_ 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     clean_html,
  20     DateRange,
  21     encodeFilename,
  22     find_xpath_attr,
  23     fix_xml_ampersands,
  24     orderedSet,
  25     OnDemandPagedList,
  26     InAdvancePagedList,
  27     parse_duration,
  28     read_batch_urls,
  29     sanitize_filename,
  30     shell_quote,
  31     smuggle_url,
  32     str_to_int,
  33     struct_unpack,
  34     timeconvert,
  35     unescapeHTML,
  36     unified_strdate,
  37     unsmuggle_url,
  38     url_basename,
  39     urlencode_postdata,
  40     xpath_with_ns,
  41     parse_iso8601,
  42     strip_jsonp,
  43     uppercase_escape,
  44     limit_length,
  45     escape_rfc3986,
  46     escape_url,
  47     js_to_json,
  48     intlist_to_bytes,
  49     args_to_str,
  50     parse_filesize,
  51     version_tuple,
  52 )
  53
  54
  55 class TestUtil(unittest.TestCase):
  56     def test_timeconvert(self):
  57         self.assertTrue(timeconvert('') is None)
  58         self.assertTrue(timeconvert('bougrg') is None)
  59
  60     def test_sanitize_filename(self):
  61         self.assertEqual(sanitize_filename('abc'), 'abc')
  62         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  63
  64         self.assertEqual(sanitize_filename('123'), '123')
  65
  66         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  67         self.assertFalse('/' in sanitize_filename('abc/de///'))
  68
  69         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  70         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  71         self.assertEqual('yes no', sanitize_filename('yes? no'))
  72         self.assertEqual('this - that', sanitize_filename('this: that'))
  73
  74         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  75         aumlaut = 'ä'
  76         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  77         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
  78         self.assertEqual(sanitize_filename(tests), tests)
  79
  80         forbidden = '"\0\\/'
  81         for fc in forbidden:
  82             for fbc in forbidden:
  83                 self.assertTrue(fbc not in sanitize_filename(fc))
  84
  85     def test_sanitize_filename_restricted(self):
  86         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  87         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  88
  89         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  90
  91         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  92         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  93
  94         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  95         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  96         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  97         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  98
  99         tests = 'a\xe4b\u4e2d\u56fd\u7684c'
 100         self.assertEqual(sanitize_filename(tests, restricted=True), 'a_b_c')
 101         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
 102
 103         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
 104         for fc in forbidden:
 105             for fbc in forbidden:
 106                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
 107
 108         # Handle a common case more neatly
 109         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
 110         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
 111         # .. but make sure the file name is never empty
 112         self.assertTrue(sanitize_filename('-', restricted=True) != '')
 113         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 114
 115     def test_sanitize_ids(self):
 116         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
 117         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
 118         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
 119
 120     def test_ordered_set(self):
 121         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
 122         self.assertEqual(orderedSet([]), [])
 123         self.assertEqual(orderedSet([1]), [1])
 124         # keep the list ordered
 125         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
 126
 127     def test_unescape_html(self):
 128         self.assertEqual(unescapeHTML('%20;'), '%20;')
 129         self.assertEqual(
 130             unescapeHTML('&eacute;'), 'é')
 131
 132     def test_daterange(self):
 133         _20century = DateRange("19000101", "20000101")
 134         self.assertFalse("17890714" in _20century)
 135         _ac = DateRange("00010101")
 136         self.assertTrue("19690721" in _ac)
 137         _firstmilenium = DateRange(end="10000101")
 138         self.assertTrue("07110427" in _firstmilenium)
 139
 140     def test_unified_dates(self):
 141         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
 142         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
 143         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
 144         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
 145         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
 146         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
 147         self.assertEqual(
 148             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
 149             '20141126')
 150
 151     def test_find_xpath_attr(self):
 152         testxml = '''<root>
