summaryrefslogtreecommitdiff
path: root/test/test_utils.py
blob: f1a748ddee81b3830bcddbebd4478393ba720d40 (plain) 
    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     age_restricted,
   20     args_to_str,
   21     encode_base_n,
   22     caesar,
   23     clean_html,
   24     clean_podcast_url,
   25     date_from_str,
   26     DateRange,
   27     detect_exe_version,
   28     determine_ext,
   29     dict_get,
   30     encode_compat_str,
   31     encodeFilename,
   32     escape_rfc3986,
   33     escape_url,
   34     extract_attributes,
   35     ExtractorError,
   36     find_xpath_attr,
   37     fix_xml_ampersands,
   38     float_or_none,
   39     get_element_by_class,
   40     get_element_by_attribute,
   41     get_elements_by_class,
   42     get_elements_by_attribute,
   43     InAdvancePagedList,
   44     int_or_none,
   45     intlist_to_bytes,
   46     is_html,
   47     js_to_json,
   48     limit_length,
   49     merge_dicts,
   50     mimetype2ext,
   51     month_by_name,
   52     multipart_encode,
   53     ohdave_rsa_encrypt,
   54     OnDemandPagedList,
   55     orderedSet,
   56     parse_age_limit,
   57     parse_duration,
   58     parse_filesize,
   59     parse_count,
   60     parse_iso8601,
   61     parse_resolution,
   62     parse_bitrate,
   63     pkcs1pad,
   64     read_batch_urls,
   65     sanitize_filename,
   66     sanitize_path,
   67     sanitize_url,
   68     expand_path,
   69     prepend_extension,
   70     replace_extension,
   71     remove_start,
   72     remove_end,
   73     remove_quotes,
   74     rot47,
   75     shell_quote,
   76     smuggle_url,
   77     str_to_int,
   78     strip_jsonp,
   79     strip_or_none,
   80     subtitles_filename,
   81     timeconvert,
   82     unescapeHTML,
   83     unified_strdate,
   84     unified_timestamp,
   85     unsmuggle_url,
   86     uppercase_escape,
   87     lowercase_escape,
   88     url_basename,
   89     url_or_none,
   90     base_url,
   91     urljoin,
   92     urlencode_postdata,
   93     urshift,
   94     update_url_query,
   95     version_tuple,
   96     xpath_with_ns,
   97     xpath_element,
   98     xpath_text,
   99     xpath_attr,
  100     render_table,
  101     match_str,
  102     parse_dfxp_time_expr,
  103     dfxp2srt,
  104     cli_option,
  105     cli_valueless_option,
  106     cli_bool_option,
  107     parse_codecs,
  108 )
  109 from youtube_dl.compat import (
  110     compat_chr,
  111     compat_etree_fromstring,
  112     compat_getenv,
  113     compat_os_name,
  114     compat_setenv,
  115     compat_urlparse,
  116     compat_parse_qs,
  117 )
  118 
  119 
  120 class TestUtil(unittest.TestCase):
  121     def test_timeconvert(self):
  122         self.assertTrue(timeconvert('') is None)
  123         self.assertTrue(timeconvert('bougrg') is None)
  124 
  125     def test_sanitize_filename(self):
  126         self.assertEqual(sanitize_filename('abc'), 'abc')
  127         self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')
  128 
  129         self.assertEqual(sanitize_filename('123'), '123')
  130 
  131         self.assertEqual('abc_de', sanitize_filename('abc/de'))
  132         self.assertFalse('/' in sanitize_filename('abc/de///'))
  133 
  134         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
  135         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
  136         self.assertEqual('yes no', sanitize_filename('yes? no'))
  137         self.assertEqual('this - that', sanitize_filename('this: that'))
  138 
  139         self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
  140         aumlaut = 'ä'
  141         self.assertEqual(sanitize_filename(aumlaut), aumlaut)
  142         tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
  143         self.assertEqual(sanitize_filename(tests), tests)
  144 
  145         self.assertEqual(
  146             sanitize_filename('New World record at 0:12:34'),
  147             'New World record at 0_12_34')
  148 
  149         self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
  150         self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
  151         self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
  152         self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
  153 
  154         forbidden = '"\0\\/'
  155         for fc in forbidden:
  156             for fbc in forbidden:
  157                 self.assertTrue(fbc not in sanitize_filename(fc))
  158 
  159     def test_sanitize_filename_restricted(self):
  160         self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
  161         self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')
  162 
  163         self.assertEqual(sanitize_filename('123', restricted=True), '123')
  164 
  165         self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
  166         self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))
  167 
  168         self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
  169         self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
  170         self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
  171         self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))
  172 
  173         tests = 'aäb\u4e2d\u56fd\u7684c'
  174         self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
  175         self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename
  176 
  177         forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
  178         for fc in forbidden:
  179             for fbc in forbidden:
  180                 self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))
  181 
  182         # Handle a common case more neatly
  183         self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
  184         self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
  185         # .. but make sure the file name is never empty
  186         self.assertTrue(sanitize_filename('-', restricted=True) != '')
  187         self.assertTrue(sanitize_filename(':', restricted=True) != '')
  188 
  189         self.assertEqual(sanitize_filename(
  190             'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
  191             'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')
  192 
  193     def test_sanitize_ids(self):
  194         self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
  195         self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
  196         self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
  197 
  198     def test_sanitize_path(self):
  199         if sys.platform != 'win32':
  200             return
  201 
  202         self.assertEqual(sanitize_path('abc'), 'abc')
  203         self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
  204         self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
  205         self.assertEqual(sanitize_path('abc|def'), 'abc#def')
  206         self.assertEqual(sanitize_path('<>:"|?*'), '#######')
  207         self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
  208         self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')
  209 
  210         self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
  211         self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')
  212 
  213         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
  214         self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
  215         self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
  216         self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
  217 
  218         self.assertEqual(
  219             sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
  220             'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')
  221 
  222         self.assertEqual(
  223             sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
  224             'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
  225         self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
  226         self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
  227         self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')
  228 
  229         self.assertEqual(sanitize_path('../abc'), '..\\abc')
  230         self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
  231         self.assertEqual(sanitize_path('./abc'), 'abc')
  232         self.assertEqual(sanitize_path('./../abc'), '..\\abc')
  233 
  234     def test_sanitize_url(self):
  235         self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
  236         self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
  237         self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
  238         self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')
  239 
  240     def test_expand_path(self):
  241         def env(var):
  242             return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)
  243 
  244         compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
  245         self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
  246         self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
  247         self.assertEqual(expand_path('~'), compat_getenv('HOME'))
  248         self.assertEqual(
  249             expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
  250             '%s/expanded' % compat_getenv('HOME'))
  251 
  252     def test_prepend_extension(self):
  253         self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
