes(';'), ';') | |
275 | self.assertEqual(remove_quotes('";'), '";') | |
276 | self.assertEqual(remove_quotes('""'), '') | |
277 | self.assertEqual(remove_quotes('";"'), ';') | |
278 | ||
279 | def test_ordered_set(self): | |
280 | self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7]) | |
281 | self.assertEqual(orderedSet([]), []) | |
282 | self.assertEqual(orderedSet([1]), [1]) | |
283 | # keep the list ordered | |
284 | self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1]) | |
285 | ||
286 | def test_unescape_html(self): | |
287 | self.assertEqual(unescapeHTML('%20;'), '%20;') | |
288 | self.assertEqual(unescapeHTML('/'), '/') | |
289 | self.assertEqual(unescapeHTML('/'), '/') | |
290 | self.assertEqual(unescapeHTML('é'), 'é') | |
291 | self.assertEqual(unescapeHTML('�'), '�') | |
292 | self.assertEqual(unescapeHTML('&a"'), '&a"') | |
293 | # HTML5 entities | |
294 | self.assertEqual(unescapeHTML('.''), '.\'') | |
295 | ||
296 | def test_date_from_str(self): | |
297 | self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day')) | |
298 | self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week')) | |
299 | self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week')) | |
300 | self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year')) | |
301 | self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month')) | |
302 | ||
303 | def test_daterange(self): | |
304 | _20century = DateRange("19000101", "20000101") | |
305 | self.assertFalse("17890714" in _20century) | |
306 | _ac = DateRange("00010101") | |
307 | self.assertTrue("19690721" in _ac) | |
308 | _firstmilenium = DateRange(end="10000101") | |
309 | self.assertTrue("07110427" in _firstmilenium) | |
310 | ||
311 | def test_unified_dates(self): | |
312 | self.assertEqual(unified_strdate('December 21, 2010'), '20101221') | |
313 | self.assertEqual(unified_strdate('8/7/2009'), '20090708') | |
314 | self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214') | |
315 | self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011') | |
316 | self.assertEqual(unified_strdate('1968 12 10'), '19681210') | |
317 | self.assertEqual(unified_strdate('1968-12-10'), '19681210') | |
318 | self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128') | |
319 | self.assertEqual( | |
320 | unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False), | |
321 | '20141126') | |
322 | self.assertEqual( | |
323 | unified_strdate('2/2/2015 6:47:40 PM', day_first=False), | |
324 | '20150202') | |
325 | self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214') | |
326 | self.assertEqual(unified_strdate('25-09-2014'), '20140925') | |
327 | self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227') | |
328 | self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None) | |
329 | self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207') | |
330 | self.assertEqual(unified_strdate('July 15th, 2013'), '20130715') | |
331 | self.assertEqual(unified_strdate('September 1st, 2013'), '20130901') | |
332 | self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902') | |
333 | ||
334 | def test_unified_timestamps(self): | |
335 | self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600) | |
336 | self.assertEqual(unified_timestamp('8/7/2009'), 1247011200) | |
337 | self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200) | |
338 | self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598) | |
339 | self.assertEqual(unified_timestamp('1968 12 10'), -33436800) | |
340 | self.assertEqual(unified_timestamp('1968-12-10'), -33436800) | |
341 | self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200) | |
342 | self.assertEqual( | |
343 | unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False), | |
344 | 1417001400) | |
345 | self.assertEqual( | |
346 | unified_timestamp('2/2/2015 6:47:40 PM', day_first=False), | |
347 | 1422902860) | |
348 | self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900) | |
349 | self.assertEqual(unified_timestamp('25-09-2014'), 1411603200) | |
350 | self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200) | |
351 | self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None) | |
352 | self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500) | |
353 | self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100) | |
354 | self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361) | |
355 | self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540) | |
356 | self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140) | |
357 | self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363) | |
358 | ||
359 | def test_determine_ext(self): | |
360 | self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4') | |
361 | self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None) | |
362 | self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None) | |
363 | self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None) | |
364 | self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8') | |
365 | self.assertEqual(determine_ext('foobar', None), None) | |
366 | ||
367 | def test_find_xpath_attr(self): | |
368 | testxml = '''<root> | |
369 | <node/> | |
370 | <node x="a"/> | |
371 | <node x="a" y="c" /> | |
372 | <node x="b" y="d" /> | |
373 | <node x="" /> | |
374 | </root>''' | |
375 | doc = compat_etree_fromstring(testxml) | |
376 | ||
377 | self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None) | |
378 | self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None) | |
379 | self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None) | |
380 | self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None) | |
381 | self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1]) | |
382 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1]) | |
383 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3]) | |
384 | self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2]) | |
385 | self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2]) | |
386 | self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3]) | |
387 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4]) | |
388 | ||
389 | def test_xpath_with_ns(self): | |
390 | testxml = '''<root xmlns:media="http://example.com/"> | |
391 | <media:song> | |
392 | <media:author>The Author</media:author> | |
393 | <url>http://server.com/download.mp3</url> | |
394 | </media:song> | |
395 | </root>''' | |
