]>
Commit | Line | Data |
---|---|---|
1 | #!/usr/bin/env python3 | |
2 | ||
3 | # Allow direct execution | |
4 | import os | |
5 | import re | |
6 | import sys | |
7 | import unittest | |
8 | ||
9 | sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) | |
10 | ||
11 | ||
12 | import contextlib | |
13 | import io | |
14 | import itertools | |
15 | import json | |
16 | import xml.etree.ElementTree | |
17 | ||
18 | from yt_dlp.compat import ( | |
19 | compat_etree_fromstring, | |
20 | compat_HTMLParseError, | |
21 | compat_os_name, | |
22 | ) | |
23 | from yt_dlp.utils import ( | |
24 | Config, | |
25 | DateRange, | |
26 | ExtractorError, | |
27 | InAdvancePagedList, | |
28 | LazyList, | |
29 | OnDemandPagedList, | |
30 | age_restricted, | |
31 | args_to_str, | |
32 | base_url, | |
33 | caesar, | |
34 | clean_html, | |
35 | clean_podcast_url, | |
36 | cli_bool_option, | |
37 | cli_option, | |
38 | cli_valueless_option, | |
39 | date_from_str, | |
40 | datetime_from_str, | |
41 | detect_exe_version, | |
42 | determine_ext, | |
43 | determine_file_encoding, | |
44 | dfxp2srt, | |
45 | dict_get, | |
46 | encode_base_n, | |
47 | encode_compat_str, | |
48 | encodeFilename, | |
49 | escape_rfc3986, | |
50 | escape_url, | |
51 | expand_path, | |
52 | extract_attributes, | |
53 | find_xpath_attr, | |
54 | fix_xml_ampersands, | |
55 | float_or_none, | |
56 | format_bytes, | |
57 | get_compatible_ext, | |
58 | get_element_by_attribute, | |
59 | get_element_by_class, | |
60 | get_element_html_by_attribute, | |
61 | get_element_html_by_class, | |
62 | get_element_text_and_html_by_tag, | |
63 | get_elements_by_attribute, | |
64 | get_elements_by_class, | |
65 | get_elements_html_by_attribute, | |
66 | get_elements_html_by_class, | |
67 | get_elements_text_and_html_by_attribute, | |
68 | int_or_none, | |
69 | intlist_to_bytes, | |
70 | iri_to_uri, | |
71 | is_html, | |
72 | js_to_json, | |
73 | limit_length, | |
74 | locked_file, | |
75 | lowercase_escape, | |
76 | match_str, | |
77 | merge_dicts, | |
78 | mimetype2ext, | |
79 | month_by_name, | |
80 | multipart_encode, | |
81 | ohdave_rsa_encrypt, | |
82 | orderedSet, | |
83 | parse_age_limit, | |
84 | parse_bitrate, | |
85 | parse_codecs, | |
86 | parse_count, | |
87 | parse_dfxp_time_expr, | |
88 | parse_duration, | |
89 | parse_filesize, | |
90 | parse_iso8601, | |
91 | parse_qs, | |
92 | parse_resolution, | |
93 | pkcs1pad, | |
94 | prepend_extension, | |
95 | read_batch_urls, | |
96 | remove_end, | |
97 | remove_quotes, | |
98 | remove_start, | |
99 | render_table, | |
100 | replace_extension, | |
101 | rot47, | |
102 | sanitize_filename, | |
103 | sanitize_path, | |
104 | sanitize_url, | |
105 | sanitized_Request, | |
106 | shell_quote, | |
107 | smuggle_url, | |
108 | str_to_int, | |
109 | strip_jsonp, | |
110 | strip_or_none, | |
111 | subtitles_filename, | |
112 | timeconvert, | |
113 | traverse_obj, | |
114 | unescapeHTML, | |
115 | unified_strdate, | |
116 | unified_timestamp, | |
117 | unsmuggle_url, | |
118 | update_url_query, | |
119 | uppercase_escape, | |
120 | url_basename, | |
121 | url_or_none, | |
122 | urlencode_postdata, | |
123 | urljoin, | |
124 | urshift, | |
125 | version_tuple, | |
126 | xpath_attr, | |
127 | xpath_element, | |
128 | xpath_text, | |
129 | xpath_with_ns, | |
130 | ) | |
131 | ||
132 | ||
133 | class TestUtil(unittest.TestCase): | |
134 | def test_timeconvert(self): | |
135 | self.assertTrue(timeconvert('') is None) | |
136 | self.assertTrue(timeconvert('bougrg') is None) | |
137 | ||
138 | def test_sanitize_filename(self): | |
139 | self.assertEqual(sanitize_filename(''), '') | |
140 | self.assertEqual(sanitize_filename('abc'), 'abc') | |
141 | self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e') | |
142 | ||
143 | self.assertEqual(sanitize_filename('123'), '123') | |
144 | ||
145 | self.assertEqual('abc⧸de', sanitize_filename('abc/de')) | |
146 | self.assertFalse('/' in sanitize_filename('abc/de///')) | |
147 | ||
148 | self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', is_id=False)) | |
149 | self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', is_id=False)) | |
150 | self.assertEqual('yes no', sanitize_filename('yes? no', is_id=False)) | |
151 | self.assertEqual('this - that', sanitize_filename('this: that', is_id=False)) | |
152 | ||
153 | self.assertEqual(sanitize_filename('AT&T'), 'AT&T') | |
154 | aumlaut = 'ä' | |
155 | self.assertEqual(sanitize_filename(aumlaut), aumlaut) | |
156 | tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430' | |
157 | self.assertEqual(sanitize_filename(tests), tests) | |
158 | ||
159 | self.assertEqual( | |
160 | sanitize_filename('New World record at 0:12:34'), | |
161 | 'New World record at 0_12_34') | |
162 | ||
163 | self.assertEqual(sanitize_filename('--gasdgf'), '--gasdgf') | |
164 | self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf') | |
165 | self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf') | |
166 | self.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf') | |
167 | self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf') | |
168 | self.assertEqual(sanitize_filename('.gasdgf', is_id=False), 'gasdgf') | |
169 | ||
170 | forbidden = '"\0\\/' | |
171 | for fc in forbidden: | |
172 | for fbc in forbidden: | |
173 | self.assertTrue(fbc not in sanitize_filename(fc)) | |
174 | ||
175 | def test_sanitize_filename_restricted(self): | |
176 | self.assertEqual(sanitize_filename('abc', restricted=True), 'abc') | |
177 | self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e') | |
178 | ||
179 | self.assertEqual(sanitize_filename('123', restricted=True), '123') | |
180 | ||
181 | self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True)) | |
182 | self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True)) | |
183 | ||
184 | self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True)) | |
185 | self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True)) | |
186 | self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True)) | |
187 | self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True)) | |
188 | ||
189 | tests = 'aäb\u4e2d\u56fd\u7684c' | |
190 | self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c') | |
191 | self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename | |
192 | ||
193 | forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#' | |
194 | for fc in forbidden: | |
195 | for fbc in forbidden: | |
196 | self.assertTrue(fbc not in sanitize_filename(fc, restricted=True)) | |
197 | ||
198 | # Handle a common case more neatly | |
199 | self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song') | |
200 | self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech') | |
201 | # .. but make sure the file name is never empty | |
202 | self.assertTrue(sanitize_filename('-', restricted=True) != '') | |
203 | self.assertTrue(sanitize_filename(':', restricted=True) != '') | |
204 | ||
205 | self.assertEqual(sanitize_filename( | |
206 | 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True), | |
207 | 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy') | |
208 | ||
209 | def test_sanitize_ids(self): | |
210 | self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw') | |
211 | self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw') | |
212 | self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI') | |
213 | ||
214 | def test_sanitize_path(self): | |
215 | if sys.platform != 'win32': | |
216 | return | |
217 | ||
218 | self.assertEqual(sanitize_path('abc'), 'abc') | |
219 | self.assertEqual(sanitize_path('abc/def'), 'abc\\def') | |
220 | self.assertEqual(sanitize_path('abc\\def'), 'abc\\def') | |
221 | self.assertEqual(sanitize_path('abc|def'), 'abc#def') | |
222 | self.assertEqual(sanitize_path('<>:"|?*'), '#######') | |
223 | self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def') | |
224 | self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def') | |
225 | ||
226 | self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc') | |
227 | self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc') | |
228 | ||
229 | self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') | |
230 | self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc') | |
231 | self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f') | |
232 | self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') | |
233 | ||
234 | self.assertEqual( | |
235 | sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'), | |
236 | 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s') | |
237 | ||
238 | self.assertEqual( | |
239 | sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'), | |
240 | 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part') | |
241 | self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#') | |
242 | self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def') | |
243 | self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#') | |
244 | ||
245 | self.assertEqual(sanitize_path('../abc'), '..\\abc') | |
246 | self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc') | |
247 | self.assertEqual(sanitize_path('./abc'), 'abc') | |
248 | self.assertEqual(sanitize_path('./../abc'), '..\\abc') | |
249 | ||
250 | def test_sanitize_url(self): | |
251 | self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar') | |
252 | self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar') | |
253 | self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar') | |
254 | self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar') | |
255 | self.assertEqual(sanitize_url('foo bar'), 'foo bar') | |
256 | ||
257 | def test_extract_basic_auth(self): | |
258 | auth_header = lambda url: sanitized_Request(url).get_header('Authorization') | |
259 | self.assertFalse(auth_header('http://foo.bar')) | |
260 | self.assertFalse(auth_header('http://:foo.bar')) | |
261 | self.assertEqual(auth_header('http://@foo.bar'), 'Basic Og==') | |
262 | self.assertEqual(auth_header('http://:pass@foo.bar'), 'Basic OnBhc3M=') | |
263 | self.assertEqual(auth_header('http://user:@foo.bar'), 'Basic dXNlcjo=') | |
264 | self.assertEqual(auth_header('http://user:pass@foo.bar'), 'Basic dXNlcjpwYXNz') | |
265 | ||
266 | def test_expand_path(self): | |
267 | def env(var): | |
268 | return f'%{var}%' if sys.platform == 'win32' else f'${var}' | |
269 | ||
270 | os.environ['yt_dlp_EXPATH_PATH'] = 'expanded' | |
271 | self.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded') | |
272 | ||
273 | old_home = os.environ.get('HOME') | |
274 | test_str = R'C:\Documents and Settings\тест\Application Data' | |
275 | try: | |
276 | os.environ['HOME'] = test_str | |
277 | self.assertEqual(expand_path(env('HOME')), os.getenv('HOME')) | |
278 | self.assertEqual(expand_path('~'), os.getenv('HOME')) | |
279 | self.assertEqual( | |
280 | expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')), | |
281 | '%s/expanded' % os.getenv('HOME')) | |
282 | finally: | |
283 | os.environ['HOME'] = old_home or '' | |
284 | ||
285 | def test_prepend_extension(self): | |
286 | self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext') | |
287 | self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext') | |
288 | self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp') | |
289 | self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp') | |
290 | self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp') | |
291 | self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext') | |
292 | ||
293 | def test_replace_extension(self): | |
294 | self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp') | |
295 | self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp') | |
296 | self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp') | |
297 | self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp') | |
298 | self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp') | |
299 | self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp') | |
300 | ||
301 | def test_subtitles_filename(self): | |
302 | self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt') | |
303 | self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt') | |
304 | self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt') | |
305 | ||
306 | def test_remove_start(self): | |
307 | self.assertEqual(remove_start(None, 'A - '), None) | |
308 | self.assertEqual(remove_start('A - B', 'A - '), 'B') | |
309 | self.assertEqual(remove_start('B - A', 'A - '), 'B - A') | |
310 | ||
311 | def test_remove_end(self): | |
312 | self.assertEqual(remove_end(None, ' - B'), None) | |
313 | self.assertEqual(remove_end('A - B', ' - B'), 'A') | |
314 | self.assertEqual(remove_end('B - A', ' - B'), 'B - A') | |
315 | ||
316 | def test_remove_quotes(self): | |
317 | self.assertEqual(remove_quotes(None), None) | |
318 | self.assertEqual(remove_quotes('"'), '"') | |