  254         self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
  255         self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
  256         self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
  257         self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
  258         self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')
  259 
  260     def test_replace_extension(self):
  261         self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
  262         self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
  263         self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
  264         self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
  265         self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
  266         self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')
  267 
  268     def test_subtitles_filename(self):
  269         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
  270         self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
  271         self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')
  272 
  273     def test_remove_start(self):
  274         self.assertEqual(remove_start(None, 'A - '), None)
  275         self.assertEqual(remove_start('A - B', 'A - '), 'B')
  276         self.assertEqual(remove_start('B - A', 'A - '), 'B - A')
  277 
  278     def test_remove_end(self):
  279         self.assertEqual(remove_end(None, ' - B'), None)
  280         self.assertEqual(remove_end('A - B', ' - B'), 'A')
  281         self.assertEqual(remove_end('B - A', ' - B'), 'B - A')
  282 
  283     def test_remove_quotes(self):
  284         self.assertEqual(remove_quotes(None), None)
  285         self.assertEqual(remove_quotes('"'), '"')
  286         self.assertEqual(remove_quotes("'"), "'")
  287         self.assertEqual(remove_quotes(';'), ';')
  288         self.assertEqual(remove_quotes('";'), '";')
  289         self.assertEqual(remove_quotes('""'), '')
  290         self.assertEqual(remove_quotes('";"'), ';')
  291 
  292     def test_ordered_set(self):
  293         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
  294         self.assertEqual(orderedSet([]), [])
  295         self.assertEqual(orderedSet([1]), [1])
  296         # keep the list ordered
  297         self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])
  298 
  299     def test_unescape_html(self):
  300         self.assertEqual(unescapeHTML('%20;'), '%20;')
  301         self.assertEqual(unescapeHTML('&#x2F;'), '/')
  302         self.assertEqual(unescapeHTML('&#47;'), '/')
  303         self.assertEqual(unescapeHTML('&eacute;'), 'é')
  304         self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
  305         self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
  306         # HTML5 entities
  307         self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')
  308 
  309     def test_date_from_str(self):
  310         self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
  311         self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
  312         self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
  313         self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
  314         self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))
  315 
  316     def test_daterange(self):
  317         _20century = DateRange("19000101", "20000101")
  318         self.assertFalse("17890714" in _20century)
  319         _ac = DateRange("00010101")
  320         self.assertTrue("19690721" in _ac)
  321         _firstmilenium = DateRange(end="10000101")
  322         self.assertTrue("07110427" in _firstmilenium)
  323 
  324     def test_unified_dates(self):
  325         self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
  326         self.assertEqual(unified_strdate('8/7/2009'), '20090708')
  327         self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
  328         self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
  329         self.assertEqual(unified_strdate('1968 12 10'), '19681210')
  330         self.assertEqual(unified_strdate('1968-12-10'), '19681210')
  331         self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
  332         self.assertEqual(
  333             unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
  334             '20141126')
  335         self.assertEqual(
  336             unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
  337             '20150202')
  338         self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
  339         self.assertEqual(unified_strdate('25-09-2014'), '20140925')
  340         self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
  341         self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
  342         self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
  343         self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
  344         self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
  345         self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
  346         self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
  347         self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')
  348 
  349     def test_unified_timestamps(self):
  350         self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
  351         self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
  352         self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
  353         self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
  354         self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
  355         self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
  356         self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
  357         self.assertEqual(
  358             unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
  359             1417001400)
  360         self.assertEqual(
  361             unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
  362             1422902860)
  363         self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
  364         self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
  365         self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
  366         self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
  367         self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
  368         self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
  369         self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
  370         self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
  371         self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
  372         self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)
  373         self.assertEqual(unified_timestamp('December 31 1969 20:00:01 EDT'), 1)
  374         self.assertEqual(unified_timestamp('Wednesday 31 December 1969 18:01:26 MDT'), 86)
  375         self.assertEqual(unified_timestamp('12/31/1969 20:01:18 EDT', False), 78)
  376 
  377     def test_determine_ext(self):
  378         self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
  379         self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
  380         self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
  381         self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
  382         self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
  383         self.assertEqual(determine_ext('foobar', None), None)
  384 
  385     def test_find_xpath_attr(self):
  386         testxml = '''<root>
  387             <node/>
  388             <node x="a"/>
  389             <node x="a" y="c" />
  390             <node x="b" y="d" />
  391             <node x="" />
  392         </root>'''
  393         doc = compat_etree_fromstring(testxml)
  394 
  395         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
  396         self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
  397         self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
  398         self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
  399         self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
  400         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
  401         self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
  402         self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
  403         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
  404         self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
  405         self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])
  406 
  407     def test_xpath_with_ns(self):
  408         testxml = '''<root xmlns:media="http://example.com/">
  409             <media:song>
  410                 <media:author>The Author</media:author>
  411                 <url>http://server.com/download.mp3</url>
  412             </media:song>
  413         </root>'''
  414         doc = compat_etree_fromstring(testxml)
  415         find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
  416         self.assertTrue(find('media:song') is not None)
  417         self.assertEqual(find('media:song/media:author').text, 'The Author')
  418         self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')
  419 
  420     def test_xpath_element(self):
  421         doc = xml.etree.ElementTree.Element('root')
  422         div = xml.etree.ElementTree.SubElement(doc, 'div')
  423         p = xml.etree.ElementTree.SubElement(div, 'p')
  424         p.text = 'Foo'