396 | doc = compat_etree_fromstring(testxml) | |
397 | find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'})) | |
398 | self.assertTrue(find('media:song') is not None) | |
399 | self.assertEqual(find('media:song/media:author').text, 'The Author') | |
400 | self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3') | |
401 | ||
402 | def test_xpath_element(self): | |
403 | doc = xml.etree.ElementTree.Element('root') | |
404 | div = xml.etree.ElementTree.SubElement(doc, 'div') | |
405 | p = xml.etree.ElementTree.SubElement(div, 'p') | |
406 | p.text = 'Foo' | |
407 | self.assertEqual(xpath_element(doc, 'div/p'), p) | |
408 | self.assertEqual(xpath_element(doc, ['div/p']), p) | |
409 | self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p) | |
410 | self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default') | |
411 | self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default') | |
412 | self.assertTrue(xpath_element(doc, 'div/bar') is None) | |
413 | self.assertTrue(xpath_element(doc, ['div/bar']) is None) | |
414 | self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None) | |
415 | self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True) | |
416 | self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True) | |
417 | self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True) | |
418 | ||
419 | def test_xpath_text(self): | |
420 | testxml = '''<root> | |
421 | <div> | |
422 | <p>Foo</p> | |
423 | </div> | |
424 | </root>''' | |
425 | doc = compat_etree_fromstring(testxml) | |
426 | self.assertEqual(xpath_text(doc, 'div/p'), 'Foo') | |
427 | self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default') | |
428 | self.assertTrue(xpath_text(doc, 'div/bar') is None) | |
429 | self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True) | |
430 | ||
431 | def test_xpath_attr(self): | |
432 | testxml = '''<root> | |
433 | <div> | |
434 | <p x="a">Foo</p> | |
435 | </div> | |
436 | </root>''' | |
437 | doc = compat_etree_fromstring(testxml) | |
438 | self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a') | |
439 | self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None) | |
440 | self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None) | |
441 | self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default') | |
442 | self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default') | |
443 | self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True) | |
444 | self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True) | |
445 | ||
446 | def test_smuggle_url(self): | |
447 | data = {"ö": "ö", "abc": [3]} | |
448 | url = 'https://foo.bar/baz?x=y#a' | |
449 | smug_url = smuggle_url(url, data) | |
450 | unsmug_url, unsmug_data = unsmuggle_url(smug_url) | |
451 | self.assertEqual(url, unsmug_url) | |
452 | self.assertEqual(data, unsmug_data) | |
453 | ||
454 | res_url, res_data = unsmuggle_url(url) | |
455 | self.assertEqual(res_url, url) | |
456 | self.assertEqual(res_data, None) | |
457 | ||
458 | smug_url = smuggle_url(url, {'a': 'b'}) | |
459 | smug_smug_url = smuggle_url(smug_url, {'c': 'd'}) | |
460 | res_url, res_data = unsmuggle_url(smug_smug_url) | |
461 | self.assertEqual(res_url, url) | |
462 | self.assertEqual(res_data, {'a': 'b', 'c': 'd'}) | |
463 | ||
464 | def test_shell_quote(self): | |
465 | args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')] | |
466 | self.assertEqual( | |
467 | shell_quote(args), | |
468 | """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''') | |
469 | ||
470 | def test_str_to_int(self): | |
471 | self.assertEqual(str_to_int('123,456'), 123456) | |
472 | self.assertEqual(str_to_int('123.456'), 123456) | |
473 | ||
474 | def test_url_basename(self): | |
475 | self.assertEqual(url_basename('http://foo.de/'), '') | |
476 | self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz') | |
477 | self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz') | |
478 | self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz') | |
479 | self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz') | |
480 | self.assertEqual( | |
481 | url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'), | |
482 | 'trailer.mp4') | |
483 | ||
484 | def test_base_url(self): | |
485 | self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/') | |
486 | self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/') | |
487 | self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/') | |
488 | self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/') | |
489 | self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/') | |
490 | ||
491 | def test_urljoin(self): | |
492 | self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
493 | self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
494 | self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
495 | self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
496 | self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt') | |
497 | self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
498 | self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
499 | self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
500 | self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
501 | self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') | |
502 | self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
503 | self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') | |
504 | self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
505 | self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
506 | self.assertEqual(urljoin('http://foo.de/', None), None) | |
507 | self.assertEqual(urljoin('http://foo.de/', ''), None) | |
508 | self.assertEqual(urljoin('http://foo.de/', ['foobar']), None) | |
509 | self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt') | |
510 | ||
511 | def test_url_or_none(self): | |
512 | self.assertEqual(url_or_none(None), None) | |
513 | self.assertEqual(url_or_none(''), None) | |
514 | self.assertEqual(url_or_none('foo'), None) | |
515 | self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de') | |
516 | self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de') | |
517 | self.assertEqual(url_or_none('http$://foo.de'), None) | |
518 | self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de') | |
519 | self.assertEqual(url_or_none('//foo.de'), '//foo.de') | |
520 | ||
521 | def test_parse_age_limit(self): | |
