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 (
81 from youtube_dl.compat import (
83 compat_etree_fromstring,
89 class TestUtil(unittest.TestCase):
90 def test_timeconvert(self):
91 self.assertTrue(timeconvert('') is None)
92 self.assertTrue(timeconvert('bougrg') is None)
94 def test_sanitize_filename(self):
95 self.assertEqual(sanitize_filename('abc'), 'abc')
96 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
98 self.assertEqual(sanitize_filename('123'), '123')
100 self.assertEqual('abc_de', sanitize_filename('abc/de'))
101 self.assertFalse('/' in sanitize_filename('abc/de///'))
103 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
104 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
105 self.assertEqual('yes no', sanitize_filename('yes? no'))
106 self.assertEqual('this - that', sanitize_filename('this: that'))
108 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
110 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
111 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
112 self.assertEqual(sanitize_filename(tests), tests)
115 sanitize_filename('New World record at 0:12:34'),
116 'New World record at 0_12_34')
118 self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
119 self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
120 self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
121 self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
125 for fbc in forbidden:
126 self.assertTrue(fbc not in sanitize_filename(fc))
128 def test_sanitize_filename_restricted(self):
129 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
130 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
132 self.assertEqual(sanitize_filename('123', restricted=True), '123')
134 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
135 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
137 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
138 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
139 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
140 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
142 tests = 'aäb\u4e2d\u56fd\u7684c'
143 self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
144 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
146 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
148 for fbc in forbidden:
149 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
151 # Handle a common case more neatly
152 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
153 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
154 # .. but make sure the file name is never empty
155 self.assertTrue(sanitize_filename('-', restricted=True) != '')
156 self.assertTrue(sanitize_filename(':', restricted=True) != '')
158 self.assertEqual(sanitize_filename(
159 'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞßàáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ', restricted=True),
160 'AAAAAAAECEEEEIIIIDNOOOOOOUUUUYPssaaaaaaaeceeeeiiiionoooooouuuuypy')
163 def test_sanitize_ids(self):
164 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
165 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
166 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
168 def test_sanitize_path(self):
169 if sys.platform != 'win32':
172 self.assertEqual(sanitize_path('abc'), 'abc')
173 self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
174 self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
175 self.assertEqual(sanitize_path('abc|def'), 'abc#def')
176 self.assertEqual(sanitize_path('<>:"|?*'), '#######')
177 self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
178 self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
180 self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
181 self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
183 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
184 self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
185 self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
186 self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
189 sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
190 'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
193 sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
194 'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
195 self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
196 self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
197 self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
199 self.assertEqual(sanitize_path('../abc'), '..\\abc')
200 self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
201 self.assertEqual(sanitize_path('./abc'), 'abc')
202 self.assertEqual(sanitize_path('./../abc'), '..\\abc')
204 def test_prepend_extension(self):
205 self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
206 self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
207 self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
208 self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
209 self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
210 self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
212 def test_replace_extension(self):
213 self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
214 self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
215 self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
216 self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
217 self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
218 self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
220 def test_remove_quotes(self):
221 self.assertEqual(remove_quotes(None), None)
222 self.assertEqual(remove_quotes('"'), '"')
223 self.assertEqual(remove_quotes("'"), "'")
224 self.assertEqual(remove_quotes(';'), ';')
225 self.assertEqual(remove_quotes('";'), '";')
226 self.assertEqual(remove_quotes('""'), '')
