_ Git - youtube-dl/blob - test/test_utils.py

   1 #!/usr/bin/env python
   2
   3 # Various small unit tests
   4
   5 import sys
   6 import unittest
   7
   8 # Allow direct execution
   9 import os
  10 sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12 #from youtube_dl.utils import htmlentity_transform
  13 from youtube_dl.utils import timeconvert
  14 from youtube_dl.utils import sanitize_filename
  15 from youtube_dl.utils import unescapeHTML
  16 from youtube_dl.utils import orderedSet
  17 from youtube_dl.utils import DateRange
  18 from youtube_dl.utils import unified_strdate
  19
  20 if sys.version_info < (3, 0):
  21     _compat_str = lambda b: b.decode('unicode-escape')
  22 else:
  23     _compat_str = lambda s: s
  24
  25
  26 class TestUtil(unittest.TestCase):
  27     def test_timeconvert(self):
  28         self.assertTrue(timeconvert('') is None)
  29         self.assertTrue(timeconvert('bougrg') is None)
  30
  31     def test_sanitize_filename(self):
  32         self.assertEqual(sanitize_filename('abc'), 'abc')
  33         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  34
  35         self.assertEqual(sanitize_filename('123'), '123')
  36
  37         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  38         self.assertFalse('/' in sanitize_filename('abc/de///'))
  39
  40         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  41         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  42         self.assertEqual('yes no', sanitize_filename('yes? no'))
  43         self.assertEqual('this - that', sanitize_filename('this: that'))
  44
  45         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  46         aumlaut = _compat_str('\xe4')
  47         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  48         tests = _compat_str('\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430')
  49         self.assertEqual(sanitize_filename(tests), tests)
  50
  51         forbidden = '"\0\\/'
  52         for fc in forbidden:
  53             for fbc in forbidden:
  54                 self.assertTrue(fbc not in sanitize_filename(fc))
  55
  56     def test_sanitize_filename_restricted(self):
  57         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  58         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  59
  60         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  61
  62         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  63         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  64
  65         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  66         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  67         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  68         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  69
  70         tests = _compat_str('a\xe4b\u4e2d\u56fd\u7684c')
  71         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
  72         self.assertTrue(sanitize_filename(_compat_str('\xf6'), restricted=True) != '')  # No empty filename
  73
  74         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  75         for fc in forbidden:
  76             for fbc in forbidden:
  77                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
  78
  79         # Handle a common case more neatly
  80         self.assertEqual(sanitize_filename(_compat_str('\u5927\u58f0\u5e26 - Song'), restricted=True), 'Song')
  81         self.assertEqual(sanitize_filename(_compat_str('\u603b\u7edf: Speech'), restricted=True), 'Speech')
  82         # .. but make sure the file name is never empty
  83         self.assertTrue(sanitize_filename('-', restricted=True) != '')
  84         self.assertTrue(sanitize_filename(':', restricted=True) != '')
  85
  86     def test_sanitize_ids(self):
  87         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
  88         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
  89         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
  90
  91     def test_ordered_set(self):
  92         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
  93         self.assertEqual(orderedSet([]), [])
  94         self.assertEqual(orderedSet([1]), [1])
  95         #keep the list ordered
  96         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
  97
  98     def test_unescape_html(self):
  99         self.assertEqual(unescapeHTML(_compat_str('%20;')), _compat_str('%20;'))
 100
 101     def test_daterange(self):
 102         _20century = DateRange("19000101","20000101")
 103         self.assertFalse("17890714" in _20century)
 104         _ac = DateRange("00010101")
 105         self.assertTrue("19690721" in _ac)
 106         _firstmilenium = DateRange(end="10000101")
 107         self.assertTrue("07110427" in _firstmilenium)
 108
 109     def test_unified_dates(self):
 110         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 111         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 112         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 113         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 114
 115 if __name__ == '__main__':
 116     unittest.main()