]> 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
Merge pull request #7242 from lalinsky/vimeo-seed-status
[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)
44fb3454 71
627dcfff 72
44fb3454 73class TestUtil(unittest.TestCase):
59ae15a5
PH		 74 def test_timeconvert(self):
75 self.assertTrue(timeconvert('') is None)
76 self.assertTrue(timeconvert('bougrg') is None)
77
78 def test_sanitize_filename(self):
79 self.assertEqual(sanitize_filename('abc'), 'abc')
80 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
81
82 self.assertEqual(sanitize_filename('123'), '123')
83
84 self.assertEqual('abc_de', sanitize_filename('abc/de'))
85 self.assertFalse('/' in sanitize_filename('abc/de///'))
86
87 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
88 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
89 self.assertEqual('yes no', sanitize_filename('yes? no'))
90 self.assertEqual('this - that', sanitize_filename('this: that'))
91
92 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
4e408e47 93 aumlaut = 'ä'
59ae15a5 94 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
4e408e47 95 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
59ae15a5
PH		 96 self.assertEqual(sanitize_filename(tests), tests)
97
2aeb06d6
PH		 98 self.assertEqual(
99 sanitize_filename('New World record at 0:12:34'),
100 'New World record at 0_12_34')
a7440261 101
5a42414b
PH		 102 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
103 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
a7440261
PH		 104 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
105 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
2aeb06d6 106
59ae15a5
PH		 107 forbidden = '"\0\\/'
108 for fc in forbidden:
109 for fbc in forbidden:
110 self.assertTrue(fbc not in sanitize_filename(fc))
111
112 def test_sanitize_filename_restricted(self):
113 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
114 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
115
116 self.assertEqual(sanitize_filename('123', restricted=True), '123')
117
118 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
119 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
120
121 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
122 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
123 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
124 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
125
4e408e47 126 tests = 'a\xe4b\u4e2d\u56fd\u7684c'
59ae15a5 127 self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
4e408e47 128 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
59ae15a5 129
627dcfff 130 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
59ae15a5
PH		 131 for fc in forbidden:
132 for fbc in forbidden:
133 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
134
135 # Handle a common case more neatly
4e408e47
PH		 136 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
137 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
59ae15a5
PH		 138 # .. but make sure the file name is never empty
139 self.assertTrue(sanitize_filename('-', restricted=True) != '')
140 self.assertTrue(sanitize_filename(':', restricted=True) != '')
141
796173d0 142 def test_sanitize_ids(self):
314d506b
PH		 143 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
144 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
145 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
796173d0 146
a2aaf4db
S		 147 def test_sanitize_path(self):
148 if sys.platform != 'win32':
149 return
150
151 self.assertEqual(sanitize_path('abc'), 'abc')
152 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
153 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
154 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
155 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
156 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
157 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
158
159 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
160 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
161
162 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
163 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
164 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
165 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
166
f18ef2d1
S		 167 self.assertEqual(
168 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
169 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
170
171 self.assertEqual(
172 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
173 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
174 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
175 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
176 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
177
2ebfeaca
S		 178 self.assertEqual(sanitize_path('../abc'), '..\\abc')
179 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
180 self.assertEqual(sanitize_path('./abc'), 'abc')
181 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
182
a4bcaad7
S		 183 def test_prepend_extension(self):
184 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
185 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
186 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
