]> jfr.im git - yt-dlp.git/blame - test/test_utils.py
projects / yt-dlp.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
[test_utils] Add tests for determine_ext
[yt-dlp.git] / test / test_utils.py
CommitLineData
e387eb5a 1#!/usr/bin/env python
9d4660ca 2# coding: utf-8
e387eb5a 3
4e408e47
PH		 4from __future__ import unicode_literals
5
44a5f171
PH		 6# Allow direct execution
7import os
dae7c920 8import sys
44fb3454 9import unittest
44a5f171 10sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
44fb3454 11
44a5f171
PH		 12
13# Various small unit tests
62e609ab 14import io
fac55558 15import json
44a5f171 16import xml.etree.ElementTree
dae7c920 17
a921f407 18from youtube_dl.utils import (
05900629 19 age_restricted,
cae97f65 20 args_to_str,
e4bdb37e 21 clean_html,
a921f407 22 DateRange,
cae97f65 23 detect_exe_version,
5035536e 24 determine_ext,
29eb5174 25 encodeFilename,
cae97f65
PH		 26 escape_rfc3986,
27 escape_url,
5379a2d4 28 ExtractorError,
a921f407 29 find_xpath_attr,
5aafe895 30 fix_xml_ampersands,
9c44d242 31 InAdvancePagedList,
cae97f65 32 intlist_to_bytes,
61ca9a80 33 is_html,
cae97f65
PH		 34 js_to_json,
35 limit_length,
36 OnDemandPagedList,
37 orderedSet,
608d11f5 38 parse_duration,
cae97f65
PH		 39 parse_filesize,
40 parse_iso8601,
62e609ab 41 read_batch_urls,
29eb5174 42 sanitize_filename,
a2aaf4db 43 sanitize_path,
a4bcaad7 44 prepend_extension,
b3ed15b7 45 replace_extension,
a6a173c2 46 shell_quote,
29eb5174 47 smuggle_url,
f53c966a 48 str_to_int,
cae97f65 49 strip_jsonp,
b53466e1 50 struct_unpack,
29eb5174
PH		 51 timeconvert,
52 unescapeHTML,
53 unified_strdate,
54 unsmuggle_url,
cae97f65 55 uppercase_escape,
0fe2ff78 56 lowercase_escape,
29eb5174 57 url_basename,
b74fa8cd 58 urlencode_postdata,
5f9b8394 59 version_tuple,
cae97f65 60 xpath_with_ns,
87f70ab3 61 xpath_element,
5379a2d4 62 xpath_text,
87f70ab3 63 xpath_attr,
cfb56d1a 64 render_table,
347de493 65 match_str,
bf6427d2
YCH		 66 parse_dfxp_time_expr,
67 dfxp2srt,
f7126449
S		 68 cli_option,
69 cli_valueless_option,
70 cli_bool_option,
a921f407 71)
36e6f62c
JMF		 72from youtube_dl.compat import (
73 compat_etree_fromstring,
74)
44fb3454 75
627dcfff 76
44fb3454 77class TestUtil(unittest.TestCase):
59ae15a5
PH		 78 def test_timeconvert(self):
79 self.assertTrue(timeconvert('') is None)
80 self.assertTrue(timeconvert('bougrg') is None)
81
82 def test_sanitize_filename(self):
83 self.assertEqual(sanitize_filename('abc'), 'abc')
84 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
85
86 self.assertEqual(sanitize_filename('123'), '123')
87
88 self.assertEqual('abc_de', sanitize_filename('abc/de'))
89 self.assertFalse('/' in sanitize_filename('abc/de///'))
90
91 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
92 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
93 self.assertEqual('yes no', sanitize_filename('yes? no'))
94 self.assertEqual('this - that', sanitize_filename('this: that'))
95
96 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
4e408e47 97 aumlaut = 'ä'
59ae15a5 98 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
4e408e47 99 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
59ae15a5
PH		 100 self.assertEqual(sanitize_filename(tests), tests)
101
2aeb06d6
PH		 102 self.assertEqual(
103 sanitize_filename('New World record at 0:12:34'),
104 'New World record at 0_12_34')
a7440261 105
5a42414b
PH		 106 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
107 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
a7440261
PH		 108 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
109 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
2aeb06d6 110
59ae15a5
PH		 111 forbidden = '"\0\\/'
112 for fc in forbidden:
113 for fbc in forbidden:
114 self.assertTrue(fbc not in sanitize_filename(fc))
115
116 def test_sanitize_filename_restricted(self):
117 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
118 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
119
120 self.assertEqual(sanitize_filename('123', restricted=True), '123')
121
122 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
123 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
124
125 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
126 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
127 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
128 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
129
4e408e47 130 tests = 'a\xe4b\u4e2d\u56fd\u7684c'
59ae15a5 131 self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
4e408e47 132 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
59ae15a5 133
627dcfff 134 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
59ae15a5
PH		 135 for fc in forbidden:
136 for fbc in forbidden:
137 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
138
