4 from __future__
import unicode_literals
6 # Allow direct execution
10 sys
.path
.insert(0, os
.path
.dirname(os
.path
.dirname(os
.path
.abspath(__file__
))))
13 # Various small unit tests
16 import xml
.etree
.ElementTree
18 from youtube_dl
.utils
import (
80 from youtube_dl
.compat
import (
82 compat_etree_fromstring
,
88 class TestUtil(unittest
.TestCase
):
89 def test_timeconvert(self
):
90 self
.assertTrue(timeconvert('') is None)
91 self
.assertTrue(timeconvert('bougrg') is None)
93 def test_sanitize_filename(self
):
94 self
.assertEqual(sanitize_filename('abc'), 'abc')
95 self
.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
97 self
.assertEqual(sanitize_filename('123'), '123')
99 self
.assertEqual('abc_de', sanitize_filename('abc/de'))
100 self
.assertFalse('/' in sanitize_filename('abc/de///'))
102 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
103 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
104 self
.assertEqual('yes no', sanitize_filename('yes? no'))
105 self
.assertEqual('this - that', sanitize_filename('this: that'))
107 self
.assertEqual(sanitize_filename('AT&T'), 'AT&T')
109 self
.assertEqual(sanitize_filename(aumlaut
), aumlaut
)
110 tests
= '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
111 self
.assertEqual(sanitize_filename(tests
), tests
)
114 sanitize_filename('New World record at 0:12:34'),
115 'New World record at 0_12_34')
117 self
.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
118 self
.assertEqual(sanitize_filename('--gasdgf', is_id
=True), '--gasdgf')
119 self
.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
120 self
.assertEqual(sanitize_filename('.gasdgf', is_id
=True), '.gasdgf')
124 for fbc
in forbidden
:
125 self
.assertTrue(fbc
not in sanitize_filename(fc
))
127 def test_sanitize_filename_restricted(self
):
128 self
.assertEqual(sanitize_filename('abc', restricted
=True), 'abc')
129 self
.assertEqual(sanitize_filename('abc_d-e', restricted
=True), 'abc_d-e')
131 self
.assertEqual(sanitize_filename('123', restricted
=True), '123')
133 self
.assertEqual('abc_de', sanitize_filename('abc/de', restricted
=True))
134 self
.assertFalse('/' in sanitize_filename('abc/de///', restricted
=True))
136 self
.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted
=True))
137 self
.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted
=True))
138 self
.assertEqual('yes_no', sanitize_filename('yes? no', restricted
=True))
139 self
.assertEqual('this_-_that', sanitize_filename('this: that', restricted
=True))
141 tests
= 'aäb\u4e2d\u56fd\u7684c'
142 self
.assertEqual(sanitize_filename(tests
, restricted
=True), 'aab_c')
143 self
.assertTrue(sanitize_filename('\xf6', restricted
=True) != '') # No empty filename
145 forbidden
= '"\0\\/&!: \'\t\n()[]{}$;`^,#'
147 for fbc
in forbidden
:
148 self
.assertTrue(fbc
not in sanitize_filename(fc
, restricted
=True))
150 # Handle a common case more neatly
151 self
.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted
=True), 'Song')
152 self
.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted
=True), 'Speech')
153 # .. but make sure the file name is never empty
154 self
.assertTrue(sanitize_filename('-', restricted
=True) != '')
155 self
.assertTrue(sanitize_filename(':', restricted
=True) != '')
157 self
.assertEqual(sanitize_filename(
158 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØŒÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøœùúûüýþÿ', restricted
=True),
159 'AAAAAAAECEEEEIIIIDNOOOOOOOEUUUUYPssaaaaaaaeceeeeiiiionoooooooeuuuuypy')
161 def test_sanitize_ids(self
):
162 self
.assertEqual(sanitize_filename('_n_cd26wFpw', is_id
=True), '_n_cd26wFpw')
163 self
.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id
=True), '_BD_eEpuzXw')
164 self
.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id
=True), 'N0Y__7-UOdI')
166 def test_sanitize_path(self
):
167 if sys
.platform
!= 'win32':
170 self
.assertEqual(sanitize_path('abc'), 'abc')
171 self
.assertEqual(sanitize_path('abc/def'), 'abc\\def')
172 self
.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
173 self
.assertEqual(sanitize_path('abc|def'), 'abc#def')
174 self
.assertEqual(sanitize_path('<>:"|?*'), '#######')
175 self
.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
176 self
.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
178 self