  425         self.assertEqual(xpath_element(doc, 'div/p'), p)
  426         self.assertEqual(xpath_element(doc, ['div/p']), p)
  427         self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
  428         self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
  429         self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
  430         self.assertTrue(xpath_element(doc, 'div/bar') is None)
  431         self.assertTrue(xpath_element(doc, ['div/bar']) is None)
  432         self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
  433         self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
  434         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
  435         self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)
  436 
  437     def test_xpath_text(self):
  438         testxml = '''<root>
  439             <div>
  440                 <p>Foo</p>
  441             </div>
  442         </root>'''
  443         doc = compat_etree_fromstring(testxml)
  444         self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
  445         self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
  446         self.assertTrue(xpath_text(doc, 'div/bar') is None)
  447         self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)
  448 
  449     def test_xpath_attr(self):
  450         testxml = '''<root>
  451             <div>
  452                 <p x="a">Foo</p>
  453             </div>
  454         </root>'''
  455         doc = compat_etree_fromstring(testxml)
  456         self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
  457         self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
  458         self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
  459         self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
  460         self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
  461         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
  462         self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)
  463 
  464     def test_smuggle_url(self):
  465         data = {"ö": "ö", "abc": [3]}
  466         url = 'https://foo.bar/baz?x=y#a'
  467         smug_url = smuggle_url(url, data)
  468         unsmug_url, unsmug_data = unsmuggle_url(smug_url)
  469         self.assertEqual(url, unsmug_url)
  470         self.assertEqual(data, unsmug_data)
  471 
  472         res_url, res_data = unsmuggle_url(url)
  473         self.assertEqual(res_url, url)
  474         self.assertEqual(res_data, None)
  475 
  476         smug_url = smuggle_url(url, {'a': 'b'})
  477         smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
  478         res_url, res_data = unsmuggle_url(smug_smug_url)
  479         self.assertEqual(res_url, url)
  480         self.assertEqual(res_data, {'a': 'b', 'c': 'd'})
  481 
  482     def test_shell_quote(self):
  483         args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
  484         self.assertEqual(
  485             shell_quote(args),
  486             """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')
  487 
  488     def test_float_or_none(self):
  489         self.assertEqual(float_or_none('42.42'), 42.42)
  490         self.assertEqual(float_or_none('42'), 42.0)
  491         self.assertEqual(float_or_none(''), None)
  492         self.assertEqual(float_or_none(None), None)
  493         self.assertEqual(float_or_none([]), None)
  494         self.assertEqual(float_or_none(set()), None)
  495 
  496     def test_int_or_none(self):
  497         self.assertEqual(int_or_none('42'), 42)
  498         self.assertEqual(int_or_none(''), None)
  499         self.assertEqual(int_or_none(None), None)
  500         self.assertEqual(int_or_none([]), None)
  501         self.assertEqual(int_or_none(set()), None)
  502 
  503     def test_str_to_int(self):
  504         self.assertEqual(str_to_int('123,456'), 123456)
  505         self.assertEqual(str_to_int('123.456'), 123456)
  506         self.assertEqual(str_to_int(523), 523)
  507         # Python 3 has no long
  508         if sys.version_info < (3, 0):
  509             eval('self.assertEqual(str_to_int(123456L), 123456)')
  510         self.assertEqual(str_to_int('noninteger'), None)
  511         self.assertEqual(str_to_int([]), None)
  512 
  513     def test_url_basename(self):
  514         self.assertEqual(url_basename('http://foo.de/'), '')
  515         self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
  516         self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
  517         self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
  518         self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
  519         self.assertEqual(
  520             url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
  521             'trailer.mp4')
  522 
  523     def test_base_url(self):
  524         self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
  525         self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
  526         self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
  527         self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
  528         self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')
  529 
  530     def test_urljoin(self):
  531         self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  532         self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  533         self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  534         self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  535         self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
  536         self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  537         self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  538         self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  539         self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  540         self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
  541         self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  542         self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
  543         self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  544         self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
  545         self.assertEqual(urljoin('http://foo.de/', None), None)
  546         self.assertEqual(urljoin('http://foo.de/', ''), None)
  547         self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
  548         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
  549         self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
  550         self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')
  551 
  552     def test_url_or_none(self):
  553         self.assertEqual(url_or_none(None), None)
  554         self.assertEqual(url_or_none(''), None)
  555         self.assertEqual(url_or_none('foo'), None)
  556         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
  557         self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
  558         self.assertEqual(url_or_none('http$://foo.de'), None)
  559         self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
  560         self.assertEqual(url_or_none('//foo.de'), '//foo.de')
  561         self.assertEqual(url_or_none('s3://foo.de'), None)
  562         self.assertEqual(url_or_none('rtmpte://foo.de'), 'rtmpte://foo.de')
  563         self.assertEqual(url_or_none('mms://foo.de'), 'mms://foo.de')
  564         self.assertEqual(url_or_none('rtspu://foo.de'), 'rtspu://foo.de')
  565         self.assertEqual(url_or_none('ftps://foo.de'), 'ftps://foo.de')
  566 
  567     def test_parse_age_limit(self):
  568         self.assertEqual(parse_age_limit(None), None)
  569         self.assertEqual(parse_age_limit(False), None)
  570         self.assertEqual(parse_age_limit('invalid'), None)
  571         self.assertEqual(parse_age_limit(0), 0)
  572         self.assertEqual(parse_age_limit(18), 18)
  573         self.assertEqual(parse_age_limit(21), 21)
  574         self.assertEqual(parse_age_limit(22), None)
  575         self.assertEqual(parse_age_limit('18'), 18)
  576         self.assertEqual(parse_age_limit('18+'), 18)
  577         self.assertEqual(parse_age_limit('PG-13'), 13)
  578         self.assertEqual(parse_age_limit('TV-14'), 14)
  579         self.assertEqual(parse_age_limit('TV-MA'), 17)
  580         self.assertEqual(parse_age_limit('TV14'), 14)
  581         self.assertEqual(parse_age_limit('TV_G'), 0)
  582 
  583     def test_parse_duration(self):
  584         self.assertEqual(parse_duration(None), None)
  585         self.assertEqual(parse_duration(False), None)
  586         self.assertEqual(parse_duration('invalid'), None)
  587         self.assertEqual(parse_duration('1'), 1)
  588         self.assertEqual(parse_duration('1337:12'), 80232)
  589         self.assertEqual(parse_duration('9:12:43'), 33163)
  590         self.assertEqual(parse_duration('12:00'), 720)
  591         self.assertEqual(parse_duration('00:01:01'), 61)
  592         self.assertEqual(parse_duration('x:y'), None)
  593         self.assertEqual(parse_duration('3h11m53s'), 11513)
  594         self.assertEqual(parse_duration('3h 11m 53s'), 11513)
  595         self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
  596         self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
  597         self.assertEqual(parse_duration('62m45s'), 3765)
  598         self.assertEqual(parse_duration('6m59s'), 419)
  599         self.assertEqual(parse_duration('49s'), 49)
