projects / yt-dlp.git / blame
| Commit | Line | Data |
|---|---|---|
| e387eb5a | 1 | #!/usr/bin/env python |
| 9d4660ca | 2 | # coding: utf-8 |
| e387eb5a | 3 | |
| 4e408e47 PH |
4 | from __future__ import unicode_literals |
| 5 | ||
| 44a5f171 PH |
6 | # Allow direct execution |
| 7 | import os | |
| dae7c920 | 8 | import sys |
| 44fb3454 | 9 | import unittest |
| 44a5f171 | 10 | sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__)))) |
| 44fb3454 | 11 | |
| 44a5f171 PH |
12 | |
| 13 | # Various small unit tests | |
| 62e609ab | 14 | import io |
| fac55558 | 15 | import json |
| 44a5f171 | 16 | import xml.etree.ElementTree |
| dae7c920 | 17 | |
| 7a5c1cfe | 18 | from yt_dlp.utils import ( |
| 05900629 | 19 | age_restricted, |
| cae97f65 | 20 | args_to_str, |
| 5eb6bdce | 21 | encode_base_n, |
| 1ced2221 | 22 | caesar, |
| e4bdb37e | 23 | clean_html, |
| 00dd0cd5 | 24 | clean_podcast_url, |
| eb9c3edd | 25 | date_from_str, |
| 9e62f283 | 26 | datetime_from_str, |
| a921f407 | 27 | DateRange, |
| cae97f65 | 28 | detect_exe_version, |
| 5035536e | 29 | determine_ext, |
| cbecc9b9 | 30 | dict_get, |
| 6b77d52b | 31 | encode_compat_str, |
| 29eb5174 | 32 | encodeFilename, |
| cae97f65 PH |
33 | escape_rfc3986, |
| 34 | escape_url, | |
| 8bb56eee | 35 | extract_attributes, |
| 5379a2d4 | 36 | ExtractorError, |
| a921f407 | 37 | find_xpath_attr, |
| 5aafe895 | 38 | fix_xml_ampersands, |
| 5e1271c5 | 39 | float_or_none, |
| 84c237fb | 40 | get_element_by_class, |
| 2af12ad9 TC |
41 | get_element_by_attribute, |
| 42 | get_elements_by_class, | |
| 43 | get_elements_by_attribute, | |
| 9c44d242 | 44 | InAdvancePagedList, |
| 5e1271c5 | 45 | int_or_none, |
| cae97f65 | 46 | intlist_to_bytes, |
| 61ca9a80 | 47 | is_html, |
| cae97f65 PH |
48 | js_to_json, |
| 49 | limit_length, | |
| 6cc62232 | 50 | merge_dicts, |
| 6562d34a | 51 | mimetype2ext, |
| f6717dec | 52 | month_by_name, |
| 0c265486 | 53 | multipart_encode, |
| 5bc880b9 | 54 | ohdave_rsa_encrypt, |
| cae97f65 PH |
55 | OnDemandPagedList, |
| 56 | orderedSet, | |
| a8795327 | 57 | parse_age_limit, |
| 608d11f5 | 58 | parse_duration, |
| cae97f65 | 59 | parse_filesize, |
| fb47597b | 60 | parse_count, |
| cae97f65 | 61 | parse_iso8601, |
| b871d7e9 | 62 | parse_resolution, |
| 0dc41787 | 63 | parse_bitrate, |
| f48409c7 | 64 | pkcs1pad, |
| 62e609ab | 65 | read_batch_urls, |
| 29eb5174 | 66 | sanitize_filename, |
| a2aaf4db | 67 | sanitize_path, |
| befa4708 | 68 | sanitize_url, |
| 5435dcf9 | 69 | sanitized_Request, |
| 51098426 | 70 | expand_path, |
| a4bcaad7 | 71 | prepend_extension, |
| b3ed15b7 | 72 | replace_extension, |
| 46bc9b7d S |
73 | remove_start, |
| 74 | remove_end, | |
| 31b2051e | 75 | remove_quotes, |
| 1ced2221 | 76 | rot47, |
| a6a173c2 | 77 | shell_quote, |
| 29eb5174 | 78 | smuggle_url, |
| f53c966a | 79 | str_to_int, |
| cae97f65 | 80 | strip_jsonp, |
| 53cd37ba | 81 | strip_or_none, |
| 824fa511 | 82 | subtitles_filename, |
| 29eb5174 PH |
83 | timeconvert, |
| 84 | unescapeHTML, | |
| 85 | unified_strdate, | |
| 46f59e89 | 86 | unified_timestamp, |
| 29eb5174 | 87 | unsmuggle_url, |
| cae97f65 | 88 | uppercase_escape, |
| 0fe2ff78 | 89 | lowercase_escape, |
| 29eb5174 | 90 | url_basename, |
| af03000a | 91 | url_or_none, |
| 02dc0a36 | 92 | base_url, |
| e34c3361 | 93 | urljoin, |
| b74fa8cd | 94 | urlencode_postdata, |
| 1143535d | 95 | urshift, |
| fb640d0a | 96 | update_url_query, |
| 5f9b8394 | 97 | version_tuple, |
| cae97f65 | 98 | xpath_with_ns, |
| 87f70ab3 | 99 | xpath_element, |
| 5379a2d4 | 100 | xpath_text, |
| 87f70ab3 | 101 | xpath_attr, |
| cfb56d1a | 102 | render_table, |
| 347de493 | 103 | match_str, |
| bf6427d2 YCH |
104 | parse_dfxp_time_expr, |
| 105 | dfxp2srt, | |
| f7126449 S |
106 | cli_option, |
| 107 | cli_valueless_option, | |
| 108 | cli_bool_option, | |
| 4f3c5e06 | 109 | parse_codecs, |
| 732044af | 110 | iri_to_uri, |
| a921f407 | 111 | ) |
| 7a5c1cfe | 112 | from yt_dlp.compat import ( |
| 8bb56eee | 113 | compat_chr, |
| 36e6f62c | 114 | compat_etree_fromstring, |
| 51098426 | 115 | compat_getenv, |
| dee2ff1d | 116 | compat_os_name, |
| 51098426 | 117 | compat_setenv, |
| fb640d0a | 118 | compat_urlparse, |
| 119 | compat_parse_qs, | |
| 36e6f62c | 120 | ) |
| 44fb3454 | 121 | |
| 627dcfff | 122 | |
| 44fb3454 | 123 | class TestUtil(unittest.TestCase): |
| 59ae15a5 PH |
124 | def test_timeconvert(self): |
| 125 | self.assertTrue(timeconvert('') is None) | |
| 126 | self.assertTrue(timeconvert('bougrg') is None) | |
| 127 | ||
| 128 | def test_sanitize_filename(self): | |
| 129 | self.assertEqual(sanitize_filename('abc'), 'abc') | |
| 130 | self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e') | |
| 131 | ||
| 132 | self.assertEqual(sanitize_filename('123'), '123') | |
| 133 | ||
| 134 | self.assertEqual('abc_de', sanitize_filename('abc/de')) | |
| 135 | self.assertFalse('/' in sanitize_filename('abc/de///')) | |
| 136 | ||
| 137 | self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de')) | |
| 138 | self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|')) | |
| 139 | self.assertEqual('yes no', sanitize_filename('yes? no')) | |
| 140 | self.assertEqual('this - that', sanitize_filename('this: that')) | |
| 141 | ||
| 142 | self.assertEqual(sanitize_filename('AT&T'), 'AT&T') | |
| 4e408e47 | 143 | aumlaut = 'ä' |
| 59ae15a5 | 144 | self.assertEqual(sanitize_filename(aumlaut), aumlaut) |
| 4e408e47 | 145 | tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430' |
| 59ae15a5 PH |
146 | self.assertEqual(sanitize_filename(tests), tests) |
| 147 | ||
| 2aeb06d6 PH |
148 | self.assertEqual( |
| 149 | sanitize_filename('New World record at 0:12:34'), | |
| 150 | 'New World record at 0_12_34') | |
| a7440261 | 151 | |
| 5a42414b PH |
152 | self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf') |
| 153 | self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf') | |
| a7440261 PH |
154 | self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf') |
| 155 | self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf') | |
| 2aeb06d6 | 156 | |
| 59ae15a5 PH |
157 | forbidden = '"\0\\/' |
| 158 | for fc in forbidden: | |
| 159 | for fbc in forbidden: | |
| 160 | self.assertTrue(fbc not in sanitize_filename(fc)) | |
| 161 | ||
| 162 | def test_sanitize_filename_restricted(self): | |
| 163 | self.assertEqual(sanitize_filename('abc', restricted=True), 'abc') | |
| 164 | self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e') | |
| 165 | ||
| 166 | self.assertEqual(sanitize_filename('123', restricted=True), '123') | |
| 167 | ||
| 168 | self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True)) | |
| 169 | self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True)) | |
| 170 | ||
| 171 | self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True)) | |
| 172 | self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True)) | |
| 173 | self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True)) | |
| 174 | self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True)) | |
| 175 | ||
| 79a2e94e AT |
176 | tests = 'aäb\u4e2d\u56fd\u7684c' |
| 177 | self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c') | |
| 4e408e47 | 178 | self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename |
| 59ae15a5 | 179 | |
| 627dcfff | 180 | forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#' |
| 59ae15a5 PH |
181 | for fc in forbidden: |
| 182 | for fbc in forbidden: | |
| 183 | self.assertTrue(fbc not in sanitize_filename(fc, restricted=True)) | |
| 184 | ||
| 185 | # Handle a common case more neatly | |
| 4e408e47 PH |
186 | self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song') |
| 187 | self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech') | |
| 59ae15a5 PH |
188 | # .. but make sure the file name is never empty |
| 189 | self.assertTrue(sanitize_filename('-', restricted=True) != '') | |
| 190 | self.assertTrue(sanitize_filename(':', restricted=True) != '') | |
| 191 | ||
| 79a2e94e | 192 | self.assertEqual(sanitize_filename( |
| b96f007e | 193 | 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True), |
| fd35d8cd | 194 | 'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy') |
| 79a2e94e | 195 | |
| 796173d0 | 196 | def test_sanitize_ids(self): |
| 314d506b PH |
197 | self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw') |
| 198 | self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw') | |
| 199 | self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI') | |
| 796173d0 | 200 | |
| a2aaf4db S |
201 | def test_sanitize_path(self): |
| 202 | if sys.platform != 'win32': | |
| 203 | return | |
| 204 | ||
| 205 | self.assertEqual(sanitize_path('abc'), 'abc') | |
| 206 | self.assertEqual(sanitize_path('abc/def'), 'abc\\def') | |
| 207 | self.assertEqual(sanitize_path('abc\\def'), 'abc\\def') | |
| 208 | self.assertEqual(sanitize_path('abc|def'), 'abc#def') | |
| 209 | self.assertEqual(sanitize_path('<>:"|?*'), '#######') | |
| 210 | self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def') | |
| 211 | self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def') | |
| 212 | ||
| 213 | self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc') | |
| 214 | self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc') | |
| 215 | ||
| 216 | self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') | |
| 217 | self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc') | |
| 218 | self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f') | |
| 219 | self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc') | |
| 220 | ||
| f18ef2d1 S |
221 | self.assertEqual( |
| 222 | sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'), | |
| 223 | 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s') | |
| 224 | ||
| 225 | self.assertEqual( | |