227 self.assertEqual(remove_quotes('";"'), ';')
229 def test_ordered_set(self):
230 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
231 self.assertEqual(orderedSet([]), [])
232 self.assertEqual(orderedSet([1]), [1])
233 # keep the list ordered
234 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
236 def test_unescape_html(self):
237 self.assertEqual(unescapeHTML('%20;'), '%20;')
238 self.assertEqual(unescapeHTML('/'), '/')
239 self.assertEqual(unescapeHTML('/'), '/')
240 self.assertEqual(unescapeHTML('é'), 'é')
241 self.assertEqual(unescapeHTML('�'), '�')
243 def test_date_from_str(self):
244 self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
245 self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
246 self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
247 self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
248 self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
250 def test_daterange(self):
251 _20century = DateRange("19000101", "20000101")
252 self.assertFalse("17890714" in _20century)
253 _ac = DateRange("00010101")
254 self.assertTrue("19690721" in _ac)
255 _firstmilenium = DateRange(end="10000101")
256 self.assertTrue("07110427" in _firstmilenium)
258 def test_unified_dates(self):
259 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
260 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
261 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
262 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
263 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
264 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
265 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
267 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
270 unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
272 self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
273 self.assertEqual(unified_strdate('25-09-2014'), '20140925')
274 self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
276 def test_determine_ext(self):
277 self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
278 self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
279 self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
280 self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
281 self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
283 def test_find_xpath_attr(self):
291 doc = compat_etree_fromstring(testxml)
293 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
294 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
295 self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
296 self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
297 self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
298 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
299 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
300 self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
301 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
302 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
303 self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
305 def test_xpath_with_ns(self):
306 testxml = '''<root xmlns:media="http://example.com/">
308 <media:author>The Author</media:author>
309 <url>http://server.com/download.mp3</url>
312 doc = compat_etree_fromstring(testxml)
313 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
314 self.assertTrue(find('media:song') is not None)
315 self.assertEqual(find('media:song/media:author').text, 'The Author')
316 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
318 def test_xpath_element(self):
319 doc = xml.etree.ElementTree.Element('root')
320 div = xml.etree.ElementTree.SubElement(doc, 'div')
321 p = xml.etree.ElementTree.SubElement(div, 'p')
323 self.assertEqual(xpath_element(doc, 'div/p'), p)
324 self.assertEqual(xpath_element(doc, ['div/p']), p)
325 self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
326 self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
327 self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
328 self.assertTrue(xpath_element(doc, 'div/bar') is None)
329 self.assertTrue(xpath_element(doc, ['div/bar']) is None)
330 self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
331 self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
332 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
333 self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
335 def test_xpath_text(self):
341 doc = compat_etree_fromstring(testxml)
342 self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
343 self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
344 self.assertTrue(xpath_text(doc, 'div/bar') is None)
345 self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
347 def test_xpath_attr(self):
353 doc = compat_etree_fromstring(testxml)
354 self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
355 self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
356 self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
357 self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
358 self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
359 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
360 self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
362 def test_smuggle_url(self):
363 data = {"ö": "ö", "abc": [3]}
364 url = 'https://foo.bar/baz?x=y#a'
365 smug_url = smuggle_url(url, data)
366 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
367 self.assertEqual(url, unsmug_url)
368 self.assertEqual(data, unsmug_data)
370 res_url, res_data = unsmuggle_url(url)
371 self.assertEqual(res_url, url)
372 self.assertEqual(res_data, None)
374 def test_shell_quote(self):
375 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
376 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
378 def test_str_to_int(self):
379 self.assertEqual(str_to_int('123,456'), 123456)