 153             <node/>
 154             <node x="a"/>
 155             <node x="a" y="c" />
 156             <node x="b" y="d" />
 157         </root>'''
 158         doc = xml.etree.ElementTree.fromstring(testxml)
 159
 160         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
 161         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
 162         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
 163
 164     def test_xpath_with_ns(self):
 165         testxml = '''<root xmlns:media="http://example.com/">
 166             <media:song>
 167                 <media:author>The Author</media:author>
 168                 <url>http://server.com/download.mp3</url>
 169             </media:song>
 170         </root>'''
 171         doc = xml.etree.ElementTree.fromstring(testxml)
 172         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
 173         self.assertTrue(find('media:song') is not None)
 174         self.assertEqual(find('media:song/media:author').text, 'The Author')
 175         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
 176
 177     def test_smuggle_url(self):
 178         data = {"ö": "ö", "abc": [3]}
 179         url = 'https://foo.bar/baz?x=y#a'
 180         smug_url = smuggle_url(url, data)
 181         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
 182         self.assertEqual(url, unsmug_url)
 183         self.assertEqual(data, unsmug_data)
 184
 185         res_url, res_data = unsmuggle_url(url)
 186         self.assertEqual(res_url, url)
 187         self.assertEqual(res_data, None)
 188
 189     def test_shell_quote(self):
 190         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
 191         self.assertEqual(shell_quote(args), """ffmpeg -i 'ñ€ß'"'"'.mp4'""")
 192
 193     def test_str_to_int(self):
 194         self.assertEqual(str_to_int('123,456'), 123456)
 195         self.assertEqual(str_to_int('123.456'), 123456)
 196
 197     def test_url_basename(self):
 198         self.assertEqual(url_basename('http://foo.de/'), '')
 199         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
 200         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
 201         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
 202         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
 203         self.assertEqual(
 204             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
 205             'trailer.mp4')
 206
 207     def test_parse_duration(self):
 208         self.assertEqual(parse_duration(None), None)
 209         self.assertEqual(parse_duration('1'), 1)
 210         self.assertEqual(parse_duration('1337:12'), 80232)
 211         self.assertEqual(parse_duration('9:12:43'), 33163)
 212         self.assertEqual(parse_duration('12:00'), 720)
 213         self.assertEqual(parse_duration('00:01:01'), 61)
 214         self.assertEqual(parse_duration('x:y'), None)
 215         self.assertEqual(parse_duration('3h11m53s'), 11513)
 216         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
 217         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
 218         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
 219         self.assertEqual(parse_duration('62m45s'), 3765)
 220         self.assertEqual(parse_duration('6m59s'), 419)
 221         self.assertEqual(parse_duration('49s'), 49)
 222         self.assertEqual(parse_duration('0h0m0s'), 0)
 223         self.assertEqual(parse_duration('0m0s'), 0)
 224         self.assertEqual(parse_duration('0s'), 0)
 225         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
 226         self.assertEqual(parse_duration('T30M38S'), 1838)
 227         self.assertEqual(parse_duration('5 s'), 5)
 228         self.assertEqual(parse_duration('3 min'), 180)
 229         self.assertEqual(parse_duration('2.5 hours'), 9000)
 230
 231     def test_fix_xml_ampersands(self):
 232         self.assertEqual(
 233             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
 234         self.assertEqual(
 235             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
 236             '"&amp;x=y&amp;wrong;&amp;z=a')
 237         self.assertEqual(
 238             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
 239             '&amp;&apos;&gt;&lt;&quot;')
 240         self.assertEqual(
 241             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
 242         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
 243
 244     def test_paged_list(self):
 245         def testPL(size, pagesize, sliceargs, expected):
 246             def get_page(pagenum):
 247                 firstid = pagenum * pagesize
 248                 upto = min(size, pagenum * pagesize + pagesize)
 249                 for i in range(firstid, upto):
 250                     yield i
 251
 252             pl = OnDemandPagedList(get_page, pagesize)
 253             got = pl.getslice(*sliceargs)
 254             self.assertEqual(got, expected)
 255
 256             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
 257             got = iapl.getslice(*sliceargs)
 258             self.assertEqual(got, expected)
 259
 260         testPL(5, 2, (), [0, 1, 2, 3, 4])
 261         testPL(5, 2, (1,), [1, 2, 3, 4])
 262         testPL(5, 2, (2,), [2, 3, 4])
 263         testPL(5, 2, (4,), [4])
 264         testPL(5, 2, (0, 3), [0, 1, 2])
 265         testPL(5, 2, (1, 4), [1, 2, 3])
 266         testPL(5, 2, (2, 99), [2, 3, 4])