  600         self.assertEqual(parse_duration('0h0m0s'), 0)
  601         self.assertEqual(parse_duration('0m0s'), 0)
  602         self.assertEqual(parse_duration('0s'), 0)
  603         self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
  604         self.assertEqual(parse_duration('T30M38S'), 1838)
  605         self.assertEqual(parse_duration('5 s'), 5)
  606         self.assertEqual(parse_duration('3 min'), 180)
  607         self.assertEqual(parse_duration('2.5 hours'), 9000)
  608         self.assertEqual(parse_duration('02:03:04'), 7384)
  609         self.assertEqual(parse_duration('01:02:03:04'), 93784)
  610         self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
  611         self.assertEqual(parse_duration('87 Min.'), 5220)
  612         self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
  613         self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
  614         self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
  615 
  616     def test_fix_xml_ampersands(self):
  617         self.assertEqual(
  618             fix_xml_ampersands('"&x=y&z=a'), '"&amp;x=y&amp;z=a')
  619         self.assertEqual(
  620             fix_xml_ampersands('"&amp;x=y&wrong;&z=a'),
  621             '"&amp;x=y&amp;wrong;&amp;z=a')
  622         self.assertEqual(
  623             fix_xml_ampersands('&amp;&apos;&gt;&lt;&quot;'),
  624             '&amp;&apos;&gt;&lt;&quot;')
  625         self.assertEqual(
  626             fix_xml_ampersands('&#1234;&#x1abC;'), '&#1234;&#x1abC;')
  627         self.assertEqual(fix_xml_ampersands('&#&#'), '&amp;#&amp;#')
  628 
  629     def test_paged_list(self):
  630         def testPL(size, pagesize, sliceargs, expected):
  631             def get_page(pagenum):
  632                 firstid = pagenum * pagesize
  633                 upto = min(size, pagenum * pagesize + pagesize)
  634                 for i in range(firstid, upto):
  635                     yield i
  636 
  637             pl = OnDemandPagedList(get_page, pagesize)
  638             got = pl.getslice(*sliceargs)
  639             self.assertEqual(got, expected)
  640 
  641             iapl = InAdvancePagedList(get_page, size // pagesize + 1, pagesize)
  642             got = iapl.getslice(*sliceargs)
  643             self.assertEqual(got, expected)
  644 
  645         testPL(5, 2, (), [0, 1, 2, 3, 4])
  646         testPL(5, 2, (1,), [1, 2, 3, 4])
  647         testPL(5, 2, (2,), [2, 3, 4])
  648         testPL(5, 2, (4,), [4])
  649         testPL(5, 2, (0, 3), [0, 1, 2])
  650         testPL(5, 2, (1, 4), [1, 2, 3])
  651         testPL(5, 2, (2, 99), [2, 3, 4])
  652         testPL(5, 2, (20, 99), [])
  653 
  654     def test_read_batch_urls(self):
  655         f = io.StringIO('''\xef\xbb\xbf foo
  656             bar\r
  657             baz
  658             # More after this line\r
  659             ; or after this
  660             bam''')
  661         self.assertEqual(read_batch_urls(f), ['foo', 'bar', 'baz', 'bam'])
  662 
  663     def test_urlencode_postdata(self):
  664         data = urlencode_postdata({'username': 'foo@bar.com', 'password': '1234'})
  665         self.assertTrue(isinstance(data, bytes))
  666 
  667     def test_update_url_query(self):
  668         def query_dict(url):
  669             return compat_parse_qs(compat_urlparse.urlparse(url).query)
  670         self.assertEqual(query_dict(update_url_query(
  671             'http://example.com/path', {'quality': ['HD'], 'format': ['mp4']})),
  672             query_dict('http://example.com/path?quality=HD&format=mp4'))
  673         self.assertEqual(query_dict(update_url_query(
  674             'http://example.com/path', {'system': ['LINUX', 'WINDOWS']})),
  675             query_dict('http://example.com/path?system=LINUX&system=WINDOWS'))
  676         self.assertEqual(query_dict(update_url_query(
  677             'http://example.com/path', {'fields': 'id,formats,subtitles'})),
  678             query_dict('http://example.com/path?fields=id,formats,subtitles'))
  679         self.assertEqual(query_dict(update_url_query(
  680             'http://example.com/path', {'fields': ('id,formats,subtitles', 'thumbnails')})),
  681             query_dict('http://example.com/path?fields=id,formats,subtitles&fields=thumbnails'))
  682         self.assertEqual(query_dict(update_url_query(
  683             'http://example.com/path?manifest=f4m', {'manifest': []})),
  684             query_dict('http://example.com/path'))
  685         self.assertEqual(query_dict(update_url_query(
  686             'http://example.com/path?system=LINUX&system=WINDOWS', {'system': 'LINUX'})),
  687             query_dict('http://example.com/path?system=LINUX'))
  688         self.assertEqual(query_dict(update_url_query(
  689             'http://example.com/path', {'fields': b'id,formats,subtitles'})),
  690             query_dict('http://example.com/path?fields=id,formats,subtitles'))
  691         self.assertEqual(query_dict(update_url_query(
  692             'http://example.com/path', {'width': 1080, 'height': 720})),
  693             query_dict('http://example.com/path?width=1080&height=720'))
  694         self.assertEqual(query_dict(update_url_query(
  695             'http://example.com/path', {'bitrate': 5020.43})),
  696             query_dict('http://example.com/path?bitrate=5020.43'))
  697         self.assertEqual(query_dict(update_url_query(
  698             'http://example.com/path', {'test': '第二行тест'})),
  699             query_dict('http://example.com/path?test=%E7%AC%AC%E4%BA%8C%E8%A1%8C%D1%82%D0%B5%D1%81%D1%82'))
  700 
  701     def test_multipart_encode(self):
  702         self.assertEqual(
  703             multipart_encode({b'field': b'value'}, boundary='AAAAAA')[0],
  704             b'--AAAAAA\r\nContent-Disposition: form-data; name="field"\r\n\r\nvalue\r\n--AAAAAA--\r\n')
  705         self.assertEqual(
  706             multipart_encode({'欄位'.encode('utf-8'): '值'.encode('utf-8')}, boundary='AAAAAA')[0],
  707             b'--AAAAAA\r\nContent-Disposition: form-data; name="\xe6\xac\x84\xe4\xbd\x8d"\r\n\r\n\xe5\x80\xbc\r\n--AAAAAA--\r\n')
  708         self.assertRaises(
  709             ValueError, multipart_encode, {b'field': b'value'}, boundary='value')
  710 
  711     def test_dict_get(self):
  712         FALSE_VALUES = {
  713             'none': None,
  714             'false': False,
  715             'zero': 0,
  716             'empty_string': '',
  717             'empty_list': [],
  718         }
  719         d = FALSE_VALUES.copy()
  720         d['a'] = 42
  721         self.assertEqual(dict_get(d, 'a'), 42)
  722         self.assertEqual(dict_get(d, 'b'), None)
  723         self.assertEqual(dict_get(d, 'b', 42), 42)
  724         self.assertEqual(dict_get(d, ('a', )), 42)
  725         self.assertEqual(dict_get(d, ('b', 'a', )), 42)
  726         self.assertEqual(dict_get(d, ('b', 'c', 'a', 'd', )), 42)
  727         self.assertEqual(dict_get(d, ('b', 'c', )), None)
  728         self.assertEqual(dict_get(d, ('b', 'c', ), 42), 42)
  729         for key, false_value in FALSE_VALUES.items():
  730             self.assertEqual(dict_get(d, ('b', 'c', key, )), None)
  731             self.assertEqual(dict_get(d, ('b', 'c', key, ), skip_false_values=False), false_value)
  732 
  733     def test_merge_dicts(self):
  734         self.assertEqual(merge_dicts({'a': 1}, {'b': 2}), {'a': 1, 'b': 2})
  735         self.assertEqual(merge_dicts({'a': 1}, {'a': 2}), {'a': 1})
  736         self.assertEqual(merge_dicts({'a': 1}, {'a': None}), {'a': 1})
  737         self.assertEqual(merge_dicts({'a': 1}, {'a': ''}), {'a': 1})
  738         self.assertEqual(merge_dicts({'a': 1}, {}), {'a': 1})
  739         self.assertEqual(merge_dicts({'a': None}, {'a': 1}), {'a': 1})
  740         self.assertEqual(merge_dicts({'a': ''}, {'a': 1}), {'a': ''})
  741         self.assertEqual(merge_dicts({'a': ''}, {'a': 'abc'}), {'a': 'abc'})
  742         self.assertEqual(merge_dicts({'a': None}, {'a': ''}, {'a': 'abc'}), {'a': 'abc'})
  743 
  744     def test_encode_compat_str(self):
  745         self.assertEqual(encode_compat_str(b'\xd1\x82\xd0\xb5\xd1\x81\xd1\x82', 'utf-8'), 'тест')
  746         self.assertEqual(encode_compat_str('тест', 'utf-8'), 'тест')
  747 
  748     def test_parse_iso8601(self):
  749         self.assertEqual(parse_iso8601('2014-03-23T23:04:26+0100'), 1395612266)
  750         self.assertEqual(parse_iso8601('2014-03-23T22:04:26+0000'), 1395612266)
  751         self.assertEqual(parse_iso8601('2014-03-23T22:04:26Z'), 1395612266)
  752         self.assertEqual(parse_iso8601('2014-03-23T22:04:26.1234Z'), 1395612266)
  753         self.assertEqual(parse_iso8601('2015-09-29T08:27:31.727'), 1443515251)
  754         self.assertEqual(parse_iso8601('2015-09-29T08-27-31.727'), None)
  755 
  756     def test_strip_jsonp(self):
  757         stripped = strip_jsonp('cb ([ {"id":"532cb",\n\n\n"x":\n3}\n]\n);')
  758         d = json.loads(stripped)
  759         self.assertEqual(d, [{"id": "532cb", "x": 3}])
  760 
  761         stripped = strip_jsonp('parseMetadata({"STATUS":"OK"})\n\n\n//epc')
  762         d = json.loads(stripped)
  763         self.assertEqual(d, {'STATUS': 'OK'})
  764 
  765         stripped = strip_jsonp('ps.embedHandler({"status": "success"});')
  766         d = json.loads(stripped)
  767         self.assertEqual(d, {'status': 'success'})
  768 
  769         stripped = strip_jsonp('window.cb && window.cb({"status": "success"});')