380 self.assertEqual(str_to_int('123.456'), 123456)
382 def test_url_basename(self):
383 self.assertEqual(url_basename('http://foo.de/'), '')
384 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
385 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
386 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
387 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
389 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
392 def test_parse_duration(self):
393 self.assertEqual(parse_duration(None), None)
394 self.assertEqual(parse_duration(False), None)
395 self.assertEqual(parse_duration('invalid'), None)
396 self.assertEqual(parse_duration('1'), 1)
397 self.assertEqual(parse_duration('1337:12'), 80232)
398 self.assertEqual(parse_duration('9:12:43'), 33163)
399 self.assertEqual(parse_duration('12:00'), 720)
400 self.assertEqual(parse_duration('00:01:01'), 61)
401 self.assertEqual(parse_duration('x:y'), None)
402 self.assertEqual(parse_duration('3h11m53s'), 11513)
403 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
404 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
405 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
406 self.assertEqual(parse_duration('62m45s'), 3765)
407 self.assertEqual(parse_duration('6m59s'), 419)
408 self.assertEqual(parse_duration('49s'), 49)
409 self.assertEqual(parse_duration('0h0m0s'), 0)
410 self.assertEqual(parse_duration('0m0s'), 0)
411 self.assertEqual(parse_duration('0s'), 0)
412 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
413 self.assertEqual(parse_duration('T30M38S'), 1838)
414 self.assertEqual(parse_duration('5 s'), 5)
415 self.assertEqual(parse_duration('3 min'), 180)
416 self.assertEqual(parse_duration('2.5 hours'), 9000)
417 self.assertEqual(parse_duration('02:03:04'), 7384)
418 self.assertEqual(parse_duration('01:02:03:04'), 93784)
419 self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
420 self.assertEqual(parse_duration('87 Min.'), 5220)
421 self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
423 def test_fix_xml_ampersands(self):
425 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
427 fix_xml_ampersands('"&x=y&wrong;&z=a'),
428 '"&x=y&wrong;&z=a')
430 fix_xml_ampersands('&'><"'),
431 '&'><"')
433 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
434 self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
436 def test_paged_list(self):
437 def testPL(size, pagesize, sliceargs, expected):
438 def get_page(pagenum):
439 firstid = pagenum * pagesize
440 upto = min(size, pagenum * pagesize + pagesize)
441 for i in range(firstid, upto):
444 pl = OnDemandPagedList(get_page, pagesize)
445 got = pl.getslice(*sliceargs)
446 self.assertEqual(got, expected)
448 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
449 got = iapl.getslice(*sliceargs)
450 self.assertEqual(got, expected)
452 testPL(5, 2, (), [0, 1, 2, 3, 4])
453 testPL(5, 2, (1,), [1, 2, 3, 4])
454 testPL(5, 2, (2,), [2, 3, 4])
455 testPL(5, 2, (4,), [4])
456 testPL(5, 2, (0, 3), [0, 1, 2])
457 testPL(5, 2, (1, 4), [1, 2, 3])
458 testPL(5, 2, (2, 99), [2, 3, 4])
459 testPL(5, 2, (20, 99), [])
461 def test_struct_unpack(self):
462 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
464 def test_read_batch_urls(self):
465 f = io.StringIO('''\xef\xbb\xbf foo
468 # More after this line\r
471 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
473 def test_urlencode_postdata(self):
474 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
475 self.assertTrue(isinstance(data, bytes))
477 def test_update_url_query(self):
479 return compat_parse_qs(compat_urlparse.urlparse(url).query)
480 self.assertEqual(query_dict(update_url_query(
481 'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
482 query_dict('http://example.com/path?quality=HD&format=mp4'))
483 self.assertEqual(query_dict(update_url_query(
484 'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
485 query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
486 self.assertEqual(query_dict(update_url_query(
487 'http://example.com/path', {'fields': 'id,formats,subtitles'})),
488 query_dict('http://example.com/path?fields=id,formats,subtitles'))
489 self.assertEqual(query_dict(update_url_query(
490 'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
491 query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
492 self.assertEqual(query_dict(update_url_query(
493 'http://example.com/path?manifest=f4m', {'manifest': []})),
494 query_dict('http://example.com/path'))
495 self.assertEqual(query_dict(update_url_query(
496 'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
497 query_dict('http://example.com/path?system=LINUX'))
498 self.assertEqual(query_dict(update_url_query(
499 'http://example.com/path', {'fields': b'id,formats,subtitles'})),
500 query_dict('http://example.com/path?fields=id,formats,subtitles'))
501 self.assertEqual(query_dict(update_url_query(
502 'http://example.com/path', {'width': 1080, 'height': 720})),
503 query_dict('http://example.com/path?width=1080&height=720'))
504 self.assertEqual(query_dict(update_url_query(
505 'http://example.com/path', {'bitrate': 5020.43})),
506 query_dict('http://example.com/path?bitrate=5020.43'))
507 self.assertEqual(query_dict(update_url_query(
508 'http://example.com/path', {'test': '第二行тест'})),
509 query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
511 def test_dict_get(self):
519 d = FALSE_VALUES.copy()
521 self.assertEqual(dict_get(d, 'a'), 42)
522 self.assertEqual(dict_get(d, 'b'), None)
523 self.assertEqual(dict_get(d, 'b', 42), 42)
524 self.assertEqual(dict_get(d, ('a', )), 42)
525 self.assertEqual(dict_get(d, ('b', 'a', )), 42)
526 self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
527 self.assertEqual(dict_get(d, ('b', 'c', )), None)
528 self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