139 # Handle a common case more neatly
4e408e47
PH		 140 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
141 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
59ae15a5
PH		 142 # .. but make sure the file name is never empty
143 self.assertTrue(sanitize_filename('-', restricted=True) != '')
144 self.assertTrue(sanitize_filename(':', restricted=True) != '')
145
796173d0 146 def test_sanitize_ids(self):
314d506b
PH		 147 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
148 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
149 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
796173d0 150
a2aaf4db
S		 151 def test_sanitize_path(self):
152 if sys.platform != 'win32':
153 return
154
155 self.assertEqual(sanitize_path('abc'), 'abc')
156 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
157 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
158 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
159 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
160 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
161 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
162
163 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
164 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
165
166 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
167 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
168 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
169 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
170
f18ef2d1
S		 171 self.assertEqual(
172 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
173 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
174
175 self.assertEqual(
176 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
177 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
178 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
179 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
180 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
181
2ebfeaca
S		 182 self.assertEqual(sanitize_path('../abc'), '..\\abc')
183 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
184 self.assertEqual(sanitize_path('./abc'), 'abc')
185 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
186
a4bcaad7
S		 187 def test_prepend_extension(self):
188 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
189 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
190 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
191 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
192 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
193 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
194
b3ed15b7
S		 195 def test_replace_extension(self):
196 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
197 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
198 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
199 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
200 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
201 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
202
59ae15a5 203 def test_ordered_set(self):
627dcfff 204 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
59ae15a5
PH		 205 self.assertEqual(orderedSet([]), [])
206 self.assertEqual(orderedSet([1]), [1])
5f6a1245 207 # keep the list ordered
627dcfff 208 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
59ae15a5
PH		 209
210 def test_unescape_html(self):
4e408e47 211 self.assertEqual(unescapeHTML('%20;'), '%20;')
91757b0f
NJ		 212 self.assertEqual(unescapeHTML('&#x2F;'), '/')
213 self.assertEqual(unescapeHTML('&#47;'), '/')
7aefc49c
S		 214 self.assertEqual(unescapeHTML('&eacute;'), 'é')
215 self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
5f6a1245 216
bd558525 217 def test_daterange(self):
5f6a1245 218 _20century = DateRange("19000101", "20000101")
bd558525
JMF		 219 self.assertFalse("17890714" in _20century)
220 _ac = DateRange("00010101")
221 self.assertTrue("19690721" in _ac)
222 _firstmilenium = DateRange(end="10000101")
223 self.assertTrue("07110427" in _firstmilenium)
37254abc 224
bf50b038
JMF		 225 def test_unified_dates(self):
226 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
227 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
228 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
229 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
a69801e2 230 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
026fcc04 231 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
99b67fec 232 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
42bdd9d0
PH		 233 self.assertEqual(
234 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
235 '20141126')
9bb8e0a3
PH		 236 self.assertEqual(
237 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
238 '20150202')
8cf70de4 239 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