| 226 | sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'), | |
| 227 | 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part') | |
| 228 | self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#') | |
| 229 | self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def') | |
| 230 | self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#') | |
| 231 | ||
| 2ebfeaca S |
232 | self.assertEqual(sanitize_path('../abc'), '..\\abc') |
| 233 | self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc') | |
| 234 | self.assertEqual(sanitize_path('./abc'), 'abc') | |
| 235 | self.assertEqual(sanitize_path('./../abc'), '..\\abc') | |
| 236 | ||
| befa4708 S |
237 | def test_sanitize_url(self): |
| 238 | self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar') | |
| 239 | self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar') | |
| 240 | self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar') | |
| 241 | self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar') | |
| 242 | ||
| 5435dcf9 HH |
243 | def test_extract_basic_auth(self): |
| 244 | auth_header = lambda url: sanitized_Request(url).get_header('Authorization') | |
| 245 | self.assertFalse(auth_header('http://foo.bar')) | |
| 246 | self.assertFalse(auth_header('http://:foo.bar')) | |
| 247 | self.assertEqual(auth_header('http://@foo.bar'), 'Basic Og==') | |
| 248 | self.assertEqual(auth_header('http://:pass@foo.bar'), 'Basic OnBhc3M=') | |
| 249 | self.assertEqual(auth_header('http://user:@foo.bar'), 'Basic dXNlcjo=') | |
| 250 | self.assertEqual(auth_header('http://user:pass@foo.bar'), 'Basic dXNlcjpwYXNz') | |
| 251 | ||
| 51098426 | 252 | def test_expand_path(self): |
| 41c5e60d S |
253 | def env(var): |
| 254 | return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var) | |
| 255 | ||
| 7a5c1cfe P |
256 | compat_setenv('yt_dlp_EXPATH_PATH', 'expanded') |
| 257 | self.assertEqual(expand_path(env('yt_dlp_EXPATH_PATH')), 'expanded') | |
| a426ef6d | 258 | self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME')) |
| 51098426 | 259 | self.assertEqual(expand_path('~'), compat_getenv('HOME')) |
| 41c5e60d | 260 | self.assertEqual( |
| 7a5c1cfe | 261 | expand_path('~/%s' % env('yt_dlp_EXPATH_PATH')), |
| 41c5e60d | 262 | '%s/expanded' % compat_getenv('HOME')) |
| 51098426 | 263 | |
| a4bcaad7 S |
264 | def test_prepend_extension(self): |
| 265 | self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext') | |
| 266 | self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext') | |
| 267 | self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp') | |
| 268 | self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp') | |
| 269 | self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp') | |
| 270 | self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext') | |
| 271 | ||
| b3ed15b7 S |
272 | def test_replace_extension(self): |
| 273 | self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp') | |
| 274 | self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp') | |
| 275 | self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp') | |
| 276 | self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp') | |
| 277 | self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp') | |
| 278 | self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp') | |
| 279 | ||
| 824fa511 S |
280 | def test_subtitles_filename(self): |
| 281 | self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt') | |
| 282 | self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt') | |
| 283 | self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt') | |
| 284 | ||
| 46bc9b7d S |
285 | def test_remove_start(self): |
| 286 | self.assertEqual(remove_start(None, 'A - '), None) | |
| 287 | self.assertEqual(remove_start('A - B', 'A - '), 'B') | |
| 288 | self.assertEqual(remove_start('B - A', 'A - '), 'B - A') | |
| 289 | ||
| 290 | def test_remove_end(self): | |
| 291 | self.assertEqual(remove_end(None, ' - B'), None) | |
| 292 | self.assertEqual(remove_end('A - B', ' - B'), 'A') | |
| 293 | self.assertEqual(remove_end('B - A', ' - B'), 'B - A') | |
| 294 | ||
| 31b2051e S |
295 | def test_remove_quotes(self): |
| 296 | self.assertEqual(remove_quotes(None), None) | |
| 297 | self.assertEqual(remove_quotes('"'), '"') | |
| 298 | self.assertEqual(remove_quotes("'"), "'") | |
| 299 | self.assertEqual(remove_quotes(';'), ';') | |
| 300 | self.assertEqual(remove_quotes('";'), '";') | |
| 301 | self.assertEqual(remove_quotes('""'), '') | |
| 302 | self.assertEqual(remove_quotes('";"'), ';') | |
| 303 | ||
| 59ae15a5 | 304 | def test_ordered_set(self): |
| 627dcfff | 305 | self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7]) |
| 59ae15a5 PH |
306 | self.assertEqual(orderedSet([]), []) |
| 307 | self.assertEqual(orderedSet([1]), [1]) | |
| 5f6a1245 | 308 | # keep the list ordered |
| 627dcfff | 309 | self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1]) |
| 59ae15a5 PH |
310 | |
| 311 | def test_unescape_html(self): | |
| 4e408e47 | 312 | self.assertEqual(unescapeHTML('%20;'), '%20;') |
| 91757b0f NJ |
313 | self.assertEqual(unescapeHTML('/'), '/') |
| 314 | self.assertEqual(unescapeHTML('/'), '/') | |
| 7aefc49c S |
315 | self.assertEqual(unescapeHTML('é'), 'é') |
| 316 | self.assertEqual(unescapeHTML('�'), '�') | |
| 95f3f7c2 | 317 | self.assertEqual(unescapeHTML('&a"'), '&a"') |
| 55b2f099 YCH |
318 | # HTML5 entities |
| 319 | self.assertEqual(unescapeHTML('.''), '.\'') | |
| 5f6a1245 | 320 | |
| eb9c3edd JMF |
321 | def test_date_from_str(self): |
| 322 | self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day')) | |
| 323 | self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week')) | |
| 324 | self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week')) | |
| 9e62f283 | 325 | self.assertEqual(date_from_str('20200229+365day'), date_from_str('20200229+1year')) |
| 326 | self.assertEqual(date_from_str('20210131+28day'), date_from_str('20210131+1month')) | |
| 327 | ||
| 328 | def test_datetime_from_str(self): | |
| 329 | self.assertEqual(datetime_from_str('yesterday', precision='day'), datetime_from_str('now-1day', precision='auto')) | |
| 330 | self.assertEqual(datetime_from_str('now+7day', precision='day'), datetime_from_str('now+1week', precision='auto')) | |
| 331 | self.assertEqual(datetime_from_str('now+14day', precision='day'), datetime_from_str('now+2week', precision='auto')) | |
| 332 | self.assertEqual(datetime_from_str('20200229+365day', precision='day'), datetime_from_str('20200229+1year', precision='auto')) | |
| 333 | self.assertEqual(datetime_from_str('20210131+28day', precision='day'), datetime_from_str('20210131+1month', precision='auto')) | |
| 334 | self.assertEqual(datetime_from_str('20210131+59day', precision='day'), datetime_from_str('20210131+2month', precision='auto')) | |
| 335 | self.assertEqual(datetime_from_str('now+1day', precision='hour'), datetime_from_str('now+24hours', precision='auto')) | |
| 336 | self.assertEqual(datetime_from_str('now+23hours', precision='hour'), datetime_from_str('now+23hours', precision='auto')) | |
| eb9c3edd | 337 | |
| bd558525 | 338 | def test_daterange(self): |
| 5f6a1245 | 339 | _20century = DateRange("19000101", "20000101") |
| bd558525 JMF |
340 | self.assertFalse("17890714" in _20century) |
| 341 | _ac = DateRange("00010101") | |
| 342 | self.assertTrue("19690721" in _ac) | |
| 343 | _firstmilenium = DateRange(end="10000101") | |
| 344 | self.assertTrue("07110427" in _firstmilenium) | |
| 37254abc | 345 | |
| bf50b038 JMF |
346 | def test_unified_dates(self): |
| 347 | self.assertEqual(unified_strdate('December 21, 2010'), '20101221') | |
| 348 | self.assertEqual(unified_strdate('8/7/2009'), '20090708') | |
| 349 | self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214') | |
| 350 | self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011') | |
| a69801e2 | 351 | self.assertEqual(unified_strdate('1968 12 10'), '19681210') |
| 026fcc04 | 352 | self.assertEqual(unified_strdate('1968-12-10'), '19681210') |
| 99b67fec | 353 | self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128') |
| 42bdd9d0 PH |
354 | self.assertEqual( |
| 355 | unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False), | |
| 356 | '20141126') | |
| 9bb8e0a3 PH |
357 | self.assertEqual( |
| 358 | unified_strdate('2/2/2015 6:47:40 PM', day_first=False), | |
| 359 | '20150202') | |
| f160785c | 360 | self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214') |
| 8cf70de4 | 361 | self.assertEqual(unified_strdate('25-09-2014'), '20140925') |
| 46f59e89 | 362 | self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227') |
| 6a750402 | 363 | self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None) |
| c6eed6b8 | 364 | self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207') |
| cb655f34 S |
365 | self.assertEqual(unified_strdate('July 15th, 2013'), '20130715') |
| 366 | self.assertEqual(unified_strdate('September 1st, 2013'), '20130901') | |
| 367 | self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902') | |
| 9d30c213 I |
368 | self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103') |
| 369 | self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023') | |
| dae7c920 | 370 | |
| 46f59e89 S |
371 | def test_unified_timestamps(self): |
| 372 | self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600) | |
| 373 | self.assertEqual(unified_timestamp('8/7/2009'), 1247011200) | |
| 374 | self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200) | |
| 375 | self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598) | |
| 376 | self.assertEqual(unified_timestamp('1968 12 10'), -33436800) | |
| 377 | self.assertEqual(unified_timestamp('1968-12-10'), -33436800) | |
| 378 | self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200) | |
| 379 | self.assertEqual( | |