529 for key, false_value in FALSE_VALUES.items():
530 self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
531 self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
533 def test_encode_compat_str(self):
534 self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
535 self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
537 def test_parse_iso8601(self):
538 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
539 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
540 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
541 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
542 self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
543 self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
545 def test_strip_jsonp(self):
546 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
547 d = json.loads(stripped)
548 self.assertEqual(d, [{"id": "532cb", "x": 3}])
550 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
551 d = json.loads(stripped)
552 self.assertEqual(d, {'STATUS': 'OK'})
554 stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
555 d = json.loads(stripped)
556 self.assertEqual(d, {'status': 'success'})
558 def test_uppercase_escape(self):
559 self.assertEqual(uppercase_escape('aä'), 'aä')
560 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
562 def test_lowercase_escape(self):
563 self.assertEqual(lowercase_escape('aä'), 'aä')
564 self.assertEqual(lowercase_escape('\\u0026'), '&')
566 def test_limit_length(self):
567 self.assertEqual(limit_length(None, 12), None)
568 self.assertEqual(limit_length('foo', 12), 'foo')
570 limit_length('foo bar baz asd', 12).startswith('foo bar'))
571 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
573 def test_escape_rfc3986(self):
574 reserved = "!*'();:@&=+$,/?#[]"
575 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
576 self.assertEqual(escape_rfc3986(reserved), reserved)
577 self.assertEqual(escape_rfc3986(unreserved), unreserved)
578 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
579 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
580 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
581 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
583 def test_escape_url(self):
585 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
586 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
589 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
590 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
593 escape_url('http://тест.рф/фрагмент'),
594 'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
597 escape_url('http://тест.рф/абв?абв=абв#абв'),
598 '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'
600 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
602 def test_js_to_json_realworld(self):
604 'clip':{'provider':'pseudo'}
606 self.assertEqual(js_to_json(inp), '''{
607 "clip":{"provider":"pseudo"}
609 json.loads(js_to_json(inp))
612 'playlist':[{'controls':{'all':null}}]
614 self.assertEqual(js_to_json(inp), '''{
615 "playlist":[{"controls":{"all":null}}]
618 inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
619 self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
621 inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
622 json_code = js_to_json(inp)
623 self.assertEqual(json.loads(json_code), json.loads(inp))
625 def test_js_to_json_edgecases(self):
626 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
627 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
629 on = js_to_json('{"abc": true}')
630 self.assertEqual(json.loads(on), {'abc': True})
632 # Ignore JavaScript code as well
639 self.assertEqual(d['x'], 1)
640 self.assertEqual(d['y'], 'a')
642 on = js_to_json('["abc", "def",]')
643 self.assertEqual(json.loads(on), ['abc', 'def'])
645 on = js_to_json('{"abc": "def",}')
646 self.assertEqual(json.loads(on), {'abc': 'def'})
648 def test_extract_attributes(self):
649 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
650 self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
651 self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
652 self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
653 self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
654 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
655 self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
656 self.assertEqual(extract_attributes('<e x="&">'), {'x': '&'}) # XML
657 self.assertEqual(extract_attributes('<e x=""">'), {'x': '"'})
658 self.assertEqual(extract_attributes('<e x="£">'), {'x': '£'}) # HTML 3.2
659 self.assertEqual(extract_attributes('<e x="λ">'), {'x': 'λ'}) # HTML 4.0
660 self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
661 self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
662 self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
663 self.assertEqual(extract_attributes('<e x >'), {'x': None})
664 self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
665 self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
666 self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
667 self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
668 self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
669 self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
670 self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
671 self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'}) # Names lowercased
672 self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
673 self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
674 self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
675 self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
676 self.assertEqual(extract_attributes('<e x="décomposé">'), {'x': 'décompose\u0301'})
677 # "Narrow" Python builds don't support unicode code points outside BMP.
