_ 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 io
  13 import json
  14 import xml.etree.ElementTree
  15
  16 #from youtube_dl.utils import htmlentity_transform
  17 from youtube_dl.utils import (
  18     DateRange,
  19     encodeFilename,
  20     find_xpath_attr,
  21     fix_xml_ampersands,
  22     get_meta_content,
  23     orderedSet,
  24     PagedList,
  25     parse_duration,
  26     read_batch_urls,
  27     sanitize_filename,
  28     shell_quote,
  29     smuggle_url,
  30     str_to_int,
  31     struct_unpack,
  32     timeconvert,
  33     unescapeHTML,
  34     unified_strdate,
  35     unsmuggle_url,
  36     url_basename,
  37     urlencode_postdata,
  38     xpath_with_ns,
  39     parse_iso8601,
  40     strip_jsonp,
  41 )
  42
  43 if sys.version_info < (3, 0):
  44     _compat_str = lambda b: b.decode('unicode-escape')
  45 else:
  46     _compat_str = lambda s: s
  47
  48
  49 class TestUtil(unittest.TestCase):
  50     def test_timeconvert(self):
  51         self.assertTrue(timeconvert('') is None)
  52         self.assertTrue(timeconvert('bougrg') is None)
  53
  54     def test_sanitize_filename(self):
  55         self.assertEqual(sanitize_filename('abc'), 'abc')
  56         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  57
  58         self.assertEqual(sanitize_filename('123'), '123')
  59
  60         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  61         self.assertFalse('/' in sanitize_filename('abc/de///'))
  62
  63         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  64         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  65         self.assertEqual('yes no', sanitize_filename('yes? no'))
  66         self.assertEqual('this - that', sanitize_filename('this: that'))
  67
  68         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  69         aumlaut = _compat_str('\xe4')
  70         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  71         tests = _compat_str('\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430')
  72         self.assertEqual(sanitize_filename(tests), tests)
  73
  74         forbidden = '"\0\\/'
  75         for fc in forbidden:
  76             for fbc in forbidden:
  77                 self.assertTrue(fbc not in sanitize_filename(fc))
  78
  79     def test_sanitize_filename_restricted(self):
  80         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  81         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  82
  83         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  84
  85         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  86         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  87
  88         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  89         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  90         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  91         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  92
  93         tests = _compat_str('a\xe4b\u4e2d\u56fd\u7684c')
  94         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
  95         self.assertTrue(sanitize_filename(_compat_str('\xf6'), restricted=True) != '')  # No empty filename
  96
  97         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  98         for fc in forbidden:
  99             for fbc in forbidden:
 100                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 101
 102         # Handle a common case more neatly
 103         self.assertEqual(sanitize_filename(_compat_str('\u5927\u58f0\u5e26 - Song'), restricted=True), 'Song')
 104         self.assertEqual(sanitize_filename(_compat_str('\u603b\u7edf: Speech'), restricted=True), 'Speech')
 105         # .. but make sure the file name is never empty
 106         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 107         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 108
 109     def test_sanitize_ids(self):
 110         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 111         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 112         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 113
 114     def test_ordered_set(self):
 115         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 116         self.assertEqual(orderedSet([]), [])
 117         self.assertEqual(orderedSet([1]), [1])
 118         #keep the list ordered
 119         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 120
 121     def test_unescape_html(self):
 122         self.assertEqual(unescapeHTML(_compat_str('%20;')), _compat_str('%20;'))
 123
 124     def test_daterange(self):
 125         _20century = DateRange("19000101","20000101")
 126         self.assertFalse("17890714" in _20century)
 127         _ac = DateRange("00010101")
 128         self.assertTrue("19690721" in _ac)
 129         _firstmilenium = DateRange(end="10000101")
 130         self.assertTrue("07110427" in _firstmilenium)
 131
 132     def test_unified_dates(self):
 133         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 134         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 135         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 136         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 137         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 138
 139     def test_find_xpath_attr(self):
 140         testxml = u'''<root>
 141             <node/>
 142             <node x="a"/>
 143             <node x="a" y="c" />
 144             <node x="b" y="d" />
 145         </root>'''
 146         doc = xml.etree.ElementTree.fromstring(testxml)
 147
 148         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 149         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 150         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 151