| 380 | unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False), | |
| 381 | 1417001400) | |
| 382 | self.assertEqual( | |
| 383 | unified_timestamp('2/2/2015 6:47:40 PM', day_first=False), | |
| 384 | 1422902860) | |
| 385 | self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900) | |
| 386 | self.assertEqual(unified_timestamp('25-09-2014'), 1411603200) | |
| 387 | self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200) | |
| 388 | self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None) | |
| 7dc2a74e | 389 | self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500) |
| c6eed6b8 | 390 | self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100) |
| deef3195 | 391 | self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361) |
| 2ae2ffda | 392 | self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540) |
| b555ae9b | 393 | self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140) |
| f226880c | 394 | self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363) |
| 46f59e89 | 395 | |
| 5035536e S |
396 | def test_determine_ext(self): |
| 397 | self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4') | |
| 398 | self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None) | |
| 9cb9a5df S |
399 | self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None) |
| 400 | self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None) | |
| 401 | self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8') | |
| 85750f89 | 402 | self.assertEqual(determine_ext('foobar', None), None) |
| 5035536e | 403 | |
| 59ae56fa | 404 | def test_find_xpath_attr(self): |
| 4e408e47 | 405 | testxml = '''<root> |
| 59ae56fa PH |
406 | <node/> |
| 407 | <node x="a"/> | |
| 408 | <node x="a" y="c" /> | |
| 409 | <node x="b" y="d" /> | |
| ee114368 | 410 | <node x="" /> |
| 59ae56fa | 411 | </root>''' |
| 36e6f62c | 412 | doc = compat_etree_fromstring(testxml) |
| 59ae56fa | 413 | |
| ee114368 | 414 | self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None) |
| 59ae56fa | 415 | self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None) |
| ee114368 S |
416 | self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None) |
| 417 | self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None) | |
| 418 | self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1]) | |
| 59ae56fa | 419 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1]) |
| ee114368 S |
420 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3]) |
| 421 | self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2]) | |
| 59ae56fa | 422 | self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2]) |
| ee114368 S |
423 | self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3]) |
| 424 | self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4]) | |
| 59ae56fa | 425 | |
| d7e66d39 | 426 | def test_xpath_with_ns(self): |
| 4e408e47 | 427 | testxml = '''<root xmlns:media="http://example.com/"> |
| d7e66d39 JMF |
428 | <media:song> |
| 429 | <media:author>The Author</media:author> | |
| 430 | <url>http://server.com/download.mp3</url> | |
| 431 | </media:song> | |
| 432 | </root>''' | |
| 36e6f62c | 433 | doc = compat_etree_fromstring(testxml) |
| d7e66d39 JMF |
434 | find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'})) |
| 435 | self.assertTrue(find('media:song') is not None) | |
| 4e408e47 PH |
436 | self.assertEqual(find('media:song/media:author').text, 'The Author') |
| 437 | self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3') | |
| d7e66d39 | 438 | |
| 87f70ab3 S |
439 | def test_xpath_element(self): |
| 440 | doc = xml.etree.ElementTree.Element('root') | |
| 441 | div = xml.etree.ElementTree.SubElement(doc, 'div') | |
| 442 | p = xml.etree.ElementTree.SubElement(div, 'p') | |
| 443 | p.text = 'Foo' | |
| 444 | self.assertEqual(xpath_element(doc, 'div/p'), p) | |
| 578c0745 S |
445 | self.assertEqual(xpath_element(doc, ['div/p']), p) |
| 446 | self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p) | |
| 87f70ab3 | 447 | self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default') |
| 578c0745 | 448 | self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default') |
| 87f70ab3 | 449 | self.assertTrue(xpath_element(doc, 'div/bar') is None) |
| 578c0745 S |
450 | self.assertTrue(xpath_element(doc, ['div/bar']) is None) |
| 451 | self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None) | |
| 87f70ab3 | 452 | self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True) |
| 578c0745 S |
453 | self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True) |
| 454 | self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True) | |
| 87f70ab3 | 455 | |
| 5379a2d4 JMF |
456 | def test_xpath_text(self): |
| 457 | testxml = '''<root> | |
| 458 | <div> | |
| 459 | <p>Foo</p> | |
| 460 | </div> | |
| 461 | </root>''' | |
| 36e6f62c | 462 | doc = compat_etree_fromstring(testxml) |
| 5379a2d4 | 463 | self.assertEqual(xpath_text(doc, 'div/p'), 'Foo') |
| 87f70ab3 | 464 | self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default') |
| 5379a2d4 JMF |
465 | self.assertTrue(xpath_text(doc, 'div/bar') is None) |
| 466 | self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True) | |
| 467 | ||
| 87f70ab3 S |
468 | def test_xpath_attr(self): |
| 469 | testxml = '''<root> | |
| 470 | <div> | |
| 471 | <p x="a">Foo</p> | |
| 472 | </div> | |
| 473 | </root>''' | |
| 36e6f62c | 474 | doc = compat_etree_fromstring(testxml) |
| 87f70ab3 S |
475 | self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a') |
| 476 | self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None) | |
| 477 | self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None) | |
| 478 | self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default') | |
| 479 | self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default') | |
| 480 | self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True) | |
| 481 | self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True) | |
| 482 | ||
| 9d4660ca | 483 | def test_smuggle_url(self): |
| e075a44a | 484 | data = {"ö": "ö", "abc": [3]} |
| 9d4660ca PH |
485 | url = 'https://foo.bar/baz?x=y#a' |
| 486 | smug_url = smuggle_url(url, data) | |
| 487 | unsmug_url, unsmug_data = unsmuggle_url(smug_url) | |
| 488 | self.assertEqual(url, unsmug_url) | |
| 489 | self.assertEqual(data, unsmug_data) | |
| 490 | ||
| 491 | res_url, res_data = unsmuggle_url(url) | |
| 492 | self.assertEqual(res_url, url) | |
| 493 | self.assertEqual(res_data, None) | |
| 494 | ||
| dfaa86b7 RA |
495 | smug_url = smuggle_url(url, {'a': 'b'}) |
| 496 | smug_smug_url = smuggle_url(smug_url, {'c': 'd'}) | |
| 497 | res_url, res_data = unsmuggle_url(smug_smug_url) | |
| 498 | self.assertEqual(res_url, url) | |
| 499 | self.assertEqual(res_data, {'a': 'b', 'c': 'd'}) | |
| 500 | ||
| a6a173c2 | 501 | def test_shell_quote(self): |
| 4e408e47 | 502 | args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')] |
| dee2ff1d S |
503 | self.assertEqual( |
| 504 | shell_quote(args), | |
| 505 | """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''') | |
| a6a173c2 | 506 | |
| 5e1271c5 S |
507 | def test_float_or_none(self): |
| 508 | self.assertEqual(float_or_none('42.42'), 42.42) | |
| 509 | self.assertEqual(float_or_none('42'), 42.0) | |
| 510 | self.assertEqual(float_or_none(''), None) | |
| 511 | self.assertEqual(float_or_none(None), None) | |
| 512 | self.assertEqual(float_or_none([]), None) | |
| 513 | self.assertEqual(float_or_none(set()), None) | |
| 514 | ||
| 515 | def test_int_or_none(self): | |
| 516 | self.assertEqual(int_or_none('42'), 42) | |
| 517 | self.assertEqual(int_or_none(''), None) | |
| 518 | self.assertEqual(int_or_none(None), None) | |
| 519 | self.assertEqual(int_or_none([]), None) | |
| 520 | self.assertEqual(int_or_none(set()), None) | |
| 521 | ||
| f53c966a JMF |
522 | def test_str_to_int(self): |
| 523 | self.assertEqual(str_to_int('123,456'), 123456) | |
| 524 | self.assertEqual(str_to_int('123.456'), 123456) | |
| 348c6bf1 | 525 | self.assertEqual(str_to_int(523), 523) |
| 42db58ec S |
526 | # Python 3 has no long |
| 527 | if sys.version_info < (3, 0): | |
| 528 | eval('self.assertEqual(str_to_int(123456L), 123456)') | |
| 529 | self.assertEqual(str_to_int('noninteger'), None) | |
| 530 | self.assertEqual(str_to_int([]), None) | |
| f53c966a | 531 | |
| 29eb5174 | 532 | def test_url_basename(self): |
| 4e408e47 PH |
533 | self.assertEqual(url_basename('http://foo.de/'), '') |
| 534 | self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz') | |
| 535 | self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz') | |
| 536 | self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz') | |
| 537 | self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz') | |
| d6c7a367 | 538 | self.assertEqual( |
| 4e408e47 PH |
539 | url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'), |
| 540 | 'trailer.mp4') | |
| 9d4660ca | 541 | |
| 02dc0a36 S |
542 | def test_base_url(self): |
| 543 | self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/') | |
| 544 | self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/') | |
| 545 | self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/') | |
| 546 | self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/') | |
| 547 | self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/') | |
| 548 | ||
| e34c3361 S |
549 | def test_urljoin(self): |
| 550 | self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| 4b5de77b S |
551 | self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') |
| 552 | self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| 553 | self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| b0c65c67 | 554 | self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt') |
| e34c3361 S |