6a750402 240 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
dae7c920 241
5035536e
S		 242 def test_determine_ext(self):
243 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
244 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
245
59ae56fa 246 def test_find_xpath_attr(self):
4e408e47 247 testxml = '''<root>
59ae56fa
PH		 248 <node/>
249 <node x="a"/>
250 <node x="a" y="c" />
251 <node x="b" y="d" />
ee114368 252 <node x="" />
59ae56fa 253 </root>'''
36e6f62c 254 doc = compat_etree_fromstring(testxml)
59ae56fa 255
ee114368 256 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
59ae56fa 257 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
ee114368
S		 258 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
259 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
260 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
59ae56fa 261 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
ee114368
S		 262 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
263 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
59ae56fa 264 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
ee114368
S		 265 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
266 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
59ae56fa 267
d7e66d39 268 def test_xpath_with_ns(self):
4e408e47 269 testxml = '''<root xmlns:media="http://example.com/">
d7e66d39
JMF		 270 <media:song>
271 <media:author>The Author</media:author>
272 <url>http://server.com/download.mp3</url>
273 </media:song>
274 </root>'''
36e6f62c 275 doc = compat_etree_fromstring(testxml)
d7e66d39
JMF		 276 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
277 self.assertTrue(find('media:song') is not None)
4e408e47
PH		 278 self.assertEqual(find('media:song/media:author').text, 'The Author')
279 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
d7e66d39 280
87f70ab3
S		 281 def test_xpath_element(self):
282 doc = xml.etree.ElementTree.Element('root')
283 div = xml.etree.ElementTree.SubElement(doc, 'div')
284 p = xml.etree.ElementTree.SubElement(div, 'p')
285 p.text = 'Foo'
286 self.assertEqual(xpath_element(doc, 'div/p'), p)
578c0745
S		 287 self.assertEqual(xpath_element(doc, ['div/p']), p)
288 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
87f70ab3 289 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
578c0745 290 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
87f70ab3 291 self.assertTrue(xpath_element(doc, 'div/bar') is None)
578c0745
S		 292 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
293 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
87f70ab3 294 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
578c0745
S		 295 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
296 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
87f70ab3 297
5379a2d4
JMF		 298 def test_xpath_text(self):
299 testxml = '''<root>
300 <div>
301 <p>Foo</p>
302 </div>
303 </root>'''
36e6f62c 304 doc = compat_etree_fromstring(testxml)
5379a2d4 305 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
87f70ab3 306 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
5379a2d4
JMF		 307 self.assertTrue(xpath_text(doc, 'div/bar') is None)
308 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
309
87f70ab3
S		 310 def test_xpath_attr(self):
311 testxml = '''<root>
312 <div>
313 <p x="a">Foo</p>
314 </div>
315 </root>'''
36e6f62c 316 doc = compat_etree_fromstring(testxml)
87f70ab3
S		 317 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
318 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
319 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
320 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
321 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
322 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
323 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
324
9d4660ca 325 def test_smuggle_url(self):
e075a44a 326 data = {"ö": "ö", "abc": [3]}
9d4660ca
PH		 327 url = 'https://foo.bar/baz?x=y#a'
328 smug_url = smuggle_url(url, data)
329 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
330 self.assertEqual(url, unsmug_url)
331 self.assertEqual(data, unsmug_data)
332
333 res_url, res_data = unsmuggle_url(url)
334 self.assertEqual(res_url, url)
335 self.assertEqual(res_data, None)
336
a6a173c2 337 def test_shell_quote(self):
4e408e47
PH		 338 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
339 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
a6a173c2 340
f53c966a
JMF		 341 def test_str_to_int(self):
342 self.assertEqual(str_to_int('123,456'), 123456)
343 self.assertEqual(str_to_int('123.456'), 123456)
344
29eb5174 345 def test_url_basename(self):
4e408e47
PH		 346 self.assertEqual(url_basename('http://foo.de/'), '')
347 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
348 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