319 | self.assertEqual(remove_quotes("'"), "'") | |
320 | self.assertEqual(remove_quotes(';'), ';') | |
321 | self.assertEqual(remove_quotes('";'), '";') | |
322 | self.assertEqual(remove_quotes('""'), '') | |
323 | self.assertEqual(remove_quotes('";"'), ';') | |
324 | ||
325 | def test_ordered_set(self): | |
326 | self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7]) | |
327 | self.assertEqual(orderedSet([]), []) | |
328 | self.assertEqual(orderedSet([1]), [1]) | |
329 | # keep the list ordered | |
330 | self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1]) | |
331 | ||
332 | def test_unescape_html(self): | |
333 | self.assertEqual(unescapeHTML('%20;'), '%20;') | |
334 | self.assertEqual(unescapeHTML('/'), '/') | |
335 | self.assertEqual(unescapeHTML('/'), '/') | |
336 | self.assertEqual(unescapeHTML('é'), 'é') | |
337 | self.assertEqual(unescapeHTML('�'), '�') | |
338 | self.assertEqual(unescapeHTML('&a"'), '&a"') | |
339 | # HTML5 entities | |
340 | self.assertEqual(unescapeHTML('.''), '.\'') | |
341 | ||
342 | def test_date_from_str(self): | |
343 | self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day')) | |
344 | self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week')) | |
345 | self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week')) | |
346 | self.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year')) | |
347 | self.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month')) | |
348 | ||
349 | def test_datetime_from_str(self): | |
350 | self.assertEqual(datetime_from_str('yesterday', precision='day'), datetime_from_str('now-1day', precision='auto')) | |
351 | self.assertEqual(datetime_from_str('now+7day', precision='day'), datetime_from_str('now+1week', precision='auto')) | |
352 | self.assertEqual(datetime_from_str('now+14day', precision='day'), datetime_from_str('now+2week', precision='auto')) | |
353 | self.assertEqual(datetime_from_str('20200229+365day', precision='day'), datetime_from_str('20200229+1year', precision='auto')) | |
354 | self.assertEqual(datetime_from_str('20210131+28day', precision='day'), datetime_from_str('20210131+1month', precision='auto')) | |
355 | self.assertEqual(datetime_from_str('20210131+59day', precision='day'), datetime_from_str('20210131+2month', precision='auto')) | |
356 | self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto')) | |
357 | self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto')) | |
358 | ||
359 | def test_daterange(self): | |
360 | _20century = DateRange("19000101", "20000101") | |
361 | self.assertFalse("17890714" in _20century) | |
362 | _ac = DateRange("00010101") | |
363 | self.assertTrue("19690721" in _ac) | |
364 | _firstmilenium = DateRange(end="10000101") | |
365 | self.assertTrue("07110427" in _firstmilenium) | |
366 | ||
367 | def test_unified_dates(self): | |
368 | self.assertEqual(unified_strdate('December 21, 2010'), '20101221') | |
369 | self.assertEqual(unified_strdate('8/7/2009'), '20090708') | |
370 | self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214') | |
371 | self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011') | |
372 | self.assertEqual(unified_strdate('1968 12 10'), '19681210') | |
373 | self.assertEqual(unified_strdate('1968-12-10'), '19681210') | |
374 | self.assertEqual(unified_strdate('31-07-2022 20:00'), '20220731') | |
375 | self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128') | |
376 | self.assertEqual( | |
377 | unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False), | |
378 | '20141126') | |
379 | self.assertEqual( | |
380 | unified_strdate('2/2/2015 6:47:40 PM', day_first=False), | |
381 | '20150202') | |
382 | self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214') | |
383 | self.assertEqual(unified_strdate('25-09-2014'), '20140925') | |
384 | self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227') | |
385 | self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None) | |
386 | self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207') | |
387 | self.assertEqual(unified_strdate('July 15th, 2013'), '20130715') | |
388 | self.assertEqual(unified_strdate('September 1st, 2013'), '20130901') | |
389 | self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902') | |
390 | self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103') | |
391 | self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023') | |
392 | ||
393 | def test_unified_timestamps(self): | |
394 | self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600) | |
395 | self.assertEqual(unified_timestamp('8/7/2009'), 1247011200) | |
396 | self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200) | |
397 | self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598) | |
398 | self.assertEqual(unified_timestamp('1968 12 10'), -33436800) | |
399 | self.assertEqual(unified_timestamp('1968-12-10'), -33436800) | |
400 | self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200) | |
401 | self.assertEqual( | |
402 | unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False), | |
403 | 1417001400) | |
404 | self.assertEqual( | |
405 | unified_timestamp('2/2/2015 6:47:40 PM', day_first=False), | |
406 | 1422902860) | |
407 | self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900) | |
408 | self.assertEqual(unified_timestamp('25-09-2014'), 1411603200) | |
409 | self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200) | |
410 | self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None) | |
411 | self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500) | |
412 | self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100) | |
413 | self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361) | |
414 | self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540) | |
415 | self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140) | |
416 | self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363) | |
417 | ||
418 | self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1) | |
419 | self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86) | |
420 | self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78) | |
421 | ||
422 | def test_determine_ext(self): | |
423 | self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4') | |
424 | self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None) | |
425 | self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None) | |
426 | self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None) | |
427 | self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8') | |
428 | self.assertEqual(determine_ext('foobar', None), None) | |
429 | ||
430 | def test_find_xpath_attr(self): | |
431 | testxml = '''<root> | |
432 | <node/> | |
433 | <node x="a"/> | |
434 | <node x="a" y="c" /> | |
435 | <node x="b" y="d" /> | |
436 | <node x="" /> | |
437 | </root>''' | |
438 | doc = compat_etree_fromstring(testxml) | |
439 | ||
440 | self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None) | |
441 | self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None) | |
442 | self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None) | |
443 | self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None) | |
444 | self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1]) | |
445 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1]) | |
446 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3]) | |
447 | self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2]) | |
448 | self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2]) | |
449 | self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3]) | |
450 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4]) | |
451 | ||
452 | def test_xpath_with_ns(self): | |
453 | testxml = '''<root xmlns:media="http://example.com/"> | |
454 | <media:song> | |
455 | <media:author>The Author</media:author> | |
456 | <url>http://server.com/download.mp3</url> | |
457 | </media:song> | |
458 | </root>''' | |
459 | doc = compat_etree_fromstring(testxml) | |
460 | find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'})) | |
461 | self.assertTrue(find('media:song') is not None) | |
462 | self.assertEqual(find('media:song/media:author').text, 'The Author') | |
463 | self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3') | |
464 | ||
465 | def test_xpath_element(self): | |
466 | doc = xml.etree.ElementTree.Element('root') | |
467 | div = xml.etree.ElementTree.SubElement(doc, 'div') | |
468 | p = xml.etree.ElementTree.SubElement(div, 'p') | |
469 | p.text = 'Foo' | |
470 | self.assertEqual(xpath_element(doc, 'div/p'), p) | |
471 | self.assertEqual(xpath_element(doc, ['div/p']), p) | |
472 | self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p) | |
473 | self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default') | |
474 | self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default') | |
475 | self.assertTrue(xpath_element(doc, 'div/bar') is None) | |
476 | self.assertTrue(xpath_element(doc, ['div/bar']) is None) | |
477 | self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None) | |
478 | self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True) | |
479 | self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True) | |
480 | self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True) | |
481 | ||
482 | def test_xpath_text(self): | |
483 | testxml = '''<root> | |
484 | <div> | |
485 | <p>Foo</p> | |
486 | </div> | |
487 | </root>''' | |
488 | doc = compat_etree_fromstring(testxml) | |
489 | self.assertEqual(xpath_text(doc, 'div/p'), 'Foo') | |
490 | self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default') | |
491 | self.assertTrue(xpath_text(doc, 'div/bar') is None) | |
492 | self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True) | |
493 | ||
494 | def test_xpath_attr(self): | |
495 | testxml = '''<root> | |
496 | <div> | |
497 | <p x="a">Foo</p> | |
498 | </div> | |
499 | </root>''' | |
500 | doc = compat_etree_fromstring(testxml) | |
501 | self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a') | |
502 | self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None) | |
503 | self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None) | |
504 | self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default') | |
505 | self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default') | |
506 | self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True) | |
507 | self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True) | |
508 | ||
509 | def test_smuggle_url(self): | |
510 | data = {"ö": "ö", "abc": [3]} | |
511 | url = 'https://foo.bar/baz?x=y#a' | |
512 | smug_url = smuggle_url(url, data) | |
513 | unsmug_url, unsmug_data = unsmuggle_url(smug_url) | |
514 | self.assertEqual(url, unsmug_url) | |
515 | self.assertEqual(data, unsmug_data) | |
516 | ||
517 | res_url, res_data = unsmuggle_url(url) | |
518 | self.assertEqual(res_url, url) | |
519 | self.assertEqual(res_data, None) | |
520 | ||
521 | smug_url = smuggle_url(url, {'a': 'b'}) | |
522 | smug_smug_url = smuggle_url(smug_url, {'c': 'd'}) | |
523 | res_url, res_data = unsmuggle_url(smug_smug_url) | |
524 | self.assertEqual(res_url, url) | |
525 | self.assertEqual(res_data, {'a': 'b', 'c': 'd'}) | |
526 | ||
527 | def test_shell_quote(self): | |
528 | args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')] | |
529 | self.assertEqual( | |
530 | shell_quote(args), | |
531 | """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''') | |
532 | ||
533 | def test_float_or_none(self): | |
534 | self.assertEqual(float_or_none('42.42'), 42.42) | |
535 | self.assertEqual(float_or_none('42'), 42.0) | |
536 | self.assertEqual(float_or_none(''), None) | |
537 | self.assertEqual(float_or_none(None), None) | |
538 | self.assertEqual(float_or_none([]), None) | |
539 | self.assertEqual(float_or_none(set()), None) | |
540 | ||
541 | def test_int_or_none(self): | |
542 | self.assertEqual(int_or_none('42'), 42) | |
543 | self.assertEqual(int_or_none(''), None) | |
544 | self.assertEqual(int_or_none(None), None) | |
545 | self.assertEqual(int_or_none([]), None) | |
546 | self.assertEqual(int_or_none(set()), None) | |
547 | ||
548 | def test_str_to_int(self): | |
549 | self.assertEqual(str_to_int('123,456'), 123456) | |
550 | self.assertEqual(str_to_int('123.456'), 123456) | |
551 | self.assertEqual(str_to_int(523), 523) | |
552 | self.assertEqual(str_to_int('noninteger'), None) | |
553 | self.assertEqual(str_to_int([]), None) | |
554 | ||
555 | def test_url_basename(self): | |
556 | self.assertEqual(url_basename('http://foo.de/'), '') | |
557 | self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz') | |
558 | self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz') | |
559 | self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz') | |
560 | self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz') | |
561 | self.assertEqual( | |