.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
179 self
.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
181 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
182 self
.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
183 self
.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
184 self
.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
187 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
188 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
191 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
192 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
193 self
.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
194 self
.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
195 self
.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
197 self
.assertEqual(sanitize_path('../abc'), '..\\abc')
198 self
.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
199 self
.assertEqual(sanitize_path('./abc'), 'abc')
200 self
.assertEqual(sanitize_path('./../abc'), '..\\abc')
202 def test_prepend_extension(self
):
203 self
.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
204 self
.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
205 self
.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
206 self
.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
207 self
.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
208 self
.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
210 def test_replace_extension(self
):
211 self
.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
212 self
.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
213 self
.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
214 self
.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
215 self
.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
216 self
.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
218 def test_remove_quotes(self
):
219 self
.assertEqual(remove_quotes(None), None)
220 self
.assertEqual(remove_quotes('"'), '"')
221 self
.assertEqual(remove_quotes("'"), "'")
222 self
.assertEqual(remove_quotes(';'), ';')
223 self
.assertEqual(remove_quotes('";'), '";')
224 self
.assertEqual(remove_quotes('""'), '')
225 self
.assertEqual(remove_quotes('";"'), ';')
227 def test_ordered_set(self
):
228 self
.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
229 self
.assertEqual(orderedSet([]), [])
230 self
.assertEqual(orderedSet([1]), [1])
231 # keep the list ordered
232 self
.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
234 def test_unescape_html(self
):
235 self
.assertEqual(unescapeHTML('%20;'), '%20;')
236 self
.assertEqual(unescapeHTML('/'), '/')
237 self
.assertEqual(unescapeHTML('/'), '/')
238 self
.assertEqual(unescapeHTML('é'), 'é')
239 self
.assertEqual(unescapeHTML('�'), '�')
241 def test_date_from_str(self
):
242 self
.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
243 self
.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
244 self
.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
245 self
.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
246 self
.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
248 def test_daterange(self
):
249 _20century
= DateRange("19000101", "20000101")
250 self
.assertFalse("17890714" in _20century
)
251 _ac
= DateRange("00010101")
252 self
.assertTrue("19690721" in _ac
)
253 _firstmilenium
= DateRange(end
="10000101")
254 self
.assertTrue("07110427" in _firstmilenium
)
256 def test_unified_dates(self
):
257 self
.assertEqual(unified_strdate('December 21, 2010'), '20101221')
258 self
.assertEqual(unified_strdate('8/7/2009'), '20090708')
259 self
.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
260 self
.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
261 self
.assertEqual(unified_strdate('1968 12 10'), '19681210')
262 self
.assertEqual(unified_strdate('1968-12-10'), '19681210')
263 self
.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
265 unified_strdate('11/26/2014 11:30:00 AM PST', day_first
=False),
268 unified_strdate('2/2/2015 6:47:40 PM', day_first
=False),
270 self
.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
271 self
.assertEqual(unified_strdate('25-09-2014'), '20140925')
272 self
.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
274 def test_determine_ext(self
):
275 self