  770         d = json.loads(stripped)
  771         self.assertEqual(d, {'status': 'success'})
  772 
  773         stripped = strip_jsonp('window.cb && cb({"status": "success"});')
  774         d = json.loads(stripped)
  775         self.assertEqual(d, {'status': 'success'})
  776 
  777         stripped = strip_jsonp('({"status": "success"});')
  778         d = json.loads(stripped)
  779         self.assertEqual(d, {'status': 'success'})
  780 
  781     def test_strip_or_none(self):
  782         self.assertEqual(strip_or_none(' abc'), 'abc')
  783         self.assertEqual(strip_or_none('abc '), 'abc')
  784         self.assertEqual(strip_or_none(' abc '), 'abc')
  785         self.assertEqual(strip_or_none('\tabc\t'), 'abc')
  786         self.assertEqual(strip_or_none('\n\tabc\n\t'), 'abc')
  787         self.assertEqual(strip_or_none('abc'), 'abc')
  788         self.assertEqual(strip_or_none(''), '')
  789         self.assertEqual(strip_or_none(None), None)
  790         self.assertEqual(strip_or_none(42), None)
  791         self.assertEqual(strip_or_none([]), None)
  792 
  793     def test_uppercase_escape(self):
  794         self.assertEqual(uppercase_escape('aä'), 'aä')
  795         self.assertEqual(uppercase_escape('\\U0001d550'), '𝕐')
  796 
  797     def test_lowercase_escape(self):
  798         self.assertEqual(lowercase_escape('aä'), 'aä')
  799         self.assertEqual(lowercase_escape('\\u0026'), '&')
  800 
  801     def test_limit_length(self):
  802         self.assertEqual(limit_length(None, 12), None)
  803         self.assertEqual(limit_length('foo', 12), 'foo')
  804         self.assertTrue(
  805             limit_length('foo bar baz asd', 12).startswith('foo bar'))
  806         self.assertTrue('...' in limit_length('foo bar baz asd', 12))
  807 
  808     def test_mimetype2ext(self):
  809         self.assertEqual(mimetype2ext(None), None)
  810         self.assertEqual(mimetype2ext('video/x-flv'), 'flv')
  811         self.assertEqual(mimetype2ext('application/x-mpegURL'), 'm3u8')
  812         self.assertEqual(mimetype2ext('text/vtt'), 'vtt')
  813         self.assertEqual(mimetype2ext('text/vtt;charset=utf-8'), 'vtt')
  814         self.assertEqual(mimetype2ext('text/html; charset=utf-8'), 'html')
  815         self.assertEqual(mimetype2ext('audio/x-wav'), 'wav')
  816         self.assertEqual(mimetype2ext('audio/x-wav;codec=pcm'), 'wav')
  817 
  818     def test_month_by_name(self):
  819         self.assertEqual(month_by_name(None), None)
  820         self.assertEqual(month_by_name('December', 'en'), 12)
  821         self.assertEqual(month_by_name('décembre', 'fr'), 12)
  822         self.assertEqual(month_by_name('December'), 12)
  823         self.assertEqual(month_by_name('décembre'), None)
  824         self.assertEqual(month_by_name('Unknown', 'unknown'), None)
  825 
  826     def test_parse_codecs(self):
  827         self.assertEqual(parse_codecs(''), {})
  828         self.assertEqual(parse_codecs('avc1.77.30, mp4a.40.2'), {
  829             'vcodec': 'avc1.77.30',
  830             'acodec': 'mp4a.40.2',
  831         })
  832         self.assertEqual(parse_codecs('mp4a.40.2'), {
  833             'vcodec': 'none',
  834             'acodec': 'mp4a.40.2',
  835         })
  836         self.assertEqual(parse_codecs('mp4a.40.5,avc1.42001e'), {
  837             'vcodec': 'avc1.42001e',
  838             'acodec': 'mp4a.40.5',
  839         })
  840         self.assertEqual(parse_codecs('avc3.640028'), {
  841             'vcodec': 'avc3.640028',
  842             'acodec': 'none',
  843         })
  844         self.assertEqual(parse_codecs(', h264,,newcodec,aac'), {
  845             'vcodec': 'h264',
  846             'acodec': 'aac',
  847         })
  848         self.assertEqual(parse_codecs('av01.0.05M.08'), {
  849             'vcodec': 'av01.0.05M.08',
  850             'acodec': 'none',
  851         })
  852         self.assertEqual(parse_codecs('theora, vorbis'), {
  853             'vcodec': 'theora',
  854             'acodec': 'vorbis',
  855         })
  856         self.assertEqual(parse_codecs('unknownvcodec, unknownacodec'), {
  857             'vcodec': 'unknownvcodec',
  858             'acodec': 'unknownacodec',
  859         })
  860         self.assertEqual(parse_codecs('unknown'), {})
  861 
  862     def test_escape_rfc3986(self):
  863         reserved = "!*'();:@&=+$,/?#[]"
  864         unreserved = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_.~'
  865         self.assertEqual(escape_rfc3986(reserved), reserved)
  866         self.assertEqual(escape_rfc3986(unreserved), unreserved)
  867         self.assertEqual(escape_rfc3986('тест'), '%D1%82%D0%B5%D1%81%D1%82')
  868         self.assertEqual(escape_rfc3986('%D1%82%D0%B5%D1%81%D1%82'), '%D1%82%D0%B5%D1%81%D1%82')
  869         self.assertEqual(escape_rfc3986('foo bar'), 'foo%20bar')
  870         self.assertEqual(escape_rfc3986('foo%20bar'), 'foo%20bar')
  871 
  872     def test_escape_url(self):
  873         self.assertEqual(
  874             escape_url('http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavré_FD.mp4'),
  875             'http://wowza.imust.org/srv/vod/telemb/new/UPLOAD/UPLOAD/20224_IncendieHavre%CC%81_FD.mp4'
  876         )
  877         self.assertEqual(
  878             escape_url('http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erklärt/Das-Erste/Video?documentId=22673108&bcastId=5290'),
  879             'http://www.ardmediathek.de/tv/Sturm-der-Liebe/Folge-2036-Zu-Mann-und-Frau-erkl%C3%A4rt/Das-Erste/Video?documentId=22673108&bcastId=5290'
  880         )
  881         self.assertEqual(
  882             escape_url('http://тест.рф/фрагмент'),
  883             'http://xn--e1aybc.xn--p1ai/%D1%84%D1%80%D0%B0%D0%B3%D0%BC%D0%B5%D0%BD%D1%82'
  884         )
  885         self.assertEqual(
  886             escape_url('http://тест.рф/абв?абв=абв#абв'),
  887             'http://xn--e1aybc.xn--p1ai/%D0%B0%D0%B1%D0%B2?%D0%B0%D0%B1%D0%B2=%D0%B0%D0%B1%D0%B2#%D0%B0%D0%B1%D0%B2'
  888         )
  889         self.assertEqual(escape_url('http://vimeo.com/56015672#at=0'), 'http://vimeo.com/56015672#at=0')
  890 
  891     def test_js_to_json_realworld(self):
  892         inp = '''{
  893             'clip':{'provider':'pseudo'}
  894         }'''
  895         self.assertEqual(js_to_json(inp), '''{
  896             "clip":{"provider":"pseudo"}
  897         }''')
  898         json.loads(js_to_json(inp))
  899 
  900         inp = '''{
  901             'playlist':[{'controls':{'all':null}}]
  902         }'''
  903         self.assertEqual(js_to_json(inp), '''{
  904             "playlist":[{"controls":{"all":null}}]
  905         }''')
  906 
  907         inp = '''"The CW\\'s \\'Crazy Ex-Girlfriend\\'"'''
  908         self.assertEqual(js_to_json(inp), '''"The CW's 'Crazy Ex-Girlfriend'"''')
  909 
  910         inp = '"SAND Number: SAND 2013-7800P\\nPresenter: Tom Russo\\nHabanero Software Training - Xyce Software\\nXyce, Sandia\\u0027s"'
  911         json_code = js_to_json(inp)
  912         self.assertEqual(json.loads(json_code), json.loads(inp))
  913 
  914         inp = '''{
  915             0:{src:'skipped', type: 'application/dash+xml'},
  916             1:{src:'skipped', type: 'application/vnd.apple.mpegURL'},
  917         }'''
  918         self.assertEqual(js_to_json(inp), '''{
  919             "0":{"src":"skipped", "type": "application/dash+xml"},
  920             "1":{"src":"skipped", "type": "application/vnd.apple.mpegURL"}
  921         }''')
  922 
  923         inp = '''{"foo":101}'''
  924         self.assertEqual(js_to_json(inp), '''{"foo":101}''')
  925 
  926         inp = '''{"duration": "00:01:07"}'''
  927         self.assertEqual(js_to_json(inp), '''{"duration": "00:01:07"}''')
  928 
  929         inp = '''{segments: [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}'''
  930         self.assertEqual(js_to_json(inp), '''{"segments": [{"offset":-3.885780586188048e-16,"duration":39.75000000000001}]}''')
  931 
  932     def test_js_to_json_edgecases(self):
  933         on = js_to_json("{abc_def:'1\\'\\\\2\\\\\\'3\"4'}")
  934         self.assertEqual(json.loads(on), {"abc_def": "1'\\2\\'3\"4"})
  935 
  936         on = js_to_json('{"abc": true}')
  937         self.assertEqual(json.loads(on), {'abc': True})
  938 
  939         # Ignore JavaScript code as well
  940         on = js_to_json('''{
  941             "x": 1,
  942             y: "a",
  943             z: some.code
  944         }''')
  945         d = json.loads(on)
  946         self.assertEqual(d['x'], 1)
  947         self.assertEqual(d['y'], 'a')
  948 
  949         # Just drop ! prefix for now though this results in a wrong value
  950         on = js_to_json('''{
  951             a: !0,
  952             b: !1,
  953             c: !!0,
  954             d: !!42.42,
  955             e: !!![],
  956             f: !"abc",
  957             g: !"",
  958             !42: 42
  959         }''')
  960         self.assertEqual(json.loads(on), {
  961             'a': 0,
  962             'b': 1,
  963             'c': 0,
  964             'd': 42.42,
  965             'e': [],
  966             'f': "abc",
  967             'g': "",
  968             '42': 42
  969         })
  970 
  971         on = js_to_json('["abc", "def",]')
  972         self.assertEqual(json.loads(on), ['abc', 'def'])
  973 
  974         on = js_to_json('[/*comment\n*/"abc"/*comment\n*/,/*comment\n*/"def",/*comment\n*/]')