555 | self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') |
| 556 | self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| 557 | self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| 558 | self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| b0c65c67 | 559 | self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') |
| e34c3361 | 560 | self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') |
| b0c65c67 | 561 | self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt') |
| e34c3361 S |
562 | self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') |
| 563 | self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt') | |
| 564 | self.assertEqual(urljoin('http://foo.de/', None), None) | |
| 565 | self.assertEqual(urljoin('http://foo.de/', ''), None) | |
| 566 | self.assertEqual(urljoin('http://foo.de/', ['foobar']), None) | |
| 7fe15920 | 567 | self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt') |
| fad4ceb5 S |
568 | self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de') |
| 569 | self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de') | |
| e34c3361 | 570 | |
| af03000a S |
571 | def test_url_or_none(self): |
| 572 | self.assertEqual(url_or_none(None), None) | |
| 573 | self.assertEqual(url_or_none(''), None) | |
| 574 | self.assertEqual(url_or_none('foo'), None) | |
| 575 | self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de') | |
| 576 | self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de') | |
| 577 | self.assertEqual(url_or_none('http$://foo.de'), None) | |
| 578 | self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de') | |
| 579 | self.assertEqual(url_or_none('//foo.de'), '//foo.de') | |
| 29f7c58a | 580 | self.assertEqual(url_or_none('s3://foo.de'), None) |
| 581 | self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de') | |
| 582 | self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de') | |
| 583 | self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de') | |
| 584 | self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de') | |
| af03000a | 585 | |
| a8795327 S |
586 | def test_parse_age_limit(self): |
| 587 | self.assertEqual(parse_age_limit(None), None) | |
| 588 | self.assertEqual(parse_age_limit(False), None) | |
| 589 | self.assertEqual(parse_age_limit('invalid'), None) | |
| 590 | self.assertEqual(parse_age_limit(0), 0) | |
| 591 | self.assertEqual(parse_age_limit(18), 18) | |
| 592 | self.assertEqual(parse_age_limit(21), 21) | |
| 593 | self.assertEqual(parse_age_limit(22), None) | |
| 594 | self.assertEqual(parse_age_limit('18'), 18) | |
| 595 | self.assertEqual(parse_age_limit('18+'), 18) | |
| 596 | self.assertEqual(parse_age_limit('PG-13'), 13) | |
| 597 | self.assertEqual(parse_age_limit('TV-14'), 14) | |
| 598 | self.assertEqual(parse_age_limit('TV-MA'), 17) | |
| 3bb3ff38 RA |
599 | self.assertEqual(parse_age_limit('TV14'), 14) |
| 600 | self.assertEqual(parse_age_limit('TV_G'), 0) | |
| a8795327 | 601 | |
| 608d11f5 PH |
602 | def test_parse_duration(self): |
| 603 | self.assertEqual(parse_duration(None), None) | |
| a5fb718c S |
604 | self.assertEqual(parse_duration(False), None) |
| 605 | self.assertEqual(parse_duration('invalid'), None) | |
| 608d11f5 PH |
606 | self.assertEqual(parse_duration('1'), 1) |
| 607 | self.assertEqual(parse_duration('1337:12'), 80232) | |
| 608 | self.assertEqual(parse_duration('9:12:43'), 33163) | |
| 2db806b4 S |
609 | self.assertEqual(parse_duration('12:00'), 720) |
| 610 | self.assertEqual(parse_duration('00:01:01'), 61) | |
| 608d11f5 | 611 | self.assertEqual(parse_duration('x:y'), None) |
| 2db806b4 | 612 | self.assertEqual(parse_duration('3h11m53s'), 11513) |
| ca7b3246 S |
613 | self.assertEqual(parse_duration('3h 11m 53s'), 11513) |
| 614 | self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513) | |
| 615 | self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513) | |
| 2db806b4 S |
616 | self.assertEqual(parse_duration('62m45s'), 3765) |
| 617 | self.assertEqual(parse_duration('6m59s'), 419) | |
| 618 | self.assertEqual(parse_duration('49s'), 49) | |
| 619 | self.assertEqual(parse_duration('0h0m0s'), 0) | |
| 620 | self.assertEqual(parse_duration('0m0s'), 0) | |
| 621 | self.assertEqual(parse_duration('0s'), 0) | |
| 7adcbe75 | 622 | self.assertEqual(parse_duration('01:02:03.05'), 3723.05) |
| 6a68bb57 | 623 | self.assertEqual(parse_duration('T30M38S'), 1838) |
| e8df5cee PH |
624 | self.assertEqual(parse_duration('5 s'), 5) |
| 625 | self.assertEqual(parse_duration('3 min'), 180) | |
| 626 | self.assertEqual(parse_duration('2.5 hours'), 9000) | |
| 8f4b58d7 PH |
627 | self.assertEqual(parse_duration('02:03:04'), 7384) |
| 628 | self.assertEqual(parse_duration('01:02:03:04'), 93784) | |
| 3e675fab | 629 | self.assertEqual(parse_duration('1 hour 3 minutes'), 3780) |
| 9c29bc69 | 630 | self.assertEqual(parse_duration('87 Min.'), 5220) |
| acaff495 | 631 | self.assertEqual(parse_duration('PT1H0.040S'), 3600.04) |
| 15846398 | 632 | self.assertEqual(parse_duration('PT00H03M30SZ'), 210) |
| 056653bb | 633 | self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88) |
| 608d11f5 | 634 | |
| 5aafe895 PH |
635 | def test_fix_xml_ampersands(self): |
| 636 | self.assertEqual( | |
| 637 | fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a') | |
| 638 | self.assertEqual( | |
| 639 | fix_xml_ampersands('"&x=y&wrong;&z=a'), | |
| 640 | '"&x=y&wrong;&z=a') | |
| 641 | self.assertEqual( | |
| 642 | fix_xml_ampersands('&'><"'), | |
| 643 | '&'><"') | |
| 644 | self.assertEqual( | |
| 645 | fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼') | |
| 646 | self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#') | |
| 647 | ||
| b7ab0590 PH |
648 | def test_paged_list(self): |
| 649 | def testPL(size, pagesize, sliceargs, expected): | |
| 650 | def get_page(pagenum): | |
| 651 | firstid = pagenum * pagesize | |
| 652 | upto = min(size, pagenum * pagesize + pagesize) | |
| 653 | for i in range(firstid, upto): | |
| 654 | yield i | |
| 655 | ||
| 9c44d242 | 656 | pl = OnDemandPagedList(get_page, pagesize) |
| b7ab0590 PH |
657 | got = pl.getslice(*sliceargs) |
| 658 | self.assertEqual(got, expected) | |
| 659 | ||
| 9c44d242 PH |
660 | iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize) |
| 661 | got = iapl.getslice(*sliceargs) | |
| 662 | self.assertEqual(got, expected) | |
| 663 | ||
| b7ab0590 PH |
664 | testPL(5, 2, (), [0, 1, 2, 3, 4]) |
| 665 | testPL(5, 2, (1,), [1, 2, 3, 4]) | |
| 666 | testPL(5, 2, (2,), [2, 3, 4]) | |
| 667 | testPL(5, 2, (4,), [4]) | |
| 668 | testPL(5, 2, (0, 3), [0, 1, 2]) | |
| 669 | testPL(5, 2, (1, 4), [1, 2, 3]) | |
| 670 | testPL(5, 2, (2, 99), [2, 3, 4]) | |
| 671 | testPL(5, 2, (20, 99), []) | |
| 672 | ||
| 62e609ab | 673 | def test_read_batch_urls(self): |
| 4e408e47 | 674 | f = io.StringIO('''\xef\xbb\xbf foo |
| 62e609ab PH |
675 | bar\r |
| 676 | baz | |
| 677 | # More after this line\r | |
| 678 | ; or after this | |
| 679 | bam''') | |
| 4e408e47 | 680 | self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam']) |
| 62e609ab | 681 | |
| b74fa8cd JMF |
682 | def test_urlencode_postdata(self): |
| 683 | data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}) | |
| 684 | self.assertTrue(isinstance(data, bytes)) | |
| 685 | ||
| fb640d0a | 686 | def test_update_url_query(self): |
| 687 | def query_dict(url): | |
| 688 | return compat_parse_qs(compat_urlparse.urlparse(url).query) | |
| 689 | self.assertEqual(query_dict(update_url_query( | |
| 690 | 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})), | |
| 691 | query_dict('http://example.com/path?quality=HD&format=mp4')) | |
| 692 | self.assertEqual(query_dict(update_url_query( | |
| 693 | 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})), | |
| 694 | query_dict('http://example.com/path?system=LINUX&system=WINDOWS')) | |
| 695 | self.assertEqual(query_dict(update_url_query( | |
| 696 | 'http://example.com/path', {'fields': 'id,formats,subtitles'})), | |
| 697 | query_dict('http://example.com/path?fields=id,formats,subtitles')) | |
| 698 | self.assertEqual(query_dict(update_url_query( | |
| 699 | 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})), | |
| 700 | query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails')) | |
| 701 | self.assertEqual(query_dict(update_url_query( | |
| 702 | 'http://example.com/path?manifest=f4m', {'manifest': []})), | |
| 703 | query_dict('http://example.com/path')) | |
| 704 | self.assertEqual(query_dict(update_url_query( | |
| 705 | 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})), | |
| 706 | query_dict('http://example.com/path?system=LINUX')) | |
| 707 | self.assertEqual(query_dict(update_url_query( | |
| 708 | 'http://example.com/path', {'fields': b'id,formats,subtitles'})), | |
| 709 | query_dict('http://example.com/path?fields=id,formats,subtitles')) | |
| 3201a67f | 710 | self.assertEqual(query_dict(update_url_query( |
| 711 | 'http://example.com/path', {'width': 1080, 'height': 720})), | |
| 712 | query_dict('http://example.com/path?width=1080&height=720')) | |
| 713 | self.assertEqual(query_dict(update_url_query( | |
| 714 | 'http://example.com/path', {'bitrate': 5020.43})), | |
| 715 | query_dict('http://example.com/path?bitrate=5020.43')) | |
| 716 | self.assertEqual(query_dict(update_url_query( | |
| 717 | 'http://example.com/path', {'test': '第二行тест'})), | |
| 718 | query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82')) | |
| fb640d0a | 719 | |
| 0c265486 YCH |
720 | def test_multipart_encode(self): |
| 721 | self.assertEqual( | |
| 722 | multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0], | |
| 723 | b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n') | |
| 724 | self.assertEqual( | |
| 725 | multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0], | |
| 726 | 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') | |
