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

   1 #!/usr/bin/env python
   2 # coding: utf-8
   3
   4 # Allow direct execution
   5 import os
   6 import sys
   7 import unittest
   8 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
   9
  10
  11 # Various small unit tests
  12 import xml.etree.ElementTree
  13
  14 #from youtube_dl.utils import htmlentity_transform
  15 from youtube_dl.utils import (
  16     DateRange,
  17     encodeFilename,
  18     find_xpath_attr,
  19     get_meta_content,
  20     orderedSet,
  21     sanitize_filename,
  22     shell_quote,
  23     smuggle_url,
  24     str_to_int,
  25     timeconvert,
  26     unescapeHTML,
  27     unified_strdate,
  28     unsmuggle_url,
  29     url_basename,
  30     xpath_with_ns,
  31 )
  32
  33 if sys.version_info < (3, 0):
  34     _compat_str = lambda b: b.decode('unicode-escape')
  35 else:
  36     _compat_str = lambda s: s
  37
  38
  39 class TestUtil(unittest.TestCase):
  40     def test_timeconvert(self):
  41         self.assertTrue(timeconvert('') is None)
  42         self.assertTrue(timeconvert('bougrg') is None)
  43
  44     def test_sanitize_filename(self):
  45         self.assertEqual(sanitize_filename('abc'), 'abc')
  46         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  47
  48         self.assertEqual(sanitize_filename('123'), '123')
  49
  50         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  51         self.assertFalse('/' in sanitize_filename('abc/de///'))
  52
  53         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  54         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  55         self.assertEqual('yes no', sanitize_filename('yes? no'))
  56         self.assertEqual('this - that', sanitize_filename('this: that'))
  57
  58         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  59         aumlaut = _compat_str('\xe4')
  60         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  61         tests = _compat_str('\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430')
  62         self.assertEqual(sanitize_filename(tests), tests)
  63
  64         forbidden = '"\0\\/'
  65         for fc in forbidden:
  66             for fbc in forbidden:
  67                 self.assertTrue(fbc not in sanitize_filename(fc))
  68
  69     def test_sanitize_filename_restricted(self):
  70         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  71         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  72
  73         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  74
  75         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  76         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  77
  78         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  79         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  80         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  81         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  82
  83         tests = _compat_str('a\xe4b\u4e2d\u56fd\u7684c')
  84         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
  85         self.assertTrue(sanitize_filename(_compat_str('\xf6'), restricted=True) != '')  # No empty filename
  86
  87         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  88         for fc in forbidden:
  89             for fbc in forbidden:
  90                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
  91
  92         # Handle a common case more neatly
  93         self.assertEqual(sanitize_filename(_compat_str('\u5927\u58f0\u5e26 - Song'), restricted=True), 'Song')
  94         self.assertEqual(sanitize_filename(_compat_str('\u603b\u7edf: Speech'), restricted=True), 'Speech')
  95         # .. but make sure the file name is never empty
  96         self.assertTrue(sanitize_filename('-', restricted=True) != '')
  97         self.assertTrue(sanitize_filename(':', restricted=True) != '')
  98
  99     def test_sanitize_ids(self):
 100         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 101         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 102         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 103
 104     def test_ordered_set(self):
 105         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 106         self.assertEqual(orderedSet([]), [])
 107         self.assertEqual(orderedSet([1]), [1])
 108         #keep the list ordered
 109         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 110
 111     def test_unescape_html(self):
 112         self.assertEqual(unescapeHTML(_compat_str('%20;')), _compat_str('%20;'))
 113
 114     def test_daterange(self):
 115         _20century = DateRange("19000101","20000101")
 116         self.assertFalse("17890714" in _20century)
 117         _ac = DateRange("00010101")
 118         self.assertTrue("19690721" in _ac)
 119         _firstmilenium = DateRange(end="10000101")
 120         self.assertTrue("07110427" in _firstmilenium)
 121
 122     def test_unified_dates(self):
 123         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 124         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 125         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 126         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 127
 128     def test_find_xpath_attr(self):
 129         testxml = u'''<root>
 130             <node/>
 131             <node x="a"/>
 132             <node x="a" y="c" />
 133             <node x="b" y="d" />
 134         </root>'''
 135         doc = xml.etree.ElementTree.fromstring(testxml)
 136
 137         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 138         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 139         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 140
 141     def test_meta_parser(self):
 142         testhtml = u'''
 143         <head>
 144             <meta name="description" content="foo &amp; bar">
 145             <meta content='Plato' name='author'/>
 146         </head>
 147         '''
 148         get_meta = lambda name: get_meta_content(name, testhtml)
 149         self.assertEqual(get_meta('description'), u'foo & bar')
 150         self.assertEqual(get_meta('author'), 'Plato')
 151
 152     def test_xpath_with_ns(self):
 153         testxml = u'''<root xmlns:media="http://example.com/">
 154             <media:song>
 155                 <media:author>The Author</media:author>
 156                 <url>http://server.com/download.mp3</url>
 157             </media:song>
 158         </root>'''
 159         doc = xml.etree.ElementTree.fromstring(testxml)
 160         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 161         self.assertTrue(find('media:song') is not None)
 162         self.assertEqual(find('media:song/media:author').text, u'The Author')
 163         self.assertEqual(find('media:song/url').text, u'http://server.com/download.mp3')
 164
 165     def test_smuggle_url(self):
 166         data = {u"ö": u"ö", u"abc": [3]}
 167         url = 'https://foo.bar/baz?x=y#a'
 168         smug_url = smuggle_url(url, data)
 169         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 170         self.assertEqual(url, unsmug_url)
 171         self.assertEqual(data, unsmug_data)
 172
 173         res_url, res_data = unsmuggle_url(url)
 174         self.assertEqual(res_url, url)
 175         self.assertEqual(res_data, None)
 176
 177     def test_shell_quote(self):
 178         args = ['ffmpeg', '-i', encodeFilename(u'ñ€ß\'.mp4')]
 179         self.assertEqual(shell_quote(args), u"""ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 180
 181     def test_str_to_int(self):
 182         self.assertEqual(str_to_int('123,456'), 123456)
 183         self.assertEqual(str_to_int('123.456'), 123456)
 184
 185     def test_url_basename(self):
 186         self.assertEqual(url_basename(u'http://foo.de/'), u'')
 187         self.assertEqual(url_basename(u'http://foo.de/bar/baz'), u'baz')
 188         self.assertEqual(url_basename(u'http://foo.de/bar/baz?x=y'), u'baz')
 189         self.assertEqual(url_basename(u'http://foo.de/bar/baz#x=y'), u'baz')
 190         self.assertEqual(url_basename(u'http://foo.de/bar/baz/'), u'baz')
 191         self.assertEqual(
 192             url_basename(u'http://media.w3.org/2010/05/sintel/trailer.mp4'),
 193             u'trailer.mp4')
 194
 195 if __name__ == '__main__':
 196     unittest.main()