562 | url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'), | |
563 | 'trailer.mp4') | |
564 | ||
565 | def test_base_url(self): | |
566 | self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/') | |
567 | self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/') | |
568 | self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/') | |
569 | self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/') | |
570 | self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/') | |
571 | self.assertEqual(base_url('http://foo.de/bar/baz&x=z&w=y/x/c'), 'http://foo.de/bar/baz&x=z&w=y/x/') | |
572 | ||
573 | def test_urljoin(self): | |
574 | self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
575 | self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
576 | self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
577 | self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
578 | self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt') | |
579 | self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
580 | self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
581 | self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
582 | self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
583 | self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') | |
584 | self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
585 | self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') | |
586 | self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
587 | self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
588 | self.assertEqual(urljoin('http://foo.de/', None), None) | |
589 | self.assertEqual(urljoin('http://foo.de/', ''), None) | |
590 | self.assertEqual(urljoin('http://foo.de/', ['foobar']), None) | |
591 | self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt') | |
592 | self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de') | |
593 | self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de') | |
594 | ||
595 | def test_url_or_none(self): | |
596 | self.assertEqual(url_or_none(None), None) | |
597 | self.assertEqual(url_or_none(''), None) | |
598 | self.assertEqual(url_or_none('foo'), None) | |
599 | self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de') | |
600 | self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de') | |
601 | self.assertEqual(url_or_none('http$://foo.de'), None) | |
602 | self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de') | |
603 | self.assertEqual(url_or_none('//foo.de'), '//foo.de') | |
604 | self.assertEqual(url_or_none('s3://foo.de'), None) | |
605 | self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de') | |
606 | self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de') | |
607 | self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de') | |
608 | self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de') | |
609 | ||
610 | def test_parse_age_limit(self): | |
611 | self.assertEqual(parse_age_limit(None), None) | |
612 | self.assertEqual(parse_age_limit(False), None) | |
613 | self.assertEqual(parse_age_limit('invalid'), None) | |
614 | self.assertEqual(parse_age_limit(0), 0) | |
615 | self.assertEqual(parse_age_limit(18), 18) | |
616 | self.assertEqual(parse_age_limit(21), 21) | |
617 | self.assertEqual(parse_age_limit(22), None) | |
618 | self.assertEqual(parse_age_limit('18'), 18) | |
619 | self.assertEqual(parse_age_limit('18+'), 18) | |
620 | self.assertEqual(parse_age_limit('PG-13'), 13) | |
621 | self.assertEqual(parse_age_limit('TV-14'), 14) | |
622 | self.assertEqual(parse_age_limit('TV-MA'), 17) | |
623 | self.assertEqual(parse_age_limit('TV14'), 14) | |
624 | self.assertEqual(parse_age_limit('TV_G'), 0) | |
625 | ||
626 | def test_parse_duration(self): | |
627 | self.assertEqual(parse_duration(None), None) | |
628 | self.assertEqual(parse_duration(False), None) | |
629 | self.assertEqual(parse_duration('invalid'), None) | |
630 | self.assertEqual(parse_duration('1'), 1) | |
631 | self.assertEqual(parse_duration('1337:12'), 80232) | |
632 | self.assertEqual(parse_duration('9:12:43'), 33163) | |
633 | self.assertEqual(parse_duration('12:00'), 720) | |
634 | self.assertEqual(parse_duration('00:01:01'), 61) | |
635 | self.assertEqual(parse_duration('x:y'), None) | |
636 | self.assertEqual(parse_duration('3h11m53s'), 11513) | |
637 | self.assertEqual(parse_duration('3h 11m 53s'), 11513) | |
638 | self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513) | |
639 | self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513) | |
640 | self.assertEqual(parse_duration('3 hours, 11 minutes, 53 seconds'), 11513) | |
641 | self.assertEqual(parse_duration('3 hours, 11 mins, 53 secs'), 11513) | |
642 | self.assertEqual(parse_duration('62m45s'), 3765) | |
643 | self.assertEqual(parse_duration('6m59s'), 419) | |
644 | self.assertEqual(parse_duration('49s'), 49) | |
645 | self.assertEqual(parse_duration('0h0m0s'), 0) | |
646 | self.assertEqual(parse_duration('0m0s'), 0) | |
647 | self.assertEqual(parse_duration('0s'), 0) | |
648 | self.assertEqual(parse_duration('01:02:03.05'), 3723.05) | |
649 | self.assertEqual(parse_duration('T30M38S'), 1838) | |
650 | self.assertEqual(parse_duration('5 s'), 5) | |
651 | self.assertEqual(parse_duration('3 min'), 180) | |
652 | self.assertEqual(parse_duration('2.5 hours'), 9000) | |
653 | self.assertEqual(parse_duration('02:03:04'), 7384) | |
654 | self.assertEqual(parse_duration('01:02:03:04'), 93784) | |
655 | self.assertEqual(parse_duration('1 hour 3 minutes'), 3780) | |
656 | self.assertEqual(parse_duration('87 Min.'), 5220) | |
657 | self.assertEqual(parse_duration('PT1H0.040S'), 3600.04) | |
658 | self.assertEqual(parse_duration('PT00H03M30SZ'), 210) | |
659 | self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88) | |
660 | self.assertEqual(parse_duration('01:02:03:050'), 3723.05) | |
661 | self.assertEqual(parse_duration('103:050'), 103.05) | |
662 | ||
663 | def test_fix_xml_ampersands(self): | |
664 | self.assertEqual( | |
665 | fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a') | |
666 | self.assertEqual( | |
667 | fix_xml_ampersands('"&x=y&wrong;&z=a'), | |
668 | '"&x=y&wrong;&z=a') | |
669 | self.assertEqual( | |
670 | fix_xml_ampersands('&'><"'), | |
671 | '&'><"') | |
672 | self.assertEqual( | |
673 | fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼') | |
674 | self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#') | |
675 | ||
676 | def test_paged_list(self): | |
677 | def testPL(size, pagesize, sliceargs, expected): | |
678 | def get_page(pagenum): | |
679 | firstid = pagenum * pagesize | |
680 | upto = min(size, pagenum * pagesize + pagesize) | |
681 | yield from range(firstid, upto) | |
682 | ||
683 | pl = OnDemandPagedList(get_page, pagesize) | |
684 | got = pl.getslice(*sliceargs) | |
685 | self.assertEqual(got, expected) | |
686 | ||
687 | iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize) | |
688 | got = iapl.getslice(*sliceargs) | |
689 | self.assertEqual(got, expected) | |
690 | ||
691 | testPL(5, 2, (), [0, 1, 2, 3, 4]) | |
692 | testPL(5, 2, (1,), [1, 2, 3, 4]) | |
693 | testPL(5, 2, (2,), [2, 3, 4]) | |
694 | testPL(5, 2, (4,), [4]) | |
695 | testPL(5, 2, (0, 3), [0, 1, 2]) | |
696 | testPL(5, 2, (1, 4), [1, 2, 3]) | |
697 | testPL(5, 2, (2, 99), [2, 3, 4]) | |
698 | testPL(5, 2, (20, 99), []) | |
699 | ||
700 | def test_read_batch_urls(self): | |
701 | f = io.StringIO('''\xef\xbb\xbf foo | |
702 | bar\r | |
703 | baz | |
704 | # More after this line\r | |
705 | ; or after this | |
706 | bam''') | |
707 | self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam']) | |
708 | ||
709 | def test_urlencode_postdata(self): | |
710 | data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}) | |
711 | self.assertTrue(isinstance(data, bytes)) | |
712 | ||
713 | def test_update_url_query(self): | |
714 | self.assertEqual(parse_qs(update_url_query( | |
715 | 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})), | |
716 | parse_qs('http://example.com/path?quality=HD&format=mp4')) | |
717 | self.assertEqual(parse_qs(update_url_query( | |
718 | 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})), | |
719 | parse_qs('http://example.com/path?system=LINUX&system=WINDOWS')) | |
720 | self.assertEqual(parse_qs(update_url_query( | |
721 | 'http://example.com/path', {'fields': 'id,formats,subtitles'})), | |
722 | parse_qs('http://example.com/path?fields=id,formats,subtitles')) | |
723 | self.assertEqual(parse_qs(update_url_query( | |
724 | 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})), | |
725 | parse_qs('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails')) | |
726 | self.assertEqual(parse_qs(update_url_query( | |
727 | 'http://example.com/path?manifest=f4m', {'manifest': []})), | |
728 | parse_qs('http://example.com/path')) | |
729 | self.assertEqual(parse_qs(update_url_query( | |
730 | 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})), | |
731 | parse_qs('http://example.com/path?system=LINUX')) | |
732 | self.assertEqual(parse_qs(update_url_query( | |
733 | 'http://example.com/path', {'fields': b'id,formats,subtitles'})), | |
734 | parse_qs('http://example.com/path?fields=id,formats,subtitles')) | |
735 | self.assertEqual(parse_qs(update_url_query( | |
736 | 'http://example.com/path', {'width': 1080, 'height': 720})), | |
737 | parse_qs('http://example.com/path?width=1080&height=720')) | |
738 | self.assertEqual(parse_qs(update_url_query( | |
739 | 'http://example.com/path', {'bitrate': 5020.43})), | |
740 | parse_qs('http://example.com/path?bitrate=5020.43')) | |
741 | self.assertEqual(parse_qs(update_url_query( | |
742 | 'http://example.com/path', {'test': '第二行тест'})), | |
743 | parse_qs('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82')) | |
744 | ||
745 | def test_multipart_encode(self): | |
746 | self.assertEqual( | |
747 | multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0], | |
748 | b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n') | |
749 | self.assertEqual( | |
750 | multipart_encode({'欄位'.encode(): '值'.encode()}, boundary='AAAAAA')[0], | |
751 | 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') | |
752 | self.assertRaises( | |
753 | ValueError, multipart_encode, {b'field': b'value'}, boundary='value') | |
754 | ||
755 | def test_dict_get(self): | |
756 | FALSE_VALUES = { | |
757 | 'none': None, | |
758 | 'false': False, | |
759 | 'zero': 0, | |
760 | 'empty_string': '', | |
761 | 'empty_list': [], | |
762 | } | |
763 | d = FALSE_VALUES.copy() | |
764 | d['a'] = 42 | |
765 | self.assertEqual(dict_get(d, 'a'), 42) | |
766 | self.assertEqual(dict_get(d, 'b'), None) | |
767 | self.assertEqual(dict_get(d, 'b', 42), 42) | |
768 | self.assertEqual(dict_get(d, ('a', )), 42) | |
769 | self.assertEqual(dict_get(d, ('b', 'a', )), 42) | |
770 | self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42) | |
771 | self.assertEqual(dict_get(d, ('b', 'c', )), None) | |
772 | self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42) | |
773 | for key, false_value in FALSE_VALUES.items(): | |
774 | self.assertEqual(dict_get(d, ('b', 'c', key, )), None) | |
775 | self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value) | |
776 | ||
777 | def test_merge_dicts(self): | |
778 | self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2}) | |
779 | self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1}) | |
780 | self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1}) | |
781 | self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1}) | |
782 | self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1}) | |
783 | self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1}) | |
784 | self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''}) | |
785 | self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'}) | |
786 | self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'}) | |
787 | ||
788 | def test_encode_compat_str(self): | |
789 | self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест') | |
790 | self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест') | |
791 | ||
792 | def test_parse_iso8601(self): | |
793 | self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266) | |
794 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266) | |
795 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266) | |
796 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266) | |
797 | self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251) | |
798 | self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None) | |
799 | ||
800 | def test_strip_jsonp(self): | |
801 | stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);') | |
802 | d = json.loads(stripped) | |
803 | self.assertEqual(d, [{"id": "532cb", "x": 3}]) | |
804 | ||
805 | stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc') | |
806 | d = json.loads(stripped) | |