.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
276 self
.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
277 self
.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
278 self
.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
279 self
.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
281 def test_find_xpath_attr(self
):
289 doc
= compat_etree_fromstring(testxml
)
291 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n'), None)
292 self
.assertEqual(find_xpath_attr(doc
, './/fourohfour', 'n', 'v'), None)
293 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n'), None)
294 self
.assertEqual(find_xpath_attr(doc
, './/node', 'n', 'v'), None)
295 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x'), doc
[1])
296 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'a'), doc
[1])
297 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', 'b'), doc
[3])
298 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y'), doc
[2])
299 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'c'), doc
[2])
300 self
.assertEqual(find_xpath_attr(doc
, './/node', 'y', 'd'), doc
[3])
301 self
.assertEqual(find_xpath_attr(doc
, './/node', 'x', ''), doc
[4])
303 def test_xpath_with_ns(self
):
304 testxml
= '''<root xmlns:media="http://example.com/">
306 <media:author>The Author</media:author>
307 <url>http://server.com/download.mp3</url>
310 doc
= compat_etree_fromstring(testxml
)
311 find
= lambda p
: doc
.find(xpath_with_ns(p
, {'media': 'http://example.com/'}
))
312 self
.assertTrue(find('media:song') is not None)
313 self
.assertEqual(find('media:song/media:author').text
, 'The Author')
314 self
.assertEqual(find('media:song/url').text
, 'http://server.com/download.mp3')
316 def test_xpath_element(self
):
317 doc
= xml
.etree
.ElementTree
.Element('root')
318 div
= xml
.etree
.ElementTree
.SubElement(doc
, 'div')
319 p
= xml
.etree
.ElementTree
.SubElement(div
, 'p')
321 self
.assertEqual(xpath_element(doc
, 'div/p'), p
)
322 self
.assertEqual(xpath_element(doc
, ['div/p']), p
)
323 self
.assertEqual(xpath_element(doc
, ['div/bar', 'div/p']), p
)
324 self
.assertEqual(xpath_element(doc
, 'div/bar', default
='default'), 'default')
325 self
.assertEqual(xpath_element(doc
, ['div/bar'], default
='default'), 'default')
326 self
.assertTrue(xpath_element(doc
, 'div/bar') is None)
327 self
.assertTrue(xpath_element(doc
, ['div/bar']) is None)
328 self
.assertTrue(xpath_element(doc
, ['div/bar'], 'div/baz') is None)
329 self
.assertRaises(ExtractorError
, xpath_element
, doc
, 'div/bar', fatal
=True)
330 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar'], fatal
=True)
331 self
.assertRaises(ExtractorError
, xpath_element
, doc
, ['div/bar', 'div/baz'], fatal
=True)
333 def test_xpath_text(self
):
339 doc
= compat_etree_fromstring(testxml
)
340 self
.assertEqual(xpath_text(doc
, 'div/p'), 'Foo')
341 self
.assertEqual(xpath_text(doc
, 'div/bar', default
='default'), 'default')
342 self
.assertTrue(xpath_text(doc
, 'div/bar') is None)
343 self
.assertRaises(ExtractorError
, xpath_text
, doc
, 'div/bar', fatal
=True)
345 def test_xpath_attr(self
):
351 doc
= compat_etree_fromstring(testxml
)
352 self
.assertEqual(xpath_attr(doc
, 'div/p', 'x'), 'a')
353 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x'), None)
354 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y'), None)
355 self
.assertEqual(xpath_attr(doc
, 'div/bar', 'x', default
='default'), 'default')
356 self
.assertEqual(xpath_attr(doc
, 'div/p', 'y', default
='default'), 'default')
357 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/bar', 'x', fatal
=True)
358 self
.assertRaises(ExtractorError
, xpath_attr
, doc
, 'div/p', 'y', fatal
=True)
360 def test_smuggle_url(self
):
361 data
= {"ö": "ö", "abc": [3]}
362 url
= 'https://foo.bar/baz?x=y#a'
363 smug_url
= smuggle_url(url
, data
)
364 unsmug_url
, unsmug_data
= unsmuggle_url(smug_url
)
365 self
.assertEqual(url
, unsmug_url
)
366 self
.assertEqual(data
, unsmug_data
)
368 res_url
, res_data
= unsmuggle_url(url
)
369 self
.assertEqual(res_url
, url
)
370 self
.assertEqual(res_data
, None)
372 def test_shell_quote(self
):
373 args
= ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
374 self
.assertEqual(shell_quote(args
), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
376 def test_str_to_int(self
):
377 self