522 | self.assertEqual(parse_age_limit(None), None) | |
523 | self.assertEqual(parse_age_limit(False), None) | |
524 | self.assertEqual(parse_age_limit('invalid'), None) | |
525 | self.assertEqual(parse_age_limit(0), 0) | |
526 | self.assertEqual(parse_age_limit(18), 18) | |
527 | self.assertEqual(parse_age_limit(21), 21) | |
528 | self.assertEqual(parse_age_limit(22), None) | |
529 | self.assertEqual(parse_age_limit('18'), 18) | |
530 | self.assertEqual(parse_age_limit('18+'), 18) | |
531 | self.assertEqual(parse_age_limit('PG-13'), 13) | |
532 | self.assertEqual(parse_age_limit('TV-14'), 14) | |
533 | self.assertEqual(parse_age_limit('TV-MA'), 17) | |
534 | self.assertEqual(parse_age_limit('TV14'), 14) | |
535 | self.assertEqual(parse_age_limit('TV_G'), 0) | |
536 | ||
537 | def test_parse_duration(self): | |
538 | self.assertEqual(parse_duration(None), None) | |
539 | self.assertEqual(parse_duration(False), None) | |
540 | self.assertEqual(parse_duration('invalid'), None) | |
541 | self.assertEqual(parse_duration('1'), 1) | |
542 | self.assertEqual(parse_duration('1337:12'), 80232) | |
543 | self.assertEqual(parse_duration('9:12:43'), 33163) | |
544 | self.assertEqual(parse_duration('12:00'), 720) | |
545 | self.assertEqual(parse_duration('00:01:01'), 61) | |
546 | self.assertEqual(parse_duration('x:y'), None) | |
547 | self.assertEqual(parse_duration('3h11m53s'), 11513) | |
548 | self.assertEqual(parse_duration('3h 11m 53s'), 11513) | |
549 | self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513) | |
550 | self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513) | |
551 | self.assertEqual(parse_duration('62m45s'), 3765) | |
552 | self.assertEqual(parse_duration('6m59s'), 419) | |
553 | self.assertEqual(parse_duration('49s'), 49) | |
554 | self.assertEqual(parse_duration('0h0m0s'), 0) | |
555 | self.assertEqual(parse_duration('0m0s'), 0) | |
556 | self.assertEqual(parse_duration('0s'), 0) | |
557 | self.assertEqual(parse_duration('01:02:03.05'), 3723.05) | |
558 | self.assertEqual(parse_duration('T30M38S'), 1838) | |
559 | self.assertEqual(parse_duration('5 s'), 5) | |
560 | self.assertEqual(parse_duration('3 min'), 180) | |
561 | self.assertEqual(parse_duration('2.5 hours'), 9000) | |
562 | self.assertEqual(parse_duration('02:03:04'), 7384) | |
563 | self.assertEqual(parse_duration('01:02:03:04'), 93784) | |
564 | self.assertEqual(parse_duration('1 hour 3 minutes'), 3780) | |
565 | self.assertEqual(parse_duration('87 Min.'), 5220) | |
566 | self.assertEqual(parse_duration('PT1H0.040S'), 3600.04) | |
567 | self.assertEqual(parse_duration('PT00H03M30SZ'), 210) | |
568 | self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88) | |
569 | ||
570 | def test_fix_xml_ampersands(self): | |
571 | self.assertEqual( | |
572 | fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a') | |
573 | self.assertEqual( | |
574 | fix_xml_ampersands('"&x=y&wrong;&z=a'), | |
575 | '"&x=y&wrong;&z=a') | |
576 | self.assertEqual( | |
577 | fix_xml_ampersands('&'><"'), | |
578 | '&'><"') | |
579 | self.assertEqual( | |
580 | fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼') | |
581 | self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#') | |
582 | ||
583 | def test_paged_list(self): | |
584 | def testPL(size, pagesize, sliceargs, expected): | |
585 | def get_page(pagenum): | |
586 | firstid = pagenum * pagesize | |
587 | upto = min(size, pagenum * pagesize + pagesize) | |
588 | for i in range(firstid, upto): | |
589 | yield i | |
590 | ||
591 | pl = OnDemandPagedList(get_page, pagesize) | |
592 | got = pl.getslice(*sliceargs) | |
593 | self.assertEqual(got, expected) | |
594 | ||
595 | iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize) | |
596 | got = iapl.getslice(*sliceargs) | |
597 | self.assertEqual(got, expected) | |
598 | ||
599 | testPL(5, 2, (), [0, 1, 2, 3, 4]) | |
600 | testPL(5, 2, (1,), [1, 2, 3, 4]) | |
601 | testPL(5, 2, (2,), [2, 3, 4]) | |
602 | testPL(5, 2, (4,), [4]) | |
603 | testPL(5, 2, (0, 3), [0, 1, 2]) | |
604 | testPL(5, 2, (1, 4), [1, 2, 3]) | |
605 | testPL(5, 2, (2, 99), [2, 3, 4]) | |
606 | testPL(5, 2, (20, 99), []) | |
607 | ||
608 | def test_read_batch_urls(self): | |
609 | f = io.StringIO('''\xef\xbb\xbf foo | |
610 | bar\r | |
611 | baz | |
612 | # More after this line\r | |
613 | ; or after this | |
614 | bam''') | |
615 | self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam']) | |
616 | ||
617 | def test_urlencode_postdata(self): | |
618 | data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}) | |
619 | self.assertTrue(isinstance(data, bytes)) | |
620 | ||
621 | def test_update_url_query(self): | |
622 | def query_dict(url): | |
623 | return compat_parse_qs(compat_urlparse.urlparse(url).query) | |
624 | self.assertEqual(query_dict(update_url_query( | |
625 | 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})), | |
626 | query_dict('http://example.com/path?quality=HD&format=mp4')) | |
627 | self.assertEqual(query_dict(update_url_query( | |
628 | 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})), | |
629 | query_dict('http://example.com/path?system=LINUX&system=WINDOWS')) | |
630 | self.assertEqual(query_dict(update_url_query( | |
631 | 'http://example.com/path', {'fields': 'id,formats,subtitles'})), | |
632 | query_dict('http://example.com/path?fields=id,formats,subtitles')) | |
633 | self.assertEqual(query_dict(update_url_query( | |
634 | 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})), | |
635 | query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails')) | |
636 | self.assertEqual(query_dict(update_url_query( | |
637 | 'http://example.com/path?manifest=f4m', {'manifest': []})), | |
638 | query_dict('http://example.com/path')) | |
639 | self.assertEqual(query_dict(update_url_query( | |
640 | 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})), | |
641 | query_dict('http://example.com/path?system=LINUX')) | |
642 | self.assertEqual(query_dict(update_url_query( | |
643 | 'http://example.com/path', {'fields': b'id,formats,subtitles'})), | |
644 | query_dict('http://example.com/path?fields=id,formats,subtitles')) | |
645 | self.assertEqual(query_dict(update_url_query( | |
646 | 'http://example.com/path', {'width': 1080, 'height': 720})), | |
647 | query_dict('http://example.com/path?width=1080&height=720')) | |
648 | self.assertEqual(query_dict(update_url_query( | |
649 | 'http://example.com/path', {'bitrate': 5020.43})), | |