| 727 | self.assertRaises( | |
| 728 | ValueError, multipart_encode, {b'field': b'value'}, boundary='value') | |
| 729 | ||
| cbecc9b9 | 730 | def test_dict_get(self): |
| 86296ad2 S |
731 | FALSE_VALUES = { |
| 732 | 'none': None, | |
| 733 | 'false': False, | |
| 734 | 'zero': 0, | |
| 735 | 'empty_string': '', | |
| 736 | 'empty_list': [], | |
| cbecc9b9 | 737 | } |
| 86296ad2 S |
738 | d = FALSE_VALUES.copy() |
| 739 | d['a'] = 42 | |
| cbecc9b9 S |
740 | self.assertEqual(dict_get(d, 'a'), 42) |
| 741 | self.assertEqual(dict_get(d, 'b'), None) | |
| 742 | self.assertEqual(dict_get(d, 'b', 42), 42) | |
| 743 | self.assertEqual(dict_get(d, ('a', )), 42) | |
| 744 | self.assertEqual(dict_get(d, ('b', 'a', )), 42) | |
| 745 | self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42) | |
| 746 | self.assertEqual(dict_get(d, ('b', 'c', )), None) | |
| 747 | self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42) | |
| 86296ad2 S |
748 | for key, false_value in FALSE_VALUES.items(): |
| 749 | self.assertEqual(dict_get(d, ('b', 'c', key, )), None) | |
| 750 | self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value) | |
| cbecc9b9 | 751 | |
| 6cc62232 S |
752 | def test_merge_dicts(self): |
| 753 | self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2}) | |
| 754 | self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1}) | |
| 755 | self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1}) | |
| 756 | self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1}) | |
| 757 | self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1}) | |
| 758 | self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1}) | |
| 759 | self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''}) | |
| 760 | self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'}) | |
| 761 | self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'}) | |
| 762 | ||
| 6b77d52b S |
763 | def test_encode_compat_str(self): |
| 764 | self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест') | |
| 765 | self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест') | |
| 766 | ||
| 912b38b4 PH |
767 | def test_parse_iso8601(self): |
| 768 | self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266) | |
| 769 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266) | |
| 770 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266) | |
| 6ad4013d | 771 | self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266) |
| 52c3a6e4 S |
772 | self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251) |
| 773 | self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None) | |
| 912b38b4 | 774 | |
| fac55558 PH |
775 | def test_strip_jsonp(self): |
| 776 | stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);') | |
| 777 | d = json.loads(stripped) | |
| 778 | self.assertEqual(d, [{"id": "532cb", "x": 3}]) | |
| 779 | ||
| 609a61e3 PH |
780 | stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc') |
| 781 | d = json.loads(stripped) | |
| 782 | self.assertEqual(d, {'STATUS': 'OK'}) | |
| 783 | ||
| 8411229b S |
784 | stripped = strip_jsonp('ps.embedHandler({"status": "success"});') |
| 785 | d = json.loads(stripped) | |
| 786 | self.assertEqual(d, {'status': 'success'}) | |
| 787 | ||
| 5552c9eb YCH |
788 | stripped = strip_jsonp('window.cb && window.cb({"status": "success"});') |
| 789 | d = json.loads(stripped) | |
| 790 | self.assertEqual(d, {'status': 'success'}) | |
| 791 | ||
| 792 | stripped = strip_jsonp('window.cb && cb({"status": "success"});') | |
| 793 | d = json.loads(stripped) | |
| 794 | self.assertEqual(d, {'status': 'success'}) | |
| e9c671d5 S |
795 | |
| 796 | stripped = strip_jsonp('({"status": "success"});') | |
| 797 | d = json.loads(stripped) | |
| 798 | self.assertEqual(d, {'status': 'success'}) | |
| 5552c9eb | 799 | |
| 53cd37ba S |
800 | def test_strip_or_none(self): |
| 801 | self.assertEqual(strip_or_none(' abc'), 'abc') | |
| 802 | self.assertEqual(strip_or_none('abc '), 'abc') | |
| 803 | self.assertEqual(strip_or_none(' abc '), 'abc') | |
| 804 | self.assertEqual(strip_or_none('\tabc\t'), 'abc') | |
| 805 | self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc') | |
| 806 | self.assertEqual(strip_or_none('abc'), 'abc') | |
| 807 | self.assertEqual(strip_or_none(''), '') | |
| 808 | self.assertEqual(strip_or_none(None), None) | |
| 809 | self.assertEqual(strip_or_none(42), None) | |
| 810 | self.assertEqual(strip_or_none([]), None) | |
| 811 | ||
| 173a7026 | 812 | def test_uppercase_escape(self): |
| 4e408e47 PH |
813 | self.assertEqual(uppercase_escape('aä'), 'aä') |
| 814 | self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐') | |
| fac55558 | 815 | |
| 0fe2ff78 YCH |
816 | def test_lowercase_escape(self): |
| 817 | self.assertEqual(lowercase_escape('aä'), 'aä') | |
| 818 | self.assertEqual(lowercase_escape('\\u0026'), '&') | |
| 819 | ||
| a020a0dc PH |
820 | def test_limit_length(self): |
| 821 | self.assertEqual(limit_length(None, 12), None) | |
| 822 | self.assertEqual(limit_length('foo', 12), 'foo') | |
| 823 | self.assertTrue( | |
| 824 | limit_length('foo bar baz asd', 12).startswith('foo bar')) | |
| 825 | self.assertTrue('...' in limit_length('foo bar baz asd', 12)) | |
| 826 | ||
| 6562d34a S |
827 | def test_mimetype2ext(self): |
| 828 | self.assertEqual(mimetype2ext(None), None) | |
| 829 | self.assertEqual(mimetype2ext('video/x-flv'), 'flv') | |
| 830 | self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8') | |
| 831 | self.assertEqual(mimetype2ext('text/vtt'), 'vtt') | |
| 832 | self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt') | |
| 833 | self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html') | |
| 39e7107d U |
834 | self.assertEqual(mimetype2ext('audio/x-wav'), 'wav') |
| 835 | self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav') | |
| 6562d34a | 836 | |
| f6717dec S |
837 | def test_month_by_name(self): |
| 838 | self.assertEqual(month_by_name(None), None) | |
| 839 | self.assertEqual(month_by_name('December', 'en'), 12) | |
| 3e4185c3 | 840 | self.assertEqual(month_by_name('décembre', 'fr'), 12) |
| f6717dec | 841 | self.assertEqual(month_by_name('December'), 12) |
| 3e4185c3 | 842 | self.assertEqual(month_by_name('décembre'), None) |
| f6717dec S |
843 | self.assertEqual(month_by_name('Unknown', 'unknown'), None) |
| 844 | ||
| 4f3c5e06 | 845 | def test_parse_codecs(self): |
| 846 | self.assertEqual(parse_codecs(''), {}) | |
| 847 | self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), { | |
| 848 | 'vcodec': 'avc1.77.30', | |
| 849 | 'acodec': 'mp4a.40.2', | |
| 850 | }) | |
| 851 | self.assertEqual(parse_codecs('mp4a.40.2'), { | |
| 852 | 'vcodec': 'none', | |
| 853 | 'acodec': 'mp4a.40.2', | |
| 854 | }) | |
| 855 | self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), { | |
| 856 | 'vcodec': 'avc1.42001e', | |
| 857 | 'acodec': 'mp4a.40.5', | |
| 858 | }) | |
| 859 | self.assertEqual(parse_codecs('avc3.640028'), { | |
| 860 | 'vcodec': 'avc3.640028', | |
| 861 | 'acodec': 'none', | |
| 862 | }) | |
| 863 | self.assertEqual(parse_codecs(', h264,,newcodec,aac'), { | |
| 864 | 'vcodec': 'h264', | |
| 865 | 'acodec': 'aac', | |
| 866 | }) | |
| 25d110be S |
867 | self.assertEqual(parse_codecs('av01.0.05M.08'), { |
| 868 | 'vcodec': 'av01.0.05M.08', | |
| 869 | 'acodec': 'none', | |
| 870 | }) | |
| 28cc2241 S |
871 | self.assertEqual(parse_codecs('theora, vorbis'), { |
| 872 | 'vcodec': 'theora', | |
| 873 | 'acodec': 'vorbis', | |
| 874 | }) | |
| 875 | self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), { | |
| 876 | 'vcodec': 'unknownvcodec', | |
| 877 | 'acodec': 'unknownacodec', | |
| 878 | }) | |
| 879 | self.assertEqual(parse_codecs('unknown'), {}) | |
| 4f3c5e06 | 880 | |
| d05cfe06 S |
881 | def test_escape_rfc3986(self): |
| 882 | reserved = "!*'();:@&=+$,/?#[]" | |
| 883 | unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~' | |
| 884 | self.assertEqual(escape_rfc3986(reserved), reserved) | |
| 885 | self.assertEqual(escape_rfc3986(unreserved), unreserved) | |
| 886 | self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82') | |
| 887 | self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82') | |
| 888 | self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar') | |
| 889 | self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar') | |
| 890 | ||
| 891 | def test_escape_url(self): | |
| 892 | self.assertEqual( | |
| 893 | escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'), | |
| 894 | 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4' | |
| 895 | ) | |
| 896 | self.assertEqual( | |
| 897 | escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'), | |
| 898 | 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290' | |
| 899 | ) | |
| 900 | self.assertEqual( | |
| 901 | escape_url('http://тест.рф/фрагмент'), | |
| 2d60465e | 902 | 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82' |
| d05cfe06 S |
903 | ) |
| 904 | self.assertEqual( | |
| 905 | escape_url('http://тест.рф/абв?абв=абв#абв'), | |
| 81f36eba | 906 | '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' |
| d05cfe06 S |
907 | ) |
| 908 | self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0') | |
| 909 | ||
| e7b6d122 | 910 | def test_js_to_json_realworld(self): |
| 410f3e73 | 911 | inp = '''{ |
| e7b6d122 | 912 | 'clip':{'provider':'pseudo'} |
| 410f3e73 PH |
913 | }''' |
| 914 | self.assertEqual(js_to_json(inp), '''{ | |
| e7b6d122 | 915 | "clip":{"provider":"pseudo"} |
| 410f3e73 PH |
916 | }''') |
| 917 | json.loads(js_to_json(inp)) | |
| 918 | ||
| e7b6d122 PH |
919 | inp = '''{ |
| 920 | 'playlist':[{'controls':{'all':null}}] | |
| 921 | }''' | |
| 922 | self.assertEqual(js_to_json(inp), '''{ | |
| 923 | "playlist":[{"controls":{"all":null}}] | |
| 924 | }''') | |
| 925 | ||
| d01949dc S |