807 | self.assertEqual(d, {'STATUS': 'OK'}) | |
808 | ||
809 | stripped = strip_jsonp('ps.embedHandler({"status": "success"});') | |
810 | d = json.loads(stripped) | |
811 | self.assertEqual(d, {'status': 'success'}) | |
812 | ||
813 | stripped = strip_jsonp('window.cb && window.cb({"status": "success"});') | |
814 | d = json.loads(stripped) | |
815 | self.assertEqual(d, {'status': 'success'}) | |
816 | ||
817 | stripped = strip_jsonp('window.cb && cb({"status": "success"});') | |
818 | d = json.loads(stripped) | |
819 | self.assertEqual(d, {'status': 'success'}) | |
820 | ||
821 | stripped = strip_jsonp('({"status": "success"});') | |
822 | d = json.loads(stripped) | |
823 | self.assertEqual(d, {'status': 'success'}) | |
824 | ||
825 | def test_strip_or_none(self): | |
826 | self.assertEqual(strip_or_none(' abc'), 'abc') | |
827 | self.assertEqual(strip_or_none('abc '), 'abc') | |
828 | self.assertEqual(strip_or_none(' abc '), 'abc') | |
829 | self.assertEqual(strip_or_none('\tabc\t'), 'abc') | |
830 | self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc') | |
831 | self.assertEqual(strip_or_none('abc'), 'abc') | |
832 | self.assertEqual(strip_or_none(''), '') | |
833 | self.assertEqual(strip_or_none(None), None) | |
834 | self.assertEqual(strip_or_none(42), None) | |
835 | self.assertEqual(strip_or_none([]), None) | |
836 | ||
837 | def test_uppercase_escape(self): | |
838 | self.assertEqual(uppercase_escape('aä'), 'aä') | |
839 | self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐') | |
840 | ||
841 | def test_lowercase_escape(self): | |
842 | self.assertEqual(lowercase_escape('aä'), 'aä') | |
843 | self.assertEqual(lowercase_escape('\\u0026'), '&') | |
844 | ||
845 | def test_limit_length(self): | |
846 | self.assertEqual(limit_length(None, 12), None) | |
847 | self.assertEqual(limit_length('foo', 12), 'foo') | |
848 | self.assertTrue( | |
849 | limit_length('foo bar baz asd', 12).startswith('foo bar')) | |
850 | self.assertTrue('...' in limit_length('foo bar baz asd', 12)) | |
851 | ||
852 | def test_mimetype2ext(self): | |
853 | self.assertEqual(mimetype2ext(None), None) | |
854 | self.assertEqual(mimetype2ext('video/x-flv'), 'flv') | |
855 | self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8') | |
856 | self.assertEqual(mimetype2ext('text/vtt'), 'vtt') | |
857 | self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt') | |
858 | self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html') | |
859 | self.assertEqual(mimetype2ext('audio/x-wav'), 'wav') | |
860 | self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav') | |
861 | ||
862 | def test_month_by_name(self): | |
863 | self.assertEqual(month_by_name(None), None) | |
864 | self.assertEqual(month_by_name('December', 'en'), 12) | |
865 | self.assertEqual(month_by_name('décembre', 'fr'), 12) | |
866 | self.assertEqual(month_by_name('December'), 12) | |
867 | self.assertEqual(month_by_name('décembre'), None) | |
868 | self.assertEqual(month_by_name('Unknown', 'unknown'), None) | |
869 | ||
870 | def test_parse_codecs(self): | |
871 | self.assertEqual(parse_codecs(''), {}) | |
872 | self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), { | |
873 | 'vcodec': 'avc1.77.30', | |
874 | 'acodec': 'mp4a.40.2', | |
875 | 'dynamic_range': None, | |
876 | }) | |
877 | self.assertEqual(parse_codecs('mp4a.40.2'), { | |
878 | 'vcodec': 'none', | |
879 | 'acodec': 'mp4a.40.2', | |
880 | 'dynamic_range': None, | |
881 | }) | |
882 | self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), { | |
883 | 'vcodec': 'avc1.42001e', | |
884 | 'acodec': 'mp4a.40.5', | |
885 | 'dynamic_range': None, | |
886 | }) | |
887 | self.assertEqual(parse_codecs('avc3.640028'), { | |
888 | 'vcodec': 'avc3.640028', | |
889 | 'acodec': 'none', | |
890 | 'dynamic_range': None, | |
891 | }) | |
892 | self.assertEqual(parse_codecs(', h264,,newcodec,aac'), { | |
893 | 'vcodec': 'h264', | |
894 | 'acodec': 'aac', | |
895 | 'dynamic_range': None, | |
896 | }) | |
897 | self.assertEqual(parse_codecs('av01.0.05M.08'), { | |
898 | 'vcodec': 'av01.0.05M.08', | |
899 | 'acodec': 'none', | |
900 | 'dynamic_range': None, | |
901 | }) | |
902 | self.assertEqual(parse_codecs('vp9.2'), { | |
903 | 'vcodec': 'vp9.2', | |
904 | 'acodec': 'none', | |
905 | 'dynamic_range': 'HDR10', | |
906 | }) | |
907 | self.assertEqual(parse_codecs('av01.0.12M.10.0.110.09.16.09.0'), { | |
908 | 'vcodec': 'av01.0.12M.10.0.110.09.16.09.0', | |
909 | 'acodec': 'none', | |
910 | 'dynamic_range': 'HDR10', | |
911 | }) | |
912 | self.assertEqual(parse_codecs('dvhe'), { | |
913 | 'vcodec': 'dvhe', | |
914 | 'acodec': 'none', | |
915 | 'dynamic_range': 'DV', | |
916 | }) | |
917 | self.assertEqual(parse_codecs('theora, vorbis'), { | |
918 | 'vcodec': 'theora', | |
919 | 'acodec': 'vorbis', | |
920 | 'dynamic_range': None, | |
921 | }) | |
922 | self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), { | |
923 | 'vcodec': 'unknownvcodec', | |
924 | 'acodec': 'unknownacodec', | |
925 | }) | |
926 | self.assertEqual(parse_codecs('unknown'), {}) | |
927 | ||
928 | def test_escape_rfc3986(self): | |
929 | reserved = "!*'();:@&=+$,/?#[]" | |
930 | unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~' | |
931 | self.assertEqual(escape_rfc3986(reserved), reserved) | |
932 | self.assertEqual(escape_rfc3986(unreserved), unreserved) | |
933 | self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82') | |
934 | self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82') | |
935 | self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar') | |
936 | self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar') | |
937 | ||
938 | def test_escape_url(self): | |
939 | self.assertEqual( | |
940 | escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'), | |
941 | 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4' | |
942 | ) | |
943 | self.assertEqual( | |
944 | escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'), | |
945 | 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290' | |
946 | ) | |
947 | self.assertEqual( | |
948 | escape_url('http://тест.рф/фрагмент'), | |
949 | 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' | |
950 | ) | |
951 | self.assertEqual( | |
952 | escape_url('http://тест.рф/абв?абв=абв#абв'), | |
953 | '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' | |
954 | ) | |
955 | self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0') | |
956 | ||
957 | def test_js_to_json_vars_strings(self): | |
958 | self.assertDictEqual( | |
959 | json.loads(js_to_json( | |
960 | '''{ | |
961 | 'null': a, | |
962 | 'nullStr': b, | |
963 | 'true': c, | |
964 | 'trueStr': d, | |
965 | 'false': e, | |
966 | 'falseStr': f, | |
967 | 'unresolvedVar': g, | |
968 | }''', | |
969 | { | |
970 | 'a': 'null', | |
971 | 'b': '"null"', | |
972 | 'c': 'true', | |
973 | 'd': '"true"', | |
974 | 'e': 'false', | |
975 | 'f': '"false"', | |
976 | 'g': 'var', | |
977 | } | |
978 | )), | |
979 | { | |
980 | 'null': None, | |
981 | 'nullStr': 'null', | |
982 | 'true': True, | |
983 | 'trueStr': 'true', | |
984 | 'false': False, | |
985 | 'falseStr': 'false', | |
986 | 'unresolvedVar': 'var' | |
987 | } | |
988 | ) | |
989 | ||
990 | self.assertDictEqual( | |
991 | json.loads(js_to_json( | |
992 | '''{ | |
993 | 'int': a, | |
994 | 'intStr': b, | |
995 | 'float': c, | |
996 | 'floatStr': d, | |
997 | }''', | |
998 | { | |
999 | 'a': '123', | |
1000 | 'b': '"123"', | |
1001 | 'c': '1.23', | |
1002 | 'd': '"1.23"', | |
1003 | } | |
1004 | )), | |
1005 | { | |
1006 | 'int': 123, | |
1007 | 'intStr': '123', | |
1008 | 'float': 1.23, | |
1009 | 'floatStr': '1.23', | |
1010 | } | |
1011 | ) | |
1012 | ||
1013 | self.assertDictEqual( | |
1014 | json.loads(js_to_json( | |
1015 | '''{ | |
1016 | 'object': a, | |
1017 | 'objectStr': b, | |
1018 | 'array': c, | |
1019 | 'arrayStr': d, | |
1020 | }''', | |
1021 | { | |
1022 | 'a': '{}', | |
1023 | 'b': '"{}"', | |
1024 | 'c': '[]', | |
1025 | 'd': '"[]"', | |
1026 | } | |
1027 | )), | |
1028 | { | |
1029 | 'object': {}, | |
1030 | 'objectStr': '{}', | |
1031 | 'array': [], | |
1032 | 'arrayStr': '[]', | |
1033 | } | |
1034 | ) | |
1035 | ||
1036 | def test_js_to_json_realworld(self): | |
1037 | inp = '''{ | |
1038 | 'clip':{'provider':'pseudo'} | |
1039 | }''' | |
1040 | self.assertEqual(js_to_json(inp), '''{ | |
1041 | "clip":{"provider":"pseudo"} | |
1042 | }''') | |
1043 | json.loads(js_to_json(inp)) | |
1044 | ||
1045 | inp = '''{ | |
1046 | 'playlist':[{'controls':{'all':null}}] | |
1047 | }''' | |
1048 | self.assertEqual(js_to_json(inp), '''{ | |
1049 | "playlist":[{"controls":{"all":null}}] | |
1050 | }''') | |
1051 | ||
1052 | inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"''' | |
1053 | self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''') | |
1054 | ||
1055 | inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"' | |
1056 | json_code = js_to_json(inp) | |
1057 | self.assertEqual(json.loads(json_code), json.loads(inp)) | |
1058 | ||
1059 | inp = '''{ | |
1060 | 0:{src:'skipped', type: 'application/dash+xml'}, | |
1061 | 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'}, | |
1062 | }''' | |
1063 | self.assertEqual(js_to_json(inp), '''{ | |
1064 | "0":{"src":"skipped", "type": "application/dash+xml"}, | |
1065 | "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"} | |
1066 | }''') | |
1067 | ||
1068 | inp = '''{"foo":101}''' | |
1069 | self.assertEqual(js_to_json(inp), '''{"foo":101}''') | |
1070 | ||
1071 | inp = '''{"duration": "00:01:07"}''' | |
1072 | self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''') | |
1073 | ||
1074 | inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''' | |
1075 | self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''') | |
1076 | ||
1077 | def test_js_to_json_edgecases(self): | |
1078 | on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}") | |
1079 | self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"}) | |
1080 | ||
1081 | on = js_to_json('{"abc": true}') | |
1082 | self.assertEqual(json.loads(on), {'abc': True}) | |
1083 | ||
1084 | # Ignore JavaScript code as well | |
1085 | on = js_to_json('''{ | |
1086 | "x": 1, | |
1087 | y: "a", | |
1088 | z: some.code | |
1089 | }''') | |
1090 | d = json.loads(on) | |
1091 | self.assertEqual(d['x'], 1) | |
1092 | self.assertEqual(d['y'], 'a') | |
1093 | ||
1094 | # Just drop ! prefix for now though this results in a wrong value | |
1095 | on = js_to_json('''{ | |
1096 | a: !0, | |
1097 | b: !1, | |
1098 | c: !!0, | |
1099 | d: !!42.42, | |
1100 | e: !!![], | |
1101 | f: !"abc", | |
1102 | g: !"", | |
1103 | !42: 42 | |
1104 | }''') | |
1105 | self.assertEqual(json.loads(on), { | |
1106 | 'a': 0, | |
1107 | 'b': 1, | |
1108 | 'c': 0, | |
1109 | 'd': 42.42, | |
1110 | 'e': [], | |
1111 | 'f': "abc", | |
1112 | 'g': "", | |
1113 | '42': 42 | |
1114 | }) | |
1115 | ||
1116 | on = js_to_json('["abc", "def",]') | |
1117 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
1118 | ||
1119 | on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]') | |
1120 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
1121 | ||
1122 | on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]') | |
1123 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
1124 | ||
1125 | on = js_to_json('{"abc": "def",}') | |
1126 | self.assertEqual(json.loads(on), {'abc': 'def'}) | |
1127 | ||
1128 | on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}') | |
1129 | self.assertEqual(json.loads(on), {'abc': 'def'}) | |
1130 | ||
1131 | on = js_to_json('{ 0: /* " \n */ ",]" , }') | |
1132 | self.assertEqual(json.loads(on), {'0': ',]'}) | |
1133 | ||
1134 | on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }') | |
1135 | self.assertEqual(json.loads(on), {'0': ',]'}) | |
1136 | ||
1137 | on = js_to_json('{ 0: // comment\n1 }') | |
1138 | self.assertEqual(json.loads(on), {'0': 1}) | |
1139 | ||
1140 | on = js_to_json(r'["<p>x<\/p>"]') | |
1141 | self.assertEqual(json.loads(on), ['<p>x</p>']) | |
1142 | ||
1143 | on = js_to_json(r'["\xaa"]') | |
1144 | self.assertEqual(json.loads(on), ['\u00aa']) | |
1145 | ||
1146 | on = js_to_json("['a\\\nb']") | |
1147 | self.assertEqual(json.loads(on), ['ab']) | |
1148 | ||
1149 | on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/") | |
1150 | self.assertEqual(json.loads(on), ['ab']) | |
1151 | ||
1152 | on = js_to_json('{0xff:0xff}') | |
1153 | self.assertEqual(json.loads(on), {'255': 255}) | |
1154 | ||
1155 | on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}') | |
1156 | self.assertEqual(json.loads(on), {'255': 255}) | |
1157 | ||
1158 | on = js_to_json('{077:077}') | |
1159 | self.assertEqual(json.loads(on), {'63': 63}) | |
1160 | ||
1161 | on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}') | |
1162 | self.assertEqual(json.loads(on), {'63': 63}) | |
1163 | ||
1164 | on = js_to_json('{42:42}') | |