650 | query_dict('http://example.com/path?bitrate=5020.43')) | |
651 | self.assertEqual(query_dict(update_url_query( | |
652 | 'http://example.com/path', {'test': '第二行тест'})), | |
653 | query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82')) | |
654 | ||
655 | def test_multipart_encode(self): | |
656 | self.assertEqual( | |
657 | multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0], | |
658 | b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n') | |
659 | self.assertEqual( | |
660 | multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0], | |
661 | b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n') | |
662 | self.assertRaises( | |
663 | ValueError, multipart_encode, {b'field': b'value'}, boundary='value') | |
664 | ||
665 | def test_dict_get(self): | |
666 | FALSE_VALUES = { | |
667 | 'none': None, | |
668 | 'false': False, | |
669 | 'zero': 0, | |
670 | 'empty_string': '', | |
671 | 'empty_list': [], | |
672 | } | |
673 | d = FALSE_VALUES.copy() | |
674 | d['a'] = 42 | |
675 | self.assertEqual(dict_get(d, 'a'), 42) | |
676 | self.assertEqual(dict_get(d, 'b'), None) | |
677 | self.assertEqual(dict_get(d, 'b', 42), 42) | |
678 | self.assertEqual(dict_get(d, ('a', )), 42) | |
679 | self.assertEqual(dict_get(d, ('b', 'a', )), 42) | |
680 | self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42) | |
681 | self.assertEqual(dict_get(d, ('b', 'c', )), None) | |
682 | self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42) | |
683 | for key, false_value in FALSE_VALUES.items(): | |
684 | self.assertEqual(dict_get(d, ('b', 'c', key, )), None) | |
685 | self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value) | |
686 | ||
687 | def test_merge_dicts(self): | |
688 | self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2}) | |
689 | self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1}) | |
690 | self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1}) | |
691 | self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1}) | |
692 | self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1}) | |
693 | self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1}) | |
694 | self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''}) | |
695 | self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'}) | |
696 | self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'}) | |
697 | ||
698 | def test_encode_compat_str(self): | |
699 | self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест') | |
700 | self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест') | |
701 | ||
702 | def test_parse_iso8601(self): | |
703 | self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266) | |
704 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266) | |
705 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266) | |
706 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266) | |
707 | self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251) | |
708 | self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None) | |
709 | ||
710 | def test_strip_jsonp(self): | |
711 | stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);') | |
712 | d = json.loads(stripped) | |
713 | self.assertEqual(d, [{"id": "532cb", "x": 3}]) | |
714 | ||
715 | stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc') | |
716 | d = json.loads(stripped) | |
717 | self.assertEqual(d, {'STATUS': 'OK'}) | |
718 | ||
719 | stripped = strip_jsonp('ps.embedHandler({"status": "success"});') | |
720 | d = json.loads(stripped) | |
721 | self.assertEqual(d, {'status': 'success'}) | |
722 | ||
723 | stripped = strip_jsonp('window.cb && window.cb({"status": "success"});') | |
724 | d = json.loads(stripped) | |
725 | self.assertEqual(d, {'status': 'success'}) | |
726 | ||
727 | stripped = strip_jsonp('window.cb && cb({"status": "success"});') | |
728 | d = json.loads(stripped) | |
729 | self.assertEqual(d, {'status': 'success'}) | |
730 | ||
731 | stripped = strip_jsonp('({"status": "success"});') | |
732 | d = json.loads(stripped) | |
733 | self.assertEqual(d, {'status': 'success'}) | |
734 | ||
735 | def test_uppercase_escape(self): | |
736 | self.assertEqual(uppercase_escape('aä'), 'aä') | |
737 | self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐') | |
738 | ||
739 | def test_lowercase_escape(self): | |
740 | self.assertEqual(lowercase_escape('aä'), 'aä') | |
741 | self.assertEqual(lowercase_escape('\\u0026'), '&') | |
742 | ||
743 | def test_limit_length(self): | |
744 | self.assertEqual(limit_length(None, 12), None) | |
745 | self.assertEqual(limit_length('foo', 12), 'foo') | |
746 | self.assertTrue( | |
747 | limit_length('foo bar baz asd', 12).startswith('foo bar')) | |
748 | self.assertTrue('...' in limit_length('foo bar baz asd', 12)) | |
749 | ||
750 | def test_mimetype2ext(self): | |
751 | self.assertEqual(mimetype2ext(None), None) | |
752 | self.assertEqual(mimetype2ext('video/x-flv'), 'flv') | |
753 | self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8') | |
754 | self.assertEqual(mimetype2ext('text/vtt'), 'vtt') | |
755 | self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt') | |
756 | self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html') | |
757 | ||
758 | def test_month_by_name(self): | |
759 | self.assertEqual(month_by_name(None), None) | |
760 | self.assertEqual(month_by_name('December', 'en'), 12) | |
761 | self.assertEqual(month_by_name('décembre', 'fr'), 12) | |
762 | self.assertEqual(month_by_name('December'), 12) | |
763 | self.assertEqual(month_by_name('décembre'), None) | |
764 | self.assertEqual(month_by_name('Unknown', 'unknown'), None) | |
765 | ||
766 | def test_parse_codecs(self): | |
767 | self.assertEqual(parse_codecs(''), {}) | |
768 | self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), { | |
769 | 'vcodec': 'avc1.77.30', | |
770 | 'acodec': 'mp4a.40.2', | |
771 | }) | |
772 | self.assertEqual(parse_codecs('mp4a.40.2'), { | |
773 | 'vcodec': 'none', | |
774 | 'acodec': 'mp4a.40.2', | |
775 | }) | |
776 | self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), { | |
777 | 'vcodec': 'avc1.42001e', | |
778 | 'acodec': 'mp4a.40.5', | |
779 | }) | |
780 | self.assertEqual(parse_codecs('avc3.640028'), { | |
781 | 'vcodec': 'avc3.640028', | |
782 | 'acodec': 'none', | |
783 | }) | |
784 | self.assertEqual(parse_codecs(', h264,,newcodec,aac'), { | |
785 | 'vcodec': 'h264', | |
786 | 'acodec': 'aac', | |