.assertEqual(str_to_int('123,456'), 123456)
378 self
.assertEqual(str_to_int('123.456'), 123456)
380 def test_url_basename(self
):
381 self
.assertEqual(url_basename('http://foo.de/'), '')
382 self
.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
383 self
.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
384 self
.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
385 self
.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
387 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
390 def test_parse_duration(self
):
391 self
.assertEqual(parse_duration(None), None)
392 self
.assertEqual(parse_duration(False), None)
393 self
.assertEqual(parse_duration('invalid'), None)
394 self
.assertEqual(parse_duration('1'), 1)
395 self
.assertEqual(parse_duration('1337:12'), 80232)
396 self
.assertEqual(parse_duration('9:12:43'), 33163)
397 self
.assertEqual(parse_duration('12:00'), 720)
398 self
.assertEqual(parse_duration('00:01:01'), 61)
399 self
.assertEqual(parse_duration('x:y'), None)
400 self
.assertEqual(parse_duration('3h11m53s'), 11513)
401 self
.assertEqual(parse_duration('3h 11m 53s'), 11513)
402 self
.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
403 self
.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
404 self
.assertEqual(parse_duration('62m45s'), 3765)
405 self
.assertEqual(parse_duration('6m59s'), 419)
406 self
.assertEqual(parse_duration('49s'), 49)
407 self
.assertEqual(parse_duration('0h0m0s'), 0)
408 self
.assertEqual(parse_duration('0m0s'), 0)
409 self
.assertEqual(parse_duration('0s'), 0)
410 self
.assertEqual(parse_duration('01:02:03.05'), 3723.05)
411 self
.assertEqual(parse_duration('T30M38S'), 1838)
412 self
.assertEqual(parse_duration('5 s'), 5)
413 self
.assertEqual(parse_duration('3 min'), 180)
414 self
.assertEqual(parse_duration('2.5 hours'), 9000)
415 self
.assertEqual(parse_duration('02:03:04'), 7384)
416 self
.assertEqual(parse_duration('01:02:03:04'), 93784)
417 self
.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
418 self
.assertEqual(parse_duration('87 Min.'), 5220)
419 self
.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
421 def test_fix_xml_ampersands(self
):
423 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
425 fix_xml_ampersands('"&x=y&wrong;&z=a'),
426 '"&x=y&wrong;&z=a')
428 fix_xml_ampersands('&'><"'),
429 '&'><"')
431 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
432 self
.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
434 def test_paged_list(self
):
435 def testPL(size
, pagesize
, sliceargs
, expected
):
436 def get_page(pagenum
):
437 firstid
= pagenum
* pagesize
438 upto
= min(size
, pagenum
* pagesize
+ pagesize
)
439 for i
in range(firstid
, upto
):
442 pl
= OnDemandPagedList(get_page
, pagesize
)
443 got
= pl
.getslice(*sliceargs
)
444 self
.assertEqual(got
, expected
)
446 iapl
= InAdvancePagedList(get_page
, size
// pagesize
+ 1, pagesize
)
447 got
= iapl
.getslice(*sliceargs
)
448 self
.assertEqual(got
, expected
)
450 testPL(5, 2, (), [0, 1, 2, 3, 4])
451 testPL(5, 2, (1,), [1, 2, 3, 4])
452 testPL(5, 2, (2,), [2, 3, 4])
453 testPL(5, 2, (4,), [4])
454 testPL(5, 2, (0, 3), [0, 1, 2])
455 testPL(5, 2, (1, 4), [1, 2, 3])
456 testPL(5, 2, (2, 99), [2, 3, 4])
457 testPL(5, 2, (20, 99), [])
459 def test_read_batch_urls(self
):
460 f
= io
.StringIO('''\xef\xbb\xbf foo
463 # More after this line\r
466 self
.assertEqual(read_batch_urls(f
), ['foo', 'bar', 'baz', 'bam'])
468 def test_urlencode_postdata(self
):
469 data
= urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'}
)
470 self
.assertTrue(isinstance(data
, bytes))
472 def test_update_url_query(self
):
474 return compat_parse_qs(compat_urlparse
.urlparse(url
).query
)
475 self
.assertEqual(query_dict(update_url_query(
476 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']}
)),
477 query_dict('http://example.com/path?quality=HD&format=mp4'))
478 self
.assertEqual(query_dict(update_url_query(
479 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']}
)),
480 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
481 self
.assertEqual(query_dict(update_url_query(
482 'http://example.com/path', {'fields': 'id,formats,subtitles'}
)),
483 query_dict('http://example.com/path?fields=id,formats,subtitles'))
484 self
.assertEqual(query_dict(update_url_query(
485 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')}