349 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
350 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
d6c7a367 351 self.assertEqual(
4e408e47
PH		 352 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
353 'trailer.mp4')
9d4660ca 354
608d11f5
PH		 355 def test_parse_duration(self):
356 self.assertEqual(parse_duration(None), None)
a5fb718c
S		 357 self.assertEqual(parse_duration(False), None)
358 self.assertEqual(parse_duration('invalid'), None)
608d11f5
PH		 359 self.assertEqual(parse_duration('1'), 1)
360 self.assertEqual(parse_duration('1337:12'), 80232)
361 self.assertEqual(parse_duration('9:12:43'), 33163)
2db806b4
S		 362 self.assertEqual(parse_duration('12:00'), 720)
363 self.assertEqual(parse_duration('00:01:01'), 61)
608d11f5 364 self.assertEqual(parse_duration('x:y'), None)
2db806b4 365 self.assertEqual(parse_duration('3h11m53s'), 11513)
ca7b3246
S		 366 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
367 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
368 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
2db806b4
S		 369 self.assertEqual(parse_duration('62m45s'), 3765)
370 self.assertEqual(parse_duration('6m59s'), 419)
371 self.assertEqual(parse_duration('49s'), 49)
372 self.assertEqual(parse_duration('0h0m0s'), 0)
373 self.assertEqual(parse_duration('0m0s'), 0)
374 self.assertEqual(parse_duration('0s'), 0)
7adcbe75 375 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
6a68bb57 376 self.assertEqual(parse_duration('T30M38S'), 1838)
e8df5cee
PH		 377 self.assertEqual(parse_duration('5 s'), 5)
378 self.assertEqual(parse_duration('3 min'), 180)
379 self.assertEqual(parse_duration('2.5 hours'), 9000)
8f4b58d7
PH		 380 self.assertEqual(parse_duration('02:03:04'), 7384)
381 self.assertEqual(parse_duration('01:02:03:04'), 93784)
3e675fab 382 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
9c29bc69 383 self.assertEqual(parse_duration('87 Min.'), 5220)
608d11f5 384
5aafe895
PH		 385 def test_fix_xml_ampersands(self):
386 self.assertEqual(
387 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
388 self.assertEqual(
389 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
390 '"&amp;x=y&amp;wrong;&amp;z=a')
391 self.assertEqual(
392 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
393 '&amp;&apos;&gt;&lt;&quot;')
394 self.assertEqual(
395 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
396 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
397
b7ab0590
PH		 398 def test_paged_list(self):
399 def testPL(size, pagesize, sliceargs, expected):
400 def get_page(pagenum):
401 firstid = pagenum * pagesize
402 upto = min(size, pagenum * pagesize + pagesize)
403 for i in range(firstid, upto):
404 yield i
405
9c44d242 406 pl = OnDemandPagedList(get_page, pagesize)
b7ab0590
PH		 407 got = pl.getslice(*sliceargs)
408 self.assertEqual(got, expected)
409
9c44d242
PH		 410 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
411 got = iapl.getslice(*sliceargs)
412 self.assertEqual(got, expected)
413
b7ab0590
PH		 414 testPL(5, 2, (), [0, 1, 2, 3, 4])
415 testPL(5, 2, (1,), [1, 2, 3, 4])
416 testPL(5, 2, (2,), [2, 3, 4])
417 testPL(5, 2, (4,), [4])
418 testPL(5, 2, (0, 3), [0, 1, 2])
419 testPL(5, 2, (1, 4), [1, 2, 3])
420 testPL(5, 2, (2, 99), [2, 3, 4])
421 testPL(5, 2, (20, 99), [])
422
b53466e1 423 def test_struct_unpack(self):
4e408e47 424 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
b53466e1 425
62e609ab 426 def test_read_batch_urls(self):
4e408e47 427 f = io.StringIO('''\xef\xbb\xbf foo
62e609ab
PH		 428 bar\r
429 baz
430 # More after this line\r
431 ; or after this
432 bam''')
4e408e47 433 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
62e609ab 434
b74fa8cd
JMF		 435 def test_urlencode_postdata(self):
436 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
437 self.assertTrue(isinstance(data, bytes))
438
912b38b4
PH		 439 def test_parse_iso8601(self):
440 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
441 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
442 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
6ad4013d 443 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
52c3a6e4
S		 444 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
445 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
912b38b4 446
fac55558
PH		 447 def test_strip_jsonp(self):
448 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
449 d = json.loads(stripped)
450 self.assertEqual(d, [{"id": "532cb", "x": 3}])
451
609a61e3
PH		 452 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
453 d = json.loads(stripped)
454 self.assertEqual(d, {'STATUS': 'OK'})
455
173a7026 456 def test_uppercase_escape(self):
4e408e47
PH		 457 self.assertEqual(uppercase_escape('aä'), 'aä')
458 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