926 | inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"''' |
| 927 | self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''') | |
| 928 | ||
| d305dd73 PH |
929 | inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"' |
| 930 | json_code = js_to_json(inp) | |
| 931 | self.assertEqual(json.loads(json_code), json.loads(inp)) | |
| 932 | ||
| 89ac4a19 S |
933 | inp = '''{ |
| 934 | 0:{src:'skipped', type: 'application/dash+xml'}, | |
| 935 | 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'}, | |
| 936 | }''' | |
| 937 | self.assertEqual(js_to_json(inp), '''{ | |
| 938 | "0":{"src":"skipped", "type": "application/dash+xml"}, | |
| 939 | "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"} | |
| 940 | }''') | |
| 941 | ||
| 47212f7b YCH |
942 | inp = '''{"foo":101}''' |
| 943 | self.assertEqual(js_to_json(inp), '''{"foo":101}''') | |
| 944 | ||
| e4659b45 YCH |
945 | inp = '''{"duration": "00:01:07"}''' |
| 946 | self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''') | |
| 947 | ||
| c384d537 MF |
948 | inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''' |
| 949 | self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''') | |
| 950 | ||
| e7b6d122 PH |
951 | def test_js_to_json_edgecases(self): |
| 952 | on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}") | |
| 953 | self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"}) | |
| 954 | ||
| 955 | on = js_to_json('{"abc": true}') | |
| 956 | self.assertEqual(json.loads(on), {'abc': True}) | |
| 957 | ||
| 8f4b58d7 PH |
958 | # Ignore JavaScript code as well |
| 959 | on = js_to_json('''{ | |
| 960 | "x": 1, | |
| 961 | y: "a", | |
| 962 | z: some.code | |
| 963 | }''') | |
| 964 | d = json.loads(on) | |
| 965 | self.assertEqual(d['x'], 1) | |
| 966 | self.assertEqual(d['y'], 'a') | |
| 967 | ||
| 8bdd16b4 | 968 | # Just drop ! prefix for now though this results in a wrong value |
| 969 | on = js_to_json('''{ | |
| 970 | a: !0, | |
| 971 | b: !1, | |
| 972 | c: !!0, | |
| 973 | d: !!42.42, | |
| 974 | e: !!![], | |
| 975 | f: !"abc", | |
| 976 | g: !"", | |
| 977 | !42: 42 | |
| 978 | }''') | |
| 979 | self.assertEqual(json.loads(on), { | |
| 980 | 'a': 0, | |
| 981 | 'b': 1, | |
| 982 | 'c': 0, | |
| 983 | 'd': 42.42, | |
| 984 | 'e': [], | |
| 985 | 'f': "abc", | |
| 986 | 'g': "", | |
| 987 | '42': 42 | |
| 988 | }) | |
| 989 | ||
| ba9e68f4 S |
990 | on = js_to_json('["abc", "def",]') |
| 991 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
| 992 | ||
| 4195096e S |
993 | on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]') |
| 994 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
| 995 | ||
| 996 | on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]') | |
| 997 | self.assertEqual(json.loads(on), ['abc', 'def']) | |
| 998 | ||
| ba9e68f4 S |
999 | on = js_to_json('{"abc": "def",}') |
| 1000 | self.assertEqual(json.loads(on), {'abc': 'def'}) | |
| 1001 | ||
| 4195096e S |
1002 | on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}') |
| 1003 | self.assertEqual(json.loads(on), {'abc': 'def'}) | |
| 1004 | ||
| bd1e4844 | 1005 | on = js_to_json('{ 0: /* " \n */ ",]" , }') |
| 1006 | self.assertEqual(json.loads(on), {'0': ',]'}) | |
| 1007 | ||
| 4195096e S |
1008 | on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }') |
| 1009 | self.assertEqual(json.loads(on), {'0': ',]'}) | |
| 1010 | ||
| b3ee552e MČ |
1011 | on = js_to_json('{ 0: // comment\n1 }') |
| 1012 | self.assertEqual(json.loads(on), {'0': 1}) | |
| 1013 | ||
| bd1e4844 | 1014 | on = js_to_json(r'["<p>x<\/p>"]') |
| 1015 | self.assertEqual(json.loads(on), ['<p>x</p>']) | |
| 1016 | ||
| 1017 | on = js_to_json(r'["\xaa"]') | |
| 1018 | self.assertEqual(json.loads(on), ['\u00aa']) | |
| 1019 | ||
| 1020 | on = js_to_json("['a\\\nb']") | |
| 1021 | self.assertEqual(json.loads(on), ['ab']) | |
| 1022 | ||
| 4195096e S |
1023 | on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/") |
| 1024 | self.assertEqual(json.loads(on), ['ab']) | |
| 1025 | ||
| 89ac4a19 S |
1026 | on = js_to_json('{0xff:0xff}') |
| 1027 | self.assertEqual(json.loads(on), {'255': 255}) | |
| 1028 | ||
| 4195096e S |
1029 | on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}') |
| 1030 | self.assertEqual(json.loads(on), {'255': 255}) | |
| 1031 | ||
| 89ac4a19 S |
1032 | on = js_to_json('{077:077}') |
| 1033 | self.assertEqual(json.loads(on), {'63': 63}) | |
| 1034 | ||
| 4195096e S |
1035 | on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}') |
| 1036 | self.assertEqual(json.loads(on), {'63': 63}) | |
| 1037 | ||
| 89ac4a19 S |
1038 | on = js_to_json('{42:42}') |
| 1039 | self.assertEqual(json.loads(on), {'42': 42}) | |
| 1040 | ||
| 4195096e S |
1041 | on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}') |
| 1042 | self.assertEqual(json.loads(on), {'42': 42}) | |
| 1043 | ||
| c384d537 MF |
1044 | on = js_to_json('{42:4.2e1}') |
| 1045 | self.assertEqual(json.loads(on), {'42': 42.0}) | |
| 1046 | ||
| 8bdd16b4 | 1047 | on = js_to_json('{ "0x40": "0x40" }') |
| 1048 | self.assertEqual(json.loads(on), {'0x40': '0x40'}) | |
| 1049 | ||
| 1050 | on = js_to_json('{ "040": "040" }') | |
| 1051 | self.assertEqual(json.loads(on), {'040': '040'}) | |
| 1052 | ||
| c707b1d8 S |
1053 | def test_js_to_json_malformed(self): |
| 1054 | self.assertEqual(js_to_json('42a1'), '42"a1"') | |
| 1055 | self.assertEqual(js_to_json('42a-1'), '42"a"-1') | |
| 1056 | ||
| 8bb56eee BF |
1057 | def test_extract_attributes(self): |
| 1058 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
| 1059 | self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}) | |
| 1060 | self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}) | |
| 1061 | self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}) | |
| 1062 | self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}) | |
| 1063 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
| 1064 | self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}) | |
| 1065 | self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML | |
| 1066 | self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}) | |
| c5229f39 S |
1067 | self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2 |
| 1068 | self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0 | |
| 8bb56eee BF |
1069 | self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}) |
| 1070 | self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}) | |
| 1071 | self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}) | |
| 1072 | self.assertEqual(extract_attributes('<e x >'), {'x': None}) | |
| 1073 | self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}) | |
| 1074 | self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}) | |
| 1075 | self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}) | |
| 1076 | self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}) | |
| 1077 | self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}) | |
| 1078 | self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}) | |
| 1079 | self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}) | |
| c5229f39 | 1080 | self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased |
| 8bb56eee BF |
1081 | self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}) |
| 1082 | self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}) | |
| 1083 | self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}) | |
| 1084 | self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}) | |
| 1085 | self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}) | |
| 1086 | # "Narrow" Python builds don't support unicode code points outside BMP. | |
| 1087 | try: | |
| 1088 | compat_chr(0x10000) | |
| 1089 | supports_outside_bmp = True | |
| 1090 | except ValueError: | |
| 1091 | supports_outside_bmp = False | |
| 1092 | if supports_outside_bmp: | |
| 1093 | self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}) | |
| b4a3d461 S |
1094 | # Malformed HTML should not break attributes extraction on older Python |
| 1095 | self.assertEqual(extract_attributes('<mal"formed/>'), {}) | |
| 8bb56eee | 1096 | |
| e4bdb37e PH |
1097 | def test_clean_html(self): |
| 1098 | self.assertEqual(clean_html('a:\nb'), 'a: b') | |
| 1099 | self.assertEqual(clean_html('a:\n "b"'), 'a: "b"') | |
| 9222d945 | 1100 | self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb') |
| e4bdb37e | 1101 | |
| 4c0924bb PH |
1102 | def test_intlist_to_bytes(self): |
| 1103 | self.assertEqual( | |
| 1104 | intlist_to_bytes([0, 1, 127, 128, 255]), | |
| 1105 | b'\x00\x01\x7f\x80\xff') | |
| 1106 | ||
| 7d4111ed PH |
1107 | def test_args_to_str(self): |
| 1108 | self.assertEqual( | |
| 1109 | args_to_str(['foo', 'ba/r', '-baz', '2 be', '']), | |
| dee2ff1d | 1110 | 'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""' |
| 7d4111ed PH |
1111 | ) |
| 1112 | ||
| be64b5b0 PH |
1113 | def test_parse_filesize(self): |
| 1114 | self.assertEqual(parse_filesize(None), None) | |
| 1115 | self.assertEqual(parse_filesize(''), None) | |
| 1116 | self.assertEqual(parse_filesize('91 B'), 91) | |
| 1117 | self.assertEqual(parse_filesize('foobar'), None) | |
| 1118 | self.assertEqual(parse_filesize('2 MiB'), 2097152) | |
| 1119 | self.assertEqual(parse_filesize('5 GB'), 5000000000) | |
| 1120 | self.assertEqual(parse_filesize('1.2Tb'), 1200000000000) | |
| 13585d76 | 1121 | self.assertEqual(parse_filesize('1.2tb'), 1200000000000) |
| 4349c07d | 1122 | self.assertEqual(parse_filesize('1,24 KB'), 1240) |
| 13585d76 | 1123 | self.assertEqual(parse_filesize('1,24 kb'), 1240) |
| 70852b47 | 1124 | self.assertEqual(parse_filesize('8.5 megabytes'), 8500000) |
| be64b5b0 | 1125 | |
| fb47597b S |
1126 | def test_parse_count(self): |
| 1127 | self.assertEqual(parse_count(None), None) | |
| 1128 | self.assertEqual(parse_count(''), None) | |