680 supports_outside_bmp = True
682 supports_outside_bmp = False
683 if supports_outside_bmp:
684 self.assertEqual(extract_attributes('<e x="Smile 😀!">'), {'x': 'Smile \U0001f600!'})
686 def test_clean_html(self):
687 self.assertEqual(clean_html('a:\nb'), 'a: b')
688 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
690 def test_intlist_to_bytes(self):
692 intlist_to_bytes([0, 1, 127, 128, 255]),
693 b'\x00\x01\x7f\x80\xff')
695 def test_args_to_str(self):
697 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
698 'foo ba/r -baz \'2 be\' \'\''
701 def test_parse_filesize(self):
702 self.assertEqual(parse_filesize(None), None)
703 self.assertEqual(parse_filesize(''), None)
704 self.assertEqual(parse_filesize('91 B'), 91)
705 self.assertEqual(parse_filesize('foobar'), None)
706 self.assertEqual(parse_filesize('2 MiB'), 2097152)
707 self.assertEqual(parse_filesize('5 GB'), 5000000000)
708 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
709 self.assertEqual(parse_filesize('1,24 KB'), 1240)
711 def test_parse_count(self):
712 self.assertEqual(parse_count(None), None)
713 self.assertEqual(parse_count(''), None)
714 self.assertEqual(parse_count('0'), 0)
715 self.assertEqual(parse_count('1000'), 1000)
716 self.assertEqual(parse_count('1.000'), 1000)
717 self.assertEqual(parse_count('1.1k'), 1100)
718 self.assertEqual(parse_count('1.1kk'), 1100000)
719 self.assertEqual(parse_count('1.1kk '), 1100000)
720 self.assertEqual(parse_count('1.1kk views'), 1100000)
722 def test_version_tuple(self):
723 self.assertEqual(version_tuple('1'), (1,))
724 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
725 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
727 def test_detect_exe_version(self):
728 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
729 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
730 configuration: --prefix=/usr --extra-'''), '1.2.1')