187 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
188 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
189 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
190
b3ed15b7
S		 191 def test_replace_extension(self):
192 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
193 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
194 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
195 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
196 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
197 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
198
59ae15a5 199 def test_ordered_set(self):
627dcfff 200 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
59ae15a5
PH		 201 self.assertEqual(orderedSet([]), [])
202 self.assertEqual(orderedSet([1]), [1])
5f6a1245 203 # keep the list ordered
627dcfff 204 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
59ae15a5
PH		 205
206 def test_unescape_html(self):
4e408e47 207 self.assertEqual(unescapeHTML('%20;'), '%20;')
91757b0f
NJ		 208 self.assertEqual(unescapeHTML('&#x2F;'), '/')
209 self.assertEqual(unescapeHTML('&#47;'), '/')
4e408e47
PH		 210 self.assertEqual(
211 unescapeHTML('&eacute;'), 'é')
5f6a1245 212
bd558525 213 def test_daterange(self):
5f6a1245 214 _20century = DateRange("19000101", "20000101")
bd558525
JMF		 215 self.assertFalse("17890714" in _20century)
216 _ac = DateRange("00010101")
217 self.assertTrue("19690721" in _ac)
218 _firstmilenium = DateRange(end="10000101")
219 self.assertTrue("07110427" in _firstmilenium)
37254abc 220
bf50b038
JMF		 221 def test_unified_dates(self):
222 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
223 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
224 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
225 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
a69801e2 226 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
026fcc04 227 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
99b67fec 228 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
42bdd9d0
PH		 229 self.assertEqual(
230 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
231 '20141126')
9bb8e0a3
PH		 232 self.assertEqual(
233 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
234 '20150202')
8cf70de4 235 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
dae7c920 236
59ae56fa 237 def test_find_xpath_attr(self):
4e408e47 238 testxml = '''<root>
59ae56fa
PH		 239 <node/>
240 <node x="a"/>
241 <node x="a" y="c" />
242 <node x="b" y="d" />
ee114368 243 <node x="" />
59ae56fa
PH		 244 </root>'''
245 doc = xml.etree.ElementTree.fromstring(testxml)
246
ee114368 247 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
59ae56fa 248 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
ee114368
S		 249 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
250 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
251 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
59ae56fa 252 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
ee114368
S		 253 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
254 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
59ae56fa 255 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
ee114368
S		 256 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
257 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
59ae56fa 258
d7e66d39 259 def test_xpath_with_ns(self):
4e408e47 260 testxml = '''<root xmlns:media="http://example.com/">
d7e66d39
JMF		 261 <media:song>
262 <media:author>The Author</media:author>
263 <url>http://server.com/download.mp3</url>
264 </media:song>
265 </root>'''
266 doc = xml.etree.ElementTree.fromstring(testxml)
267 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
268 self.assertTrue(find('media:song') is not None)
4e408e47
PH		 269 self.assertEqual(find('media:song/media:author').text, 'The Author')
270 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
d7e66d39 271
87f70ab3
S		 272 def test_xpath_element(self):
273 doc = xml.etree.ElementTree.Element('root')
274 div = xml.etree.ElementTree.SubElement(doc, 'div')
275 p = xml.etree.ElementTree.SubElement(div, 'p')
276 p.text = 'Foo'
277 self.assertEqual(xpath_element(doc, 'div/p'), p)
278 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
279 self.assertTrue(xpath_element(doc, 'div/bar') is None)
280 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
281
5379a2d4
JMF		 282 def test_xpath_text(self):
283 testxml = '''<root>
284 <div>
285 <p>Foo</p>
286 </div>
287 </root>'''
288 doc = xml.etree.ElementTree.fromstring(testxml)
289 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
87f70ab3 290 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
5379a2d4
JMF		 291 self.assertTrue(xpath_text(doc, 'div/bar') is None)
292 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
293
87f70ab3
S		 294 def test_xpath_attr(self):
295 testxml = '''<root>
296 <div>
297 <p x="a">Foo</p>
298 </div>
299 </root>'''
300 doc = xml.etree.ElementTree.fromstring(testxml)
301 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
302 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
303 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
304 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