| 1129 | self.assertEqual(parse_count('0'), 0) | |
| 1130 | self.assertEqual(parse_count('1000'), 1000) | |
| 1131 | self.assertEqual(parse_count('1.000'), 1000) | |
| 1132 | self.assertEqual(parse_count('1.1k'), 1100) | |
| 1133 | self.assertEqual(parse_count('1.1kk'), 1100000) | |
| 782b1b5b JMF |
1134 | self.assertEqual(parse_count('1.1kk '), 1100000) |
| 1135 | self.assertEqual(parse_count('1.1kk views'), 1100000) | |
| fb47597b | 1136 | |
| b871d7e9 S |
1137 | def test_parse_resolution(self): |
| 1138 | self.assertEqual(parse_resolution(None), {}) | |
| 1139 | self.assertEqual(parse_resolution(''), {}) | |
| 1140 | self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080}) | |
| 1141 | self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080}) | |
| 1142 | self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080}) | |
| 1143 | self.assertEqual(parse_resolution('720p'), {'height': 720}) | |
| 1144 | self.assertEqual(parse_resolution('4k'), {'height': 2160}) | |
| 1145 | self.assertEqual(parse_resolution('8K'), {'height': 4320}) | |
| 1146 | ||
| 0dc41787 S |
1147 | def test_parse_bitrate(self): |
| 1148 | self.assertEqual(parse_bitrate(None), None) | |
| 1149 | self.assertEqual(parse_bitrate(''), None) | |
| 1150 | self.assertEqual(parse_bitrate('300kbps'), 300) | |
| 1151 | self.assertEqual(parse_bitrate('1500kbps'), 1500) | |
| 1152 | self.assertEqual(parse_bitrate('300 kbps'), 300) | |
| 1153 | ||
| 5f9b8394 PH |
1154 | def test_version_tuple(self): |
| 1155 | self.assertEqual(version_tuple('1'), (1,)) | |
| 1156 | self.assertEqual(version_tuple('10.23.344'), (10, 23, 344)) | |
| 47d7c642 | 1157 | self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style |
| 5f9b8394 | 1158 | |
| cae97f65 PH |
1159 | def test_detect_exe_version(self): |
| 1160 | self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1 | |
| 1161 | built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4) | |
| 1162 | configuration: --prefix=/usr --extra-'''), '1.2.1') | |
| 1163 | self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685 | |
| 1164 | built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685') | |
| 1165 | self.assertEqual(detect_exe_version('''X server found. dri2 connection failed! | |
| 1166 | Trying to open render node... | |
| 1167 | Success at /dev/dri/renderD128. | |
| 1168 | ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4') | |
| 1169 | ||
| 05900629 PH |
1170 | def test_age_restricted(self): |
| 1171 | self.assertFalse(age_restricted(None, 10)) # unrestricted content | |
| 1172 | self.assertFalse(age_restricted(1, None)) # unrestricted policy | |
| 1173 | self.assertFalse(age_restricted(8, 10)) | |
| 1174 | self.assertTrue(age_restricted(18, 14)) | |
| 1175 | self.assertFalse(age_restricted(18, 18)) | |
| 1176 | ||
| 61ca9a80 PH |
1177 | def test_is_html(self): |
| 1178 | self.assertFalse(is_html(b'\x49\x44\x43<html')) | |
| 1179 | self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa')) | |
| 1180 | self.assertTrue(is_html( # UTF-8 with BOM | |
| 1181 | b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa')) | |
| 1182 | self.assertTrue(is_html( # UTF-16-LE | |
| 1183 | b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00' | |
| 1184 | )) | |
| 1185 | self.assertTrue(is_html( # UTF-16-BE | |
| 1186 | b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4' | |
| 1187 | )) | |
| 1188 | self.assertTrue(is_html( # UTF-32-BE | |
| 1189 | 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')) | |
| 1190 | self.assertTrue(is_html( # UTF-32-LE | |
| 1191 | 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')) | |
| 1192 | ||
| cfb56d1a PH |
1193 | def test_render_table(self): |
| 1194 | self.assertEqual( | |
| 1195 | render_table( | |
| 1196 | ['a', 'bcd'], | |
| 1197 | [[123, 4], [9999, 51]]), | |
| 1198 | 'a bcd\n' | |
| 1199 | '123 4\n' | |
| 1200 | '9999 51') | |
| 1201 | ||
| 347de493 PH |
1202 | def test_match_str(self): |
| 1203 | self.assertRaises(ValueError, match_str, 'xy>foobar', {}) | |
| 1204 | self.assertFalse(match_str('xy', {'x': 1200})) | |
| 1205 | self.assertTrue(match_str('!xy', {'x': 1200})) | |
| 1206 | self.assertTrue(match_str('x', {'x': 1200})) | |
| 1207 | self.assertFalse(match_str('!x', {'x': 1200})) | |
| 1208 | self.assertTrue(match_str('x', {'x': 0})) | |
| 1209 | self.assertFalse(match_str('x>0', {'x': 0})) | |
| 1210 | self.assertFalse(match_str('x>0', {})) | |
| 1211 | self.assertTrue(match_str('x>?0', {})) | |
| 1212 | self.assertTrue(match_str('x>1K', {'x': 1200})) | |
| 1213 | self.assertFalse(match_str('x>2K', {'x': 1200})) | |
| 1214 | self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200})) | |
| 1215 | self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200})) | |
| 1216 | self.assertFalse(match_str('y=a212', {'y': 'foobar42'})) | |
| 1217 | self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'})) | |
| 1218 | self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'})) | |
| 1219 | self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'})) | |
| 1220 | self.assertFalse(match_str( | |
| 1221 | 'like_count > 100 & dislike_count <? 50 & description', | |
| 1222 | {'like_count': 90, 'description': 'foo'})) | |
| 1223 | self.assertTrue(match_str( | |
| 1224 | 'like_count > 100 & dislike_count <? 50 & description', | |
| 1225 | {'like_count': 190, 'description': 'foo'})) | |
| 1226 | self.assertFalse(match_str( | |
| 1227 | 'like_count > 100 & dislike_count <? 50 & description', | |
| 1228 | {'like_count': 190, 'dislike_count': 60, 'description': 'foo'})) | |
| 1229 | self.assertFalse(match_str( | |
| 1230 | 'like_count > 100 & dislike_count <? 50 & description', | |
| 1231 | {'like_count': 190, 'dislike_count': 10})) | |
| 1cc47c66 S |
1232 | self.assertTrue(match_str('is_live', {'is_live': True})) |
| 1233 | self.assertFalse(match_str('is_live', {'is_live': False})) | |
| 1234 | self.assertFalse(match_str('is_live', {'is_live': None})) | |
| 1235 | self.assertFalse(match_str('is_live', {})) | |
| 1236 | self.assertFalse(match_str('!is_live', {'is_live': True})) | |
| 1237 | self.assertTrue(match_str('!is_live', {'is_live': False})) | |
| 1238 | self.assertTrue(match_str('!is_live', {'is_live': None})) | |
| 1239 | self.assertTrue(match_str('!is_live', {})) | |
| 1240 | self.assertTrue(match_str('title', {'title': 'abc'})) | |
| 1241 | self.assertTrue(match_str('title', {'title': ''})) | |
| 1242 | self.assertFalse(match_str('!title', {'title': 'abc'})) | |
| 1243 | self.assertFalse(match_str('!title', {'title': ''})) | |
| 347de493 | 1244 | |
| bf6427d2 | 1245 | def test_parse_dfxp_time_expr(self): |
| d631d5f9 YCH |
1246 | self.assertEqual(parse_dfxp_time_expr(None), None) |
| 1247 | self.assertEqual(parse_dfxp_time_expr(''), None) | |
| bf6427d2 YCH |
1248 | self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1) |
| 1249 | self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1) | |
| 1250 | self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0) | |
| 1251 | self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1) | |
| db2fe38b | 1252 | self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1) |
| bf6427d2 YCH |
1253 | |
| 1254 | def test_dfxp2srt(self): | |
| 1255 | dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?> | |
| 1256 | <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
| 1257 | <body> | |
| 1258 | <div xml:lang="en"> | |
| 1259 | <p begin="0" end="1">The following line contains Chinese characters and special symbols</p> | |
| 1260 | <p begin="1" end="2">第二行<br/>♪♪</p> | |
| 7dff0363 | 1261 | <p begin="2" dur="1"><span>Third<br/>Line</span></p> |
| d631d5f9 YCH |
1262 | <p begin="3" end="-1">Lines with invalid timestamps are ignored</p> |
| 1263 | <p begin="-1" end="-1">Ignore, two</p> | |
| 1264 | <p begin="3" dur="-1">Ignored, three</p> | |
| bf6427d2 YCH |
1265 | </div> |
| 1266 | </body> | |
| 3869028f | 1267 | </tt>'''.encode('utf-8') |
| bf6427d2 YCH |
1268 | srt_data = '''1 |
| 1269 | 00:00:00,000 --> 00:00:01,000 | |
| 1270 | The following line contains Chinese characters and special symbols | |
| 1271 | ||
| 1272 | 2 | |
| 1273 | 00:00:01,000 --> 00:00:02,000 | |
| 1274 | 第二行 | |
| 1275 | ♪♪ | |
| 1276 | ||
| 1277 | 3 | |
| 1278 | 00:00:02,000 --> 00:00:03,000 | |
| 1279 | Third | |
| 1280 | Line | |
| 1281 | ||
| 1282 | ''' | |
| 1283 | self.assertEqual(dfxp2srt(dfxp_data), srt_data) | |
| 1284 | ||
| 1b0427e6 YCH |
1285 | dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?> |
| 1286 | <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
| 1287 | <body> | |
| 1288 | <div xml:lang="en"> | |
| 1289 | <p begin="0" end="1">The first line</p> | |
| 1290 | </div> | |
| 1291 | </body> | |
| 3869028f | 1292 | </tt>'''.encode('utf-8') |
| 1b0427e6 YCH |
1293 | srt_data = '''1 |
| 1294 | 00:00:00,000 --> 00:00:01,000 | |
| 1295 | The first line | |
| 1296 | ||
| 1297 | ''' | |
| 1298 | self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data) | |
| 1299 | ||
| 5b995f71 RA |
1300 | dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?> |
| 1301 | <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"> | |
| 1302 | <head> | |
| 1303 | <styling> | |
| 1304 | <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" /> | |
| 1305 | <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" /> | |
| 1306 | <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" /> | |
| 1307 | <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" /> | |
| 1308 | </styling> | |
| 1309 | </head> | |
| 1310 | <body tts:textAlign="center" style="s0"> | |
| 1311 | <div> | |
| 1312 | <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p> | |
| 1313 | <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> | |
| 1314 | <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p> | |
| 1315 | <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> | |