)),
486 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
487 self
.assertEqual(query_dict(update_url_query(
488 'http://example.com/path?manifest=f4m', {'manifest': []}
)),
489 query_dict('http://example.com/path'))
490 self
.assertEqual(query_dict(update_url_query(
491 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'}
)),
492 query_dict('http://example.com/path?system=LINUX'))
493 self
.assertEqual(query_dict(update_url_query(
494 'http://example.com/path', {'fields': b'id,formats,subtitles'}
)),
495 query_dict('http://example.com/path?fields=id,formats,subtitles'))
496 self
.assertEqual(query_dict(update_url_query(
497 'http://example.com/path', {'width': 1080, 'height': 720}
)),
498 query_dict('http://example.com/path?width=1080&height=720'))
499 self
.assertEqual(query_dict(update_url_query(
500 'http://example.com/path', {'bitrate': 5020.43}
)),
501 query_dict('http://example.com/path?bitrate=5020.43'))
502 self
.assertEqual(query_dict(update_url_query(
503 'http://example.com/path', {'test': '第二行тест'}
)),
504 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
506 def test_dict_get(self
):
514 d
= FALSE_VALUES
.copy()
516 self
.assertEqual(dict_get(d
, 'a'), 42)
517 self
.assertEqual(dict_get(d
, 'b'), None)
518 self
.assertEqual(dict_get(d
, 'b', 42), 42)
519 self
.assertEqual(dict_get(d
, ('a', )), 42)
520 self
.assertEqual(dict_get(d
, ('b', 'a', )), 42)
521 self
.assertEqual(dict_get(d
, ('b', 'c', 'a', 'd', )), 42)
522 self
.assertEqual(dict_get(d
, ('b', 'c', )), None)
523 self
.assertEqual(dict_get(d
, ('b', 'c', ), 42), 42)
524 for key
, false_value
in FALSE_VALUES
.items():
525 self
.assertEqual(dict_get(d
, ('b', 'c', key
, )), None)
526 self
.assertEqual(dict_get(d
, ('b', 'c', key
, ), skip_false_values
=False), false_value
)
528 def test_encode_compat_str(self
):
529 self
.assertEqual(encode_compat_str(b
'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
530 self
.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
532 def test_parse_iso8601(self
):
533 self
.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
534 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
535 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
536 self
.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
537 self
.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
538 self
.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
540 def test_strip_jsonp(self
):
541 stripped
= strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
542 d
= json
.loads(stripped
)
543 self
.assertEqual(d
, [{"id": "532cb", "x": 3}
])
545 stripped
= strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
546 d
= json
.loads(stripped
)
547 self
.assertEqual(d
, {'STATUS': 'OK'}
)
549 stripped
= strip_jsonp('ps.embedHandler({"status": "success"});')
550 d
= json
.loads(stripped
)
551 self
.assertEqual(d
, {'status': 'success'}
)
553 def test_uppercase_escape(self
):
554 self
.assertEqual(uppercase_escape('aä'), 'aä')
555 self
.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
557 def test_lowercase_escape(self
):
558 self
.assertEqual(lowercase_escape('aä'), 'aä')
559 self
.assertEqual(lowercase_escape('\\u0026'), '&')
561 def test_limit_length(self
):
562 self
.assertEqual(limit_length(None, 12), None)
563 self
.assertEqual(limit_length('foo', 12), 'foo')
565 limit_length('foo bar baz asd', 12).startswith('foo bar'))
566 self
.assertTrue('...' in limit_length('foo bar baz asd', 12))
568 def test_escape_rfc3986(self
):
569 reserved
= "!*'();:@&=+$,/?#[]"
570 unreserved
= 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
571 self
.assertEqual(escape_rfc3986(reserved
), reserved
)
572 self
.assertEqual(escape_rfc3986(unreserved
), unreserved
)
573 self
.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
574 self
.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
575 self
.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
576 self
.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
578 def test_escape_url(self
):
580 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
581 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
584 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
585 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
588 escape_url('http://тест.рф/фрагмент'),
589 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