787 | }) | |
788 | ||
789 | def test_escape_rfc3986(self): | |
790 | reserved = "!*'();:@&=+$,/?#[]" | |
791 | unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~' | |
792 | self.assertEqual(escape_rfc3986(reserved), reserved) | |
793 | self.assertEqual(escape_rfc3986(unreserved), unreserved) | |
794 | self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82') | |
795 | self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82') | |
796 | self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar') | |
797 | self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar') | |
798 | ||
799 | def test_escape_url(self): | |
800 | self.assertEqual( | |
801 | escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'), | |
802 | 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4' | |
803 | ) | |
804 | self.assertEqual( | |
805 | escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'), | |
806 | 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290' | |
807 | ) | |
808 | self.assertEqual( | |
809 | escape_url('http://тест.рф/фрагмент'), | |
810 | 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' | |
811 | ) | |
812 | self.assertEqual( | |
813 | escape_url('http://тест.рф/абв?абв=абв#абв'), | |
814 | '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' | |
815 | ) | |
816 | self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0') | |
817 | ||
818 | def test_js_to_json_realworld(self): | |
819 | inp = '''{ | |
820 | 'clip':{'provider':'pseudo'} | |
821 | }''' | |
822 | self.assertEqual(js_to_json(inp), '''{ | |
823 | "clip":{"provider":"pseudo"} | |
824 | }''') | |
825 | json.loads(js_to_json(inp)) | |
826 | ||
827 | inp = '''{ | |
828 | 'playlist':[{'controls':{'all':null}}] | |
829 | }''' | |
830 | self.assertEqual(js_to_json(inp), '''{ | |
831 | "playlist":[{"controls":{"all":null}}] | |
832 | }''') | |
833 | ||
834 | inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"''' | |
835 | self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''') | |
836 | ||
837 | inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"' | |
838 | json_code = js_to_json(inp) | |
839 | self.assertEqual(json.loads(json_code), json.loads(inp)) | |
840 | ||
841 | inp = '''{ | |
842 | 0:{src:'skipped', type: 'application/dash+xml'}, | |
843 | 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'}, | |
844 | }''' | |
845 | self.assertEqual(js_to_json(inp), '''{ | |
846 | "0":{"src":"skipped", "type": "application/dash+xml"}, | |
847 | "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"} | |
848 | }''') | |
849 | ||
850 | inp = '''{"foo":101}''' | |
851 | self.assertEqual(js_to_json(inp), '''{"foo":101}''') | |
852 | ||
853 | inp = '''{"duration": "00:01:07"}''' | |
854 | self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''') | |
855 | ||
856 | inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''' | |
857 | self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''') | |
858 | ||
859 | def test_js_to_json_edgecases(self): | |
860 | on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}") | |
861 | self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"}) | |
862 | ||
863 | on = js_to_json('{"abc": true}') | |
864 | self.assertEqual(json.loads(on), {'abc': True}) | |
865 | ||
866 | # Ignore JavaScript code as well | |
867 | on = js_to_json('''{ | |
868 | "x": 1, | |
869 | y: "a", | |
870 | z: some.code | |
871 | }''') | |
872 | d = json.loads(on) | |
873 | self.assertEqual(d['x'], 1) | |
874 | self.assertEqual(d['y'], 'a') | |
875 | ||
876 | on = js_to_json('["abc", "def",]') | |
877 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
878 | ||
879 | on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]') | |
880 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
881 | ||
882 | on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]') | |
883 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
884 | ||
885 | on = js_to_json('{"abc": "def",}') | |
886 | self.assertEqual(json.loads(on), {'abc': 'def'}) | |
887 | ||
888 | on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}') | |
889 | self.assertEqual(json.loads(on), {'abc': 'def'}) | |
890 | ||
891 | on = js_to_json('{ 0: /* " \n */ ",]" , }') | |
892 | self.assertEqual(json.loads(on), {'0': ',]'}) | |
893 | ||
894 | on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }') | |
895 | self.assertEqual(json.loads(on), {'0': ',]'}) | |
896 | ||
897 | on = js_to_json('{ 0: // comment\n1 }') | |
898 | self.assertEqual(json.loads(on), {'0': 1}) | |
899 | ||
900 | on = js_to_json(r'["<p>x<\/p>"]') | |
901 | self.assertEqual(json.loads(on), ['<p>x</p>']) | |
902 | ||
903 | on = js_to_json(r'["\xaa"]') | |
904 | self.assertEqual(json.loads(on), ['\u00aa']) | |
905 | ||
906 | on = js_to_json("['a\\\nb']") | |
907 | self.assertEqual(json.loads(on), ['ab']) | |
908 | ||
909 | on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/") | |
910 | self.assertEqual(json.loads(on), ['ab']) | |
911 | ||
912 | on = js_to_json('{0xff:0xff}') | |
913 | self.assertEqual(json.loads(on), {'255': 255}) | |
914 | ||
915 | on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}') | |
916 | self.assertEqual(json.loads(on), {'255': 255}) | |
917 | ||
918 | on = js_to_json('{077:077}') | |
919 | self.assertEqual(json.loads(on), {'63': 63}) | |
920 | ||
921 | on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}') | |
922 | self.assertEqual(json.loads(on), {'63': 63}) | |
923 | ||
924 | on = js_to_json('{42:42}') | |
925 | self.assertEqual(json.loads(on), {'42': 42}) | |
926 | ||
927 | on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}') | |
928 | self.assertEqual(json.loads(on), {'42': 42}) | |
929 | ||
930 | on = js_to_json('{42:4.2e1}') | |
931 | self.assertEqual(json.loads(on), {'42': 42.0}) | |
932 | ||
933 | def test_js_to_json_malformed(self): | |
934 | self.assertEqual(js_to_json('42a1'), '42"a1"') | |
935 | self.assertEqual(js_to_json('42a-1'), '42"a"-1') | |
936 | ||
937 | def test_extract_attributes(self): | |
938 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
939 | self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}) | |
940 | self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}) | |
941 | self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}) | |
942 | self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}) | |
943 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
