]> jfr.im git - yt-dlp.git/blame - test/test_utils.py
[videofy.me] fix info extraction
[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,
29eb5174 24 encodeFilename,
cae97f65
PH
25 escape_rfc3986,
26 escape_url,
5379a2d4 27 ExtractorError,
a921f407 28 find_xpath_attr,
5aafe895 29 fix_xml_ampersands,
9c44d242 30 InAdvancePagedList,
cae97f65 31 intlist_to_bytes,
61ca9a80 32 is_html,
cae97f65
PH
33 js_to_json,
34 limit_length,
35 OnDemandPagedList,
36 orderedSet,
608d11f5 37 parse_duration,
cae97f65
PH
38 parse_filesize,
39 parse_iso8601,
62e609ab 40 read_batch_urls,
29eb5174 41 sanitize_filename,
a2aaf4db 42 sanitize_path,
a4bcaad7 43 prepend_extension,
b3ed15b7 44 replace_extension,
a6a173c2 45 shell_quote,
29eb5174 46 smuggle_url,
f53c966a 47 str_to_int,
cae97f65 48 strip_jsonp,
b53466e1 49 struct_unpack,
29eb5174
PH
50 timeconvert,
51 unescapeHTML,
52 unified_strdate,
53 unsmuggle_url,
cae97f65 54 uppercase_escape,
0fe2ff78 55 lowercase_escape,
29eb5174 56 url_basename,
b74fa8cd 57 urlencode_postdata,
5f9b8394 58 version_tuple,
cae97f65 59 xpath_with_ns,
87f70ab3 60 xpath_element,
5379a2d4 61 xpath_text,
87f70ab3 62 xpath_attr,
cfb56d1a 63 render_table,
347de493 64 match_str,
bf6427d2
YCH
65 parse_dfxp_time_expr,
66 dfxp2srt,
f7126449
S
67 cli_option,
68 cli_valueless_option,
69 cli_bool_option,
a921f407 70)
36e6f62c
JMF
71from youtube_dl.compat import (
72 compat_etree_fromstring,
73)
44fb3454 74
627dcfff 75
44fb3454 76class TestUtil(unittest.TestCase):
59ae15a5
PH
77 def test_timeconvert(self):
78 self.assertTrue(timeconvert('') is None)
79 self.assertTrue(timeconvert('bougrg') is None)
80
81 def test_sanitize_filename(self):
82 self.assertEqual(sanitize_filename('abc'), 'abc')
83 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
84
85 self.assertEqual(sanitize_filename('123'), '123')
86
87 self.assertEqual('abc_de', sanitize_filename('abc/de'))
88 self.assertFalse('/' in sanitize_filename('abc/de///'))
89
90 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
91 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
92 self.assertEqual('yes no', sanitize_filename('yes? no'))
93 self.assertEqual('this - that', sanitize_filename('this: that'))
94
95 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
4e408e47 96 aumlaut = 'ä'
59ae15a5 97 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
4e408e47 98 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
59ae15a5
PH
99 self.assertEqual(sanitize_filename(tests), tests)
100
2aeb06d6
PH
101 self.assertEqual(
102 sanitize_filename('New World record at 0:12:34'),
103 'New World record at 0_12_34')
a7440261 104
5a42414b
PH
105 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
106 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
a7440261
PH
107 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
108 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
2aeb06d6 109
59ae15a5
PH
110 forbidden = '"\0\\/'
111 for fc in forbidden:
112 for fbc in forbidden:
113 self.assertTrue(fbc not in sanitize_filename(fc))
114
115 def test_sanitize_filename_restricted(self):
116 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
117 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
118
119 self.assertEqual(sanitize_filename('123', restricted=True), '123')
120
121 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
122 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
123
124 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
125 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
126 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
127 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
128
4e408e47 129 tests = 'a\xe4b\u4e2d\u56fd\u7684c'
59ae15a5 130 self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
4e408e47 131 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
59ae15a5 132
627dcfff 133 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
59ae15a5
PH
134 for fc in forbidden:
135 for fbc in forbidden:
136 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
137
138 # Handle a common case more neatly
4e408e47
PH
139 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
140 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
59ae15a5
PH
141 # .. but make sure the file name is never empty
142 self.assertTrue(sanitize_filename('-', restricted=True) != '')
143 self.assertTrue(sanitize_filename(':', restricted=True) != '')
144
796173d0 145 def test_sanitize_ids(self):
314d506b
PH
146 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
147 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