| 1316 | </div> | |
| 1317 | </body> | |
| 3869028f | 1318 | </tt>'''.encode('utf-8') |
| 5b995f71 RA |
1319 | srt_data = '''1 |
| 1320 | 00:00:02,080 --> 00:00:05,839 | |
| 1321 | <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font> | |
| 1322 | ||
| 1323 | 2 | |
| 1324 | 00:00:02,080 --> 00:00:05,839 | |
| 1325 | <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1 | |
| 1326 | </font>part 2</font></b> | |
| 1327 | ||
| 1328 | 3 | |
| 1329 | 00:00:05,839 --> 00:00:09,560 | |
| 1330 | <u><font color="lime">line 3 | |
| 1331 | part 3</font></u> | |
| 1332 | ||
| 1333 | 4 | |
| 1334 | 00:00:09,560 --> 00:00:12,359 | |
| 1335 | <i><u><font color="yellow"><font color="lime">inner | |
| 1336 | </font>style</font></u></i> | |
| 1337 | ||
| 1338 | ''' | |
| 1339 | self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data) | |
| 1340 | ||
| 3869028f YCH |
1341 | dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?> |
| 1342 | <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter"> | |
| 1343 | <body> | |
| 1344 | <div xml:lang="en"> | |
| 1345 | <p begin="0" end="1">Line 1</p> | |
| 1346 | <p begin="1" end="2">第二行</p> | |
| 1347 | </div> | |
| 1348 | </body> | |
| 1349 | </tt>'''.encode('utf-16') | |
| 1350 | srt_data = '''1 | |
| 1351 | 00:00:00,000 --> 00:00:01,000 | |
| 1352 | Line 1 | |
| 1353 | ||
| 1354 | 2 | |
| 1355 | 00:00:01,000 --> 00:00:02,000 | |
| 1356 | 第二行 | |
| 1357 | ||
| 1358 | ''' | |
| 1359 | self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data) | |
| 1360 | ||
| f7126449 S |
1361 | def test_cli_option(self): |
| 1362 | self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128']) | |
| 1363 | self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), []) | |
| 1364 | self.assertEqual(cli_option({}, '--proxy', 'proxy'), []) | |
| 5f2c2b79 | 1365 | self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10']) |
| f7126449 S |
1366 | |
| 1367 | def test_cli_valueless_option(self): | |
| 1368 | self.assertEqual(cli_valueless_option( | |
| 1369 | {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader']) | |
| 1370 | self.assertEqual(cli_valueless_option( | |
| 1371 | {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), []) | |
| 1372 | self.assertEqual(cli_valueless_option( | |
| 1373 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate']) | |
| 1374 | self.assertEqual(cli_valueless_option( | |
| 1375 | {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), []) | |
| 1376 | self.assertEqual(cli_valueless_option( | |
| 1377 | {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), []) | |
| 1378 | self.assertEqual(cli_valueless_option( | |
| 1379 | {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate']) | |
| 1380 | ||
| 1381 | def test_cli_bool_option(self): | |
| 1382 | self.assertEqual( | |
| 1383 | cli_bool_option( | |
| 1384 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), | |
| 1385 | ['--no-check-certificate', 'true']) | |
| 1386 | self.assertEqual( | |
| 1387 | cli_bool_option( | |
| 1388 | {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='), | |
| 1389 | ['--no-check-certificate=true']) | |
| 1390 | self.assertEqual( | |
| 1391 | cli_bool_option( | |
| 1392 | {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'), | |
| 1393 | ['--check-certificate', 'false']) | |
| 1394 | self.assertEqual( | |
| 1395 | cli_bool_option( | |
| 1396 | {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
| 1397 | ['--check-certificate=false']) | |
| 1398 | self.assertEqual( | |
| 1399 | cli_bool_option( | |
| 1400 | {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'), | |
| 1401 | ['--check-certificate', 'true']) | |
| 1402 | self.assertEqual( | |
| 1403 | cli_bool_option( | |
| 1404 | {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
| 1405 | ['--check-certificate=true']) | |
| 5b232f46 S |
1406 | self.assertEqual( |
| 1407 | cli_bool_option( | |
| 1408 | {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='), | |
| 1409 | []) | |
| f7126449 | 1410 | |
| 5bc880b9 YCH |
1411 | def test_ohdave_rsa_encrypt(self): |
| 1412 | N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd | |
| 1413 | e = 65537 | |
| 1414 | ||
| 1415 | self.assertEqual( | |
| 1416 | ohdave_rsa_encrypt(b'aa111222', e, N), | |
| 1417 | '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881') | |
| cfb56d1a | 1418 | |
| f48409c7 YCH |
1419 | def test_pkcs1pad(self): |
| 1420 | data = [1, 2, 3] | |
| 1421 | padded_data = pkcs1pad(data, 32) | |
| 1422 | self.assertEqual(padded_data[:2], [0, 2]) | |
| 1423 | self.assertEqual(padded_data[28:], [0, 1, 2, 3]) | |
| 1424 | ||
| 1425 | self.assertRaises(ValueError, pkcs1pad, data, 8) | |
| 1426 | ||
| 5eb6bdce YCH |
1427 | def test_encode_base_n(self): |
| 1428 | self.assertEqual(encode_base_n(0, 30), '0') | |
| 1429 | self.assertEqual(encode_base_n(80, 30), '2k') | |
| 1430 | ||
| 1431 | custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA' | |
| 1432 | self.assertEqual(encode_base_n(0, 30, custom_table), '9') | |
| 1433 | self.assertEqual(encode_base_n(80, 30, custom_table), '7P') | |
| 1434 | ||
| 1435 | self.assertRaises(ValueError, encode_base_n, 0, 70) | |
| 1436 | self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table) | |
| 1437 | ||
| 1ced2221 S |
1438 | def test_caesar(self): |
| 1439 | self.assertEqual(caesar('ace', 'abcdef', 2), 'cea') | |
| 1440 | self.assertEqual(caesar('cea', 'abcdef', -2), 'ace') | |
| 1441 | self.assertEqual(caesar('ace', 'abcdef', -2), 'eac') | |
| 1442 | self.assertEqual(caesar('eac', 'abcdef', 2), 'ace') | |
| 1443 | self.assertEqual(caesar('ace', 'abcdef', 0), 'ace') | |
| 1444 | self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz') | |
| 1445 | self.assertEqual(caesar('abc', 'acegik', 2), 'ebg') | |
| 1446 | self.assertEqual(caesar('ebg', 'acegik', -2), 'abc') | |
| 1447 | ||
| 1448 | def test_rot47(self): | |
| 7a5c1cfe P |
1449 | self.assertEqual(rot47('yt-dlp'), r'JE\5=A') |
| 1450 | self.assertEqual(rot47('YT-DLP'), r'*%\s{!') | |
| 1ced2221 | 1451 | |
| 1143535d YCH |
1452 | def test_urshift(self): |
| 1453 | self.assertEqual(urshift(3, 1), 1) | |
| 1454 | self.assertEqual(urshift(-3, 1), 2147483646) | |
| 1455 | ||
| 84c237fb YCH |
1456 | def test_get_element_by_class(self): |
| 1457 | html = ''' | |
| 1458 | <span class="foo bar">nice</span> | |
| 1459 | ''' | |
| 1460 | ||
| 1461 | self.assertEqual(get_element_by_class('foo', html), 'nice') | |
| 1462 | self.assertEqual(get_element_by_class('no-such-class', html), None) | |
| 1463 | ||
| 2af12ad9 TC |
1464 | def test_get_element_by_attribute(self): |
| 1465 | html = ''' | |
| 1466 | <span class="foo bar">nice</span> | |
| 1467 | ''' | |
| 1468 | ||
| 1469 | self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice') | |
| 1470 | self.assertEqual(get_element_by_attribute('class', 'foo', html), None) | |
| 1471 | self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None) | |
| 1472 | ||
| 609ff8ca YCH |
1473 | html = ''' |
| 1474 | <div itemprop="author" itemscope>foo</div> | |
| 1475 | ''' | |
| 1476 | ||
| 1477 | self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo') | |
| 1478 | ||
| 2af12ad9 TC |
1479 | def test_get_elements_by_class(self): |
| 1480 | html = ''' | |
| 1481 | <span class="foo bar">nice</span><span class="foo bar">also nice</span> | |
| 1482 | ''' | |
| 1483 | ||
| 1484 | self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice']) | |
| 1485 | self.assertEqual(get_elements_by_class('no-such-class', html), []) | |
| 1486 | ||
| 1487 | def test_get_elements_by_attribute(self): | |
| 1488 | html = ''' | |
| 1489 | <span class="foo bar">nice</span><span class="foo bar">also nice</span> | |
| 1490 | ''' | |
| 1491 | ||
| 1492 | self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice']) | |
| 1493 | self.assertEqual(get_elements_by_attribute('class', 'foo', html), []) | |
| 1494 | self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), []) | |
| 1495 | ||
| 732044af | 1496 | def test_iri_to_uri(self): |
| 1497 | self.assertEqual( | |
| 1498 | iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'), | |
| 1499 | 'https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b') # Same | |
| 1500 | self.assertEqual( | |
| 1501 | iri_to_uri('https://www.google.com/search?q=Käsesoßenrührlöffel'), # German for cheese sauce stirring spoon | |
| 1502 | 'https://www.google.com/search?q=K%C3%A4seso%C3%9Fenr%C3%BChrl%C3%B6ffel') | |
| 1503 | self.assertEqual( | |
| 1504 | iri_to_uri('https://www.google.com/search?q=lt<+gt>+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#'), | |
| 1505 | 'https://www.google.com/search?q=lt%3C+gt%3E+eq%3D+amp%26+percent%25+hash%23+colon%3A+tilde~#trash=?&garbage=#') | |
| 1506 | self.assertEqual( | |
| 1507 | iri_to_uri('http://правозащита38.рф/category/news/'), | |
| 1508 | 'http://xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/') | |
| 1509 | self.assertEqual( | |
| 1510 | iri_to_uri('http://www.правозащита38.рф/category/news/'), | |
| 1511 | 'http://www.xn--38-6kcaak9aj5chl4a3g.xn--p1ai/category/news/') | |
| 1512 | self.assertEqual( | |
| 1513 | iri_to_uri('https://i❤.ws/emojidomain/👍👏🤝💪'), | |
| 1514 | 'https://xn--i-7iq.ws/emojidomain/%F0%9F%91%8D%F0%9F%91%8F%F0%9F%A4%9D%F0%9F%92%AA') | |
| 1515 | self.assertEqual( | |
| 1516 | iri_to_uri('http://日本語.jp/'), | |
| 1517 | 'http://xn--wgv71a119e.jp/') | |
| 1518 | self.assertEqual( | |
| 1519 | iri_to_uri('http://导航.中国/'), | |
| 1520 | 'http://xn--fet810g.xn--fiqs8s/') | |
| 1521 | ||
| 00dd0cd5 | 1522 | def test_clean_podcast_url(self): |
| 1523 | 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') | |
| 1524 | 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') | |
| 1525 | ||
| 582be358 | 1526 | |
| dae7c920 | 1527 | if __name__ == '__main__': |
| 59ae15a5 | 1528 | unittest.main() |