1165 | self.assertEqual(json.loads(on), {'42': 42}) | |
1166 | ||
1167 | on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}') | |
1168 | self.assertEqual(json.loads(on), {'42': 42}) | |
1169 | ||
1170 | on = js_to_json('{42:4.2e1}') | |
1171 | self.assertEqual(json.loads(on), {'42': 42.0}) | |
1172 | ||
1173 | on = js_to_json('{ "0x40": "0x40" }') | |
1174 | self.assertEqual(json.loads(on), {'0x40': '0x40'}) | |
1175 | ||
1176 | on = js_to_json('{ "040": "040" }') | |
1177 | self.assertEqual(json.loads(on), {'040': '040'}) | |
1178 | ||
1179 | on = js_to_json('[1,//{},\n2]') | |
1180 | self.assertEqual(json.loads(on), [1, 2]) | |
1181 | ||
1182 | on = js_to_json(R'"\^\$\#"') | |
1183 | self.assertEqual(json.loads(on), R'^$#', msg='Unnecessary escapes should be stripped') | |
1184 | ||
1185 | on = js_to_json('\'"\\""\'') | |
1186 | self.assertEqual(json.loads(on), '"""', msg='Unnecessary quote escape should be escaped') | |
1187 | ||
1188 | def test_js_to_json_malformed(self): | |
1189 | self.assertEqual(js_to_json('42a1'), '42"a1"') | |
1190 | self.assertEqual(js_to_json('42a-1'), '42"a"-1') | |
1191 | ||
1192 | def test_extract_attributes(self): | |
1193 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
1194 | self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}) | |
1195 | self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}) | |
1196 | self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}) | |
1197 | self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}) | |
1198 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
1199 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
1200 | self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML | |
1201 | self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}) | |
1202 | self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2 | |
1203 | self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0 | |
1204 | self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}) | |
1205 | self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}) | |
1206 | self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}) | |
1207 | self.assertEqual(extract_attributes('<e x >'), {'x': None}) | |
1208 | self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}) | |
1209 | self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}) | |
1210 | self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}) | |
1211 | self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}) | |
1212 | self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}) | |
1213 | self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}) | |
1214 | self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}) | |
1215 | self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased | |
1216 | self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}) | |
1217 | self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}) | |
1218 | self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}) | |
1219 | self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}) | |
1220 | self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}) | |
1221 | # "Narrow" Python builds don't support unicode code points outside BMP. | |
1222 | try: | |
1223 | chr(0x10000) | |
1224 | supports_outside_bmp = True | |
1225 | except ValueError: | |
1226 | supports_outside_bmp = False | |
1227 | if supports_outside_bmp: | |
1228 | self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}) | |
1229 | # Malformed HTML should not break attributes extraction on older Python | |
1230 | self.assertEqual(extract_attributes('<mal"formed/>'), {}) | |
1231 | ||
1232 | def test_clean_html(self): | |
1233 | self.assertEqual(clean_html('a:\nb'), 'a: b') | |
1234 | self.assertEqual(clean_html('a:\n "b"'), 'a: "b"') | |
1235 | self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb') | |
1236 | ||
1237 | def test_intlist_to_bytes(self): | |
1238 | self.assertEqual( | |
1239 | intlist_to_bytes([0, 1, 127, 128, 255]), | |
1240 | b'\x00\x01\x7f\x80\xff') | |
1241 | ||
1242 | def test_args_to_str(self): | |
1243 | self.assertEqual( | |
1244 | args_to_str(['foo', 'ba/r', '-baz', '2 be', '']), | |
1245 | 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""' | |
1246 | ) | |
1247 | ||
1248 | def test_parse_filesize(self): | |
1249 | self.assertEqual(parse_filesize(None), None) | |
1250 | self.assertEqual(parse_filesize(''), None) | |
1251 | self.assertEqual(parse_filesize('91 B'), 91) | |
1252 | self.assertEqual(parse_filesize('foobar'), None) | |
1253 | self.assertEqual(parse_filesize('2 MiB'), 2097152) | |
1254 | self.assertEqual(parse_filesize('5 GB'), 5000000000) | |
1255 | self.assertEqual(parse_filesize('1.2Tb'), 1200000000000) | |
1256 | self.assertEqual(parse_filesize('1.2tb'), 1200000000000) | |
1257 | self.assertEqual(parse_filesize('1,24 KB'), 1240) | |
1258 | self.assertEqual(parse_filesize('1,24 kb'), 1240) | |
1259 | self.assertEqual(parse_filesize('8.5 megabytes'), 8500000) | |
1260 | ||
1261 | def test_parse_count(self): | |
1262 | self.assertEqual(parse_count(None), None) | |
1263 | self.assertEqual(parse_count(''), None) | |
1264 | self.assertEqual(parse_count('0'), 0) | |
1265 | self.assertEqual(parse_count('1000'), 1000) | |
1266 | self.assertEqual(parse_count('1.000'), 1000) | |
1267 | self.assertEqual(parse_count('1.1k'), 1100) | |
1268 | self.assertEqual(parse_count('1.1 k'), 1100) | |
1269 | self.assertEqual(parse_count('1,1 k'), 1100) | |
1270 | self.assertEqual(parse_count('1.1kk'), 1100000) | |
1271 | self.assertEqual(parse_count('1.1kk '), 1100000) | |
1272 | self.assertEqual(parse_count('1,1kk'), 1100000) | |
1273 | self.assertEqual(parse_count('100 views'), 100) | |
1274 | self.assertEqual(parse_count('1,100 views'), 1100) | |
1275 | self.assertEqual(parse_count('1.1kk views'), 1100000) | |
1276 | self.assertEqual(parse_count('10M views'), 10000000) | |
1277 | self.assertEqual(parse_count('has 10M views'), 10000000) | |
1278 | ||
1279 | def test_parse_resolution(self): | |
1280 | self.assertEqual(parse_resolution(None), {}) | |
1281 | self.assertEqual(parse_resolution(''), {}) | |
1282 | self.assertEqual(parse_resolution(' 1920x1080'), {'width': 1920, 'height': 1080}) | |
1283 | self.assertEqual(parse_resolution('1920×1080 '), {'width': 1920, 'height': 1080}) | |
1284 | self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}) | |
1285 | self.assertEqual(parse_resolution('720p'), {'height': 720}) | |
1286 | self.assertEqual(parse_resolution('4k'), {'height': 2160}) | |
1287 | self.assertEqual(parse_resolution('8K'), {'height': 4320}) | |
1288 | self.assertEqual(parse_resolution('pre_1920x1080_post'), {'width': 1920, 'height': 1080}) | |
1289 | self.assertEqual(parse_resolution('ep1x2'), {}) | |
1290 | self.assertEqual(parse_resolution('1920, 1080'), {'width': 1920, 'height': 1080}) | |
1291 | ||
1292 | def test_parse_bitrate(self): | |
1293 | self.assertEqual(parse_bitrate(None), None) | |
1294 | self.assertEqual(parse_bitrate(''), None) | |
1295 | self.assertEqual(parse_bitrate('300kbps'), 300) | |
1296 | self.assertEqual(parse_bitrate('1500kbps'), 1500) | |
1297 | self.assertEqual(parse_bitrate('300 kbps'), 300) | |
1298 | ||
1299 | def test_version_tuple(self): | |
1300 | self.assertEqual(version_tuple('1'), (1,)) | |
1301 | self.assertEqual(version_tuple('10.23.344'), (10, 23, 344)) | |
1302 | self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style | |
1303 | ||
1304 | def test_detect_exe_version(self): | |
1305 | self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1 | |
1306 | built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4) | |
1307 | configuration: --prefix=/usr --extra-'''), '1.2.1') | |
1308 | self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685 | |
1309 | built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685') | |
1310 | self.assertEqual(detect_exe_version('''X server found. dri2 connection failed! | |
1311 | Trying to open render node... | |
1312 | Success at /dev/dri/renderD128. | |
1313 | ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4') | |
1314 | ||
1315 | def test_age_restricted(self): | |
1316 | self.assertFalse(age_restricted(None, 10)) # unrestricted content | |
1317 | self.assertFalse(age_restricted(1, None)) # unrestricted policy | |
1318 | self.assertFalse(age_restricted(8, 10)) | |
1319 | self.assertTrue(age_restricted(18, 14)) | |
1320 | self.assertFalse(age_restricted(18, 18)) | |
1321 | ||
1322 | def test_is_html(self): | |
1323 | self.assertFalse(is_html(b'\x49\x44\x43<html')) | |
1324 | self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa')) | |
1325 | self.assertTrue(is_html( # UTF-8 with BOM | |
1326 | b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa')) | |
1327 | self.assertTrue(is_html( # UTF-16-LE | |
1328 | b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00' | |
1329 | )) | |
1330 | self.assertTrue(is_html( # UTF-16-BE | |
1331 | b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4' | |
1332 | )) | |
1333 | self.assertTrue(is_html( # UTF-32-BE | |
1334 | 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')) | |
1335 | self.assertTrue(is_html( # UTF-32-LE | |
1336 | 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')) | |
1337 | ||
1338 | def test_render_table(self): | |
1339 | self.assertEqual( | |
1340 | render_table( | |
1341 | ['a', 'empty', 'bcd'], | |
1342 | [[123, '', 4], [9999, '', 51]]), | |
1343 | 'a empty bcd\n' | |
1344 | '123 4\n' | |
1345 | '9999 51') | |
1346 | ||
1347 | self.assertEqual( | |
1348 | render_table( | |
1349 | ['a', 'empty', 'bcd'], | |
1350 | [[123, '', 4], [9999, '', 51]], | |
1351 | hide_empty=True), | |
1352 | 'a bcd\n' | |
1353 | '123 4\n' | |
1354 | '9999 51') | |
1355 | ||
1356 | self.assertEqual( | |
1357 | render_table( | |
1358 | ['\ta', 'bcd'], | |
1359 | [['1\t23', 4], ['\t9999', 51]]), | |
1360 | ' a bcd\n' | |
1361 | '1 23 4\n' | |
1362 | '9999 51') | |
1363 | ||
1364 | self.assertEqual( | |
1365 | render_table( | |
1366 | ['a', 'bcd'], | |
1367 | [[123, 4], [9999, 51]], | |
1368 | delim='-'), | |
1369 | 'a bcd\n' | |
1370 | '--------\n' | |
1371 | '123 4\n' | |
1372 | '9999 51') | |
1373 | ||
1374 | self.assertEqual( | |
1375 | render_table( | |
1376 | ['a', 'bcd'], | |
1377 | [[123, 4], [9999, 51]], | |
1378 | delim='-', extra_gap=2), | |
1379 | 'a bcd\n' | |
1380 | '----------\n' | |
1381 | '123 4\n' | |
1382 | '9999 51') | |
1383 | ||
1384 | def test_match_str(self): | |
1385 | # Unary | |
1386 | self.assertFalse(match_str('xy', {'x': 1200})) | |
1387 | self.assertTrue(match_str('!xy', {'x': 1200})) | |
1388 | self.assertTrue(match_str('x', {'x': 1200})) | |
1389 | self.assertFalse(match_str('!x', {'x': 1200})) | |
1390 | self.assertTrue(match_str('x', {'x': 0})) | |
1391 | self.assertTrue(match_str('is_live', {'is_live': True})) | |
1392 | self.assertFalse(match_str('is_live', {'is_live': False})) | |
1393 | self.assertFalse(match_str('is_live', {'is_live': None})) | |
1394 | self.assertFalse(match_str('is_live', {})) | |
1395 | self.assertFalse(match_str('!is_live', {'is_live': True})) | |
1396 | self.assertTrue(match_str('!is_live', {'is_live': False})) | |
1397 | self.assertTrue(match_str('!is_live', {'is_live': None})) | |
1398 | self.assertTrue(match_str('!is_live', {})) | |
1399 | self.assertTrue(match_str('title', {'title': 'abc'})) | |
1400 | self.assertTrue(match_str('title', {'title': ''})) | |
1401 | self.assertFalse(match_str('!title', {'title': 'abc'})) | |
1402 | self.assertFalse(match_str('!title', {'title': ''})) | |
1403 | ||
1404 | # Numeric | |
1405 | self.assertFalse(match_str('x>0', {'x': 0})) | |
1406 | self.assertFalse(match_str('x>0', {})) | |
1407 | self.assertTrue(match_str('x>?0', {})) | |
1408 | self.assertTrue(match_str('x>1K', {'x': 1200})) | |
1409 | self.assertFalse(match_str('x>2K', {'x': 1200})) | |
1410 | self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200})) | |
1411 | self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200})) | |
1412 | self.assertTrue(match_str('x > 1:0:0', {'x': 3700})) | |
1413 | ||
1414 | # String | |
1415 | self.assertFalse(match_str('y=a212', {'y': 'foobar42'})) | |
1416 | self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'})) | |
1417 | self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'})) | |
1418 | self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'})) | |
1419 | self.assertTrue(match_str('y^=foo', {'y': 'foobar42'})) | |
1420 | self.assertFalse(match_str('y!^=foo', {'y': 'foobar42'})) | |
1421 | self.assertFalse(match_str('y^=bar', {'y': 'foobar42'})) | |
1422 | self.assertTrue(match_str('y!^=bar', {'y': 'foobar42'})) | |
1423 | self.assertRaises(ValueError, match_str, 'x^=42', {'x': 42}) | |
1424 | self.assertTrue(match_str('y*=bar', {'y': 'foobar42'})) | |
1425 | self.assertFalse(match_str('y!*=bar', {'y': 'foobar42'})) | |