148 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
796173d0 149
a2aaf4db
S
150 def test_sanitize_path(self):
151 if sys.platform != 'win32':
152 return
153
154 self.assertEqual(sanitize_path('abc'), 'abc')
155 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
156 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
157 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
158 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
159 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
160 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
161
162 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
163 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
164
165 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
166 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
167 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
168 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
169
f18ef2d1
S
170 self.assertEqual(
171 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
172 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
173
174 self.assertEqual(
175 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
176 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
177 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
178 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
179 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
180
2ebfeaca
S
181 self.assertEqual(sanitize_path('../abc'), '..\\abc')
182 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
183 self.assertEqual(sanitize_path('./abc'), 'abc')
184 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
185
a4bcaad7
S
186 def test_prepend_extension(self):
187 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
188 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
189 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
190 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
191 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
192 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
193
b3ed15b7
S
194 def test_replace_extension(self):
195 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
196 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
197 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
198 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
199 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
200 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
201
59ae15a5 202 def test_ordered_set(self):
627dcfff 203 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
59ae15a5
PH
204 self.assertEqual(orderedSet([]), [])
205 self.assertEqual(orderedSet([1]), [1])
5f6a1245 206 # keep the list ordered
627dcfff 207 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
59ae15a5
PH
208
209 def test_unescape_html(self):
4e408e47 210 self.assertEqual(unescapeHTML('%20;'), '%20;')
91757b0f
NJ
211 self.assertEqual(unescapeHTML('&#x2F;'), '/')
212 self.assertEqual(unescapeHTML('&#47;'), '/')
4e408e47
PH
213 self.assertEqual(
214 unescapeHTML('&eacute;'), 'é')
5f6a1245 215
bd558525 216 def test_daterange(self):
5f6a1245 217 _20century = DateRange("19000101", "20000101")
bd558525
JMF
218 self.assertFalse("17890714" in _20century)
219 _ac = DateRange("00010101")
220 self.assertTrue("19690721" in _ac)
221 _firstmilenium = DateRange(end="10000101")
222 self.assertTrue("07110427" in _firstmilenium)
37254abc 223
bf50b038
JMF
224 def test_unified_dates(self):
225 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
226 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
227 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
228 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
a69801e2 229 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
026fcc04 230 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
99b67fec 231 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
42bdd9d0
PH
232 self.assertEqual(
233 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
234 '20141126')
9bb8e0a3
PH
235 self.assertEqual(
236 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
237 '20150202')
8cf70de4 238 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
dae7c920 239
59ae56fa 240 def test_find_xpath_attr(self):
4e408e47 241 testxml = '''<root>
59ae56fa
PH
242 <node/>
243 <node x="a"/>
244 <node x="a" y="c" />
245 <node x="b" y="d" />
ee114368 246 <node x="" />
59ae56fa 247 </root>'''
36e6f62c 248 doc = compat_etree_fromstring(testxml)
59ae56fa 249
ee114368 250 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