592 escape_url('http://тест.рф/абв?абв=абв#абв'),
593 'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
595 self
.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
597 def test_js_to_json_realworld(self
):
599 'clip':{'provider':'pseudo'}
601 self
.assertEqual(js_to_json(inp
), '''{
602 "clip":{"provider":"pseudo"}
604 json
.loads(js_to_json(inp
))
607 'playlist':[{'controls':{'all':null}}]
609 self
.assertEqual(js_to_json(inp
), '''{
610 "playlist":[{"controls":{"all":null}}]
613 inp
= '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
614 self
.assertEqual(js_to_json(inp
), '''"The CW's 'Crazy Ex-Girlfriend'"''')
616 inp
= '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
617 json_code
= js_to_json(inp
)
618 self
.assertEqual(json
.loads(json_code
), json
.loads(inp
))
621 0:{src:'skipped', type: 'application/dash+xml'},
622 1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
624 self
.assertEqual(js_to_json(inp
), '''{
625 "0":{"src":"skipped", "type": "application/dash+xml"},
626 "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
629 def test_js_to_json_edgecases(self
):
630 on
= js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
631 self
.assertEqual(json
.loads(on
), {"abc_def": "1'\\2\\'3\"4"}
)
633 on
= js_to_json('{"abc": true}')
634 self
.assertEqual(json
.loads(on
), {'abc': True}
)
636 # Ignore JavaScript code as well
643 self
.assertEqual(d
['x'], 1)
644 self
.assertEqual(d
['y'], 'a')
646 on
= js_to_json('["abc", "def",]')
647 self
.assertEqual(json
.loads(on
), ['abc', 'def'])
649 on
= js_to_json('{"abc": "def",}')
650 self
.assertEqual(json
.loads(on
), {'abc': 'def'}
)
652 on
= js_to_json('{ 0: /* " \n */ ",]" , }')
653 self
.assertEqual(json
.loads(on
), {'0': ',]'}
)
655 on
= js_to_json(r
'["<p>x<\/p>"]')
656 self
.assertEqual(json
.loads(on
), ['<p>x</p>'])
658 on
= js_to_json(r
'["\xaa"]')
659 self
.assertEqual(json
.loads(on
), ['\u00aa'])
661 on
= js_to_json("['a\\\nb']")
662 self
.assertEqual(json
.loads(on
), ['ab'])
664 on
= js_to_json('{0xff:0xff}')
665 self
.assertEqual(json
.loads(on
), {'255': 255}
)
667 on
= js_to_json('{077:077}')
668 self
.assertEqual(json
.loads(on
), {'63': 63}
)
670 on
= js_to_json('{42:42}')
671 self
.assertEqual(json
.loads(on
), {'42': 42}
)
673 def test_extract_attributes(self
):
674 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
675 self
.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'}
)
676 self
.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'}
)
677 self
.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"}
)
678 self
.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'}
)
679 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
680 self
.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'}
)
681 self
.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}
) # XML
682 self
.assertEqual(extract_attributes('<e x=""">'), {'x': '"'}
)
683 self
.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}
) # HTML 3.2
684 self
.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}
) # HTML 4.0
685 self
.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'}
)
686 self
.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"}
)
687 self
.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'}
)
688 self
.assertEqual(extract_attributes('<e x >'), {'x': None}
)
689 self
.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None}
)
690 self
.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'}
)
691 self
.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'}
)
692 self
.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'}
)
693 self
.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'}
)
694 self
.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'}
)
695 self
.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'}
)
696 self
.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}
) # Names lowercased
697 self
.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'}
)
698 self
.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'}
)
699 self
.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'}
)
700 self
.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'}
)
701 self
.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'}
)
702 # "Narrow" Python builds don't support unicode code points outside BMP.