 152     def test_meta_parser(self):
 153         testhtml = u'''
 154         <head>
 155             <meta name="description" content="foo &amp; bar">
 156             <meta content='Plato' name='author'/>
 157         </head>
 158         '''
 159         get_meta = lambda name: get_meta_content(name, testhtml)
 160         self.assertEqual(get_meta('description'), u'foo & bar')
 161         self.assertEqual(get_meta('author'), 'Plato')
 162
 163     def test_xpath_with_ns(self):
 164         testxml = u'''<root xmlns:media="http://example.com/">
 165             <media:song>
 166                 <media:author>The Author</media:author>
 167                 <url>http://server.com/download.mp3</url>
 168             </media:song>
 169         </root>'''
 170         doc = xml.etree.ElementTree.fromstring(testxml)
 171         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 172         self.assertTrue(find('media:song') is not None)
 173         self.assertEqual(find('media:song/media:author').text, u'The Author')
 174         self.assertEqual(find('media:song/url').text, u'http://server.com/download.mp3')
 175
 176     def test_smuggle_url(self):
 177         data = {u"ö": u"ö", u"abc": [3]}
 178         url = 'https://foo.bar/baz?x=y#a'
 179         smug_url = smuggle_url(url, data)
 180         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 181         self.assertEqual(url, unsmug_url)
 182         self.assertEqual(data, unsmug_data)
 183
 184         res_url, res_data = unsmuggle_url(url)
 185         self.assertEqual(res_url, url)
 186         self.assertEqual(res_data, None)
 187
 188     def test_shell_quote(self):
 189         args = ['ffmpeg', '-i', encodeFilename(u'ñ€ß\'.mp4')]
 190         self.assertEqual(shell_quote(args), u"""ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 191
 192     def test_str_to_int(self):
 193         self.assertEqual(str_to_int('123,456'), 123456)
 194         self.assertEqual(str_to_int('123.456'), 123456)
 195
 196     def test_url_basename(self):
 197         self.assertEqual(url_basename(u'http://foo.de/'), u'')
 198         self.assertEqual(url_basename(u'http://foo.de/bar/baz'), u'baz')
 199         self.assertEqual(url_basename(u'http://foo.de/bar/baz?x=y'), u'baz')
 200         self.assertEqual(url_basename(u'http://foo.de/bar/baz#x=y'), u'baz')
 201         self.assertEqual(url_basename(u'http://foo.de/bar/baz/'), u'baz')
 202         self.assertEqual(
 203             url_basename(u'http://media.w3.org/2010/05/sintel/trailer.mp4'),
 204             u'trailer.mp4')
 205
 206     def test_parse_duration(self):
 207         self.assertEqual(parse_duration(None), None)
 208         self.assertEqual(parse_duration('1'), 1)
 209         self.assertEqual(parse_duration('1337:12'), 80232)
 210         self.assertEqual(parse_duration('9:12:43'), 33163)
 211         self.assertEqual(parse_duration('12:00'), 720)
 212         self.assertEqual(parse_duration('00:01:01'), 61)
 213         self.assertEqual(parse_duration('x:y'), None)
 214         self.assertEqual(parse_duration('3h11m53s'), 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
 222     def test_fix_xml_ampersands(self):
 223         self.assertEqual(
 224             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 225         self.assertEqual(
 226             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 227             '"&amp;x=y&amp;wrong;&amp;z=a')
 228         self.assertEqual(
 229             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 230             '&amp;&apos;&gt;&lt;&quot;')
 231         self.assertEqual(
 232             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 233         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 234
 235     def test_paged_list(self):
 236         def testPL(size, pagesize, sliceargs, expected):
 237             def get_page(pagenum):
 238                 firstid = pagenum * pagesize
 239                 upto = min(size, pagenum * pagesize + pagesize)
 240                 for i in range(firstid, upto):
 241                     yield i
 242
 243             pl = PagedList(get_page, pagesize)
 244             got = pl.getslice(*sliceargs)
 245             self.assertEqual(got, expected)
 246
 247         testPL(5, 2, (), [0, 1, 2, 3, 4])
 248         testPL(5, 2, (1,), [1, 2, 3, 4])
 249         testPL(5, 2, (2,), [2, 3, 4])
 250         testPL(5, 2, (4,), [4])
 251         testPL(5, 2, (0, 3), [0, 1, 2])
 252         testPL(5, 2, (1, 4), [1, 2, 3])
 253         testPL(5, 2, (2, 99), [2, 3, 4])
 254         testPL(5, 2, (20, 99), [])
 255
 256     def test_struct_unpack(self):
 257         self.assertEqual(struct_unpack(u'!B', b'\x00'), (0,))
 258
 259     def test_read_batch_urls(self):
 260         f = io.StringIO(u'''\xef\xbb\xbf foo
 261             bar\r
 262             baz
 263             # More after this line\r
 264             ; or after this
 265             bam''')
 266         self.assertEqual(read_batch_urls(f), [u'foo', u'bar', u'baz', u'bam'])
 267
 268     def test_urlencode_postdata(self):
 269         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 270         self.assertTrue(isinstance(data, bytes))
 271
 272     def test_parse_iso8601(self):
 273         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 274         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 275         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 276
 277     def test_strip_jsonp(self):
 278         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 279         d = json.loads(stripped)
 280         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 281
 282
 283 if __name__ == '__main__':
 284     unittest.main()