59ae56fa 251 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
ee114368
S
252 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
253 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
254 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
59ae56fa 255 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
ee114368
S
256 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
257 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
59ae56fa 258 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
ee114368
S
259 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
260 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
59ae56fa 261
d7e66d39 262 def test_xpath_with_ns(self):
4e408e47 263 testxml = '''<root xmlns:media="http://example.com/">
d7e66d39
JMF
264 <media:song>
265 <media:author>The Author</media:author>
266 <url>http://server.com/download.mp3</url>
267 </media:song>
268 </root>'''
36e6f62c 269 doc = compat_etree_fromstring(testxml)
d7e66d39
JMF
270 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
271 self.assertTrue(find('media:song') is not None)
4e408e47
PH
272 self.assertEqual(find('media:song/media:author').text, 'The Author')
273 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
d7e66d39 274
87f70ab3
S
275 def test_xpath_element(self):
276 doc = xml.etree.ElementTree.Element('root')
277 div = xml.etree.ElementTree.SubElement(doc, 'div')
278 p = xml.etree.ElementTree.SubElement(div, 'p')
279 p.text = 'Foo'
280 self.assertEqual(xpath_element(doc, 'div/p'), p)
578c0745
S
281 self.assertEqual(xpath_element(doc, ['div/p']), p)
282 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
87f70ab3 283 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
578c0745 284 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
87f70ab3 285 self.assertTrue(xpath_element(doc, 'div/bar') is None)
578c0745
S
286 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
287 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
87f70ab3 288 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
578c0745
S
289 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
290 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
87f70ab3 291
5379a2d4
JMF
292 def test_xpath_text(self):
293 testxml = '''<root>
294 <div>
295 <p>Foo</p>
296 </div>
297 </root>'''
36e6f62c 298 doc = compat_etree_fromstring(testxml)
5379a2d4 299 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
87f70ab3 300 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
5379a2d4
JMF
301 self.assertTrue(xpath_text(doc, 'div/bar') is None)
302 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
303
87f70ab3
S
304 def test_xpath_attr(self):
305 testxml = '''<root>
306 <div>
307 <p x="a">Foo</p>
308 </div>
309 </root>'''
36e6f62c 310 doc = compat_etree_fromstring(testxml)
87f70ab3
S
311 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
312 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
313 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
314 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
315 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
316 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
317 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
318
9d4660ca 319 def test_smuggle_url(self):
e075a44a 320 data = {"ö": "ö", "abc": [3]}
9d4660ca
PH
321 url = 'https://foo.bar/baz?x=y#a'
322 smug_url = smuggle_url(url, data)
323 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
324 self.assertEqual(url, unsmug_url)
325 self.assertEqual(data, unsmug_data)
326
327 res_url, res_data = unsmuggle_url(url)
328 self.assertEqual(res_url, url)
329 self.assertEqual(res_data, None)
330
a6a173c2 331 def test_shell_quote(self):
4e408e47
PH
332 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
333 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
a6a173c2 334
f53c966a
JMF
335 def test_str_to_int(self):
336 self.assertEqual(str_to_int('123,456'), 123456)
337 self.assertEqual(str_to_int('123.456'), 123456)
338
29eb5174 339 def test_url_basename(self):
4e408e47
PH
340 self.assertEqual(url_basename('http://foo.de/'), '')
341 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
342 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
343 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
344 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
d6c7a367 345 self.assertEqual(
4e408e47
PH
346 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
347 'trailer.mp4')
9d4660ca 348
608d11f5
PH
349 def test_parse_duration(self):
350 self.assertEqual(parse_duration(None), None)
a5fb718c
S
