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

   1 #!/usr/bin/env python
   2 # coding: utf-8
   3
   4 from __future__ import unicode_literals
   5
   6 # Allow direct execution
   7 import os
   8 import sys
   9 import unittest
  10 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
  11
  12
  13 # Various small unit tests
  14 import io
  15 import json
  16 import xml.etree.ElementTree
  17
  18 from youtube_dl.utils import (
  19     DateRange,
  20     encodeFilename,
  21     find_xpath_attr,
  22     fix_xml_ampersands,
  23     get_meta_content,
  24     orderedSet,
  25     PagedList,
  26     parse_duration,
  27     read_batch_urls,
  28     sanitize_filename,
  29     shell_quote,
  30     smuggle_url,
  31     str_to_int,
  32     struct_unpack,
  33     timeconvert,
  34     unescapeHTML,
  35     unified_strdate,
  36     unsmuggle_url,
  37     url_basename,
  38     urlencode_postdata,
  39     xpath_with_ns,
  40     parse_iso8601,
  41     strip_jsonp,
  42     uppercase_escape,
  43     limit_length,
  44 )
  45
  46
  47 class TestUtil(unittest.TestCase):
  48     def test_timeconvert(self):
  49         self.assertTrue(timeconvert('') is None)
  50         self.assertTrue(timeconvert('bougrg') is None)
  51
  52     def test_sanitize_filename(self):
  53         self.assertEqual(sanitize_filename('abc'), 'abc')
  54         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  55
  56         self.assertEqual(sanitize_filename('123'), '123')
  57
  58         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  59         self.assertFalse('/' in sanitize_filename('abc/de///'))
  60
  61         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  62         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  63         self.assertEqual('yes no', sanitize_filename('yes? no'))
  64         self.assertEqual('this - that', sanitize_filename('this: that'))
  65
  66         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  67         aumlaut = 'ä'
  68         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  69         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
  70         self.assertEqual(sanitize_filename(tests), tests)
  71
  72         forbidden = '"\0\\/'
  73         for fc in forbidden:
  74             for fbc in forbidden:
  75                 self.assertTrue(fbc not in sanitize_filename(fc))
  76
  77     def test_sanitize_filename_restricted(self):
  78         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  79         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  80
  81         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  82
  83         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  84         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  85
  86         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  87         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  88         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  89         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  90
  91         tests = 'a\xe4b\u4e2d\u56fd\u7684c'
  92         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
  93         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
  94
  95         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  96         for fc in forbidden:
  97             for fbc in forbidden:
  98                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
  99
 100         # Handle a common case more neatly
 101         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 102         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 103         # .. but make sure the file name is never empty
 104         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 105         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 106
 107     def test_sanitize_ids(self):
 108         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 109         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 110         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 111
 112     def test_ordered_set(self):
 113         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 114         self.assertEqual(orderedSet([]), [])
 115         self.assertEqual(orderedSet([1]), [1])
 116         #keep the list ordered
 117         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 118
 119     def test_unescape_html(self):
 120         self.assertEqual(unescapeHTML('%20;'), '%20;')
 121         self.assertEqual(
 122             unescapeHTML('&eacute;'), 'é')
 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 = '''<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 = '''
 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'), 'foo & bar')
 161         self.assertEqual(get_meta('author'), 'Plato')
 162
 163     def test_xpath_with_ns(self):
 164         testxml = '''<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, 'The Author')
 174         self.assertEqual(find('media:song/url').text, '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('ñ€ß\'.mp4')]
 190         self.assertEqual(shell_quote(args), """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('http://foo.de/'), '')
 198         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 199         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 200         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 201         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 202         self.assertEqual(
 203             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 204             '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('3h 11m 53s'), 11513)
 216         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 217         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 218         self.assertEqual(parse_duration('62m45s'), 3765)
 219         self.assertEqual(parse_duration('6m59s'), 419)
 220         self.assertEqual(parse_duration('49s'), 49)
 221         self.assertEqual(parse_duration('0h0m0s'), 0)
 222         self.assertEqual(parse_duration('0m0s'), 0)
 223         self.assertEqual(parse_duration('0s'), 0)
 224         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 225
 226     def test_fix_xml_ampersands(self):
 227         self.assertEqual(
 228             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 229         self.assertEqual(
 230             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 231             '"&amp;x=y&amp;wrong;&amp;z=a')
 232         self.assertEqual(
 233             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 234             '&amp;&apos;&gt;&lt;&quot;')
 235         self.assertEqual(
 236             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 237         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 238
 239     def test_paged_list(self):
 240         def testPL(size, pagesize, sliceargs, expected):
 241             def get_page(pagenum):
 242                 firstid = pagenum * pagesize
 243                 upto = min(size, pagenum * pagesize + pagesize)
 244                 for i in range(firstid, upto):
 245                     yield i
 246
 247             pl = PagedList(get_page, pagesize)
 248             got = pl.getslice(*sliceargs)
 249             self.assertEqual(got, expected)
 250
 251         testPL(5, 2, (), [0, 1, 2, 3, 4])
 252         testPL(5, 2, (1,), [1, 2, 3, 4])
 253         testPL(5, 2, (2,), [2, 3, 4])
 254         testPL(5, 2, (4,), [4])
 255         testPL(5, 2, (0, 3), [0, 1, 2])
 256         testPL(5, 2, (1, 4), [1, 2, 3])
 257         testPL(5, 2, (2, 99), [2, 3, 4])
 258         testPL(5, 2, (20, 99), [])
 259
 260     def test_struct_unpack(self):
 261         self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
 262
 263     def test_read_batch_urls(self):
 264         f = io.StringIO('''\xef\xbb\xbf foo
 265             bar\r
 266             baz
 267             # More after this line\r
 268             ; or after this
 269             bam''')
 270         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 271
 272     def test_urlencode_postdata(self):
 273         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 274         self.assertTrue(isinstance(data, bytes))
 275
 276     def test_parse_iso8601(self):
 277         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 278         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 279         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 280
 281     def test_strip_jsonp(self):
 282         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 283         d = json.loads(stripped)
 284         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 285
 286     def test_uppercase_escape(self):
 287         self.assertEqual(uppercase_escape('aä'), 'aä')
 288         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 289
 290     def test_limit_length(self):
 291         self.assertEqual(limit_length(None, 12), None)
 292         self.assertEqual(limit_length('foo', 12), 'foo')
 293         self.assertTrue(
 294             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 295         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 296
 297 if __name__ == '__main__':
 298     unittest.main()