944 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
945 | self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML | |
946 | self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}) | |
947 | self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2 | |
948 | self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0 | |
949 | self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}) | |
950 | self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}) | |
951 | self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}) | |
952 | self.assertEqual(extract_attributes('<e x >'), {'x': None}) | |
953 | self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}) | |
954 | self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}) | |
955 | self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}) | |
956 | self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}) | |
957 | self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}) | |
958 | self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}) | |
959 | self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}) | |
960 | self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased | |
961 | self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}) | |
962 | self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}) | |
963 | self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}) | |
964 | self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}) | |
965 | self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}) | |
966 | # "Narrow" Python builds don't support unicode code points outside BMP. | |
967 | try: | |
968 | compat_chr(0x10000) | |
969 | supports_outside_bmp = True | |
970 | except ValueError: | |
971 | supports_outside_bmp = False | |
972 | if supports_outside_bmp: | |
973 | self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}) | |
974 | # Malformed HTML should not break attributes extraction on older Python | |
975 | self.assertEqual(extract_attributes('<mal"formed/>'), {}) | |
976 | ||
977 | def test_clean_html(self): | |
978 | self.assertEqual(clean_html('a:\nb'), 'a: b') | |
979 | self.assertEqual(clean_html('a:\n "b"'), 'a: "b"') | |
980 | self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb') | |
981 | ||
982 | def test_intlist_to_bytes(self): | |
983 | self.assertEqual( | |
984 | intlist_to_bytes([0, 1, 127, 128, 255]), | |
985 | b'\x00\x01\x7f\x80\xff') | |
986 | ||
987 | def test_args_to_str(self): | |
988 | self.assertEqual( | |
989 | args_to_str(['foo', 'ba/r', '-baz', '2 be', '']), | |
990 | 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""' | |
991 | ) | |
992 | ||
993 | def test_parse_filesize(self): | |
994 | self.assertEqual(parse_filesize(None), None) | |
995 | self.assertEqual(parse_filesize(''), None) | |
996 | self.assertEqual(parse_filesize('91 B'), 91) | |
997 | self.assertEqual(parse_filesize('foobar'), None) | |
998 | self.assertEqual(parse_filesize('2 MiB'), 2097152) | |
999 | self.assertEqual(parse_filesize('5 GB'), 5000000000) | |
1000 | self.assertEqual(parse_filesize('1.2Tb'), 1200000000000) | |
1001 | self.assertEqual(parse_filesize('1.2tb'), 1200000000000) | |
1002 | self.assertEqual(parse_filesize('1,24 KB'), 1240) | |
1003 | self.assertEqual(parse_filesize('1,24 kb'), 1240) | |
1004 | self.assertEqual(parse_filesize('8.5 megabytes'), 8500000) | |
1005 | ||
1006 | def test_parse_count(self): | |
1007 | self.assertEqual(parse_count(None), None) | |
1008 | self.assertEqual(parse_count(''), None) | |
1009 | self.assertEqual(parse_count('0'), 0) | |
1010 | self.assertEqual(parse_count('1000'), 1000) | |
1011 | self.assertEqual(parse_count('1.000'), 1000) | |
1012 | self.assertEqual(parse_count('1.1k'), 1100) | |
1013 | self.assertEqual(parse_count('1.1kk'), 1100000) | |
1014 | self.assertEqual(parse_count('1.1kk '), 1100000) | |
1015 | self.assertEqual(parse_count('1.1kk views'), 1100000) | |
1016 | ||
1017 | def test_parse_resolution(self): | |
1018 | self.assertEqual(parse_resolution(None), {}) | |
1019 | self.assertEqual(parse_resolution(''), {}) | |
1020 | self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080}) | |
1021 | self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080}) | |
1022 | self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}) | |
1023 | self.assertEqual(parse_resolution('720p'), {'height': 720}) | |
1024 | self.assertEqual(parse_resolution('4k'), {'height': 2160}) | |
1025 | self.assertEqual(parse_resolution('8K'), {'height': 4320}) | |
1026 | ||
1027 | def test_version_tuple(self): | |
1028 | self.assertEqual(version_tuple('1'), (1,)) | |
1029 | self.assertEqual(version_tuple('10.23.344'), (10, 23, 344)) | |
1030 | self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style | |
1031 | ||
1032 | def test_detect_exe_version(self): | |
1033 | self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1 | |
1034 | built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4) | |
1035 | configuration: --prefix=/usr --extra-'''), '1.2.1') | |
1036 | self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685 | |
1037 | built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685') | |
1038 | self.assertEqual(detect_exe_version('''X server found. dri2 connection failed! | |
1039 | Trying to open render node... | |
1040 | Success at /dev/dri/renderD128. | |
1041 | ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4') | |
1042 | ||
1043 | def test_age_restricted(self): | |
1044 | self.assertFalse(age_restricted(None, 10)) # unrestricted content | |
1045 | self.assertFalse(age_restricted(1, None)) # unrestricted policy | |
1046 | self.assertFalse(age_restricted(8, 10)) | |
1047 | self.assertTrue(age_restricted(18, 14)) | |
1048 | self.assertFalse(age_restricted(18, 18)) | |
1049 | ||
1050 | def test_is_html(self): | |
1051 | self.assertFalse(is_html(b'\x49\x44\x43<html')) | |
1052 | self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa')) | |
1053 | self.assertTrue(is_html( # UTF-8 with BOM | |
1054 | b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa')) | |
1055 | self.assertTrue(is_html( # UTF-16-LE | |
1056 | b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00' | |
1057 | )) | |
1058 | self.assertTrue(is_html( # UTF-16-BE | |
1059 | b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4' | |
1060 | )) | |
1061 | self.assertTrue(is_html( # UTF-32-BE | |
1062 | 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')) | |