  975         self.assertEqual(json.loads(on), ['abc', 'def'])
  976 
  977         on = js_to_json('[//comment\n"abc" //comment\n,//comment\n"def",//comment\n]')
  978         self.assertEqual(json.loads(on), ['abc', 'def'])
  979 
  980         on = js_to_json('{"abc": "def",}')
  981         self.assertEqual(json.loads(on), {'abc': 'def'})
  982 
  983         on = js_to_json('{/*comment\n*/"abc"/*comment\n*/:/*comment\n*/"def"/*comment\n*/,/*comment\n*/}')
  984         self.assertEqual(json.loads(on), {'abc': 'def'})
  985 
  986         on = js_to_json('{ 0: /* " \n */ ",]" , }')
  987         self.assertEqual(json.loads(on), {'0': ',]'})
  988 
  989         on = js_to_json('{ /*comment\n*/0/*comment\n*/: /* " \n */ ",]" , }')
  990         self.assertEqual(json.loads(on), {'0': ',]'})
  991 
  992         on = js_to_json('{ 0: // comment\n1 }')
  993         self.assertEqual(json.loads(on), {'0': 1})
  994 
  995         on = js_to_json(r'["<p>x<\/p>"]')
  996         self.assertEqual(json.loads(on), ['<p>x</p>'])
  997 
  998         on = js_to_json(r'["\xaa"]')
  999         self.assertEqual(json.loads(on), ['\u00aa'])
 1000 
 1001         on = js_to_json("['a\\\nb']")
 1002         self.assertEqual(json.loads(on), ['ab'])
 1003 
 1004         on = js_to_json("/*comment\n*/[/*comment\n*/'a\\\nb'/*comment\n*/]/*comment\n*/")
 1005         self.assertEqual(json.loads(on), ['ab'])
 1006 
 1007         on = js_to_json('{0xff:0xff}')
 1008         self.assertEqual(json.loads(on), {'255': 255})
 1009 
 1010         on = js_to_json('{/*comment\n*/0xff/*comment\n*/:/*comment\n*/0xff/*comment\n*/}')
 1011         self.assertEqual(json.loads(on), {'255': 255})
 1012 
 1013         on = js_to_json('{077:077}')
 1014         self.assertEqual(json.loads(on), {'63': 63})
 1015 
 1016         on = js_to_json('{/*comment\n*/077/*comment\n*/:/*comment\n*/077/*comment\n*/}')
 1017         self.assertEqual(json.loads(on), {'63': 63})
 1018 
 1019         on = js_to_json('{42:42}')
 1020         self.assertEqual(json.loads(on), {'42': 42})
 1021 
 1022         on = js_to_json('{/*comment\n*/42/*comment\n*/:/*comment\n*/42/*comment\n*/}')
 1023         self.assertEqual(json.loads(on), {'42': 42})
 1024 
 1025         on = js_to_json('{42:4.2e1}')
 1026         self.assertEqual(json.loads(on), {'42': 42.0})
 1027 
 1028         on = js_to_json('{ "0x40": "0x40" }')
 1029         self.assertEqual(json.loads(on), {'0x40': '0x40'})
 1030 
 1031         on = js_to_json('{ "040": "040" }')
 1032         self.assertEqual(json.loads(on), {'040': '040'})
 1033 
 1034     def test_js_to_json_malformed(self):
 1035         self.assertEqual(js_to_json('42a1'), '42"a1"')
 1036         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
 1037 
 1038     def test_extract_attributes(self):
 1039         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
 1040         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})
 1041         self.assertEqual(extract_attributes('<e x=y>'), {'x': 'y'})
 1042         self.assertEqual(extract_attributes('<e x="a \'b\' c">'), {'x': "a 'b' c"})
 1043         self.assertEqual(extract_attributes('<e x=\'a "b" c\'>'), {'x': 'a "b" c'})
 1044         self.assertEqual(extract_attributes('<e x="&#121;">'), {'x': 'y'})
 1045         self.assertEqual(extract_attributes('<e x="&#x79;">'), {'x': 'y'})
 1046         self.assertEqual(extract_attributes('<e x="&amp;">'), {'x': '&'})  # XML
 1047         self.assertEqual(extract_attributes('<e x="&quot;">'), {'x': '"'})
 1048         self.assertEqual(extract_attributes('<e x="&pound;">'), {'x': '£'})  # HTML 3.2
 1049         self.assertEqual(extract_attributes('<e x="&lambda;">'), {'x': 'λ'})  # HTML 4.0
 1050         self.assertEqual(extract_attributes('<e x="&foo">'), {'x': '&foo'})
 1051         self.assertEqual(extract_attributes('<e x="\'">'), {'x': "'"})
 1052         self.assertEqual(extract_attributes('<e x=\'"\'>'), {'x': '"'})
 1053         self.assertEqual(extract_attributes('<e x >'), {'x': None})
 1054         self.assertEqual(extract_attributes('<e x=y a>'), {'x': 'y', 'a': None})
 1055         self.assertEqual(extract_attributes('<e x= y>'), {'x': 'y'})
 1056         self.assertEqual(extract_attributes('<e x=1 y=2 x=3>'), {'y': '2', 'x': '3'})
 1057         self.assertEqual(extract_attributes('<e \nx=\ny\n>'), {'x': 'y'})
 1058         self.assertEqual(extract_attributes('<e \nx=\n"y"\n>'), {'x': 'y'})
 1059         self.assertEqual(extract_attributes("<e \nx=\n'y'\n>"), {'x': 'y'})
 1060         self.assertEqual(extract_attributes('<e \nx="\ny\n">'), {'x': '\ny\n'})
 1061         self.assertEqual(extract_attributes('<e CAPS=x>'), {'caps': 'x'})  # Names lowercased
 1062         self.assertEqual(extract_attributes('<e x=1 X=2>'), {'x': '2'})
 1063         self.assertEqual(extract_attributes('<e X=1 x=2>'), {'x': '2'})
 1064         self.assertEqual(extract_attributes('<e _:funny-name1=1>'), {'_:funny-name1': '1'})
 1065         self.assertEqual(extract_attributes('<e x="Fáilte 世界 \U0001f600">'), {'x': 'Fáilte 世界 \U0001f600'})
 1066         self.assertEqual(extract_attributes('<e x="décompose&#769;">'), {'x': 'décompose\u0301'})
 1067         # "Narrow" Python builds don't support unicode code points outside BMP.
 1068         try:
 1069             compat_chr(0x10000)
 1070             supports_outside_bmp = True
 1071         except ValueError:
 1072             supports_outside_bmp = False
 1073         if supports_outside_bmp:
 1074             self.assertEqual(extract_attributes('<e x="Smile &#128512;!">'), {'x': 'Smile \U0001f600!'})
 1075         # Malformed HTML should not break attributes extraction on older Python
 1076         self.assertEqual(extract_attributes('<mal"formed/>'), {})
 1077 
 1078     def test_clean_html(self):
 1079         self.assertEqual(clean_html('a:\nb'), 'a: b')
 1080         self.assertEqual(clean_html('a:\n   "b"'), 'a:    "b"')
 1081         self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
 1082 
 1083     def test_intlist_to_bytes(self):
 1084         self.assertEqual(
 1085             intlist_to_bytes([0, 1, 127, 128, 255]),
 1086             b'\x00\x01\x7f\x80\xff')
 1087 
 1088     def test_args_to_str(self):
 1089         self.assertEqual(
 1090             args_to_str(['foo', 'ba/r', '-baz', '2 be', '']),
 1091             'foo ba/r -baz \'2 be\' \'\'' if compat_os_name != 'nt' else 'foo ba/r -baz "2 be" ""'
 1092         )
 1093 
 1094     def test_parse_filesize(self):
 1095         self.assertEqual(parse_filesize(None), None)
 1096         self.assertEqual(parse_filesize(''), None)
 1097         self.assertEqual(parse_filesize('91 B'), 91)
 1098         self.assertEqual(parse_filesize('foobar'), None)
 1099         self.assertEqual(parse_filesize('2 MiB'), 2097152)
 1100         self.assertEqual(parse_filesize('5 GB'), 5000000000)
 1101         self.assertEqual(parse_filesize('1.2Tb'), 1200000000000)
 1102         self.assertEqual(parse_filesize('1.2tb'), 1200000000000)
 1103         self.assertEqual(parse_filesize('1,24 KB'), 1240)
 1104         self.assertEqual(parse_filesize('1,24 kb'), 1240)
 1105         self.assertEqual(parse_filesize('8.5 megabytes'), 8500000)
 1106 
 1107     def test_parse_count(self):
 1108         self.assertEqual(parse_count(None), None)
 1109         self.assertEqual(parse_count(''), None)
 1110         self.assertEqual(parse_count('0'), 0)
 1111         self.assertEqual(parse_count('1000'), 1000)
 1112         self.assertEqual(parse_count('1.000'), 1000)
 1113         self.assertEqual(parse_count('1.1k'), 1100)
 1114         self.assertEqual(parse_count('1.1kk'), 1100000)
 1115         self.assertEqual(parse_count('1.1kk '), 1100000)
 1116         self.assertEqual(parse_count('1.1kk views'), 1100000)
 1117 
 1118     def test_parse_resolution(self):
 1119         self.assertEqual(parse_resolution(None), {})
 1120         self.assertEqual(parse_resolution(''), {})
 1121         self.assertEqual(parse_resolution('1920x1080'), {'width': 1920, 'height': 1080})
 1122         self.assertEqual(parse_resolution('1920×1080'), {'width': 1920, 'height': 1080})
 1123         self.assertEqual(parse_resolution('1920 x 1080'), {'width': 1920, 'height': 1080})
 1124         self.assertEqual(parse_resolution('720p'), {'height': 720})
 1125         self.assertEqual(parse_resolution('4k'), {'height': 2160})
 1126         self.assertEqual(parse_resolution('8K'), {'height': 4320})
 1127 
 1128     def test_parse_bitrate(self):
 1129         self.assertEqual(parse_bitrate(None), None)
 1130         self.assertEqual(parse_bitrate(''), None)
 1131         self.assertEqual(parse_bitrate('300kbps'), 300)
 1132         self.assertEqual(parse_bitrate('1500kbps'), 1500)
 1133         self.assertEqual(parse_bitrate('300 kbps'), 300)
 1134 
 1135     def test_version_tuple(self):
 1136         self.assertEqual(version_tuple('1'), (1,))
 1137         self.assertEqual(version_tuple('10.23.344'), (10, 23, 344))
 1138         self.assertEqual(version_tuple('10.1-6'), (10, 1, 6))  # avconv style
 1139 
 1140     def test_detect_exe_version(self):
 1141         self.assertEqual(detect_exe_version('''ffmpeg version 1.2.1
 1142 built on May 27 2013 08:37:26 with gcc 4.7 (Debian 4.7.3-4)
 1143 configuration: --prefix=/usr --extra-'''), '1.2.1')
 1144         self.assertEqual(detect_exe_version('''ffmpeg version N-63176-g1fb4685
 1145 built on May 15 2014 22:09:06 with gcc 4.8.2 (GCC)'''), 'N-63176-g1fb4685')
 1146         self.assertEqual(detect_exe_version('''X server found. dri2 connection failed!
