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     PagedList,
  23     parse_duration,
  24     sanitize_filename,
  25     shell_quote,
  26     smuggle_url,
  27     str_to_int,
  28     timeconvert,
  29     unescapeHTML,
  30     unified_strdate,
  31     unsmuggle_url,
  32     url_basename,
  33     xpath_with_ns,
  34 )
  35
  36 if sys.version_info < (3, 0):
  37     _compat_str = lambda b: b.decode('unicode-escape')
  38 else:
  39     _compat_str = lambda s: s
  40
  41
  42 class TestUtil(unittest.TestCase):
  43     def test_timeconvert(self):
  44         self.assertTrue(timeconvert('') is None)
  45         self.assertTrue(timeconvert('bougrg') is None)
  46
  47     def test_sanitize_filename(self):
  48         self.assertEqual(sanitize_filename('abc'), 'abc')
  49         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  50
  51         self.assertEqual(sanitize_filename('123'), '123')
  52
  53         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  54         self.assertFalse('/' in sanitize_filename('abc/de///'))
  55
  56         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  57         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  58         self.assertEqual('yes no', sanitize_filename('yes? no'))
  59         self.assertEqual('this - that', sanitize_filename('this: that'))
  60
  61         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  62         aumlaut = _compat_str('\xe4')
  63         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  64         tests = _compat_str('\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430')
  65         self.assertEqual(sanitize_filename(tests), tests)
  66
  67         forbidden = '"\0\\/'
  68         for fc in forbidden:
  69             for fbc in forbidden:
  70                 self.assertTrue(fbc not in sanitize_filename(fc))
  71
  72     def test_sanitize_filename_restricted(self):
  73         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  74         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  75
  76         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  77
  78         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  79         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  80
  81         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  82         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  83         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  84         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  85
  86         tests = _compat_str('a\xe4b\u4e2d\u56fd\u7684c')
  87         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
  88         self.assertTrue(sanitize_filename(_compat_str('\xf6'), restricted=True) != '')  # No empty filename
  89
  90         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  91         for fc in forbidden:
  92             for fbc in forbidden:
  93                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
  94
  95         # Handle a common case more neatly
  96         self.assertEqual(sanitize_filename(_compat_str('\u5927\u58f0\u5e26 - Song'), restricted=True), 'Song')
  97         self.assertEqual(sanitize_filename(_compat_str('\u603b\u7edf: Speech'), restricted=True), 'Speech')
  98         # .. but make sure the file name is never empty
  99         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 100         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 101
 102     def test_sanitize_ids(self):
 103         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 104         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 105         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 106
 107     def test_ordered_set(self):
 108         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 109         self.assertEqual(orderedSet([]), [])
 110         self.assertEqual(orderedSet([1]), [1])
 111         #keep the list ordered
 112         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 113
 114     def test_unescape_html(self):
 115         self.assertEqual(unescapeHTML(_compat_str('%20;')), _compat_str('%20;'))
 116
 117     def test_daterange(self):
 118         _20century = DateRange("19000101","20000101")
 119         self.assertFalse("17890714" in _20century)
 120         _ac = DateRange("00010101")
 121         self.assertTrue("19690721" in _ac)
 122         _firstmilenium = DateRange(end="10000101")
 123         self.assertTrue("07110427" in _firstmilenium)
 124
 125     def test_unified_dates(self):
 126         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 127         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 128         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 129         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 130