fac55558 459
0fe2ff78
YCH		 460 def test_lowercase_escape(self):
461 self.assertEqual(lowercase_escape('aä'), 'aä')
462 self.assertEqual(lowercase_escape('\\u0026'), '&')
463
a020a0dc
PH		 464 def test_limit_length(self):
465 self.assertEqual(limit_length(None, 12), None)
466 self.assertEqual(limit_length('foo', 12), 'foo')
467 self.assertTrue(
468 limit_length('foo bar baz asd', 12).startswith('foo bar'))
469 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
470
d05cfe06
S		 471 def test_escape_rfc3986(self):
472 reserved = "!*'();:@&=+$,/?#[]"
473 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
474 self.assertEqual(escape_rfc3986(reserved), reserved)
475 self.assertEqual(escape_rfc3986(unreserved), unreserved)
476 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
477 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
478 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
479 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
480
481 def test_escape_url(self):
482 self.assertEqual(
483 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
484 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
485 )
486 self.assertEqual(
487 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
488 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
489 )
490 self.assertEqual(
491 escape_url('http://тест.рф/фрагмент'),
492 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
493 )
494 self.assertEqual(
495 escape_url('http://тест.рф/абв?абв=абв#абв'),
496 'http://тест.рф/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
497 )
498 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
499
e7b6d122 500 def test_js_to_json_realworld(self):
410f3e73 501 inp = '''{
e7b6d122 502 'clip':{'provider':'pseudo'}
410f3e73
PH		 503 }'''
504 self.assertEqual(js_to_json(inp), '''{
e7b6d122 505 "clip":{"provider":"pseudo"}
410f3e73
PH		 506 }''')
507 json.loads(js_to_json(inp))
508
e7b6d122
PH		 509 inp = '''{
510 'playlist':[{'controls':{'all':null}}]
511 }'''
512 self.assertEqual(js_to_json(inp), '''{
513 "playlist":[{"controls":{"all":null}}]
514 }''')
515
d01949dc
S		 516 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
517 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
518
d305dd73
PH		 519 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
520 json_code = js_to_json(inp)
521 self.assertEqual(json.loads(json_code), json.loads(inp))
522
e7b6d122
PH		 523 def test_js_to_json_edgecases(self):
524 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
525 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
526
527 on = js_to_json('{"abc": true}')
528 self.assertEqual(json.loads(on), {'abc': True})
529
8f4b58d7
PH		 530 # Ignore JavaScript code as well
531 on = js_to_json('''{
532 "x": 1,
533 y: "a",
534 z: some.code
535 }''')
536 d = json.loads(on)
537 self.assertEqual(d['x'], 1)
538 self.assertEqual(d['y'], 'a')
539
ba9e68f4
S		 540 on = js_to_json('["abc", "def",]')
541 self.assertEqual(json.loads(on), ['abc', 'def'])
542
543 on = js_to_json('{"abc": "def",}')
544 self.assertEqual(json.loads(on), {'abc': 'def'})
545
e4bdb37e
PH		 546 def test_clean_html(self):
547 self.assertEqual(clean_html('a:\nb'), 'a: b')
548 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
549
4c0924bb
PH		 550 def test_intlist_to_bytes(self):
551 self.assertEqual(
552 intlist_to_bytes([0, 1, 127, 128, 255]),
553 b'\x00\x01\x7f\x80\xff')
554
7d4111ed
PH		 555 def test_args_to_str(self):
556 self.assertEqual(
557 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
558 'foo ba/r -baz \'2 be\' \'\''
559 )
560
be64b5b0
PH		 561 def test_parse_filesize(self):
562 self.assertEqual(parse_filesize(None), None)
563 self.assertEqual(parse_filesize(''), None)
564 self.assertEqual(parse_filesize('91 B'), 91)
565 self.assertEqual(parse_filesize('foobar'), None)
566 self.assertEqual(parse_filesize('2 MiB'), 2097152)
567 self.assertEqual(parse_filesize('5 GB'), 5000000000)
568 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
4349c07d 569 self.assertEqual(parse_filesize('1,24 KB'), 1240)
be64b5b0 570
5f9b8394
PH		 571 def test_version_tuple(self):
572 self.assertEqual(version_tuple('1'), (1,))
573 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
47d7c642 574 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
5f9b8394 575
cae97f65
PH		 576 def test_detect_exe_version(self):
577 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
578built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
579configuration: --prefix=/usr --extra-'''), '1.2.1')
580 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
581built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
582 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
583Trying to open render node...
584Success at /dev/dri/renderD128.
585ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
586
05900629
PH		 587 def test_age_restricted(self):
588 self.assertFalse(age_restricted(None, 10)) # unrestricted content
589 self.assertFalse(age_restricted(1, None)) # unrestricted policy
590 self.assertFalse(age_restricted(8, 10))
591 self.assertTrue(age_restricted(18, 14))
592 self.assertFalse(age_restricted(18, 18))
593
61ca9a80
PH		 594 def test_is_html(self):
595 self.assertFalse(is_html(b'\x49\x44\x43<html'))
596 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
597 self.assertTrue(is_html( # UTF-8 with BOM
598 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
599 self.assertTrue(is_html( # UTF-16-LE
600 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
601 ))
602 self.assertTrue(is_html( # UTF-16-BE
603 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
604 ))
605 self.assertTrue(is_html( # UTF-32-BE
606 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'))
607 self.assertTrue(is_html( # UTF-32-LE
608 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'))
609
cfb56d1a
PH		 610 def test_render_table(self):
611 self.assertEqual(
612 render_table(
613 ['a', 'bcd'],
614 [[123, 4], [9999, 51]]),
615 'a bcd\n'
616 '123 4\n'
617 '9999 51')
618
347de493
PH		 619 def test_match_str(self):
620 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
621 self.assertFalse(match_str('xy', {'x': 1200}))
622 self.assertTrue(match_str('!xy', {'x': 1200}))
623 self.assertTrue(match_str('x', {'x': 1200}))
624 self.assertFalse(match_str('!x', {'x': 1200}))
625 self.assertTrue(match_str('x', {'x': 0}))
626 self.assertFalse(match_str('x>0', {'x': 0}))
627 self.assertFalse(match_str('x>0', {}))
628 self.assertTrue(match_str('x>?0', {}))
629 self.assertTrue(match_str('x>1K', {'x': 1200}))
630 self.assertFalse(match_str('x>2K', {'x': 1200}))
631 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
632 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
633 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
634 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
635 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
636 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
637 self.assertFalse(match_str(
638 'like_count > 100 & dislike_count <? 50 & description',
639 {'like_count': 90, 'description': 'foo'}))
640 self.assertTrue(match_str(
641 'like_count > 100 & dislike_count <? 50 & description',
642 {'like_count': 190, 'description': 'foo'}))
643 self.assertFalse(match_str(
644 'like_count > 100 & dislike_count <? 50 & description',
645 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
646 self.assertFalse(match_str(
647 'like_count > 100 & dislike_count <? 50 & description',
648 {'like_count': 190, 'dislike_count': 10}))
649
bf6427d2
YCH		 650 def test_parse_dfxp_time_expr(self):
651 self.assertEqual(parse_dfxp_time_expr(None), 0.0)
652 self.assertEqual(parse_dfxp_time_expr(''), 0.0)
653 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
654 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
655 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
656 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
657
658 def test_dfxp2srt(self):
659 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
660 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
661 <body>
662 <div xml:lang="en">
663 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
664 <p begin="1" end="2">第二行<br/>♪♪</p>
7dff0363 665 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
bf6427d2
YCH		 666 </div>
667 </body>
668 </tt>'''
669 srt_data = '''1
67000:00:00,000 --> 00:00:01,000
671The following line contains Chinese characters and special symbols
672
6732
67400:00:01,000 --> 00:00:02,000
675第二行
676♪♪
677
6783
67900:00:02,000 --> 00:00:03,000
680Third
681Line
682
683'''
684 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
685
1b0427e6
YCH		 686 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
687 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
688 <body>
689 <div xml:lang="en">
690 <p begin="0" end="1">The first line</p>
691 </div>
692 </body>
693 </tt>'''
694 srt_data = '''1
69500:00:00,000 --> 00:00:01,000
696The first line
697
698'''
699 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
700
f7126449
S		 701 def test_cli_option(self):
702 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
703 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
704 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
705
706 def test_cli_valueless_option(self):
707 self.assertEqual(cli_valueless_option(
708 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
709 self.assertEqual(cli_valueless_option(
710 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
711 self.assertEqual(cli_valueless_option(
712 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
713 self.assertEqual(cli_valueless_option(
714 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
715 self.assertEqual(cli_valueless_option(
716 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
717 self.assertEqual(cli_valueless_option(
718 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
719
720 def test_cli_bool_option(self):
721 self.assertEqual(
722 cli_bool_option(
723 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
724 ['--no-check-certificate', 'true'])
725 self.assertEqual(
726 cli_bool_option(
727 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
728 ['--no-check-certificate=true'])
729 self.assertEqual(
730 cli_bool_option(
731 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
732 ['--check-certificate', 'false'])
733 self.assertEqual(
734 cli_bool_option(
735 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
736 ['--check-certificate=false'])
737 self.assertEqual(
738 cli_bool_option(
739 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
740 ['--check-certificate', 'true'])
741 self.assertEqual(
742 cli_bool_option(
743 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
744 ['--check-certificate=true'])
745
cfb56d1a 746
dae7c920 747if __name__ == '__main__':
59ae15a5 748 unittest.main()
fork of https://github.com/yt-dlp/yt-dlp/