 1147 Trying to open render node...
 1148 Success at /dev/dri/renderD128.
 1149 ffmpeg version 2.4.4 Copyright (c) 2000-2014 the FFmpeg ...'''), '2.4.4')
 1150 
 1151     def test_age_restricted(self):
 1152         self.assertFalse(age_restricted(None, 10))  # unrestricted content
 1153         self.assertFalse(age_restricted(1, None))  # unrestricted policy
 1154         self.assertFalse(age_restricted(8, 10))
 1155         self.assertTrue(age_restricted(18, 14))
 1156         self.assertFalse(age_restricted(18, 18))
 1157 
 1158     def test_is_html(self):
 1159         self.assertFalse(is_html(b'\x49\x44\x43<html'))
 1160         self.assertTrue(is_html(b'<!DOCTYPE foo>\xaaa'))
 1161         self.assertTrue(is_html(  # UTF-8 with BOM
 1162             b'\xef\xbb\xbf<!DOCTYPE foo>\xaaa'))
 1163         self.assertTrue(is_html(  # UTF-16-LE
 1164             b'\xff\xfe<\x00h\x00t\x00m\x00l\x00>\x00\xe4\x00'
 1165         ))
 1166         self.assertTrue(is_html(  # UTF-16-BE
 1167             b'\xfe\xff\x00<\x00h\x00t\x00m\x00l\x00>\x00\xe4'
 1168         ))
 1169         self.assertTrue(is_html(  # UTF-32-BE
 1170             b'\x00\x00\xFE\xFF\x00\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4'))
 1171         self.assertTrue(is_html(  # UTF-32-LE
 1172             b'\xFF\xFE\x00\x00<\x00\x00\x00h\x00\x00\x00t\x00\x00\x00m\x00\x00\x00l\x00\x00\x00>\x00\x00\x00\xe4\x00\x00\x00'))
 1173 
 1174     def test_render_table(self):
 1175         self.assertEqual(
 1176             render_table(
 1177                 ['a', 'bcd'],
 1178                 [[123, 4], [9999, 51]]),
 1179             'a    bcd\n'
 1180             '123  4\n'
 1181             '9999 51')
 1182 
 1183     def test_match_str(self):
 1184         self.assertRaises(ValueError, match_str, 'xy>foobar', {})
 1185         self.assertFalse(match_str('xy', {'x': 1200}))
 1186         self.assertTrue(match_str('!xy', {'x': 1200}))
 1187         self.assertTrue(match_str('x', {'x': 1200}))
 1188         self.assertFalse(match_str('!x', {'x': 1200}))
 1189         self.assertTrue(match_str('x', {'x': 0}))
 1190         self.assertFalse(match_str('x>0', {'x': 0}))
 1191         self.assertFalse(match_str('x>0', {}))
 1192         self.assertTrue(match_str('x>?0', {}))
 1193         self.assertTrue(match_str('x>1K', {'x': 1200}))
 1194         self.assertFalse(match_str('x>2K', {'x': 1200}))
 1195         self.assertTrue(match_str('x>=1200 & x < 1300', {'x': 1200}))
 1196         self.assertFalse(match_str('x>=1100 & x < 1200', {'x': 1200}))
 1197         self.assertFalse(match_str('y=a212', {'y': 'foobar42'}))
 1198         self.assertTrue(match_str('y=foobar42', {'y': 'foobar42'}))
 1199         self.assertFalse(match_str('y!=foobar42', {'y': 'foobar42'}))
 1200         self.assertTrue(match_str('y!=foobar2', {'y': 'foobar42'}))
 1201         self.assertFalse(match_str(
 1202             'like_count > 100 & dislike_count <? 50 & description',
 1203             {'like_count': 90, 'description': 'foo'}))
 1204         self.assertTrue(match_str(
 1205             'like_count > 100 & dislike_count <? 50 & description',
 1206             {'like_count': 190, 'description': 'foo'}))
 1207         self.assertFalse(match_str(
 1208             'like_count > 100 & dislike_count <? 50 & description',
 1209             {'like_count': 190, 'dislike_count': 60, 'description': 'foo'}))
 1210         self.assertFalse(match_str(
 1211             'like_count > 100 & dislike_count <? 50 & description',
 1212             {'like_count': 190, 'dislike_count': 10}))
 1213         self.assertTrue(match_str('is_live', {'is_live': True}))
 1214         self.assertFalse(match_str('is_live', {'is_live': False}))
 1215         self.assertFalse(match_str('is_live', {'is_live': None}))
 1216         self.assertFalse(match_str('is_live', {}))
 1217         self.assertFalse(match_str('!is_live', {'is_live': True}))
 1218         self.assertTrue(match_str('!is_live', {'is_live': False}))
 1219         self.assertTrue(match_str('!is_live', {'is_live': None}))
 1220         self.assertTrue(match_str('!is_live', {}))
 1221         self.assertTrue(match_str('title', {'title': 'abc'}))
 1222         self.assertTrue(match_str('title', {'title': ''}))
 1223         self.assertFalse(match_str('!title', {'title': 'abc'}))
 1224         self.assertFalse(match_str('!title', {'title': ''}))
 1225 
 1226     def test_parse_dfxp_time_expr(self):
 1227         self.assertEqual(parse_dfxp_time_expr(None), None)
 1228         self.assertEqual(parse_dfxp_time_expr(''), None)
 1229         self.assertEqual(parse_dfxp_time_expr('0.1'), 0.1)
 1230         self.assertEqual(parse_dfxp_time_expr('0.1s'), 0.1)
 1231         self.assertEqual(parse_dfxp_time_expr('00:00:01'), 1.0)
 1232         self.assertEqual(parse_dfxp_time_expr('00:00:01.100'), 1.1)
 1233         self.assertEqual(parse_dfxp_time_expr('00:00:01:100'), 1.1)
 1234 
 1235     def test_dfxp2srt(self):
 1236         dfxp_data = '''<?xml version="1.0" encoding="UTF-8"?>
 1237             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 1238             <body>
 1239                 <div xml:lang="en">
 1240                     <p begin="0" end="1">The following line contains Chinese characters and special symbols</p>
 1241                     <p begin="1" end="2">第二行<br/>♪♪</p>
 1242                     <p begin="2" dur="1"><span>Third<br/>Line</span></p>
 1243                     <p begin="3" end="-1">Lines with invalid timestamps are ignored</p>
 1244                     <p begin="-1" end="-1">Ignore, two</p>
 1245                     <p begin="3" dur="-1">Ignored, three</p>
 1246                 </div>
 1247             </body>
 1248             </tt>'''.encode('utf-8')
 1249         srt_data = '''1
 1250 00:00:00,000 --> 00:00:01,000
 1251 The following line contains Chinese characters and special symbols
 1252 
 1253 2
 1254 00:00:01,000 --> 00:00:02,000
 1255 第二行
 1256 ♪♪
 1257 
 1258 3
 1259 00:00:02,000 --> 00:00:03,000
 1260 Third
 1261 Line
 1262 
 1263 '''
 1264         self.assertEqual(dfxp2srt(dfxp_data), srt_data)
 1265 
 1266         dfxp_data_no_default_namespace = '''<?xml version="1.0" encoding="UTF-8"?>
 1267             <tt xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 1268             <body>
 1269                 <div xml:lang="en">
 1270                     <p begin="0" end="1">The first line</p>
 1271                 </div>
 1272             </body>
 1273             </tt>'''.encode('utf-8')
 1274         srt_data = '''1
 1275 00:00:00,000 --> 00:00:01,000
 1276 The first line
 1277 
 1278 '''
 1279         self.assertEqual(dfxp2srt(dfxp_data_no_default_namespace), srt_data)
 1280 
 1281         dfxp_data_with_style = '''<?xml version="1.0" encoding="utf-8"?>
 1282 <tt xmlns="http://www.w3.org/2006/10/ttaf1" xmlns:ttp="http://www.w3.org/2006/10/ttaf1#parameter" ttp:timeBase="media" xmlns:tts="http://www.w3.org/2006/10/ttaf1#style" xml:lang="en" xmlns:ttm="http://www.w3.org/2006/10/ttaf1#metadata">
 1283   <head>
 1284     <styling>
 1285       <style id="s2" style="s0" tts:color="cyan" tts:fontWeight="bold" />
 1286       <style id="s1" style="s0" tts:color="yellow" tts:fontStyle="italic" />
 1287       <style id="s3" style="s0" tts:color="lime" tts:textDecoration="underline" />
 1288       <style id="s0" tts:backgroundColor="black" tts:fontStyle="normal" tts:fontSize="16" tts:fontFamily="sansSerif" tts:color="white" />
 1289     </styling>
 1290   </head>
 1291   <body tts:textAlign="center" style="s0">
 1292     <div>
 1293       <p begin="00:00:02.08" id="p0" end="00:00:05.84">default style<span tts:color="red">custom style</span></p>
 1294       <p style="s2" begin="00:00:02.08" id="p0" end="00:00:05.84"><span tts:color="lime">part 1<br /></span><span tts:color="cyan">part 2</span></p>
 1295       <p style="s3" begin="00:00:05.84" id="p1" end="00:00:09.56">line 3<br />part 3</p>
 1296       <p style="s1" tts:textDecoration="underline" begin="00:00:09.56" id="p2" end="00:00:12.36"><span style="s2" tts:color="lime">inner<br /> </span>style</p>
 1297     </div>
 1298   </body>
 1299 </tt>'''.encode('utf-8')
 1300         srt_data = '''1
 1301 00:00:02,080 --> 00:00:05,839
 1302 <font color="white" face="sansSerif" size="16">default style<font color="red">custom style</font></font>
 1303 
 1304 2
 1305 00:00:02,080 --> 00:00:05,839
 1306 <b><font color="cyan" face="sansSerif" size="16"><font color="lime">part 1
 1307 </font>part 2</font></b>
