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 (
59 class TestUtil(unittest.TestCase):
60 def test_timeconvert(self):
61 self.assertTrue(timeconvert('') is None)
62 self.assertTrue(timeconvert('bougrg') is None)
64 def test_sanitize_filename(self):
65 self.assertEqual(sanitize_filename('abc'), 'abc')
66 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
68 self.assertEqual(sanitize_filename('123'), '123')
70 self.assertEqual('abc_de', sanitize_filename('abc/de'))
71 self.assertFalse('/' in sanitize_filename('abc/de///'))
73 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
74 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
75 self.assertEqual('yes no', sanitize_filename('yes? no'))
76 self.assertEqual('this - that', sanitize_filename('this: that'))
78 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
80 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
81 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
82 self.assertEqual(sanitize_filename(tests), tests)
85 sanitize_filename('New World record at 0:12:34'),
86 'New World record at 0_12_34')
91 self.assertTrue(fbc not in sanitize_filename(fc))
93 def test_sanitize_filename_restricted(self):
94 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
95 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
97 self.assertEqual(sanitize_filename('123', restricted=True), '123')
99 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
100 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
102 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
103 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
104 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
105 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
107 tests = 'a\xe4b\u4e2d\u56fd\u7684c'
108 self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
109 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
111 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
113 for fbc in forbidden:
114 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
116 # Handle a common case more neatly
117 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
118 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
119 # .. but make sure the file name is never empty
120 self.assertTrue(sanitize_filename('-', restricted=True) != '')
121 self.assertTrue(sanitize_filename(':', restricted=True) != '')
123 def test_sanitize_ids(self):
124 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
125 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
126 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
128 def test_ordered_set(self):
129 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
130 self.assertEqual(orderedSet([]), [])
131 self.assertEqual(orderedSet([1]), [1])
132 # keep the list ordered
133 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
135 def test_unescape_html(self):
136 self.assertEqual(unescapeHTML('%20;'), '%20;')
138 unescapeHTML('é'), 'é')
140 def test_daterange(self):
141 _20century = DateRange("19000101", "20000101")
142 self.assertFalse("17890714" in _20century)
143 _ac = DateRange("00010101")
144 self.assertTrue("19690721" in _ac)
145 _firstmilenium = DateRange(end="10000101")
146 self.assertTrue("07110427" in _firstmilenium)
148 def test_unified_dates(self):
149 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
150 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
151 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
152 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
153 self.assertEqual(unified_strdate('1968 12 10'), '19681210')
154 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
155 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
157 unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
160 def test_find_xpath_attr(self):
167 doc = xml.etree.ElementTree.fromstring(testxml)
169 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
170 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
171 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
173 def test_xpath_with_ns(self):
174 testxml = '''<root xmlns:media="http://example.com/">
176 <media:author>The Author</media:author>
177 <url>http://server.com/download.mp3</url>
180 doc = xml.etree.ElementTree.fromstring(testxml)
181 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
182 self.assertTrue(find('media:song') is not None)
183 self.assertEqual(find('media:song/media:author').text, 'The Author')
184 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
186 def test_smuggle_url(self):
187 data = {"ö": "ö", "abc": [3]}
188 url = 'https://foo.bar/baz?x=y#a'
189 smug_url = smuggle_url(url, data)
190 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
191 self.assertEqual(url, unsmug_url)
192 self.assertEqual(data, unsmug_data)
194 res_url, res_data = unsmuggle_url(url)
195 self.assertEqual(res_url, url)
196 self.assertEqual(res_data, None)
198 def test_shell_quote(self):
199 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
200 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
202 def test_str_to_int(self):
203 self.assertEqual(str_to_int('123,456'), 123456)
204 self.assertEqual(str_to_int('123.456'), 123456)
206 def test_url_basename(self):
207 self.assertEqual(url_basename('http://foo.de/'), '')
208 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
209 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
210 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
211 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
213 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
216 def test_parse_duration(self):
217 self.assertEqual(parse_duration(None), None)
218 self.assertEqual(parse_duration(False), None)
219 self.assertEqual(parse_duration('invalid'), None)
220 self.assertEqual(parse_duration('1'), 1)
221 self.assertEqual(parse_duration('1337:12'), 80232)
222 self.assertEqual(parse_duration('9:12:43'), 33163)
223 self.assertEqual(parse_duration('12:00'), 720)
224 self.assertEqual(parse_duration('00:01:01'), 61)
225 self.assertEqual(parse_duration('x:y'), None)
226 self.assertEqual(parse_duration('3h11m53s'), 11513)
227 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
228 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
229 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
230 self.assertEqual(parse_duration('62m45s'), 3765)
231 self.assertEqual(parse_duration('6m59s'), 419)
232 self.assertEqual(parse_duration('49s'), 49)
233 self.assertEqual(parse_duration('0h0m0s'), 0)
234 self.assertEqual(parse_duration('0m0s'), 0)
235 self.assertEqual(parse_duration('0s'), 0)
236 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
237 self.assertEqual(parse_duration('T30M38S'), 1838)
238 self.assertEqual(parse_duration('5 s'), 5)