731 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
732 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
733 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
734 Trying to open render node...
735 Success at /dev/dri/renderD128.
736 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
738 def test_age_restricted(self):
739 self.assertFalse(age_restricted(None, 10)) # unrestricted content
740 self.assertFalse(age_restricted(1, None)) # unrestricted policy
741 self.assertFalse(age_restricted(8, 10))
742 self.assertTrue(age_restricted(18, 14))
743 self.assertFalse(age_restricted(18, 18))
745 def test_is_html(self):
746 self.assertFalse(is_html(b'\x49\x44\x43<html'))
747 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
748 self.assertTrue(is_html( # UTF-8 with BOM
749 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
750 self.assertTrue(is_html( # UTF-16-LE
751 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
753 self.assertTrue(is_html( # UTF-16-BE
754 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
756 self.assertTrue(is_html( # UTF-32-BE
757 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'))
758 self.assertTrue(is_html( # UTF-32-LE
759 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'))
761 def test_render_table(self):
765 [[123, 4], [9999, 51]]),
770 def test_match_str(self):
771 self.assertRaises(ValueError, match_str, 'xy>foobar', {})
772 self.assertFalse(match_str('xy', {'x': 1200}))
773 self.assertTrue(match_str('!xy', {'x': 1200}))
774 self.assertTrue(match_str('x', {'x': 1200}))
775 self.assertFalse(match_str('!x', {'x': 1200}))
776 self.assertTrue(match_str('x', {'x': 0}))
777 self.assertFalse(match_str('x>0', {'x': 0}))
778 self.assertFalse(match_str('x>0', {}))
779 self.assertTrue(match_str('x>?0', {}))
780 self.assertTrue(match_str('x>1K', {'x': 1200}))
781 self.assertFalse(match_str('x>2K', {'x': 1200}))
782 self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
783 self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
784 self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
785 self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
786 self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
787 self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
788 self.assertFalse(match_str(
789 'like_count > 100 & dislike_count <? 50 & description',
790 {'like_count': 90, 'description': 'foo'}))
791 self.assertTrue(match_str(
792 'like_count > 100 & dislike_count <? 50 & description',
793 {'like_count': 190, 'description': 'foo'}))
794 self.assertFalse(match_str(
795 'like_count > 100 & dislike_count <? 50 & description',
796 {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
797 self.assertFalse(match_str(
798 'like_count > 100 & dislike_count <? 50 & description',
799 {'like_count': 190, 'dislike_count': 10}))
801 def test_parse_dfxp_time_expr(self):
802 self.assertEqual(parse_dfxp_time_expr(None), None)
803 self.assertEqual(parse_dfxp_time_expr(''), None)
804 self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
805 self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
806 self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
807 self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
808 self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
810 def test_dfxp2srt(self):
811 dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
812 <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
815 <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
816 <p begin="1" end="2">第二行<br/>♪♪</p>
817 <p begin="2" dur="1"><span>Third<br/>Line</span></p>
818 <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
819 <p begin="-1" end="-1">Ignore, two</p>
820 <p begin="3" dur="-1">Ignored, three</p>
825 00:00:00,000 --> 00:00:01,000
826 The following line contains Chinese characters and special symbols
829 00:00:01,000 --> 00:00:02,000
834 00:00:02,000 --> 00:00:03,000
839 self.assertEqual(dfxp2srt(dfxp_data), srt_data)
841 dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
842 <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
845 <p begin="0" end="1">The first line</p>
850 00:00:00,000 --> 00:00:01,000
854 self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
856 def test_cli_option(self):
857 self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
858 self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
859 self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
861 def test_cli_valueless_option(self):
862 self.assertEqual(cli_valueless_option(
863 {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
864 self.assertEqual(cli_valueless_option(
865 {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
866 self.assertEqual(cli_valueless_option(
867 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
868 self.assertEqual(cli_valueless_option(
869 {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
870 self.assertEqual(cli_valueless_option(
871 {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
872 self.assertEqual(cli_valueless_option(
873 {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
875 def test_cli_bool_option(self):
878 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
879 ['--no-check-certificate', 'true'])
882 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
883 ['--no-check-certificate=true'])
886 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
887 ['--check-certificate', 'false'])
890 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
891 ['--check-certificate=false'])
894 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
895 ['--check-certificate', 'true'])
898 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
899 ['--check-certificate=true'])
901 def test_ohdave_rsa_encrypt(self):
902 N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
906 ohdave_rsa_encrypt(b'aa111222', e, N),
907 '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
909 def test_encode_base_n(self):
910 self.assertEqual(encode_base_n(0, 30), '0')
911 self.assertEqual(encode_base_n(80, 30), '2k')
913 custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
914 self.assertEqual(encode_base_n(0, 30, custom_table), '9')
915 self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
917 self.assertRaises(ValueError, encode_base_n, 0, 70)
918 self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
920 if __name__ == '__main__':