 1308 
 1309 3
 1310 00:00:05,839 --> 00:00:09,560
 1311 <u><font color="lime">line 3
 1312 part 3</font></u>
 1313 
 1314 4
 1315 00:00:09,560 --> 00:00:12,359
 1316 <i><u><font color="yellow"><font color="lime">inner
 1317  </font>style</font></u></i>
 1318 
 1319 '''
 1320         self.assertEqual(dfxp2srt(dfxp_data_with_style), srt_data)
 1321 
 1322         dfxp_data_non_utf8 = '''<?xml version="1.0" encoding="UTF-16"?>
 1323             <tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en" xmlns:tts="http://www.w3.org/ns/ttml#parameter">
 1324             <body>
 1325                 <div xml:lang="en">
 1326                     <p begin="0" end="1">Line 1</p>
 1327                     <p begin="1" end="2">第二行</p>
 1328                 </div>
 1329             </body>
 1330             </tt>'''.encode('utf-16')
 1331         srt_data = '''1
 1332 00:00:00,000 --> 00:00:01,000
 1333 Line 1
 1334 
 1335 2
 1336 00:00:01,000 --> 00:00:02,000
 1337 第二行
 1338 
 1339 '''
 1340         self.assertEqual(dfxp2srt(dfxp_data_non_utf8), srt_data)
 1341 
 1342     def test_cli_option(self):
 1343         self.assertEqual(cli_option({'proxy': '127.0.0.1:3128'}, '--proxy', 'proxy'), ['--proxy', '127.0.0.1:3128'])
 1344         self.assertEqual(cli_option({'proxy': None}, '--proxy', 'proxy'), [])
 1345         self.assertEqual(cli_option({}, '--proxy', 'proxy'), [])
 1346         self.assertEqual(cli_option({'retries': 10}, '--retries', 'retries'), ['--retries', '10'])
 1347 
 1348     def test_cli_valueless_option(self):
 1349         self.assertEqual(cli_valueless_option(
 1350             {'downloader': 'external'}, '--external-downloader', 'downloader', 'external'), ['--external-downloader'])
 1351         self.assertEqual(cli_valueless_option(
 1352             {'downloader': 'internal'}, '--external-downloader', 'downloader', 'external'), [])
 1353         self.assertEqual(cli_valueless_option(
 1354             {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'), ['--no-check-certificate'])
 1355         self.assertEqual(cli_valueless_option(
 1356             {'nocheckcertificate': False}, '--no-check-certificate', 'nocheckcertificate'), [])
 1357         self.assertEqual(cli_valueless_option(
 1358             {'checkcertificate': True}, '--no-check-certificate', 'checkcertificate', False), [])
 1359         self.assertEqual(cli_valueless_option(
 1360             {'checkcertificate': False}, '--no-check-certificate', 'checkcertificate', False), ['--no-check-certificate'])
 1361 
 1362     def test_cli_bool_option(self):
 1363         self.assertEqual(
 1364             cli_bool_option(
 1365                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate'),
 1366             ['--no-check-certificate', 'true'])
 1367         self.assertEqual(
 1368             cli_bool_option(
 1369                 {'nocheckcertificate': True}, '--no-check-certificate', 'nocheckcertificate', separator='='),
 1370             ['--no-check-certificate=true'])
 1371         self.assertEqual(
 1372             cli_bool_option(
 1373                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
 1374             ['--check-certificate', 'false'])
 1375         self.assertEqual(
 1376             cli_bool_option(
 1377                 {'nocheckcertificate': True}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 1378             ['--check-certificate=false'])
 1379         self.assertEqual(
 1380             cli_bool_option(
 1381                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true'),
 1382             ['--check-certificate', 'true'])
 1383         self.assertEqual(
 1384             cli_bool_option(
 1385                 {'nocheckcertificate': False}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 1386             ['--check-certificate=true'])
 1387         self.assertEqual(
 1388             cli_bool_option(
 1389                 {}, '--check-certificate', 'nocheckcertificate', 'false', 'true', '='),
 1390             [])
 1391 
 1392     def test_ohdave_rsa_encrypt(self):
 1393         N = 0xab86b6371b5318aaa1d3c9e612a9f1264f372323c8c0f19875b5fc3b3fd3afcc1e5bec527aa94bfa85bffc157e4245aebda05389a5357b75115ac94f074aefcd
 1394         e = 65537
 1395 
 1396         self.assertEqual(
 1397             ohdave_rsa_encrypt(b'aa111222', e, N),
 1398             '726664bd9a23fd0c70f9f1b84aab5e3905ce1e45a584e9cbcf9bcc7510338fc1986d6c599ff990d923aa43c51c0d9013cd572e13bc58f4ae48f2ed8c0b0ba881')
 1399 
 1400     def test_pkcs1pad(self):
 1401         data = [1, 2, 3]
 1402         padded_data = pkcs1pad(data, 32)
 1403         self.assertEqual(padded_data[:2], [0, 2])
 1404         self.assertEqual(padded_data[28:], [0, 1, 2, 3])
 1405 
 1406         self.assertRaises(ValueError, pkcs1pad, data, 8)
 1407 
 1408     def test_encode_base_n(self):
 1409         self.assertEqual(encode_base_n(0, 30), '0')
 1410         self.assertEqual(encode_base_n(80, 30), '2k')
 1411 
 1412         custom_table = '9876543210ZYXWVUTSRQPONMLKJIHGFEDCBA'
 1413         self.assertEqual(encode_base_n(0, 30, custom_table), '9')
 1414         self.assertEqual(encode_base_n(80, 30, custom_table), '7P')
 1415 
 1416         self.assertRaises(ValueError, encode_base_n, 0, 70)
 1417         self.assertRaises(ValueError, encode_base_n, 0, 60, custom_table)
 1418 
 1419     def test_caesar(self):
 1420         self.assertEqual(caesar('ace', 'abcdef', 2), 'cea')
 1421         self.assertEqual(caesar('cea', 'abcdef', -2), 'ace')
 1422         self.assertEqual(caesar('ace', 'abcdef', -2), 'eac')
 1423         self.assertEqual(caesar('eac', 'abcdef', 2), 'ace')
 1424         self.assertEqual(caesar('ace', 'abcdef', 0), 'ace')
 1425         self.assertEqual(caesar('xyz', 'abcdef', 2), 'xyz')
 1426         self.assertEqual(caesar('abc', 'acegik', 2), 'ebg')
 1427         self.assertEqual(caesar('ebg', 'acegik', -2), 'abc')
 1428 
 1429     def test_rot47(self):
 1430         self.assertEqual(rot47('youtube-dl'), r'J@FEF36\5=')
 1431         self.assertEqual(rot47('YOUTUBE-DL'), r'*~&%&qt\s{')
 1432 
 1433     def test_urshift(self):
 1434         self.assertEqual(urshift(3, 1), 1)
 1435         self.assertEqual(urshift(-3, 1), 2147483646)
 1436 
 1437     def test_get_element_by_class(self):
 1438         html = '''
 1439             <span class="foo bar">nice</span>
 1440         '''
 1441 
 1442         self.assertEqual(get_element_by_class('foo', html), 'nice')
 1443         self.assertEqual(get_element_by_class('no-such-class', html), None)
 1444 
 1445     def test_get_element_by_attribute(self):
 1446         html = '''
 1447             <span class="foo bar">nice</span>
 1448         '''
 1449 
 1450         self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
 1451         self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
 1452         self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
 1453 
 1454         html = '''
 1455             <div itemprop="author" itemscope>foo</div>
 1456         '''
 1457 
 1458         self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
 1459 
 1460     def test_get_elements_by_class(self):
 1461         html = '''
 1462             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
 1463         '''
 1464 
 1465         self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
 1466         self.assertEqual(get_elements_by_class('no-such-class', html), [])
 1467 
 1468     def test_get_elements_by_attribute(self):
 1469         html = '''
 1470             <span class="foo bar">nice</span><span class="foo bar">also nice</span>
 1471         '''
 1472 
 1473         self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
 1474         self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
 1475         self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
 1476 
 1477     def test_clean_podcast_url(self):
 1478         self.assertEqual(clean_podcast_url('https://www.podtrac.com/pts/redirect.mp3/chtbl.com/track/5899E/traffic.megaphone.fm/HSW7835899191.mp3'), 'https://traffic.megaphone.fm/HSW7835899191.mp3')
 1479         self.assertEqual(clean_podcast_url('https://play.podtrac.com/npr-344098539/edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3'), 'https://edge1.pod.npr.org/anon.npr-podcasts/podcast/npr/waitwait/2020/10/20201003_waitwait_wwdtmpodcast201003-015621a5-f035-4eca-a9a1-7c118d90bc3c.mp3')
 1480 
 1481 
 1482 if __name__ == '__main__':
 1483     unittest.main()

Generated by cgit