351 self.assertEqual(parse_duration(False), None)
352 self.assertEqual(parse_duration('invalid'), None)
608d11f5
PH
353 self.assertEqual(parse_duration('1'), 1)
354 self.assertEqual(parse_duration('1337:12'), 80232)
355 self.assertEqual(parse_duration('9:12:43'), 33163)
2db806b4
S
356 self.assertEqual(parse_duration('12:00'), 720)
357 self.assertEqual(parse_duration('00:01:01'), 61)
608d11f5 358 self.assertEqual(parse_duration('x:y'), None)
2db806b4 359 self.assertEqual(parse_duration('3h11m53s'), 11513)
ca7b3246
S
360 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
361 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
362 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
2db806b4
S
363 self.assertEqual(parse_duration('62m45s'), 3765)
364 self.assertEqual(parse_duration('6m59s'), 419)
365 self.assertEqual(parse_duration('49s'), 49)
366 self.assertEqual(parse_duration('0h0m0s'), 0)
367 self.assertEqual(parse_duration('0m0s'), 0)
368 self.assertEqual(parse_duration('0s'), 0)
7adcbe75 369 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
6a68bb57 370 self.assertEqual(parse_duration('T30M38S'), 1838)
e8df5cee
PH
371 self.assertEqual(parse_duration('5 s'), 5)
372 self.assertEqual(parse_duration('3 min'), 180)
373 self.assertEqual(parse_duration('2.5 hours'), 9000)
8f4b58d7
PH
374 self.assertEqual(parse_duration('02:03:04'), 7384)
375 self.assertEqual(parse_duration('01:02:03:04'), 93784)
3e675fab 376 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
9c29bc69 377 self.assertEqual(parse_duration('87 Min.'), 5220)
608d11f5 378
5aafe895
PH
379 def test_fix_xml_ampersands(self):
380 self.assertEqual(
381 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
382 self.assertEqual(
383 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
384 '"&amp;x=y&amp;wrong;&amp;z=a')
385 self.assertEqual(
386 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
387 '&amp;&apos;&gt;&lt;&quot;')
388 self.assertEqual(
389 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
390 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
391
b7ab0590
PH
392 def test_paged_list(self):
393 def testPL(size, pagesize, sliceargs, expected):
394 def get_page(pagenum):
395 firstid = pagenum * pagesize
396 upto = min(size, pagenum * pagesize + pagesize)
397 for i in range(firstid, upto):
398 yield i
399
9c44d242 400 pl = OnDemandPagedList(get_page, pagesize)
b7ab0590
PH
401 got = pl.getslice(*sliceargs)
402 self.assertEqual(got, expected)
403
9c44d242
PH
404 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
405 got = iapl.getslice(*sliceargs)
406 self.assertEqual(got, expected)
407
b7ab0590
PH
408 testPL(5, 2, (), [0, 1, 2, 3, 4])
409 testPL(5, 2, (1,), [1, 2, 3, 4])
410 testPL(5, 2, (2,), [2, 3, 4])
411 testPL(5, 2, (4,), [4])
412 testPL(5, 2, (0, 3), [0, 1, 2])
413 testPL(5, 2, (1, 4), [1, 2, 3])
414 testPL(5, 2, (2, 99), [2, 3, 4])
415 testPL(5, 2, (20, 99), [])
416
b53466e1 417 def test_struct_unpack(self):
4e408e47 418 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
b53466e1 419
62e609ab 420 def test_read_batch_urls(self):
4e408e47 421 f = io.StringIO('''\xef\xbb\xbf foo
62e609ab
PH
422 bar\r
423 baz
424 # More after this line\r
425 ; or after this
426 bam''')
4e408e47 427 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
62e609ab 428
b74fa8cd
JMF
429 def test_urlencode_postdata(self):
430 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
431 self.assertTrue(isinstance(data, bytes))
432
912b38b4
PH
433 def test_parse_iso8601(self):
434 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
435 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
436 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
6ad4013d 437 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
52c3a6e4
S
438 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
439 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
912b38b4 440
fac55558
PH
441 def test_strip_jsonp(self):
442 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
443 d = json.loads(stripped)
444 self.assertEqual(d, [{"id": "532cb", "x": 3}])
445
609a61e3
PH
446 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
447 d = json.loads(stripped)
448 self.assertEqual(d, {'STATUS': 'OK'})
449
173a7026 450 def test_uppercase_escape(self):
4e408e47
PH
451 self.assertEqual(uppercase_escape('aä'), 'aä')
452 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
fac55558 453
0fe2ff78
YCH
454 def test_lowercase_escape(self):
455 self.assertEqual(lowercase_escape('aä'), 'aä')
456 self.assertEqual(lowercase_escape('\\u0026'), '&')
457