1426 | self.assertFalse(match_str('y*=baz', {'y': 'foobar42'})) | |
1427 | self.assertTrue(match_str('y!*=baz', {'y': 'foobar42'})) | |
1428 | self.assertTrue(match_str('y$=42', {'y': 'foobar42'})) | |
1429 | self.assertFalse(match_str('y$=43', {'y': 'foobar42'})) | |
1430 | ||
1431 | # And | |
1432 | self.assertFalse(match_str( | |
1433 | 'like_count > 100 & dislike_count <? 50 & description', | |
1434 | {'like_count': 90, 'description': 'foo'})) | |
1435 | self.assertTrue(match_str( | |
1436 | 'like_count > 100 & dislike_count <? 50 & description', | |
1437 | {'like_count': 190, 'description': 'foo'})) | |
1438 | self.assertFalse(match_str( | |
1439 | 'like_count > 100 & dislike_count <? 50 & description', | |
1440 | {'like_count': 190, 'dislike_count': 60, 'description': 'foo'})) | |
1441 | self.assertFalse(match_str( | |
1442 | 'like_count > 100 & dislike_count <? 50 & description', | |
1443 | {'like_count': 190, 'dislike_count': 10})) | |
1444 | ||
1445 | # Regex | |
1446 | self.assertTrue(match_str(r'x~=\bbar', {'x': 'foo bar'})) | |
1447 | self.assertFalse(match_str(r'x~=\bbar.+', {'x': 'foo bar'})) | |
1448 | self.assertFalse(match_str(r'x~=^FOO', {'x': 'foo bar'})) | |
1449 | self.assertTrue(match_str(r'x~=(?i)^FOO', {'x': 'foo bar'})) | |
1450 | ||
1451 | # Quotes | |
1452 | self.assertTrue(match_str(r'x^="foo"', {'x': 'foo "bar"'})) | |
1453 | self.assertFalse(match_str(r'x^="foo "', {'x': 'foo "bar"'})) | |
1454 | self.assertFalse(match_str(r'x$="bar"', {'x': 'foo "bar"'})) | |
1455 | self.assertTrue(match_str(r'x$=" \"bar\""', {'x': 'foo "bar"'})) | |
1456 | ||
1457 | # Escaping & | |
1458 | self.assertFalse(match_str(r'x=foo & bar', {'x': 'foo & bar'})) | |
1459 | self.assertTrue(match_str(r'x=foo \& bar', {'x': 'foo & bar'})) | |
1460 | self.assertTrue(match_str(r'x=foo \& bar & x^=foo', {'x': 'foo & bar'})) | |
1461 | self.assertTrue(match_str(r'x="foo \& bar" & x^=foo', {'x': 'foo & bar'})) | |
1462 | ||
1463 | # Example from docs | |
1464 | self.assertTrue(match_str( | |
1465 | r"!is_live & like_count>?100 & description~='(?i)\bcats \& dogs\b'", | |
1466 | {'description': 'Raining Cats & Dogs'})) | |
1467 | ||
1468 | # Incomplete | |
1469 | self.assertFalse(match_str('id!=foo', {'id': 'foo'}, True)) | |
1470 | self.assertTrue(match_str('x', {'id': 'foo'}, True)) | |
1471 | self.assertTrue(match_str('!x', {'id': 'foo'}, True)) | |
1472 | self.assertFalse(match_str('x', {'id': 'foo'}, False)) | |
1473 | ||
1474 | def test_parse_dfxp_time_expr(self): | |
1475 | self.assertEqual(parse_dfxp_time_expr(None), None) | |
1476 | self.assertEqual(parse_dfxp_time_expr(''), None) | |
1477 | self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1) | |
1478 | self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1) | |
1479 | self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0) | |
1480 | self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1) | |
1481 | self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1) | |
1482 | ||
1483 | def test_dfxp2srt(self): | |
1484 | dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?> | |
1485 | <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
1486 | <body> | |
1487 | <div xml:lang="en"> | |
1488 | <p begin="0" end="1">The following line contains Chinese characters and special symbols</p> | |
1489 | <p begin="1" end="2">第二行<br/>♪♪</p> | |
1490 | <p begin="2" dur="1"><span>Third<br/>Line</span></p> | |
1491 | <p begin="3" end="-1">Lines with invalid timestamps are ignored</p> | |
1492 | <p begin="-1" end="-1">Ignore, two</p> | |
1493 | <p begin="3" dur="-1">Ignored, three</p> | |
1494 | </div> | |
1495 | </body> | |
1496 | </tt>'''.encode() | |
1497 | srt_data = '''1 | |
1498 | 00:00:00,000 --> 00:00:01,000 | |
1499 | The following line contains Chinese characters and special symbols | |
1500 | ||
1501 | 2 | |
1502 | 00:00:01,000 --> 00:00:02,000 | |
1503 | 第二行 | |
1504 | ♪♪ | |
1505 | ||
1506 | 3 | |
1507 | 00:00:02,000 --> 00:00:03,000 | |
1508 | Third | |
1509 | Line | |
1510 | ||
1511 | ''' | |
1512 | self.assertEqual(dfxp2srt(dfxp_data), srt_data) | |
1513 | ||
1514 | dfxp_data_no_default_namespace = b'''<?xml version="1.0" encoding="UTF-8"?> | |
1515 | <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
1516 | <body> | |
1517 | <div xml:lang="en"> | |
1518 | <p begin="0" end="1">The first line</p> | |
1519 | </div> | |
1520 | </body> | |
1521 | </tt>''' | |
1522 | srt_data = '''1 | |
1523 | 00:00:00,000 --> 00:00:01,000 | |
1524 | The first line | |
1525 | ||
1526 | ''' | |
1527 | self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data) | |
1528 | ||
1529 | dfxp_data_with_style = b'''<?xml version="1.0" encoding="utf-8"?> | |
1530 | <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"> | |
1531 | <head> | |
1532 | <styling> | |
1533 | <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" /> | |
1534 | <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" /> | |
1535 | <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" /> | |
1536 | <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" /> | |
1537 | </styling> | |
1538 | </head> | |
1539 | <body tts:textAlign="center" style="s0"> | |
1540 | <div> | |
1541 | <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p> | |
1542 | <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> | |
1543 | <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p> | |
1544 | <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> | |
1545 | </div> | |
1546 | </body> | |
1547 | </tt>''' | |
1548 | srt_data = '''1 | |
1549 | 00:00:02,080 --> 00:00:05,840 | |
1550 | <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font> | |
1551 | ||
1552 | 2 | |
1553 | 00:00:02,080 --> 00:00:05,840 | |
1554 | <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1 | |
1555 | </font>part 2</font></b> | |
1556 | ||
1557 | 3 | |
1558 | 00:00:05,840 --> 00:00:09,560 | |
1559 | <u><font color="lime">line 3 | |
1560 | part 3</font></u> | |
1561 | ||
1562 | 4 | |
1563 | 00:00:09,560 --> 00:00:12,360 | |
1564 | <i><u><font color="yellow"><font color="lime">inner | |
1565 | </font>style</font></u></i> | |
1566 | ||
1567 | ''' | |
1568 | self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data) | |
1569 | ||
1570 | dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?> | |
1571 | <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
1572 | <body> | |
1573 | <div xml:lang="en"> | |
1574 | <p begin="0" end="1">Line 1</p> | |
1575 | <p begin="1" end="2">第二行</p> | |
1576 | </div> | |
1577 | </body> | |
1578 | </tt>'''.encode('utf-16') | |
1579 | srt_data = '''1 | |
1580 | 00:00:00,000 --> 00:00:01,000 | |
1581 | Line 1 | |
1582 | ||
1583 | 2 | |
1584 | 00:00:01,000 --> 00:00:02,000 | |
1585 | 第二行 | |
1586 | ||
1587 | ''' | |
1588 | self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data) | |
1589 | ||
1590 | def test_cli_option(self): | |
1591 | self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128']) | |
1592 | self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), []) | |
1593 | self.assertEqual(cli_option({}, '--proxy', 'proxy'), []) | |
1594 | self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10']) | |
1595 | ||
1596 | def test_cli_valueless_option(self): | |
1597 | self.assertEqual(cli_valueless_option( | |
1598 | {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader']) | |
1599 | self.assertEqual(cli_valueless_option( | |
1600 | {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), []) | |
1601 | self.assertEqual(cli_valueless_option( | |
1602 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate']) | |
1603 | self.assertEqual(cli_valueless_option( | |
1604 | {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), []) | |
1605 | self.assertEqual(cli_valueless_option( | |
1606 | {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), []) | |
1607 | self.assertEqual(cli_valueless_option( | |
1608 | {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate']) | |
1609 | ||
1610 | def test_cli_bool_option(self): | |
1611 | self.assertEqual( | |
1612 | cli_bool_option( | |
1613 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), | |
1614 | ['--no-check-certificate', 'true']) | |
1615 | self.assertEqual( | |
1616 | cli_bool_option( | |
1617 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='), | |
1618 | ['--no-check-certificate=true']) | |
1619 | self.assertEqual( | |
1620 | cli_bool_option( | |
1621 | {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'), | |
1622 | ['--check-certificate', 'false']) | |
1623 | self.assertEqual( | |
1624 | cli_bool_option( | |
1625 | {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
1626 | ['--check-certificate=false']) | |
1627 | self.assertEqual( | |
1628 | cli_bool_option( | |
1629 | {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'), | |
1630 | ['--check-certificate', 'true']) | |
1631 | self.assertEqual( | |
1632 | cli_bool_option( | |
1633 | {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
1634 | ['--check-certificate=true']) | |
1635 | self.assertEqual( | |
1636 | cli_bool_option( | |
1637 | {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
1638 | []) | |
1639 | ||
1640 | def test_ohdave_rsa_encrypt(self): | |
1641 | N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd | |
1642 | e = 65537 | |
1643 | ||
1644 | self.assertEqual( | |
1645 | ohdave_rsa_encrypt(b'aa111222', e, N), | |
1646 | '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881') | |
1647 | ||
1648 | def test_pkcs1pad(self): | |
1649 | data = [1, 2, 3] | |
1650 | padded_data = pkcs1pad(data, 32) | |
1651 | self.assertEqual(padded_data[:2], [0, 2]) | |
1652 | self.assertEqual(padded_data[28:], [0, 1, 2, 3]) | |
1653 | ||
1654 | self.assertRaises(ValueError, pkcs1pad, data, 8) | |
1655 | ||
1656 | def test_encode_base_n(self): | |
1657 | self.assertEqual(encode_base_n(0, 30), '0') | |
1658 | self.assertEqual(encode_base_n(80, 30), '2k') | |
1659 | ||
1660 | custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA' | |
1661 | self.assertEqual(encode_base_n(0, 30, custom_table), '9') | |
1662 | self.assertEqual(encode_base_n(80, 30, custom_table), '7P') | |
1663 | ||
1664 | self.assertRaises(ValueError, encode_base_n, 0, 70) | |
1665 | self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table) | |
1666 | ||
1667 | def test_caesar(self): | |
1668 | self.assertEqual(caesar('ace', 'abcdef', 2), 'cea') | |
1669 | self.assertEqual(caesar('cea', 'abcdef', -2), 'ace') | |
1670 | self.assertEqual(caesar('ace', 'abcdef', -2), 'eac') | |
1671 | self.assertEqual(caesar('eac', 'abcdef', 2), 'ace') | |
1672 | self.assertEqual(caesar('ace', 'abcdef', 0), 'ace') | |
1673 | self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz') | |
1674 | self.assertEqual(caesar('abc', 'acegik', 2), 'ebg') | |
1675 | self.assertEqual(caesar('ebg', 'acegik', -2), 'abc') | |
1676 | ||
1677 | def test_rot47(self): | |
1678 | self.assertEqual(rot47('yt-dlp'), r'JE\5=A') | |
1679 | self.assertEqual(rot47('YT-DLP'), r'*%\s{!') | |
1680 | ||
1681 | def test_urshift(self): | |
1682 | self.assertEqual(urshift(3, 1), 1) | |
1683 | self.assertEqual(urshift(-3, 1), 2147483646) | |
1684 | ||
1685 | GET_ELEMENT_BY_CLASS_TEST_STRING = ''' | |
1686 | <span class="foo bar">nice</span> | |
1687 | ''' | |
1688 | ||
1689 | def test_get_element_by_class(self): | |
1690 | html = self.GET_ELEMENT_BY_CLASS_TEST_STRING | |
1691 | ||
1692 | self.assertEqual(get_element_by_class('foo', html), 'nice') | |
1693 | self.assertEqual(get_element_by_class('no-such-class', html), None) | |
1694 | ||
1695 | def test_get_element_html_by_class(self): | |
1696 | html = self.GET_ELEMENT_BY_CLASS_TEST_STRING | |
1697 | ||
1698 | self.assertEqual(get_element_html_by_class('foo', html), html.strip()) | |
1699 | self.assertEqual(get_element_by_class('no-such-class', html), None) | |
1700 | ||
1701 | GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = ''' | |
1702 | <div itemprop="author" itemscope>foo</div> | |
1703 | ''' | |
1704 | ||
1705 | def test_get_element_by_attribute(self): | |
1706 | html = self.GET_ELEMENT_BY_CLASS_TEST_STRING | |
1707 | ||
1708 | self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice') | |
1709 | self.assertEqual(get_element_by_attribute('class', 'foo', html), None) | |
1710 | self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None) | |
1711 | ||
1712 | html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING | |
1713 | ||
1714 | self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo') | |
1715 | ||
1716 | def test_get_element_html_by_attribute(self): | |
1717 | html = self.GET_ELEMENT_BY_CLASS_TEST_STRING | |
1718 | ||
1719 | self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip()) | |
1720 | self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None) | |
1721 | self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None) | |
1722 | ||
1723 | html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING | |
1724 | ||
1725 | self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip()) | |
1726 | ||
1727 | GET_ELEMENTS_BY_CLASS_TEST_STRING = ''' | |
1728 | <span class="foo bar">nice</span><span class="foo bar">also nice</span> | |
1729 | ''' | |
1730 | GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>'] | |
1731 | ||
1732 | def test_get_elements_by_class(self): | |
1733 | html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING | |
1734 | ||
1735 | self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice']) | |
1736 | self.assertEqual(get_elements_by_class('no-such-class', html), []) | |
1737 | ||
1738 | def test_get_elements_html_by_class(self): | |
1739 | html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING | |
1740 | ||
1741 | self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES) | |
1742 | self.assertEqual(get_elements_html_by_class('no-such-class', html), []) | |
1743 | ||
1744 | def test_get_elements_by_attribute(self): | |
1745 | html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING | |
1746 | ||
1747 | self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice']) | |
1748 | self.assertEqual(get_elements_by_attribute('class', 'foo', html), []) | |
1749 | self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), []) | |
1750 | ||
1751 | def test_get_elements_html_by_attribute(self): | |
1752 | html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING | |
1753 | ||
1754 | self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES) | |
1755 | self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), []) | |
1756 | self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), []) | |
1757 | ||
1758 | def test_get_elements_text_and_html_by_attribute(self): | |
1759 | html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING | |
1760 | ||
1761 | self.assertEqual( | |
1762 | list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)), | |
1763 | list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES))) | |
1764 | self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), []) | |
1765 | self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), []) | |
1766 | ||
1767 | self.assertEqual(list(get_elements_text_and_html_by_attribute( | |
1768 | 'class', 'foo', '<a class="foo">nice</a><span class="foo">nice</span>', tag='a')), [('nice', '<a class="foo">nice</a>')]) | |
1769 | ||
1770 | GET_ELEMENT_BY_TAG_TEST_STRING = ''' | |
1771 | random text lorem ipsum</p> | |
1772 | <div> | |
1773 | this should be returned | |
1774 | <span>this should also be returned</span> | |
1775 | <div> | |
1776 | this should also be returned | |
1777 | </div> | |
1778 | closing tag above should not trick, so this should also be returned | |
1779 | </div> | |
1780 | but this text should not be returned | |
1781 | ''' | |
1782 | GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276] | |
1783 | GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6] | |
1784 | GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119] | |
1785 | GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7] | |
1786 | ||
1787 | def test_get_element_text_and_html_by_tag(self): | |
1788 | html = self.GET_ELEMENT_BY_TAG_TEST_STRING | |
1789 | ||
1790 | self.assertEqual( | |
1791 | get_element_text_and_html_by_tag('div', html), | |
1792 | (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML)) | |
1793 | self.assertEqual( | |
1794 | get_element_text_and_html_by_tag('span', html), | |
1795 | (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML)) | |
1796 | self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html) | |
1797 | ||
1798 | def test_iri_to_uri(self): | |
1799 | self.assertEqual( | |
1800 | iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'), | |
1801 | 'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b') # Same | |
1802 | self.assertEqual( | |
1803 | iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'), # German for cheese sauce stirring spoon | |
1804 | 'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel') | |
1805 | self.assertEqual( | |
1806 | iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'), | |
1807 | 'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#') | |
1808 | self.assertEqual( | |
1809 | iri_to_uri('http://правозащита38.рф/category/news/'), | |
1810 | 'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/') | |
1811 | self.assertEqual( | |
1812 | iri_to_uri('http://www.правозащита38.рф/category/news/'), | |
1813 | 'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/') | |
1814 | self.assertEqual( | |
1815 | iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'), | |
1816 | 'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA') | |
1817 | self.assertEqual( | |
1818 | iri_to_uri('http://日本語.jp/'), | |
1819 | 'http://xn--wgv71a119e.jp/') | |
1820 | self.assertEqual( | |
1821 | iri_to_uri('http://导航.中国/'), | |
1822 | 'http://xn--fet810g.xn--fiqs8s/') | |
1823 | ||
1824 | def test_clean_podcast_url(self): | |
1825 | self.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3') | |
1826 | self.assertEqual(clean_podcast_url('https://play.podtrac.com/npr-344098539/edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3'), 'https://edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3') | |
1827 | ||
1828 | def test_LazyList(self): | |
1829 | it = list(range(10)) | |
1830 | ||
1831 | self.assertEqual(list(LazyList(it)), it) | |
1832 | self.assertEqual(LazyList(it).exhaust(), it) | |
1833 | self.assertEqual(LazyList(it)[5], it[5]) | |
1834 | ||
1835 | self.assertEqual(LazyList(it)[5:], it[5:]) | |
1836 | self.assertEqual(LazyList(it)[:5], it[:5]) | |
1837 | self.assertEqual(LazyList(it)[::2], it[::2]) | |
1838 | self.assertEqual(LazyList(it)[1::2], it[1::2]) | |
1839 | self.assertEqual(LazyList(it)[5::-1], it[5::-1]) | |
1840 | self.assertEqual(LazyList(it)[6:2:-2], it[6:2:-2]) | |
1841 | self.assertEqual(LazyList(it)[::-1], it[::-1]) | |
1842 | ||
1843 | self.assertTrue(LazyList(it)) | |
1844 | self.assertFalse(LazyList(range(0))) | |
1845 | self.assertEqual(len(LazyList(it)), len(it)) | |
1846 | self.assertEqual(repr(LazyList(it)), repr(it)) | |
1847 | self.assertEqual(str(LazyList(it)), str(it)) | |
1848 | ||
1849 | self.assertEqual(list(LazyList(it, reverse=True)), it[::-1]) | |
1850 | self.assertEqual(list(reversed(LazyList(it))[::-1]), it) | |
1851 | self.assertEqual(list(reversed(LazyList(it))[1:3:7]), it[::-1][1:3:7]) | |
1852 | ||
1853 | def test_LazyList_laziness(self): | |
1854 | ||
1855 | def test(ll, idx, val, cache): | |
1856 | self.assertEqual(ll[idx], val) | |
1857 | self.assertEqual(ll._cache, list(cache)) | |
1858 | ||
1859 | ll = LazyList(range(10)) | |
1860 | test(ll, 0, 0, range(1)) | |
1861 | test(ll, 5, 5, range(6)) | |
1862 | test(ll, -3, 7, range(10)) | |
1863 | ||
1864 | ll = LazyList(range(10), reverse=True) | |
1865 | test(ll, -1, 0, range(1)) | |
1866 | test(ll, 3, 6, range(10)) | |
1867 | ||
1868 | ll = LazyList(itertools.count()) | |
1869 | test(ll, 10, 10, range(11)) | |
1870 | ll = reversed(ll) | |
1871 | test(ll, -15, 14, range(15)) | |
1872 | ||
1873 | def test_format_bytes(self): | |
1874 | self.assertEqual(format_bytes(0), '0.00B') | |
1875 | self.assertEqual(format_bytes(1000), '1000.00B') | |
1876 | self.assertEqual(format_bytes(1024), '1.00KiB') | |
1877 | self.assertEqual(format_bytes(1024**2), '1.00MiB') | |
1878 | self.assertEqual(format_bytes(1024**3), '1.00GiB') | |
1879 | self.assertEqual(format_bytes(1024**4), '1.00TiB') | |
1880 | self.assertEqual(format_bytes(1024**5), '1.00PiB') | |
1881 | self.assertEqual(format_bytes(1024**6), '1.00EiB') | |
1882 | self.assertEqual(format_bytes(1024**7), '1.00ZiB') | |
1883 | self.assertEqual(format_bytes(1024**8), '1.00YiB') | |
1884 | self.assertEqual(format_bytes(1024**9), '1024.00YiB') | |
1885 | ||
1886 | def test_hide_login_info(self): | |
1887 | self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']), | |
1888 | ['-u', 'PRIVATE', '-p', 'PRIVATE']) | |
1889 | self.assertEqual(Config.hide_login_info(['-u']), ['-u']) | |
1890 | self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']), | |
1891 | ['-u', 'PRIVATE', '-u', 'PRIVATE']) | |
1892 | self.assertEqual(Config.hide_login_info(['--username=foo']), | |
1893 | ['--username=PRIVATE']) | |
1894 | ||
1895 | def test_locked_file(self): | |
1896 | TEXT = 'test_locked_file\n' | |
1897 | FILE = 'test_locked_file.ytdl' | |
1898 | MODES = 'war' # Order is important | |
1899 | ||
1900 | try: | |
1901 | for lock_mode in MODES: | |
1902 | with locked_file(FILE, lock_mode, False) as f: | |
1903 | if lock_mode == 'r': | |
1904 | self.assertEqual(f.read(), TEXT * 2, 'Wrong file content') | |
1905 | else: | |
1906 | f.write(TEXT) | |
1907 | for test_mode in MODES: | |
1908 | testing_write = test_mode != 'r' | |
1909 | try: | |
1910 | with locked_file(FILE, test_mode, False): | |
1911 | pass | |
1912 | except (BlockingIOError, PermissionError): | |
1913 | if not testing_write: # FIXME | |
1914 | print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})') | |
1915 | continue | |
1916 | self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}') | |
1917 | else: | |
1918 | self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}') | |
1919 | finally: | |
1920 | with contextlib.suppress(OSError): | |
1921 | os.remove(FILE) | |
1922 | ||
1923 | def test_determine_file_encoding(self): | |
1924 | self.assertEqual(determine_file_encoding(b''), (None, 0)) | |
1925 | self.assertEqual(determine_file_encoding(b'--verbose -x --audio-format mkv\n'), (None, 0)) | |
1926 | ||
1927 | self.assertEqual(determine_file_encoding(b'\xef\xbb\xbf'), ('utf-8', 3)) | |
1928 | self.assertEqual(determine_file_encoding(b'\x00\x00\xfe\xff'), ('utf-32-be', 4)) | |
1929 | self.assertEqual(determine_file_encoding(b'\xff\xfe'), ('utf-16-le', 2)) | |
1930 | ||
1931 | self.assertEqual(determine_file_encoding(b'\xff\xfe# coding: utf-8\n--verbose'), ('utf-16-le', 2)) | |
1932 | ||
1933 | self.assertEqual(determine_file_encoding(b'# coding: utf-8\n--verbose'), ('utf-8', 0)) | |
1934 | self.assertEqual(determine_file_encoding(b'# coding: someencodinghere-12345\n--verbose'), ('someencodinghere-12345', 0)) | |
1935 | ||
1936 | self.assertEqual(determine_file_encoding(b'#coding:utf-8\n--verbose'), ('utf-8', 0)) | |
1937 | self.assertEqual(determine_file_encoding(b'# coding: utf-8 \r\n--verbose'), ('utf-8', 0)) | |
1938 | ||
1939 | self.assertEqual(determine_file_encoding('# coding: utf-32-be'.encode('utf-32-be')), ('utf-32-be', 0)) | |
1940 | self.assertEqual(determine_file_encoding('# coding: utf-16-le'.encode('utf-16-le')), ('utf-16-le', 0)) | |
1941 | ||
1942 | def test_get_compatible_ext(self): | |
1943 | self.assertEqual(get_compatible_ext( | |
1944 | vcodecs=[None], acodecs=[None, None], vexts=['mp4'], aexts=['m4a', 'm4a']), 'mkv') | |
1945 | self.assertEqual(get_compatible_ext( | |
1946 | vcodecs=[None], acodecs=[None], vexts=['flv'], aexts=['flv']), 'flv') | |
1947 | ||
1948 | self.assertEqual(get_compatible_ext( | |
1949 | vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['m4a']), 'mp4') | |
1950 | self.assertEqual(get_compatible_ext( | |
1951 | vcodecs=[None], acodecs=[None], vexts=['mp4'], aexts=['webm']), 'mkv') | |
1952 | self.assertEqual(get_compatible_ext( | |
1953 | vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['m4a']), 'mkv') | |
1954 | self.assertEqual(get_compatible_ext( | |
1955 | vcodecs=[None], acodecs=[None], vexts=['webm'], aexts=['webm']), 'webm') | |
1956 | ||
1957 | self.assertEqual(get_compatible_ext( | |
1958 | vcodecs=['h264'], acodecs=['mp4a'], vexts=['mov'], aexts=['m4a']), 'mp4') | |
1959 | self.assertEqual(get_compatible_ext( | |
1960 | vcodecs=['av01.0.12M.08'], acodecs=['opus'], vexts=['mp4'], aexts=['webm']), 'webm') | |
1961 | ||
1962 | self.assertEqual(get_compatible_ext( | |
1963 | vcodecs=['vp9'], acodecs=['opus'], vexts=['webm'], aexts=['webm'], preferences=['flv', 'mp4']), 'mp4') | |
1964 | self.assertEqual(get_compatible_ext( | |
1965 | vcodecs=['av1'], acodecs=['mp4a'], vexts=['webm'], aexts=['m4a'], preferences=('webm', 'mkv')), 'mkv') | |
1966 | ||
1967 | def test_traverse_obj(self): | |
1968 | _TEST_DATA = { | |
1969 | 100: 100, | |
1970 | 1.2: 1.2, | |
1971 | 'str': 'str', | |
1972 | 'None': None, | |
1973 | '...': ..., | |
1974 | 'urls': [ | |
1975 | {'index': 0, 'url': 'https://www.example.com/0'}, | |
1976 | {'index': 1, 'url': 'https://www.example.com/1'}, | |
1977 | ], | |
1978 | 'data': ( | |
1979 | {'index': 2}, | |
1980 | {'index': 3}, | |
1981 | ), | |
1982 | 'dict': {}, | |
1983 | } | |
1984 | ||
1985 | # Test base functionality | |
1986 | self.assertEqual(traverse_obj(_TEST_DATA, ('str',)), 'str', | |
1987 | msg='allow tuple path') | |
1988 | self.assertEqual(traverse_obj(_TEST_DATA, ['str']), 'str', | |
1989 | msg='allow list path') | |