 267         testPL(5, 2, (20, 99), [])
 268
 269     def test_struct_unpack(self):
 270         self.assertEqual(struct_unpack('!B', b'\x00'), (0,))
 271
 272     def test_read_batch_urls(self):
 273         f = io.StringIO('''\xef\xbb\xbf foo
 274             bar\r
 275             baz
 276             # More after this line\r
 277             ; or after this
 278             bam''')
 279         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
 280
 281     def test_urlencode_postdata(self):
 282         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
 283         self.assertTrue(isinstance(data, bytes))
 284
 285     def test_parse_iso8601(self):
 286         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
 287         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
 288         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
 289         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
 290
 291     def test_strip_jsonp(self):
 292         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
 293         d = json.loads(stripped)
 294         self.assertEqual(d, [{"id": "532cb", "x": 3}])
 295
 296         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
 297         d = json.loads(stripped)
 298         self.assertEqual(d, {'STATUS': 'OK'})
 299
 300     def test_uppercase_escape(self):
 301         self.assertEqual(uppercase_escape('aä'), 'aä')
 302         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
 303
 304     def test_limit_length(self):
 305         self.assertEqual(limit_length(None, 12), None)
 306         self.assertEqual(limit_length('foo', 12), 'foo')
 307         self.assertTrue(
 308             limit_length('foo bar baz asd', 12).startswith('foo bar'))
 309         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
 310
 311     def test_escape_rfc3986(self):
 312         reserved = "!*'();:@&=+$,/?#[]"
 313         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
 314         self.assertEqual(escape_rfc3986(reserved), reserved)
 315         self.assertEqual(escape_rfc3986(unreserved), unreserved)
 316         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
 317         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
 318         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
 319         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
 320
 321     def test_escape_url(self):
 322         self.assertEqual(
 323             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
 324             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
 325         )
 326         self.assertEqual(
 327             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
 328             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
 329         )
 330         self.assertEqual(
 331             escape_url('http://тест.рф/фрагмент'),
 332             'http://тест.рф/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
 333         )
 334         self.assertEqual(
 335             escape_url('http://тест.рф/абв?абв=абв#абв'),
 336             'http://тест.рф/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
 337         )
 338         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
 339
 340     def test_js_to_json_realworld(self):
 341         inp = '''{
 342             'clip':{'provider':'pseudo'}
 343         }'''
 344         self.assertEqual(js_to_json(inp), '''{
 345             "clip":{"provider":"pseudo"}
 346         }''')
 347         json.loads(js_to_json(inp))
 348
 349         inp = '''{
 350             'playlist':[{'controls':{'all':null}}]
 351         }'''
 352         self.assertEqual(js_to_json(inp), '''{
 353             "playlist":[{"controls":{"all":null}}]
 354         }''')
 355
 356     def test_js_to_json_edgecases(self):
 357         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
 358         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
 359
 360         on = js_to_json('{"abc": true}')
 361         self.assertEqual(json.loads(on), {'abc': True})
 362
 363     def test_clean_html(self):
 364         self.assertEqual(clean_html('a:\nb'), 'a: b')
 365         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 366
 367     def test_intlist_to_bytes(self):
 368         self.assertEqual(
 369             intlist_to_bytes([0, 1, 127, 128, 255]),
 370             b'\x00\x01\x7f\x80\xff')
 371
 372     def test_args_to_str(self):
 373         self.assertEqual(
 374             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 375             'foo ba/r -baz \'2 be\' \'\''
 376         )
 377
 378     def test_parse_filesize(self):
 379         self.assertEqual(parse_filesize(None), None)
 380         self.assertEqual(parse_filesize(''), None)
 381         self.assertEqual(parse_filesize('91 B'), 91)
 382         self.assertEqual(parse_filesize('foobar'), None)
 383         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 384         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 385         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 386         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 387
 388     def test_version_tuple(self):
 389         self.assertEqual(version_tuple('1'), (1,))
 390         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 391         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 392
 393 if __name__ == '__main__':
 394     unittest.main()