_ 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     fix_xml_ampersands,
  20     get_meta_content,
  21     orderedSet,
  22     parse_duration,
  23     sanitize_filename,
  24     shell_quote,
  25     smuggle_url,
  26     str_to_int,
  27     timeconvert,
  28     unescapeHTML,
  29     unified_strdate,
  30     unsmuggle_url,
  31     url_basename,
  32     xpath_with_ns,
  33 )
  34
  35 if sys.version_info < (3, 0):
  36     _compat_str = lambda b: b.decode('unicode-escape')
  37 else:
  38     _compat_str = lambda s: s
  39
  40
  41 class TestUtil(unittest.TestCase):
  42     def test_timeconvert(self):
  43         self.assertTrue(timeconvert('') is None)
  44         self.assertTrue(timeconvert('bougrg') is None)
  45
  46     def test_sanitize_filename(self):
  47         self.assertEqual(sanitize_filename('abc'), 'abc')
  48         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  49
  50         self.assertEqual(sanitize_filename('123'), '123')
  51
  52         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  53         self.assertFalse('/' in sanitize_filename('abc/de///'))
  54
  55         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  56         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  57         self.assertEqual('yes no', sanitize_filename('yes? no'))
  58         self.assertEqual('this - that', sanitize_filename('this: that'))
  59
  60         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  61         aumlaut = _compat_str('\xe4')
  62         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  63         tests = _compat_str('\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430')
  64         self.assertEqual(sanitize_filename(tests), tests)
  65
  66         forbidden = '"\0\\/'
  67         for fc in forbidden:
  68             for fbc in forbidden:
  69                 self.assertTrue(fbc not in sanitize_filename(fc))
  70
  71     def test_sanitize_filename_restricted(self):
  72         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  73         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  74
  75         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  76
  77         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  78         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  79
  80         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  81         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  82         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  83         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  84
  85         tests = _compat_str('a\xe4b\u4e2d\u56fd\u7684c')
  86         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
  87         self.assertTrue(sanitize_filename(_compat_str('\xf6'), restricted=True) != '')  # No empty filename
  88
  89         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  90         for fc in forbidden:
  91             for fbc in forbidden:
  92                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
  93
  94         # Handle a common case more neatly
  95         self.assertEqual(sanitize_filename(_compat_str('\u5927\u58f0\u5e26 - Song'), restricted=True), 'Song')
  96         self.assertEqual(sanitize_filename(_compat_str('\u603b\u7edf: Speech'), restricted=True), 'Speech')
  97         # .. but make sure the file name is never empty
  98         self.assertTrue(sanitize_filename('-', restricted=True) != '')
  99         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 100
 101     def test_sanitize_ids(self):
 102         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 103         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 104         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 105
 106     def test_ordered_set(self):
 107         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 108         self.assertEqual(orderedSet([]), [])
 109         self.assertEqual(orderedSet([1]), [1])
 110         #keep the list ordered
 111         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 112
 113     def test_unescape_html(self):
 114         self.assertEqual(unescapeHTML(_compat_str('%20;')), _compat_str('%20;'))
 115
 116     def test_daterange(self):
 117         _20century = DateRange("19000101","20000101")
 118         self.assertFalse("17890714" in _20century)
 119         _ac = DateRange("00010101")
 120         self.assertTrue("19690721" in _ac)
 121         _firstmilenium = DateRange(end="10000101")
 122         self.assertTrue("07110427" in _firstmilenium)
 123
 124     def test_unified_dates(self):
 125         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 126         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 127         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 128         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 129
 130     def test_find_xpath_attr(self):
 131         testxml = u'''<root>
 132             <node/>
 133             <node x="a"/>
 134             <node x="a" y="c" />
 135             <node x="b" y="d" />
 136         </root>'''
 137         doc = xml.etree.ElementTree.fromstring(testxml)
 138
 139         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 140         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 141         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 142
 143     def test_meta_parser(self):
 144         testhtml = u'''
 145         <head>
 146             <meta name="description" content="foo &amp; bar">
 147             <meta content='Plato' name='author'/>
 148         </head>
 149         '''
 150         get_meta = lambda name: get_meta_content(name, testhtml)
 151         self.assertEqual(get_meta('description'), u'foo & bar')
 152         self.assertEqual(get_meta('author'), 'Plato')
 153
 154     def test_xpath_with_ns(self):
 155         testxml = u'''<root xmlns:media="http://example.com/">
 156             <media:song>
 157                 <media:author>The Author</media:author>
 158                 <url>http://server.com/download.mp3</url>
 159             </media:song>
 160         </root>'''
 161         doc = xml.etree.ElementTree.fromstring(testxml)
 162         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 163         self.assertTrue(find('media:song') is not None)
 164         self.assertEqual(find('media:song/media:author').text, u'The Author')
 165         self.assertEqual(find('media:song/url').text, u'http://server.com/download.mp3')
 166
 167     def test_smuggle_url(self):
 168         data = {u"ö": u"ö", u"abc": [3]}
 169         url = 'https://foo.bar/baz?x=y#a'
 170         smug_url = smuggle_url(url, data)
 171         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 172         self.assertEqual(url, unsmug_url)
 173         self.assertEqual(data, unsmug_data)
 174
 175         res_url, res_data = unsmuggle_url(url)
 176         self.assertEqual(res_url, url)
 177         self.assertEqual(res_data, None)
 178
 179     def test_shell_quote(self):
 180         args = ['ffmpeg', '-i', encodeFilename(u'ñ€ß\'.mp4')]
 181         self.assertEqual(shell_quote(args), u"""ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 182
 183     def test_str_to_int(self):
 184         self.assertEqual(str_to_int('123,456'), 123456)
 185         self.assertEqual(str_to_int('123.456'), 123456)
 186
 187     def test_url_basename(self):
 188         self.assertEqual(url_basename(u'http://foo.de/'), u'')
 189         self.assertEqual(url_basename(u'http://foo.de/bar/baz'), u'baz')
 190         self.assertEqual(url_basename(u'http://foo.de/bar/baz?x=y'), u'baz')
 191         self.assertEqual(url_basename(u'http://foo.de/bar/baz#x=y'), u'baz')
 192         self.assertEqual(url_basename(u'http://foo.de/bar/baz/'), u'baz')
 193         self.assertEqual(
 194             url_basename(u'http://media.w3.org/2010/05/sintel/trailer.mp4'),
 195             u'trailer.mp4')
 196
 197     def test_parse_duration(self):
 198         self.assertEqual(parse_duration(None), None)
 199         self.assertEqual(parse_duration('1'), 1)
 200         self.assertEqual(parse_duration('1337:12'), 80232)
 201         self.assertEqual(parse_duration('9:12:43'), 33163)
 202         self.assertEqual(parse_duration('x:y'), None)
 203
 204     def test_fix_xml_ampersands(self):
 205         self.assertEqual(
 206             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 207         self.assertEqual(
 208             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 209             '"&amp;x=y&amp;wrong;&amp;z=a')
 210         self.assertEqual(
 211             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 212             '&amp;&apos;&gt;&lt;&quot;')
 213         self.assertEqual(
 214             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 215         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 216
 217 if __name__ == '__main__':
 218     unittest.main()