1990 | self.assertEqual(traverse_obj(_TEST_DATA, (value for value in ("str",))), 'str', | |
1991 | msg='allow iterable path') | |
1992 | self.assertEqual(traverse_obj(_TEST_DATA, 'str'), 'str', | |
1993 | msg='single items should be treated as a path') | |
1994 | self.assertEqual(traverse_obj(_TEST_DATA, None), _TEST_DATA) | |
1995 | self.assertEqual(traverse_obj(_TEST_DATA, 100), 100) | |
1996 | self.assertEqual(traverse_obj(_TEST_DATA, 1.2), 1.2) | |
1997 | ||
1998 | # Test Ellipsis behavior | |
1999 | self.assertCountEqual(traverse_obj(_TEST_DATA, ...), | |
2000 | (item for item in _TEST_DATA.values() if item is not None), | |
2001 | msg='`...` should give all values except `None`') | |
2002 | self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', 0, ...)), _TEST_DATA['urls'][0].values(), | |
2003 | msg='`...` selection for dicts should select all values') | |
2004 | self.assertEqual(traverse_obj(_TEST_DATA, (..., ..., 'url')), | |
2005 | ['https://www.example.com/0', 'https://www.example.com/1'], | |
2006 | msg='nested `...` queries should work') | |
2007 | self.assertCountEqual(traverse_obj(_TEST_DATA, (..., ..., 'index')), range(4), | |
2008 | msg='`...` query result should be flattened') | |
2009 | ||
2010 | # Test function as key | |
2011 | self.assertEqual(traverse_obj(_TEST_DATA, lambda x, y: x == 'urls' and isinstance(y, list)), | |
2012 | [_TEST_DATA['urls']], | |
2013 | msg='function as query key should perform a filter based on (key, value)') | |
2014 | self.assertCountEqual(traverse_obj(_TEST_DATA, lambda _, x: isinstance(x[0], str)), {'str'}, | |
2015 | msg='exceptions in the query function should be catched') | |
2016 | ||
2017 | # Test alternative paths | |
2018 | self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'str'), 'str', | |
2019 | msg='multiple `paths` should be treated as alternative paths') | |
2020 | self.assertEqual(traverse_obj(_TEST_DATA, 'str', 100), 'str', | |
2021 | msg='alternatives should exit early') | |
2022 | self.assertEqual(traverse_obj(_TEST_DATA, 'fail', 'fail'), None, | |
2023 | msg='alternatives should return `default` if exhausted') | |
2024 | self.assertEqual(traverse_obj(_TEST_DATA, (..., 'fail'), 100), 100, | |
2025 | msg='alternatives should track their own branching return') | |
2026 | self.assertEqual(traverse_obj(_TEST_DATA, ('dict', ...), ('data', ...)), list(_TEST_DATA['data']), | |
2027 | msg='alternatives on empty objects should search further') | |
2028 | ||
2029 | # Test branch and path nesting | |
2030 | self.assertEqual(traverse_obj(_TEST_DATA, ('urls', (3, 0), 'url')), ['https://www.example.com/0'], | |
2031 | msg='tuple as key should be treated as branches') | |
2032 | self.assertEqual(traverse_obj(_TEST_DATA, ('urls', [3, 0], 'url')), ['https://www.example.com/0'], | |
2033 | msg='list as key should be treated as branches') | |
2034 | self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ((1, 'fail'), (0, 'url')))), ['https://www.example.com/0'], | |
2035 | msg='double nesting in path should be treated as paths') | |
2036 | self.assertEqual(traverse_obj(['0', [1, 2]], [(0, 1), 0]), [1], | |
2037 | msg='do not fail early on branching') | |
2038 | self.assertCountEqual(traverse_obj(_TEST_DATA, ('urls', ((1, ('fail', 'url')), (0, 'url')))), | |
2039 | ['https://www.example.com/0', 'https://www.example.com/1'], | |
2040 | msg='tripple nesting in path should be treated as branches') | |
2041 | self.assertEqual(traverse_obj(_TEST_DATA, ('urls', ('fail', (..., 'url')))), | |
2042 | ['https://www.example.com/0', 'https://www.example.com/1'], | |
2043 | msg='ellipsis as branch path start gets flattened') | |
2044 | ||
2045 | # Test dictionary as key | |
2046 | self.assertEqual(traverse_obj(_TEST_DATA, {0: 100, 1: 1.2}), {0: 100, 1: 1.2}, | |
2047 | msg='dict key should result in a dict with the same keys') | |
2048 | self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', 0, 'url')}), | |
2049 | {0: 'https://www.example.com/0'}, | |
2050 | msg='dict key should allow paths') | |
2051 | self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', (3, 0), 'url')}), | |
2052 | {0: ['https://www.example.com/0']}, | |
2053 | msg='tuple in dict path should be treated as branches') | |
2054 | self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, 'fail'), (0, 'url')))}), | |
2055 | {0: ['https://www.example.com/0']}, | |
2056 | msg='double nesting in dict path should be treated as paths') | |
2057 | self.assertEqual(traverse_obj(_TEST_DATA, {0: ('urls', ((1, ('fail', 'url')), (0, 'url')))}), | |
2058 | {0: ['https://www.example.com/1', 'https://www.example.com/0']}, | |
2059 | msg='tripple nesting in dict path should be treated as branches') | |
2060 | self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}), {}, | |
2061 | msg='remove `None` values when dict key') | |
2062 | self.assertEqual(traverse_obj(_TEST_DATA, {0: 'fail'}, default=...), {0: ...}, | |
2063 | msg='do not remove `None` values if `default`') | |
2064 | self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}), {0: {}}, | |
2065 | msg='do not remove empty values when dict key') | |
2066 | self.assertEqual(traverse_obj(_TEST_DATA, {0: 'dict'}, default=...), {0: {}}, | |
2067 | msg='do not remove empty values when dict key and a default') | |
2068 | self.assertEqual(traverse_obj(_TEST_DATA, {0: ('dict', ...)}), {0: []}, | |
2069 | msg='if branch in dict key not successful, return `[]`') | |
2070 | ||
2071 | # Testing default parameter behavior | |
2072 | _DEFAULT_DATA = {'None': None, 'int': 0, 'list': []} | |
2073 | self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail'), None, | |
2074 | msg='default value should be `None`') | |
2075 | self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', 'fail', default=...), ..., | |
2076 | msg='chained fails should result in default') | |
2077 | self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', 'int'), 0, | |
2078 | msg='should not short cirquit on `None`') | |
2079 | self.assertEqual(traverse_obj(_DEFAULT_DATA, 'fail', default=1), 1, | |
2080 | msg='invalid dict key should result in `default`') | |
2081 | self.assertEqual(traverse_obj(_DEFAULT_DATA, 'None', default=1), 1, | |
2082 | msg='`None` is a deliberate sentinel and should become `default`') | |
2083 | self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', 10)), None, | |
2084 | msg='`IndexError` should result in `default`') | |
2085 | self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=1), 1, | |
2086 | msg='if branched but not successful return `default` if defined, not `[]`') | |
2087 | self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail'), default=None), None, | |
2088 | msg='if branched but not successful return `default` even if `default` is `None`') | |
2089 | self.assertEqual(traverse_obj(_DEFAULT_DATA, (..., 'fail')), [], | |
2090 | msg='if branched but not successful return `[]`, not `default`') | |
2091 | self.assertEqual(traverse_obj(_DEFAULT_DATA, ('list', ...)), [], | |
2092 | msg='if branched but object is empty return `[]`, not `default`') | |
2093 | ||
2094 | # Testing expected_type behavior | |
2095 | _EXPECTED_TYPE_DATA = {'str': 'str', 'int': 0} | |
2096 | self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=str), 'str', | |
2097 | msg='accept matching `expected_type` type') | |
2098 | self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', expected_type=int), None, | |
2099 | msg='reject non matching `expected_type` type') | |
2100 | self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'int', expected_type=lambda x: str(x)), '0', | |
2101 | msg='transform type using type function') | |
2102 | self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, 'str', | |
2103 | expected_type=lambda _: 1 / 0), None, | |
2104 | msg='wrap expected_type fuction in try_call') | |
2105 | self.assertEqual(traverse_obj(_EXPECTED_TYPE_DATA, ..., expected_type=str), ['str'], | |
2106 | msg='eliminate items that expected_type fails on') | |
2107 | ||
2108 | # Test get_all behavior | |
2109 | _GET_ALL_DATA = {'key': [0, 1, 2]} | |
2110 | self.assertEqual(traverse_obj(_GET_ALL_DATA, ('key', ...), get_all=False), 0, | |
2111 | msg='if not `get_all`, return only first matching value') | |
2112 | self.assertEqual(traverse_obj(_GET_ALL_DATA, ..., get_all=False), [0, 1, 2], | |
2113 | msg='do not overflatten if not `get_all`') | |
2114 | ||
2115 | # Test casesense behavior | |
2116 | _CASESENSE_DATA = { | |
2117 | 'KeY': 'value0', | |
2118 | 0: { | |
2119 | 'KeY': 'value1', | |
2120 | 0: {'KeY': 'value2'}, | |
2121 | }, | |
2122 | } | |
2123 | self.assertEqual(traverse_obj(_CASESENSE_DATA, 'key'), None, | |
2124 | msg='dict keys should be case sensitive unless `casesense`') | |
2125 | self.assertEqual(traverse_obj(_CASESENSE_DATA, 'keY', | |
2126 | casesense=False), 'value0', | |
2127 | msg='allow non matching key case if `casesense`') | |
2128 | self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ('keY',)), | |
2129 | casesense=False), ['value1'], | |
2130 | msg='allow non matching key case in branch if `casesense`') | |
2131 | self.assertEqual(traverse_obj(_CASESENSE_DATA, (0, ((0, 'keY'),)), | |
2132 | casesense=False), ['value2'], | |
2133 | msg='allow non matching key case in branch path if `casesense`') | |
2134 | ||
2135 | # Test traverse_string behavior | |
2136 | _TRAVERSE_STRING_DATA = {'str': 'str', 1.2: 1.2} | |
2137 | self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0)), None, | |
2138 | msg='do not traverse into string if not `traverse_string`') | |
2139 | self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', 0), | |
2140 | traverse_string=True), 's', | |
2141 | msg='traverse into string if `traverse_string`') | |
2142 | self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, (1.2, 1), | |
2143 | traverse_string=True), '.', | |
2144 | msg='traverse into converted data if `traverse_string`') | |
2145 | self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', ...), | |
2146 | traverse_string=True), list('str'), | |
2147 | msg='`...` branching into string should result in list') | |
2148 | self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', (0, 2)), | |
2149 | traverse_string=True), ['s', 'r'], | |
2150 | msg='branching into string should result in list') | |
2151 | self.assertEqual(traverse_obj(_TRAVERSE_STRING_DATA, ('str', lambda _, x: x), | |
2152 | traverse_string=True), list('str'), | |
2153 | msg='function branching into string should result in list') | |
2154 | ||
2155 | # Test is_user_input behavior | |
2156 | _IS_USER_INPUT_DATA = {'range8': list(range(8))} | |
2157 | self.assertEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3'), | |
2158 | is_user_input=True), 3, | |
2159 | msg='allow for string indexing if `is_user_input`') | |
2160 | self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', '3:'), | |
2161 | is_user_input=True), tuple(range(8))[3:], | |
2162 | msg='allow for string slice if `is_user_input`') | |
2163 | self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':4:2'), | |
2164 | is_user_input=True), tuple(range(8))[:4:2], | |
2165 | msg='allow step in string slice if `is_user_input`') | |
2166 | self.assertCountEqual(traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':'), | |
2167 | is_user_input=True), range(8), | |
2168 | msg='`:` should be treated as `...` if `is_user_input`') | |
2169 | with self.assertRaises(TypeError, msg='too many params should result in error'): | |
2170 | traverse_obj(_IS_USER_INPUT_DATA, ('range8', ':::'), is_user_input=True) | |
2171 | ||
2172 | # Test re.Match as input obj | |
2173 | mobj = re.fullmatch(r'0(12)(?P<group>3)(4)?', '0123') | |
2174 | self.assertEqual(traverse_obj(mobj, ...), [x for x in mobj.groups() if x is not None], | |
2175 | msg='`...` on a `re.Match` should give its `groups()`') | |
2176 | self.assertEqual(traverse_obj(mobj, lambda k, _: k in (0, 2)), ['0123', '3'], | |
2177 | msg='function on a `re.Match` should give groupno, value starting at 0') | |
2178 | self.assertEqual(traverse_obj(mobj, 'group'), '3', | |
2179 | msg='str key on a `re.Match` should give group with that name') | |
2180 | self.assertEqual(traverse_obj(mobj, 2), '3', | |
2181 | msg='int key on a `re.Match` should give group with that name') | |
2182 | self.assertEqual(traverse_obj(mobj, 'gRoUp', casesense=False), '3', | |
2183 | msg='str key on a `re.Match` should respect casesense') | |
2184 | self.assertEqual(traverse_obj(mobj, 'fail'), None, | |
2185 | msg='failing str key on a `re.Match` should return `default`') | |
2186 | self.assertEqual(traverse_obj(mobj, 'gRoUpS', casesense=False), None, | |
2187 | msg='failing str key on a `re.Match` should return `default`') | |
2188 | self.assertEqual(traverse_obj(mobj, 8), None, | |
2189 | msg='failing int key on a `re.Match` should return `default`') | |
2190 | ||
2191 | ||
2192 | if __name__ == '__main__': | |
2193 | unittest.main() |