305 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
306 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
307 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
308
9d4660ca 309 def test_smuggle_url(self):
e075a44a 310 data = {"ö": "ö", "abc": [3]}
9d4660ca
PH		 311 url = 'https://foo.bar/baz?x=y#a'
312 smug_url = smuggle_url(url, data)
313 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
314 self.assertEqual(url, unsmug_url)
315 self.assertEqual(data, unsmug_data)
316
317 res_url, res_data = unsmuggle_url(url)
318 self.assertEqual(res_url, url)
319 self.assertEqual(res_data, None)
320
a6a173c2 321 def test_shell_quote(self):
4e408e47
PH		 322 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
323 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
a6a173c2 324
f53c966a
JMF		 325 def test_str_to_int(self):
326 self.assertEqual(str_to_int('123,456'), 123456)
327 self.assertEqual(str_to_int('123.456'), 123456)
328
29eb5174 329 def test_url_basename(self):
4e408e47
PH		 330 self.assertEqual(url_basename('http://foo.de/'), '')
331 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
332 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
333 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
334 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
d6c7a367 335 self.assertEqual(
4e408e47
PH		 336 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
337 'trailer.mp4')
9d4660ca 338
608d11f5
PH		 339 def test_parse_duration(self):
340 self.assertEqual(parse_duration(None), None)
a5fb718c
S		 341 self.assertEqual(parse_duration(False), None)
342 self.assertEqual(parse_duration('invalid'), None)
608d11f5
PH		 343 self.assertEqual(parse_duration('1'), 1)
344 self.assertEqual(parse_duration('1337:12'), 80232)
345 self.assertEqual(parse_duration('9:12:43'), 33163)
2db806b4
S		 346 self.assertEqual(parse_duration('12:00'), 720)
347 self.assertEqual(parse_duration('00:01:01'), 61)
608d11f5 348 self.assertEqual(parse_duration('x:y'), None)
2db806b4 349 self.assertEqual(parse_duration('3h11m53s'), 11513)
ca7b3246
S		 350 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
351 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
352 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
2db806b4
S		 353 self.assertEqual(parse_duration('62m45s'), 3765)
354 self.assertEqual(parse_duration('6m59s'), 419)
355 self.assertEqual(parse_duration('49s'), 49)
356 self.assertEqual(parse_duration('0h0m0s'), 0)
357 self.assertEqual(parse_duration('0m0s'), 0)
358 self.assertEqual(parse_duration('0s'), 0)
7adcbe75 359 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
6a68bb57 360 self.assertEqual(parse_duration('T30M38S'), 1838)
e8df5cee
PH		 361 self.assertEqual(parse_duration('5 s'), 5)
362 self.assertEqual(parse_duration('3 min'), 180)
363 self.assertEqual(parse_duration('2.5 hours'), 9000)
8f4b58d7
PH		 364 self.assertEqual(parse_duration('02:03:04'), 7384)
365 self.assertEqual(parse_duration('01:02:03:04'), 93784)
3e675fab 366 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
9c29bc69 367 self.assertEqual(parse_duration('87 Min.'), 5220)
608d11f5 368
5aafe895
PH		 369 def test_fix_xml_ampersands(self):
370 self.assertEqual(
371 fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
372 self.assertEqual(
373 fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
374 '"&amp;x=y&amp;wrong;&amp;z=a')
375 self.assertEqual(
376 fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
377 '&amp;&apos;&gt;&lt;&quot;')
378 self.assertEqual(
379 fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
380 self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
381
b7ab0590
PH		 382 def test_paged_list(self):
383 def testPL(size, pagesize, sliceargs, expected):
384 def get_page(pagenum):
385 firstid = pagenum * pagesize
386 upto = min(size, pagenum * pagesize + pagesize)
387 for i in range(firstid, upto):
388 yield i
389
9c44d242 390 pl = OnDemandPagedList(get_page, pagesize)
b7ab0590
PH		 391 got = pl.getslice(*sliceargs)
392 self.assertEqual(got, expected)
393
9c44d242
PH		 394 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
395 got = iapl.getslice(*sliceargs)
396 self.assertEqual(got, expected)
397
b7ab0590
PH		 398 testPL(5, 2, (), [0, 1, 2, 3, 4])
399 testPL(5, 2, (1,), [1, 2, 3, 4])
400 testPL(5, 2, (2,), [2, 3, 4])
401 testPL(5, 2, (4,), [4])
402 testPL(5, 2, (0, 3), [0, 1, 2])
403 testPL(5, 2, (1, 4), [1, 2, 3])
404 testPL(5, 2, (2, 99), [2, 3, 4])
405 testPL(5, 2, (20, 99), [])
406
b53466e1 407 def test_struct_unpack(self):
4e408e47 408 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
b53466e1 409
62e609ab 410 def test_read_batch_urls(self):
4e408e47 411 f = io.StringIO('''\xef\xbb\xbf foo
62e609ab
PH		 412 bar\r
413 baz
414 # More after this line\r
415 ; or after this
416 bam''')
4e408e47 417 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
62e609ab 418
b74fa8cd
JMF		 419 def test_urlencode_postdata(self):
420 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
421 self.assertTrue(isinstance(data, bytes))
422
912b38b4
PH		 423 def test_parse_iso8601(self):
424 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
425 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
426 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