 131     def test_find_xpath_attr(self):
 132         testxml = u'''<root>
 133             <node/>
 134             <node x="a"/>
 135             <node x="a" y="c" />
 136             <node x="b" y="d" />
 137         </root>'''
 138         doc = xml.etree.ElementTree.fromstring(testxml)
 139
 140         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 141         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 142         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 143
 144     def test_meta_parser(self):
 145         testhtml = u'''
 146         <head>
 147             <meta name="description" content="foo &amp; bar">
 148             <meta content='Plato' name='author'/>
 149         </head>
 150         '''
 151         get_meta = lambda name: get_meta_content(name, testhtml)
 152         self.assertEqual(get_meta('description'), u'foo & bar')
 153         self.assertEqual(get_meta('author'), 'Plato')
 154
 155     def test_xpath_with_ns(self):
 156         testxml = u'''<root xmlns:media="http://example.com/">
 157             <media:song>
 158                 <media:author>The Author</media:author>
 159                 <url>http://server.com/download.mp3</url>
 160             </media:song>
 161         </root>'''
 162         doc = xml.etree.ElementTree.fromstring(testxml)
 163         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 164         self.assertTrue(find('media:song') is not None)
 165         self.assertEqual(find('media:song/media:author').text, u'The Author')
 166         self.assertEqual(find('media:song/url').text, u'http://server.com/download.mp3')
 167
 168     def test_smuggle_url(self):
 169         data = {u"ö": u"ö", u"abc": [3]}
 170         url = 'https://foo.bar/baz?x=y#a'
 171         smug_url = smuggle_url(url, data)
 172         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 173         self.assertEqual(url, unsmug_url)
 174         self.assertEqual(data, unsmug_data)
 175
 176         res_url, res_data = unsmuggle_url(url)
 177         self.assertEqual(res_url, url)
 178         self.assertEqual(res_data, None)
 179
 180     def test_shell_quote(self):
 181         args = ['ffmpeg', '-i', encodeFilename(u'ñ€ß\'.mp4')]
 182         self.assertEqual(shell_quote(args), u"""ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 183
 184     def test_str_to_int(self):
 185         self.assertEqual(str_to_int('123,456'), 123456)
 186         self.assertEqual(str_to_int('123.456'), 123456)
 187
 188     def test_url_basename(self):
 189         self.assertEqual(url_basename(u'http://foo.de/'), u'')
 190         self.assertEqual(url_basename(u'http://foo.de/bar/baz'), 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#x=y'), u'baz')
 193         self.assertEqual(url_basename(u'http://foo.de/bar/baz/'), u'baz')
 194         self.assertEqual(
 195             url_basename(u'http://media.w3.org/2010/05/sintel/trailer.mp4'),
 196             u'trailer.mp4')
 197
 198     def test_parse_duration(self):
 199         self.assertEqual(parse_duration(None), None)
 200         self.assertEqual(parse_duration('1'), 1)
 201         self.assertEqual(parse_duration('1337:12'), 80232)
 202         self.assertEqual(parse_duration('9:12:43'), 33163)
 203         self.assertEqual(parse_duration('x:y'), None)
 204
 205     def test_fix_xml_ampersands(self):
 206         self.assertEqual(
 207             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 208         self.assertEqual(
 209             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 210             '"&amp;x=y&amp;wrong;&amp;z=a')
 211         self.assertEqual(
 212             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 213             '&amp;&apos;&gt;&lt;&quot;')
 214         self.assertEqual(
 215             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 216         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 217
 218     def test_paged_list(self):
 219         def testPL(size, pagesize, sliceargs, expected):
 220             def get_page(pagenum):
 221                 firstid = pagenum * pagesize
 222                 upto = min(size, pagenum * pagesize + pagesize)
 223                 for i in range(firstid, upto):
 224                     yield i
 225
 226             pl = PagedList(get_page, pagesize)
 227             got = pl.getslice(*sliceargs)
 228             self.assertEqual(got, expected)
 229
 230         testPL(5, 2, (), [0, 1, 2, 3, 4])
 231         testPL(5, 2, (1,), [1, 2, 3, 4])
 232         testPL(5, 2, (2,), [2, 3, 4])
 233         testPL(5, 2, (4,), [4])
 234         testPL(5, 2, (0, 3), [0, 1, 2])
 235         testPL(5, 2, (1, 4), [1, 2, 3])
 236         testPL(5, 2, (2, 99), [2, 3, 4])
 237         testPL(5, 2, (20, 99), [])
 238
 239 if __name__ == '__main__':
 240     unittest.main()