705 supports_outside_bmp
= True
707 supports_outside_bmp
= False
708 if supports_outside_bmp
:
709 self
.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'}
)
711 def test_clean_html(self
):
712 self
.assertEqual(clean_html('a:\nb'), 'a: b')
713 self
.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
715 def test_intlist_to_bytes(self
):
717 intlist_to_bytes([0, 1, 127, 128, 255]),
718 b
'\x00\x01\x7f\x80\xff')
720 def test_args_to_str(self
):
722 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
723 'foo ba/r -baz \'2 be\' \'\''
726 def test_parse_filesize(self
):
727 self
.assertEqual(parse_filesize(None), None)
728 self
.assertEqual(parse_filesize(''), None)
729 self
.assertEqual(parse_filesize('91 B'), 91)
730 self
.assertEqual(parse_filesize('foobar'), None)
731 self
.assertEqual(parse_filesize('2 MiB'), 2097152)
732 self
.assertEqual(parse_filesize('5 GB'), 5000000000)
733 self
.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
734 self
.assertEqual(parse_filesize('1,24 KB'), 1240)
736 def test_parse_count(self
):
737 self
.assertEqual(parse_count(None), None)
738 self
.assertEqual(parse_count(''), None)
739 self
.assertEqual(parse_count('0'), 0)
740 self
.assertEqual(parse_count('1000'), 1000)
741 self
.assertEqual(parse_count('1.000'), 1000)
742 self
.assertEqual(parse_count('1.1k'), 1100)
743 self
.assertEqual(parse_count('1.1kk'), 1100000)
744 self
.assertEqual(parse_count('1.1kk '), 1100000)
745 self
.assertEqual(parse_count('1.1kk views'), 1100000)
747 def test_version_tuple(self
):
748 self
.assertEqual(version_tuple('1'), (1,))
749 self
.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
750 self
.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
752 def test_detect_exe_version(self
):
753 self
.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
754 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
755 configuration: --prefix=/usr --extra-'''), '1.2.1')
756 self
.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
757 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
758 self
.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
759 Trying to open render node...
760 Success at /dev/dri/renderD128.
761 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
763 def test_age_restricted(self
):
764 self
.assertFalse(age_restricted(None, 10)) # unrestricted content
765 self
.assertFalse(age_restricted(1, None)) # unrestricted policy
766 self
.assertFalse(age_restricted(8, 10))
767 self
.assertTrue(age_restricted(18, 14))
768 self
.assertFalse(age_restricted(18, 18))
770 def test_is_html(self
):
771 self
.assertFalse(is_html(b
'\x49\x44\x43<html'))
772 self
.assertTrue(is_html(b
'<!DOCTYPE foo>\xaaa'))
773 self
.assertTrue(is_html( # UTF-8 with BOM
774 b
'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
775 self
.assertTrue(is_html( # UTF-16-LE
776 b
'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
778 self
.assertTrue(is_html( # UTF-16-BE
779 b
'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
781 self
.assertTrue(is_html( # UTF-32-BE
782 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'))
783 self
.assertTrue(is_html( # UTF-32-LE
784 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'))
786 def test_render_table(self
):
790 [[123, 4], [9999, 51]]),
795 def test_match_str(self
):
796 self
.assertRaises(ValueError, match_str
, 'xy>foobar', {})
797 self
.assertFalse(match_str('xy', {'x': 1200}
))
798 self
.assertTrue(match_str('!xy', {'x': 1200}
))
799 self
.assertTrue(match_str('x', {'x': 1200}
))
800 self
.assertFalse(match_str('!x', {'x': 1200}
))
801 self
.assertTrue(match_str('x', {'x': 0}
))
802 self
.assertFalse(match_str('x>0', {'x': 0}
))
803 self
.assertFalse(match_str('x>0', {}))
804 self
.assertTrue(match_str('x>?0', {}))
805 self
.assertTrue(match_str('x>1K', {'x': 1200}
))
806 self
.assertFalse(match_str('x>2K', {'x': 1200}
))
807 self
.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}
))
808 self
.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}
))
809 self
.assertFalse(match_str('y=a212', {'y': 'foobar42'}
))
810 self
.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}
))
811 self
.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}
))
812 self
.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}
))
813 self
.assertFalse(match_str(
814 'like_count > 100 & dislike_count <? 50 & description',
815 {'like_count': 90, 'description': 'foo'}