a020a0dc
PH
458 def test_limit_length(self):
459 self.assertEqual(limit_length(None, 12), None)
460 self.assertEqual(limit_length('foo', 12), 'foo')
461 self.assertTrue(
462 limit_length('foo bar baz asd', 12).startswith('foo bar'))
463 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
464
d05cfe06
S
465 def test_escape_rfc3986(self):
466 reserved = "!*'();:@&=+$,/?#[]"
467 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
468 self.assertEqual(escape_rfc3986(reserved), reserved)
469 self.assertEqual(escape_rfc3986(unreserved), unreserved)
470 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
471 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
472 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
473 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
474
475 def test_escape_url(self):
476 self.assertEqual(
477 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
478 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
479 )
480 self.assertEqual(
481 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
482 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
483 )
484 self.assertEqual(
485 escape_url('http://тест.рф/фрагмент'),
486 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
487 )
488 self.assertEqual(
489 escape_url('http://тест.рф/абв?абв=абв#абв'),
490 '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'
491 )
492 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
493
e7b6d122 494 def test_js_to_json_realworld(self):
410f3e73 495 inp = '''{
e7b6d122 496 'clip':{'provider':'pseudo'}
410f3e73
PH
497 }'''
498 self.assertEqual(js_to_json(inp), '''{
e7b6d122 499 "clip":{"provider":"pseudo"}
410f3e73
PH
500 }''')
501 json.loads(js_to_json(inp))
502
e7b6d122
PH
503 inp = '''{
504 'playlist':[{'controls':{'all':null}}]
505 }'''
506 self.assertEqual(js_to_json(inp), '''{
507 "playlist":[{"controls":{"all":null}}]
508 }''')
509
d01949dc
S
510 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
511 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
512
d305dd73
PH
513 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
514 json_code = js_to_json(inp)
515 self.assertEqual(json.loads(json_code), json.loads(inp))
516
e7b6d122
PH
517 def test_js_to_json_edgecases(self):
518 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
519 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
520
521 on = js_to_json('{"abc": true}')
522 self.assertEqual(json.loads(on), {'abc': True})
523
8f4b58d7
PH
524 # Ignore JavaScript code as well
525 on = js_to_json('''{
526 "x": 1,
527 y: "a",
528 z: some.code
529 }''')
530 d = json.loads(on)
531 self.assertEqual(d['x'], 1)
532 self.assertEqual(d['y'], 'a')
533
ba9e68f4
S
534 on = js_to_json('["abc", "def",]')
535 self.assertEqual(json.loads(on), ['abc', 'def'])
536
537 on = js_to_json('{"abc": "def",}')
538 self.assertEqual(json.loads(on), {'abc': 'def'})
539
e4bdb37e
PH
540 def test_clean_html(self):
541 self.assertEqual(clean_html('a:\nb'), 'a: b')
542 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
543
4c0924bb
PH
544 def test_intlist_to_bytes(self):
545 self.assertEqual(
546 intlist_to_bytes([0, 1, 127, 128, 255]),
547 b'\x00\x01\x7f\x80\xff')
548
7d4111ed
PH
549 def test_args_to_str(self):
550 self.assertEqual(
551 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
552 'foo ba/r -baz \'2 be\' \'\''
553 )
554
be64b5b0
PH
555 def test_parse_filesize(self):
556 self.assertEqual(parse_filesize(None), None)
557 self.assertEqual(parse_filesize(''), None)
558 self.assertEqual(parse_filesize('91 B'), 91)
559 self.assertEqual(parse_filesize('foobar'), None)
560 self.assertEqual(parse_filesize('2 MiB'), 2097152)
561 self.assertEqual(parse_filesize('5 GB'), 5000000000)
562 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
4349c07d 563 self.assertEqual(parse_filesize('1,24 KB'), 1240)
be64b5b0 564
5f9b8394
PH
565 def test_version_tuple(self):
566 self.assertEqual(version_tuple('1'), (1,))
567 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
47d7c642 568 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
5f9b8394 569
cae97f65
PH
570 def test_detect_exe_version(self):
571 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
572built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
573configuration: --prefix=/usr --extra-'''), '1.2.1')
574 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
575built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
576 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
577Trying to open render node...
578Success at /dev/dri/renderD128.
579ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
580
05900629
PH
581 def test_age_restricted(self):
582 self.assertFalse(age_restricted(None, 10)) # unrestricted content
583 self.assertFalse(age_restricted(1, None)) # unrestricted policy
584 self.assertFalse(age_restricted(8, 10))
585 self.assertTrue(age_restricted(18, 14))
586 self.assertFalse(age_restricted(18, 18))
587
61ca9a80
PH
588 def test_is_html(self):
589 self.assertFalse(is_html(b'\x49\x44\x43<html'))
590 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
591 self.assertTrue(is_html( # UTF-8 with BOM
592 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
593 self.assertTrue(is_html( # UTF-16-LE
594 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
595 ))
596 self.assertTrue(is_html( # UTF-16-BE
597 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
598 ))
599 self.assertTrue(is_html( # UTF-32-BE
600 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'))
601 self.assertTrue(is_html( # UTF-32-LE
602 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'))
603
cfb56d1a
PH
604 def test_render_table(self):
605 self.assertEqual(
606 render_table(
607 ['a', 'bcd'],
608 [[123, 4], [9999, 51]]),
609 'a bcd\n'
610 '123 4\n'
611 '9999 51')
612
347de493
PH
613 def test_match_str(self):
614 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
615 self.assertFalse(match_str('xy', {'x': 1200}))
616 self.assertTrue(match_str('!xy', {'x': 1200}))
617 self.assertTrue(match_str('x', {'x': 1200}))
618 self.assertFalse(match_str('!x', {'x': 1200}))
619 self.assertTrue(match_str('x', {'x': 0}))
620 self.assertFalse(match_str('x>0', {'x': 0}))
621 self.assertFalse(match_str('x>0', {}))
622 self.assertTrue(match_str('x>?0', {}))
623 self.assertTrue(match_str('x>1K', {'x': 1200}))
624 self.assertFalse(match_str('x>2K', {'x': 1200}))
625 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
626 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
627 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
628 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
629 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
630 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
631 self.assertFalse(match_str(
632 'like_count > 100 & dislike_count <? 50 & description',
633 {'like_count': 90, 'description': 'foo'}))
634 self.assertTrue(match_str(
635 'like_count > 100 & dislike_count <? 50 & description',
636 {'like_count': 190, 'description': 'foo'}))
637 self.assertFalse(match_str(
638 'like_count > 100 & dislike_count <? 50 & description',
639 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
640 self.assertFalse(match_str(
641 'like_count > 100 & dislike_count <? 50 & description',
642 {'like_count': 190, 'dislike_count': 10}))
643
bf6427d2
YCH
644 def test_parse_dfxp_time_expr(self):
645 self.assertEqual(parse_dfxp_time_expr(None), 0.0)
646 self.assertEqual(parse_dfxp_time_expr(''), 0.0)
647 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
648 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
649 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
650 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
651
652 def test_dfxp2srt(self):
653 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
654 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
655 <body>
656 <div xml:lang="en">
657 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
658 <p begin="1" end="2">第二行<br/>♪♪</p>
7dff0363 659 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
bf6427d2
YCH
660 </div>
661 </body>
662 </tt>'''
663 srt_data = '''1
66400:00:00,000 --> 00:00:01,000
665The following line contains Chinese characters and special symbols
666
6672
66800:00:01,000 --> 00:00:02,000
669第二行
670♪♪
671
6723
67300:00:02,000 --> 00:00:03,000
674Third
675Line
676
677'''
678 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
679
1b0427e6
YCH
680 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
681 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
682 <body>
683 <div xml:lang="en">
684 <p begin="0" end="1">The first line</p>
685 </div>
686 </body>
687 </tt>'''
688 srt_data = '''1
68900:00:00,000 --> 00:00:01,000
690The first line
691
692'''
693 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
694
f7126449
S
695 def test_cli_option(self):
696 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
697 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
698 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
699
700 def test_cli_valueless_option(self):
701 self.assertEqual(cli_valueless_option(
702 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
703 self.assertEqual(cli_valueless_option(
704 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
705 self.assertEqual(cli_valueless_option(
706 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
707 self.assertEqual(cli_valueless_option(
708 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
709 self.assertEqual(cli_valueless_option(
710 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
711 self.assertEqual(cli_valueless_option(
712 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
713
714 def test_cli_bool_option(self):
715 self.assertEqual(
716 cli_bool_option(
717 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
718 ['--no-check-certificate', 'true'])
719 self.assertEqual(
720 cli_bool_option(
721 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
722 ['--no-check-certificate=true'])
723 self.assertEqual(
724 cli_bool_option(
725 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
726 ['--check-certificate', 'false'])
727 self.assertEqual(
728 cli_bool_option(
729 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
730 ['--check-certificate=false'])
731 self.assertEqual(
732 cli_bool_option(
733 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
734 ['--check-certificate', 'true'])
735 self.assertEqual(
736 cli_bool_option(
737 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
738 ['--check-certificate=true'])
739
cfb56d1a 740
dae7c920 741if __name__ == '__main__':
59ae15a5 742 unittest.main()