239 self.assertEqual(parse_duration('3 min'), 180)
240 self.assertEqual(parse_duration('2.5 hours'), 9000)
242 def test_fix_xml_ampersands(self):
244 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
246 fix_xml_ampersands('"&x=y&wrong;&z=a'),
247 '"&x=y&wrong;&z=a')
249 fix_xml_ampersands('&'><"'),
250 '&'><"')
252 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
253 self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
255 def test_paged_list(self):
256 def testPL(size, pagesize, sliceargs, expected):
257 def get_page(pagenum):
258 firstid = pagenum * pagesize
259 upto = min(size, pagenum * pagesize + pagesize)
260 for i in range(firstid, upto):
263 pl = OnDemandPagedList(get_page, pagesize)
264 got = pl.getslice(*sliceargs)
265 self.assertEqual(got, expected)
267 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
268 got = iapl.getslice(*sliceargs)
269 self.assertEqual(got, expected)
271 testPL(5, 2, (), [0, 1, 2, 3, 4])
272 testPL(5, 2, (1,), [1, 2, 3, 4])
273 testPL(5, 2, (2,), [2, 3, 4])
274 testPL(5, 2, (4,), [4])
275 testPL(5, 2, (0, 3), [0, 1, 2])
276 testPL(5, 2, (1, 4), [1, 2, 3])
277 testPL(5, 2, (2, 99), [2, 3, 4])
278 testPL(5, 2, (20, 99), [])
280 def test_struct_unpack(self):
281 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
283 def test_read_batch_urls(self):
284 f = io.StringIO('''\xef\xbb\xbf foo
287 # More after this line\r
290 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
292 def test_urlencode_postdata(self):
293 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
294 self.assertTrue(isinstance(data, bytes))
296 def test_parse_iso8601(self):
297 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
298 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
299 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
300 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
302 def test_strip_jsonp(self):
303 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
304 d = json.loads(stripped)
305 self.assertEqual(d, [{"id": "532cb", "x": 3}])
307 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
308 d = json.loads(stripped)
309 self.assertEqual(d, {'STATUS': 'OK'})
311 def test_uppercase_escape(self):
312 self.assertEqual(uppercase_escape('aä'), 'aä')
313 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
315 def test_limit_length(self):
316 self.assertEqual(limit_length(None, 12), None)
317 self.assertEqual(limit_length('foo', 12), 'foo')
319 limit_length('foo bar baz asd', 12).startswith('foo bar'))
320 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
322 def test_escape_rfc3986(self):
323 reserved = "!*'();:@&=+$,/?#[]"
324 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
325 self.assertEqual(escape_rfc3986(reserved), reserved)
326 self.assertEqual(escape_rfc3986(unreserved), unreserved)
327 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
328 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
329 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
330 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
332 def test_escape_url(self):
334 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
335 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
338 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
339 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
342 escape_url('http://тест.рф/фрагмент'),
343 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
346 escape_url('http://тест.рф/абв?абв=абв#абв'),
347 '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'
349 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
351 def test_js_to_json_realworld(self):
353 'clip':{'provider':'pseudo'}
355 self.assertEqual(js_to_json(inp), '''{
356 "clip":{"provider":"pseudo"}
358 json.loads(js_to_json(inp))
361 'playlist':[{'controls':{'all':null}}]
363 self.assertEqual(js_to_json(inp), '''{
364 "playlist":[{"controls":{"all":null}}]
367 def test_js_to_json_edgecases(self):
368 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
369 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
371 on = js_to_json('{"abc": true}')
372 self.assertEqual(json.loads(on), {'abc': True})
374 def test_clean_html(self):
375 self.assertEqual(clean_html('a:\nb'), 'a: b')
376 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
378 def test_intlist_to_bytes(self):
380 intlist_to_bytes([0, 1, 127, 128, 255]),
381 b'\x00\x01\x7f\x80\xff')
383 def test_args_to_str(self):
385 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
386 'foo ba/r -baz \'2 be\' \'\''
389 def test_parse_filesize(self):
390 self.assertEqual(parse_filesize(None), None)
391 self.assertEqual(parse_filesize(''), None)
392 self.assertEqual(parse_filesize('91 B'), 91)
393 self.assertEqual(parse_filesize('foobar'), None)
394 self.assertEqual(parse_filesize('2 MiB'), 2097152)
395 self.assertEqual(parse_filesize('5 GB'), 5000000000)
396 self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
397 self.assertEqual(parse_filesize('1,24 KB'), 1240)
399 def test_version_tuple(self):
400 self.assertEqual(version_tuple('1'), (1,))
401 self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
402 self.assertEqual(version_tuple('10.1-6'), (10, 1, 6)) # avconv style
404 def test_detect_exe_version(self):
405 self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
406 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
407 configuration: --prefix=/usr --extra-'''), '1.2.1')
408 self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
409 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
410 self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
411 Trying to open render node...
412 Success at /dev/dri/renderD128.
413 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
415 def test_age_restricted(self):
416 self.assertFalse(age_restricted(None, 10)) # unrestricted content
417 self.assertFalse(age_restricted(1, None)) # unrestricted policy
418 self.assertFalse(age_restricted(8, 10))
419 self.assertTrue(age_restricted(18, 14))
420 self.assertFalse(age_restricted(18, 18))
422 def test_is_html(self):
423 self.assertFalse(is_html(b'\x49\x44\x43<html'))
424 self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
425 self.assertTrue(is_html( # UTF-8 with BOM
426 b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
427 self.assertTrue(is_html( # UTF-16-LE
428 b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
430 self.assertTrue(is_html( # UTF-16-BE
431 b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
433 self.assertTrue(is_html( # UTF-32-BE
434 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'))
435 self.assertTrue(is_html( # UTF-32-LE
436 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'))
438 def test_render_table(self):
442 [[123, 4], [9999, 51]]),
448 if __name__ == '__main__':