))
816 self
.assertTrue(match_str(
817 'like_count > 100 & dislike_count <? 50 & description',
818 {'like_count': 190, 'description': 'foo'}
))
819 self
.assertFalse(match_str(
820 'like_count > 100 & dislike_count <? 50 & description',
821 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}
))
822 self
.assertFalse(match_str(
823 'like_count > 100 & dislike_count <? 50 & description',
824 {'like_count': 190, 'dislike_count': 10}
))
826 def test_parse_dfxp_time_expr(self
):
827 self
.assertEqual(parse_dfxp_time_expr(None), None)
828 self
.assertEqual(parse_dfxp_time_expr(''), None)
829 self
.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
830 self
.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
831 self
.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
832 self
.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
833 self
.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
835 def test_dfxp2srt(self
):
836 dfxp_data
= '''<?xml version="1.0" encoding="UTF-8"?>
837 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
840 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
841 <p begin="1" end="2">第二行<br/>♪♪</p>
842 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
843 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
844 <p begin="-1" end="-1">Ignore, two</p>
845 <p begin="3" dur="-1">Ignored, three</p>
850 00:00:00,000 --> 00:00:01,000
851 The following line contains Chinese characters and special symbols
854 00:00:01,000 --> 00:00:02,000
859 00:00:02,000 --> 00:00:03,000
864 self
.assertEqual(dfxp2srt(dfxp_data
), srt_data
)
866 dfxp_data_no_default_namespace
= '''<?xml version="1.0" encoding="UTF-8"?>
867 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
870 <p begin="0" end="1">The first line</p>
875 00:00:00,000 --> 00:00:01,000
879 self
.assertEqual(dfxp2srt(dfxp_data_no_default_namespace
), srt_data
)
881 def test_cli_option(self
):
882 self
.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}
, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
883 self
.assertEqual(cli_option({'proxy': None}
, '--proxy', 'proxy'), [])
884 self
.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
886 def test_cli_valueless_option(self
):
887 self
.assertEqual(cli_valueless_option(
888 {'downloader': 'external'}
, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
889 self
.assertEqual(cli_valueless_option(
890 {'downloader': 'internal'}
, '--external-downloader', 'downloader', 'external'), [])
891 self
.assertEqual(cli_valueless_option(
892 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
893 self
.assertEqual(cli_valueless_option(
894 {'nocheckcertificate': False}
, '--no-check-certificate', 'nocheckcertificate'), [])
895 self
.assertEqual(cli_valueless_option(
896 {'checkcertificate': True}
, '--no-check-certificate', 'checkcertificate', False), [])
897 self
.assertEqual(cli_valueless_option(
898 {'checkcertificate': False}
, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
900 def test_cli_bool_option(self
):
903 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate'),
904 ['--no-check-certificate', 'true'])
907 {'nocheckcertificate': True}
, '--no-check-certificate', 'nocheckcertificate', separator
='='),
908 ['--no-check-certificate=true'])
911 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
912 ['--check-certificate', 'false'])
915 {'nocheckcertificate': True}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
916 ['--check-certificate=false'])
919 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
920 ['--check-certificate', 'true'])
923 {'nocheckcertificate': False}
, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
924 ['--check-certificate=true'])
926 def test_ohdave_rsa_encrypt(self
):
927 N
= 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
931 ohdave_rsa_encrypt(b
'aa111222', e
, N
),
932 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
934 def test_encode_base_n(self
):
935 self
.assertEqual(encode_base_n(0, 30), '0')
936 self
.assertEqual(encode_base_n(80, 30), '2k')
938 custom_table
= '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
939 self
.assertEqual(encode_base_n(0, 30, custom_table
), '9')
940 self
.assertEqual(encode_base_n(80, 30, custom_table
), '7P')
942 self
.assertRaises(ValueError, encode_base_n
, 0, 70)
943 self
.assertRaises(ValueError, encode_base_n
, 0, 60, custom_table
)
945 if __name__
== '__main__':