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 (
48 get_filesystem_encoding,
54 class TestUtil(unittest.TestCase):
55 def test_timeconvert(self):
56 self.assertTrue(timeconvert('') is None)
57 self.assertTrue(timeconvert('bougrg') is None)
59 def test_sanitize_filename(self):
60 self.assertEqual(sanitize_filename('abc'), 'abc')
61 self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
63 self.assertEqual(sanitize_filename('123'), '123')
65 self.assertEqual('abc_de', sanitize_filename('abc/de'))
66 self.assertFalse('/' in sanitize_filename('abc/de///'))
68 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
69 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
70 self.assertEqual('yes no', sanitize_filename('yes? no'))
71 self.assertEqual('this - that', sanitize_filename('this: that'))
73 self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
75 self.assertEqual(sanitize_filename(aumlaut), aumlaut)
76 tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
77 self.assertEqual(sanitize_filename(tests), tests)
82 self.assertTrue(fbc not in sanitize_filename(fc))
84 def test_sanitize_filename_restricted(self):
85 self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
86 self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
88 self.assertEqual(sanitize_filename('123', restricted=True), '123')
90 self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
91 self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
93 self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
94 self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
95 self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
96 self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
98 tests = 'a\xe4b\u4e2d\u56fd\u7684c'
99 self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
100 self.assertTrue(sanitize_filename('\xf6', restricted=True) != '') # No empty filename
102 forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
104 for fbc in forbidden:
105 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
107 # Handle a common case more neatly
108 self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
109 self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
110 # .. but make sure the file name is never empty
111 self.assertTrue(sanitize_filename('-', restricted=True) != '')
112 self.assertTrue(sanitize_filename(':', restricted=True) != '')
114 def test_sanitize_ids(self):
115 self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
116 self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
117 self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
119 def test_ordered_set(self):
120 self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
121 self.assertEqual(orderedSet([]), [])
122 self.assertEqual(orderedSet([1]), [1])
123 #keep the list ordered
124 self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
126 def test_unescape_html(self):
127 self.assertEqual(unescapeHTML('%20;'), '%20;')
129 unescapeHTML('é'), 'é')
131 def test_daterange(self):
132 _20century = DateRange("19000101","20000101")
133 self.assertFalse("17890714" in _20century)
134 _ac = DateRange("00010101")
135 self.assertTrue("19690721" in _ac)
136 _firstmilenium = DateRange(end="10000101")
137 self.assertTrue("07110427" in _firstmilenium)
139 def test_unified_dates(self):
140 self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
141 self.assertEqual(unified_strdate('8/7/2009'), '20090708')
142 self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
143 self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
144 self.assertEqual(unified_strdate('1968-12-10'), '19681210')
145 self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
147 def test_find_xpath_attr(self):
154 doc = xml.etree.ElementTree.fromstring(testxml)
156 self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
157 self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
158 self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
160 def test_xpath_with_ns(self):
161 testxml = '''<root xmlns:media="http://example.com/">
163 <media:author>The Author</media:author>
164 <url>http://server.com/download.mp3</url>
167 doc = xml.etree.ElementTree.fromstring(testxml)
168 find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
169 self.assertTrue(find('media:song') is not None)
170 self.assertEqual(find('media:song/media:author').text, 'The Author')
171 self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
173 def test_smuggle_url(self):
174 data = {u"ö": u"ö", u"abc": [3]}
175 url = 'https://foo.bar/baz?x=y#a'
176 smug_url = smuggle_url(url, data)
177 unsmug_url, unsmug_data = unsmuggle_url(smug_url)
178 self.assertEqual(url, unsmug_url)
179 self.assertEqual(data, unsmug_data)
181 res_url, res_data = unsmuggle_url(url)
182 self.assertEqual(res_url, url)
183 self.assertEqual(res_data, None)
185 def test_shell_quote(self):
186 args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
187 self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
189 def test_str_to_int(self):
190 self.assertEqual(str_to_int('123,456'), 123456)
191 self.assertEqual(str_to_int('123.456'), 123456)
193 def test_url_basename(self):
194 self.assertEqual(url_basename('http://foo.de/'), '')
195 self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
196 self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
197 self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
198 self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
200 url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