6ad4013d 427 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
912b38b4 428
fac55558
PH		 429 def test_strip_jsonp(self):
430 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
431 d = json.loads(stripped)
432 self.assertEqual(d, [{"id": "532cb", "x": 3}])
433
609a61e3
PH		 434 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
435 d = json.loads(stripped)
436 self.assertEqual(d, {'STATUS': 'OK'})
437
173a7026 438 def test_uppercase_escape(self):
4e408e47
PH		 439 self.assertEqual(uppercase_escape('aä'), 'aä')
440 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
fac55558 441
0fe2ff78
YCH		 442 def test_lowercase_escape(self):
443 self.assertEqual(lowercase_escape('aä'), 'aä')
444 self.assertEqual(lowercase_escape('\\u0026'), '&')
445
a020a0dc
PH		 446 def test_limit_length(self):
447 self.assertEqual(limit_length(None, 12), None)
448 self.assertEqual(limit_length('foo', 12), 'foo')
449 self.assertTrue(
450 limit_length('foo bar baz asd', 12).startswith('foo bar'))
451 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
452
d05cfe06
S		 453 def test_escape_rfc3986(self):
454 reserved = "!*'();:@&=+$,/?#[]"
455 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
456 self.assertEqual(escape_rfc3986(reserved), reserved)
457 self.assertEqual(escape_rfc3986(unreserved), unreserved)
458 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
459 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
460 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
461 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
462
463 def test_escape_url(self):
464 self.assertEqual(
465 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
466 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
467 )
468 self.assertEqual(
469 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
470 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
471 )
472 self.assertEqual(
473 escape_url('http://тест.рф/фрагмент'),
474 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
475 )
476 self.assertEqual(
477 escape_url('http://тест.рф/абв?абв=абв#абв'),
478 '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'
479 )
480 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
481
e7b6d122 482 def test_js_to_json_realworld(self):
410f3e73 483 inp = '''{
e7b6d122 484 'clip':{'provider':'pseudo'}
410f3e73
PH		 485 }'''
486 self.assertEqual(js_to_json(inp), '''{
e7b6d122 487 "clip":{"provider":"pseudo"}
410f3e73
PH		 488 }''')
489 json.loads(js_to_json(inp))
490
e7b6d122
PH		 491 inp = '''{
492 'playlist':[{'controls':{'all':null}}]
493 }'''
494 self.assertEqual(js_to_json(inp), '''{
495 "playlist":[{"controls":{"all":null}}]
496 }''')
497
d305dd73
PH		 498 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
499 json_code = js_to_json(inp)
500 self.assertEqual(json.loads(json_code), json.loads(inp))
501
e7b6d122
PH		 502 def test_js_to_json_edgecases(self):
503 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
504 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
505
506 on = js_to_json('{"abc": true}')
507 self.assertEqual(json.loads(on), {'abc': True})
508
8f4b58d7
PH		 509 # Ignore JavaScript code as well
510 on = js_to_json('''{
511 "x": 1,
512 y: "a",
513 z: some.code
514 }''')
515 d = json.loads(on)
516 self.assertEqual(d['x'], 1)
517 self.assertEqual(d['y'], 'a')
518
ba9e68f4
S		 519 on = js_to_json('["abc", "def",]')
520 self.assertEqual(json.loads(on), ['abc', 'def'])
521
522 on = js_to_json('{"abc": "def",}')
523 self.assertEqual(json.loads(on), {'abc': 'def'})
524
e4bdb37e
PH		 525 def test_clean_html(self):
526 self.assertEqual(clean_html('a:\nb'), 'a: b')
527 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
528
4c0924bb
PH		 529 def test_intlist_to_bytes(self):
530 self.assertEqual(
531 intlist_to_bytes([0, 1, 127, 128, 255]),
532 b'\x00\x01\x7f\x80\xff')
533
7d4111ed
PH		 534 def test_args_to_str(self):
535 self.assertEqual(
536 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
537 'foo ba/r -baz \'2 be\' \'\''
538 )
539
be64b5b0
PH		 540 def test_parse_filesize(self):
541 self.assertEqual(parse_filesize(None), None)
542 self.assertEqual(parse_filesize(''), None)
543 self.assertEqual(parse_filesize('91 B'), 91)
544 self.assertEqual(parse_filesize('foobar'), None)
545 self.assertEqual(parse_filesize('2 MiB'), 2097152)
546 self.assertEqual(parse_filesize('5 GB'), 5000000000)
547 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
4349c07d 548 self.assertEqual(parse_filesize('1,24 KB'), 1240)
be64b5b0 549
5f9b8394
PH		 550 def test_version_tuple(self):
551 self.assertEqual(version_tuple('1'), (1,))
552 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
47d7c642 553 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
5f9b8394 554
cae97f65
PH		 555 def test_detect_exe_version(self):
556 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
557built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
558configuration: --prefix=/usr --extra-'''), '1.2.1')
559 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
560built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
561 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
562Trying to open render node...
563Success at /dev/dri/renderD128.
564ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
565
05900629
PH		 566 def test_age_restricted(self):
567 self.assertFalse(age_restricted(None, 10)) # unrestricted content
568 self.assertFalse(age_restricted(1, None)) # unrestricted policy
569 self.assertFalse(age_restricted(8, 10))
570 self.assertTrue(age_restricted(18, 14))
571 self.assertFalse(age_restricted(18, 18))
572
61ca9a80
PH		 573 def test_is_html(self):
574 self.assertFalse(is_html(b'\x49\x44\x43<html'))
575 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
576 self.assertTrue(is_html( # UTF-8 with BOM
577 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
578 self.assertTrue(is_html( # UTF-16-LE
579 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
580 ))
581 self.assertTrue(is_html( # UTF-16-BE
582 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
583 ))
584 self.assertTrue(is_html( # UTF-32-BE
585 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'))
586 self.assertTrue(is_html( # UTF-32-LE
587 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'))
588
cfb56d1a
PH		 589 def test_render_table(self):
590 self.assertEqual(
591 render_table(
592 ['a', 'bcd'],
593 [[123, 4], [9999, 51]]),
594 'a bcd\n'
595 '123 4\n'
596 '9999 51')
597
347de493
PH		 598 def test_match_str(self):
599 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
600 self.assertFalse(match_str('xy', {'x': 1200}))
601 self.assertTrue(match_str('!xy', {'x': 1200}))
602 self.assertTrue(match_str('x', {'x': 1200}))
603 self.assertFalse(match_str('!x', {'x': 1200}))
604 self.assertTrue(match_str('x', {'x': 0}))
605 self.assertFalse(match_str('x>0', {'x': 0}))
606 self.assertFalse(match_str('x>0', {}))
607 self.assertTrue(match_str('x>?0', {}))
608 self.assertTrue(match_str('x>1K', {'x': 1200}))
609 self.assertFalse(match_str('x>2K', {'x': 1200}))
610 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
611 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
612 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
613 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
614 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
615 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
616 self.assertFalse(match_str(
617 'like_count > 100 & dislike_count <? 50 & description',
618 {'like_count': 90, 'description': 'foo'}))
619 self.assertTrue(match_str(
620 'like_count > 100 & dislike_count <? 50 & description',
621 {'like_count': 190, 'description': 'foo'}))
622 self.assertFalse(match_str(
623 'like_count > 100 & dislike_count <? 50 & description',
624 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
625 self.assertFalse(match_str(
626 'like_count > 100 & dislike_count <? 50 & description',
627 {'like_count': 190, 'dislike_count': 10}))
628
bf6427d2
YCH		 629 def test_parse_dfxp_time_expr(self):
630 self.assertEqual(parse_dfxp_time_expr(None), 0.0)
631 self.assertEqual(parse_dfxp_time_expr(''), 0.0)
632 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
633 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
634 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
635 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
636
637 def test_dfxp2srt(self):
638 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
639 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
640 <body>
641 <div xml:lang="en">
642 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
643 <p begin="1" end="2">第二行<br/>♪♪</p>
7dff0363 644 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
bf6427d2
YCH		 645 </div>
646 </body>
647 </tt>'''
648 srt_data = '''1
64900:00:00,000 --> 00:00:01,000
650The following line contains Chinese characters and special symbols
651
6522
65300:00:01,000 --> 00:00:02,000
654第二行
655♪♪
656
6573
65800:00:02,000 --> 00:00:03,000
659Third
660Line
661
662'''
663 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
664
1b0427e6
YCH		 665 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
666 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
667 <body>
668 <div xml:lang="en">
669 <p begin="0" end="1">The first line</p>
670 </div>
671 </body>
672 </tt>'''
673 srt_data = '''1
67400:00:00,000 --> 00:00:01,000
675The first line
676
677'''
678 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
679
f7126449
S		 680 def test_cli_option(self):
681 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
682 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
683 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
684
685 def test_cli_valueless_option(self):
686 self.assertEqual(cli_valueless_option(
687 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
688 self.assertEqual(cli_valueless_option(
689 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
690 self.assertEqual(cli_valueless_option(
691 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
692 self.assertEqual(cli_valueless_option(
693 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
694 self.assertEqual(cli_valueless_option(
695 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
696 self.assertEqual(cli_valueless_option(
697 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
698
699 def test_cli_bool_option(self):
700 self.assertEqual(
701 cli_bool_option(
702 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
703 ['--no-check-certificate', 'true'])
704 self.assertEqual(
705 cli_bool_option(
706 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
707 ['--no-check-certificate=true'])
708 self.assertEqual(
709 cli_bool_option(
710 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
711 ['--check-certificate', 'false'])
712 self.assertEqual(
713 cli_bool_option(
714 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
715 ['--check-certificate=false'])
716 self.assertEqual(
717 cli_bool_option(
718 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
719 ['--check-certificate', 'true'])
720 self.assertEqual(
721 cli_bool_option(
722 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
723 ['--check-certificate=true'])
724
cfb56d1a 725
dae7c920 726if __name__ == '__main__':
59ae15a5 727 unittest.main()
fork of https://github.com/yt-dlp/yt-dlp/