1063 | self.assertTrue(is_html( # UTF-32-LE | |
1064 | 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')) | |
1065 | ||
1066 | def test_render_table(self): | |
1067 | self.assertEqual( | |
1068 | render_table( | |
1069 | ['a', 'bcd'], | |
1070 | [[123, 4], [9999, 51]]), | |
1071 | 'a bcd\n' | |
1072 | '123 4\n' | |
1073 | '9999 51') | |
1074 | ||
1075 | def test_match_str(self): | |
1076 | self.assertRaises(ValueError, match_str, 'xy>foobar', {}) | |
1077 | self.assertFalse(match_str('xy', {'x': 1200})) | |
1078 | self.assertTrue(match_str('!xy', {'x': 1200})) | |
1079 | self.assertTrue(match_str('x', {'x': 1200})) | |
1080 | self.assertFalse(match_str('!x', {'x': 1200})) | |
1081 | self.assertTrue(match_str('x', {'x': 0})) | |
1082 | self.assertFalse(match_str('x>0', {'x': 0})) | |
1083 | self.assertFalse(match_str('x>0', {})) | |
1084 | self.assertTrue(match_str('x>?0', {})) | |
1085 | self.assertTrue(match_str('x>1K', {'x': 1200})) | |
1086 | self.assertFalse(match_str('x>2K', {'x': 1200})) | |
1087 | self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200})) | |
1088 | self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200})) | |
1089 | self.assertFalse(match_str('y=a212', {'y': 'foobar42'})) | |
1090 | self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'})) | |
1091 | self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'})) | |
1092 | self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'})) | |
1093 | self.assertFalse(match_str( | |
1094 | 'like_count > 100 & dislike_count <? 50 & description', | |
1095 | {'like_count': 90, 'description': 'foo'})) | |
1096 | self.assertTrue(match_str( | |
1097 | 'like_count > 100 & dislike_count <? 50 & description', | |
1098 | {'like_count': 190, 'description': 'foo'})) | |
1099 | self.assertFalse(match_str( | |
1100 | 'like_count > 100 & dislike_count <? 50 & description', | |
1101 | {'like_count': 190, 'dislike_count': 60, 'description': 'foo'})) | |
1102 | self.assertFalse(match_str( | |
1103 | 'like_count > 100 & dislike_count <? 50 & description', | |
1104 | {'like_count': 190, 'dislike_count': 10})) | |
1105 | self.assertTrue(match_str('is_live', {'is_live': True})) | |
1106 | self.assertFalse(match_str('is_live', {'is_live': False})) | |
1107 | self.assertFalse(match_str('is_live', {'is_live': None})) | |
1108 | self.assertFalse(match_str('is_live', {})) | |
1109 | self.assertFalse(match_str('!is_live', {'is_live': True})) | |
1110 | self.assertTrue(match_str('!is_live', {'is_live': False})) | |
1111 | self.assertTrue(match_str('!is_live', {'is_live': None})) | |
1112 | self.assertTrue(match_str('!is_live', {})) | |
1113 | self.assertTrue(match_str('title', {'title': 'abc'})) | |
1114 | self.assertTrue(match_str('title', {'title': ''})) | |
1115 | self.assertFalse(match_str('!title', {'title': 'abc'})) | |
1116 | self.assertFalse(match_str('!title', {'title': ''})) | |
1117 | ||
1118 | def test_parse_dfxp_time_expr(self): | |
1119 | self.assertEqual(parse_dfxp_time_expr(None), None) | |
1120 | self.assertEqual(parse_dfxp_time_expr(''), None) | |
1121 | self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1) | |
1122 | self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1) | |
1123 | self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0) | |
1124 | self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1) | |
1125 | self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1) | |
1126 | ||
1127 | def test_dfxp2srt(self): | |
1128 | dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?> | |
1129 | <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
1130 | <body> | |
1131 | <div xml:lang="en"> | |
1132 | <p begin="0" end="1">The following line contains Chinese characters and special symbols</p> | |
1133 | <p begin="1" end="2">第二行<br/>♪♪</p> | |
1134 | <p begin="2" dur="1"><span>Third<br/>Line</span></p> | |
1135 | <p begin="3" end="-1">Lines with invalid timestamps are ignored</p> | |
1136 | <p begin="-1" end="-1">Ignore, two</p> | |
1137 | <p begin="3" dur="-1">Ignored, three</p> | |
1138 | </div> | |
1139 | </body> | |
1140 | </tt>'''.encode('utf-8') | |
1141 | srt_data = '''1 | |
1142 | 00:00:00,000 --> 00:00:01,000 | |
1143 | The following line contains Chinese characters and special symbols | |
1144 | ||
1145 | 2 | |
1146 | 00:00:01,000 --> 00:00:02,000 | |
1147 | 第二行 | |
1148 | ♪♪ | |
1149 | ||
1150 | 3 | |
1151 | 00:00:02,000 --> 00:00:03,000 | |
1152 | Third | |
1153 | Line | |
1154 | ||
1155 | ''' | |
1156 | self.assertEqual(dfxp2srt(dfxp_data), srt_data) | |
1157 | ||
1158 | dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?> | |
1159 | <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
1160 | <body> | |
1161 | <div xml:lang="en"> | |
1162 | <p begin="0" end="1">The first line</p> | |
1163 | </div> | |
1164 | </body> | |
1165 | </tt>'''.encode('utf-8') | |
1166 | srt_data = '''1 | |
1167 | 00:00:00,000 --> 00:00:01,000 | |
1168 | The first line | |
1169 | ||
1170 | ''' | |
1171 | self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data) | |
1172 | ||
1173 | dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?> | |
1174 | <tt xmlns="http://www.w3.org/2006/10/ttaf1" xmlns:ttp="http://www.w3.org/2006/10/ttaf1#parameter" ttp:timeBase="media" xmlns:tts="http://www.w3.org/2006/10/ttaf1#style" xml:lang="en" xmlns:ttm="http://www.w3.org/2006/10/ttaf1#metadata"> | |
1175 | <head> | |
1176 | <styling> | |
1177 | <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" /> | |
1178 | <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" /> | |
1179 | <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" /> | |
1180 | <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" /> | |
1181 | </styling> | |
1182 | </head> | |
1183 | <body tts:textAlign="center" style="s0"> | |
1184 | <div> | |
1185 | <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p> | |
1186 | <p style="s2" begin="00:00:02.08" id="p0" end="00:00:05.84"><span tts:color="lime">part 1<br /></span><span tts:color="cyan">part 2</span></p> | |
1187 | <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p> | |
1188 | <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p> | |
1189 | </div> | |
1190 | </body> | |
1191 | </tt>'''.encode('utf-8') | |
1192 | srt_data = '''1 | |
1193 | 00:00:02,080 --> 00:00:05,839 | |
1194 | <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font> | |
1195 | ||
1196 | 2 | |
1197 | 00:00:02,080 --> 00:00:05,839 | |