203 def test_parse_duration(self):
204 self.assertEqual(parse_duration(None), None)
205 self.assertEqual(parse_duration('1'), 1)
206 self.assertEqual(parse_duration('1337:12'), 80232)
207 self.assertEqual(parse_duration('9:12:43'), 33163)
208 self.assertEqual(parse_duration('12:00'), 720)
209 self.assertEqual(parse_duration('00:01:01'), 61)
210 self.assertEqual(parse_duration('x:y'), None)
211 self.assertEqual(parse_duration('3h11m53s'), 11513)
212 self.assertEqual(parse_duration('3h 11m 53s'), 11513)
213 self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
214 self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
215 self.assertEqual(parse_duration('62m45s'), 3765)
216 self.assertEqual(parse_duration('6m59s'), 419)
217 self.assertEqual(parse_duration('49s'), 49)
218 self.assertEqual(parse_duration('0h0m0s'), 0)
219 self.assertEqual(parse_duration('0m0s'), 0)
220 self.assertEqual(parse_duration('0s'), 0)
221 self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
222 self.assertEqual(parse_duration('T30M38S'), 1838)
224 def test_fix_xml_ampersands(self):
226 fix_xml_ampersands('"&x=y&z=a'), '"&x=y&z=a')
228 fix_xml_ampersands('"&x=y&wrong;&z=a'),
229 '"&x=y&wrong;&z=a')
231 fix_xml_ampersands('&'><"'),
232 '&'><"')
234 fix_xml_ampersands('Ӓ᪼'), 'Ӓ᪼')
235 self.assertEqual(fix_xml_ampersands('&#&#'), '&#&#')
237 def test_paged_list(self):
238 def testPL(size, pagesize, sliceargs, expected):
239 def get_page(pagenum):
240 firstid = pagenum * pagesize
241 upto = min(size, pagenum * pagesize + pagesize)
242 for i in range(firstid, upto):
245 pl = OnDemandPagedList(get_page, pagesize)
246 got = pl.getslice(*sliceargs)
247 self.assertEqual(got, expected)
249 iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
250 got = iapl.getslice(*sliceargs)
251 self.assertEqual(got, expected)
253 testPL(5, 2, (), [0, 1, 2, 3, 4])
254 testPL(5, 2, (1,), [1, 2, 3, 4])
255 testPL(5, 2, (2,), [2, 3, 4])
256 testPL(5, 2, (4,), [4])
257 testPL(5, 2, (0, 3), [0, 1, 2])
258 testPL(5, 2, (1, 4), [1, 2, 3])
259 testPL(5, 2, (2, 99), [2, 3, 4])
260 testPL(5, 2, (20, 99), [])
262 def test_struct_unpack(self):
263 self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
265 def test_read_batch_urls(self):
266 f = io.StringIO('''\xef\xbb\xbf foo
269 # More after this line\r
272 self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
274 def test_urlencode_postdata(self):
275 data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
276 self.assertTrue(isinstance(data, bytes))
278 def test_parse_iso8601(self):
279 self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
280 self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
281 self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
282 self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
284 def test_strip_jsonp(self):
285 stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
286 d = json.loads(stripped)
287 self.assertEqual(d, [{"id": "532cb", "x": 3}])
289 stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
290 d = json.loads(stripped)
291 self.assertEqual(d, {'STATUS': 'OK'})
293 def test_uppercase_escape(self):
294 self.assertEqual(uppercase_escape('aä'), 'aä')
295 self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
297 def test_limit_length(self):
298 self.assertEqual(limit_length(None, 12), None)
299 self.assertEqual(limit_length('foo', 12), 'foo')
301 limit_length('foo bar baz asd', 12).startswith('foo bar'))
302 self.assertTrue('...' in limit_length('foo bar baz asd', 12))
304 def test_escape_rfc3986(self):
305 reserved = "!*'();:@&=+$,/?#[]"
306 unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
307 self.assertEqual(escape_rfc3986(reserved), reserved)
308 self.assertEqual(escape_rfc3986(unreserved), unreserved)
309 self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
310 self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
311 self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
312 self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
314 def test_escape_url(self):
316 escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
317 'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
320 escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
321 'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
324 escape_url('http://тест.рф/фрагмент'),
325 'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
328 escape_url('http://тест.рф/абв?абв=абв#абв'),
329 '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'
331 self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
333 def test_js_to_json_realworld(self):
335 'clip':{'provider':'pseudo'}
337 self.assertEqual(js_to_json(inp), '''{
338 "clip":{"provider":"pseudo"}
340 json.loads(js_to_json(inp))
343 'playlist':[{'controls':{'all':null}}]
345 self.assertEqual(js_to_json(inp), '''{
346 "playlist":[{"controls":{"all":null}}]
349 def test_js_to_json_edgecases(self):
350 on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
351 self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
353 on = js_to_json('{"abc": true}')
354 self.assertEqual(json.loads(on), {'abc': True})
356 def test_clean_html(self):
357 self.assertEqual(clean_html('a:\nb'), 'a: b')
358 self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
360 def test_intlist_to_bytes(self):
362 intlist_to_bytes([0, 1, 127, 128, 255]),
363 b'\x00\x01\x7f\x80\xff')
365 def test_args_to_str(self):
367 args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
368 'foo ba/r -baz \'2 be\' \'\''
371 if __name__ == '__main__':