1198 | <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1 | |
1199 | </font>part 2</font></b> | |
1200 | ||
1201 | 3 | |
1202 | 00:00:05,839 --> 00:00:09,560 | |
1203 | <u><font color="lime">line 3 | |
1204 | part 3</font></u> | |
1205 | ||
1206 | 4 | |
1207 | 00:00:09,560 --> 00:00:12,359 | |
1208 | <i><u><font color="yellow"><font color="lime">inner | |
1209 | </font>style</font></u></i> | |
1210 | ||
1211 | ''' | |
1212 | self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data) | |
1213 | ||
1214 | dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?> | |
1215 | <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
1216 | <body> | |
1217 | <div xml:lang="en"> | |
1218 | <p begin="0" end="1">Line 1</p> | |
1219 | <p begin="1" end="2">第二行</p> | |
1220 | </div> | |
1221 | </body> | |
1222 | </tt>'''.encode('utf-16') | |
1223 | srt_data = '''1 | |
1224 | 00:00:00,000 --> 00:00:01,000 | |
1225 | Line 1 | |
1226 | ||
1227 | 2 | |
1228 | 00:00:01,000 --> 00:00:02,000 | |
1229 | 第二行 | |
1230 | ||
1231 | ''' | |
1232 | self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data) | |
1233 | ||
1234 | def test_cli_option(self): | |
1235 | self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128']) | |
1236 | self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), []) | |
1237 | self.assertEqual(cli_option({}, '--proxy', 'proxy'), []) | |
1238 | self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10']) | |
1239 | ||
1240 | def test_cli_valueless_option(self): | |
1241 | self.assertEqual(cli_valueless_option( | |
1242 | {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader']) | |
1243 | self.assertEqual(cli_valueless_option( | |
1244 | {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), []) | |
1245 | self.assertEqual(cli_valueless_option( | |
1246 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate']) | |
1247 | self.assertEqual(cli_valueless_option( | |
1248 | {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), []) | |
1249 | self.assertEqual(cli_valueless_option( | |
1250 | {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), []) | |
1251 | self.assertEqual(cli_valueless_option( | |
1252 | {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate']) | |
1253 | ||
1254 | def test_cli_bool_option(self): | |
1255 | self.assertEqual( | |
1256 | cli_bool_option( | |
1257 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), | |
1258 | ['--no-check-certificate', 'true']) | |
1259 | self.assertEqual( | |
1260 | cli_bool_option( | |
1261 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='), | |
1262 | ['--no-check-certificate=true']) | |
1263 | self.assertEqual( | |
1264 | cli_bool_option( | |
1265 | {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'), | |
1266 | ['--check-certificate', 'false']) | |
1267 | self.assertEqual( | |
1268 | cli_bool_option( | |
1269 | {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
1270 | ['--check-certificate=false']) | |
1271 | self.assertEqual( | |
1272 | cli_bool_option( | |
1273 | {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'), | |
1274 | ['--check-certificate', 'true']) | |
1275 | self.assertEqual( | |
1276 | cli_bool_option( | |
1277 | {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
1278 | ['--check-certificate=true']) | |
1279 | self.assertEqual( | |
1280 | cli_bool_option( | |
1281 | {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
1282 | []) | |
1283 | ||
1284 | def test_ohdave_rsa_encrypt(self): | |
1285 | N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd | |
1286 | e = 65537 | |
1287 | ||
1288 | self.assertEqual( | |
1289 | ohdave_rsa_encrypt(b'aa111222', e, N), | |
1290 | '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881') | |
1291 | ||
1292 | def test_pkcs1pad(self): | |
1293 | data = [1, 2, 3] | |
1294 | padded_data = pkcs1pad(data, 32) | |
1295 | self.assertEqual(padded_data[:2], [0, 2]) | |
1296 | self.assertEqual(padded_data[28:], [0, 1, 2, 3]) | |
1297 | ||
1298 | self.assertRaises(ValueError, pkcs1pad, data, 8) | |
1299 | ||
1300 | def test_encode_base_n(self): | |
1301 | self.assertEqual(encode_base_n(0, 30), '0') | |
1302 | self.assertEqual(encode_base_n(80, 30), '2k') | |
1303 | ||
1304 | custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA' | |
1305 | self.assertEqual(encode_base_n(0, 30, custom_table), '9') | |
1306 | self.assertEqual(encode_base_n(80, 30, custom_table), '7P') | |
1307 | ||
1308 | self.assertRaises(ValueError, encode_base_n, 0, 70) | |
1309 | self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table) | |
1310 | ||
1311 | def test_urshift(self): | |
1312 | self.assertEqual(urshift(3, 1), 1) | |
1313 | self.assertEqual(urshift(-3, 1), 2147483646) | |
1314 | ||
1315 | def test_get_element_by_class(self): | |
1316 | html = ''' | |
1317 | <span class="foo bar">nice</span> | |
1318 | ''' | |
1319 | ||
1320 | self.assertEqual(get_element_by_class('foo', html), 'nice') | |
1321 | self.assertEqual(get_element_by_class('no-such-class', html), None) | |
1322 | ||
1323 | def test_get_element_by_attribute(self): | |
1324 | html = ''' | |
1325 | <span class="foo bar">nice</span> | |
1326 | ''' | |
1327 | ||
1328 | self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice') | |
1329 | self.assertEqual(get_element_by_attribute('class', 'foo', html), None) | |
1330 | self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None) | |
1331 | ||
1332 | html = ''' | |
1333 | <div itemprop="author" itemscope>foo</div> | |
1334 | ''' | |
1335 | ||
1336 | self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo') | |
1337 | ||
1338 | def test_get_elements_by_class(self): | |
1339 | html = ''' | |
1340 | <span class="foo bar">nice</span><span class="foo bar">also nice</span> | |
1341 | ''' | |
1342 | ||
1343 | self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice']) | |
1344 | self.assertEqual(get_elements_by_class('no-such-class', html), []) | |
1345 | ||
1346 | def test_get_elements_by_attribute(self): | |
1347 | html = ''' | |
1348 | <span class="foo bar">nice</span><span class="foo bar">also nice</span> | |
1349 | ''' | |
1350 | ||
1351 | self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice']) | |
1352 | self.assertEqual(get_elements_by_attribute('class', 'foo', html), []) | |
1353 | self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), []) | |
1354 | ||
1355 | ||
1356 | if __name__ == '__main__': | |
1357 | unittest.main() |