youtube-dl/test/test_utils.py

#!/usr/bin/env python
# coding: utf-8

from __future__ import unicode_literals

# Allow direct execution
import os
import sys
import unittest
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))


# Various small unit tests
import io
import json
import xml.etree.ElementTree

from youtube_dl.utils import (
    age_restricted,
    args_to_str,
    encode_base_n,
    caesar,
    clean_html,
    date_from_str,
    DateRange,
    detect_exe_version,
    determine_ext,
    dict_get,
    encode_compat_str,
    encodeFilename,
    escape_rfc3986,
    escape_url,
    extract_attributes,
    ExtractorError,
    find_xpath_attr,
    fix_xml_ampersands,
    float_or_none,
    get_element_by_class,
    get_element_by_attribute,
    get_elements_by_class,
    get_elements_by_attribute,
    InAdvancePagedList,
    int_or_none,
    intlist_to_bytes,
    is_html,
    js_to_json,
    limit_length,
    merge_dicts,
    mimetype2ext,
    month_by_name,
    multipart_encode,
    ohdave_rsa_encrypt,
    OnDemandPagedList,
    orderedSet,
    parse_age_limit,
    parse_duration,
    parse_filesize,
    parse_count,
    parse_iso8601,
    parse_resolution,
    parse_bitrate,
    pkcs1pad,
    read_batch_urls,
    sanitize_filename,
    sanitize_path,
    sanitize_url,
    expand_path,
    prepend_extension,
    replace_extension,
    remove_start,
    remove_end,
    remove_quotes,
    rot47,
    shell_quote,
    smuggle_url,
    str_to_int,
    strip_jsonp,
    strip_or_none,
    subtitles_filename,
    timeconvert,
    unescapeHTML,
    unified_strdate,
    unified_timestamp,
    unsmuggle_url,
    uppercase_escape,
    lowercase_escape,
    url_basename,
    url_or_none,
    base_url,
    urljoin,
    urlencode_postdata,
    urshift,
    update_url_query,
    version_tuple,
    xpath_with_ns,
    xpath_element,
    xpath_text,
    xpath_attr,
    render_table,
    match_str,
    parse_dfxp_time_expr,
    dfxp2srt,
    cli_option,
    cli_valueless_option,
    cli_bool_option,
    parse_codecs,
)
from youtube_dl.compat import (
    compat_chr,
    compat_etree_fromstring,
    compat_getenv,
    compat_os_name,
    compat_setenv,
    compat_urlparse,
    compat_parse_qs,
)


class TestUtil(unittest.TestCase):
    def test_timeconvert(self):
        self.assertTrue(timeconvert('') is None)
        self.assertTrue(timeconvert('bougrg') is None)

    def test_sanitize_filename(self):
        self.assertEqual(sanitize_filename('abc'), 'abc')
        self.assertEqual(sanitize_filename('abc_d-e'), 'abc_d-e')

        self.assertEqual(sanitize_filename('123'), '123')

        self.assertEqual('abc_de', sanitize_filename('abc/de'))
        self.assertFalse('/' in sanitize_filename('abc/de///'))

        self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de'))
        self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|'))
        self.assertEqual('yes no', sanitize_filename('yes? no'))
        self.assertEqual('this - that', sanitize_filename('this: that'))

        self.assertEqual(sanitize_filename('AT&T'), 'AT&T')
        aumlaut = 'ä'
        self.assertEqual(sanitize_filename(aumlaut), aumlaut)
        tests = '\u043a\u0438\u0440\u0438\u043b\u043b\u0438\u0446\u0430'
        self.assertEqual(sanitize_filename(tests), tests)

        self.assertEqual(
            sanitize_filename('New World record at 0:12:34'),
            'New World record at 0_12_34')

        self.assertEqual(sanitize_filename('--gasdgf'), '_-gasdgf')
        self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
        self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
        self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')

        forbidden = '"\0\\/'
        for fc in forbidden:
            for fbc in forbidden:
                self.assertTrue(fbc not in sanitize_filename(fc))

    def test_sanitize_filename_restricted(self):
        self.assertEqual(sanitize_filename('abc', restricted=True), 'abc')
        self.assertEqual(sanitize_filename('abc_d-e', restricted=True), 'abc_d-e')

        self.assertEqual(sanitize_filename('123', restricted=True), '123')

        self.assertEqual('abc_de', sanitize_filename('abc/de', restricted=True))
        self.assertFalse('/' in sanitize_filename('abc/de///', restricted=True))

        self.assertEqual('abc_de', sanitize_filename('abc/<>\\*|de', restricted=True))
        self.assertEqual('xxx', sanitize_filename('xxx/<>\\*|', restricted=True))
        self.assertEqual('yes_no', sanitize_filename('yes? no', restricted=True))
        self.assertEqual('this_-_that', sanitize_filename('this: that', restricted=True))

        tests = 'aäb\u4e2d\u56fd\u7684c'
        self.assertEqual(sanitize_filename(tests, restricted=True), 'aab_c')
        self.assertTrue(sanitize_filename('\xf6', restricted=True) != '')  # No empty filename

        forbidden = '"\0\\/&!: \'\t\n()[]{}$;`^,#'
        for fc in forbidden:
            for fbc in forbidden:
                self.assertTrue(fbc not in sanitize_filename(fc, restricted=True))

        # Handle a common case more neatly
        self.assertEqual(sanitize_filename('\u5927\u58f0\u5e26 - Song', restricted=True), 'Song')
        self.assertEqual(sanitize_filename('\u603b\u7edf: Speech', restricted=True), 'Speech')
        # .. but make sure the file name is never empty
        self.assertTrue(sanitize_filename('-', restricted=True) != '')
        self.assertTrue(sanitize_filename(':', restricted=True) != '')

        self.assertEqual(sanitize_filename(
            'ÂÃÄÀÁÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖŐØŒÙÚÛÜŰÝÞßàáâãäåæçèéêëìíîïðñòóôõöőøœùúûüűýþÿ', restricted=True),
            'AAAAAAAECEEEEIIIIDNOOOOOOOOEUUUUUYTHssaaaaaaaeceeeeiiiionooooooooeuuuuuythy')

    def test_sanitize_ids(self):
        self.assertEqual(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
        self.assertEqual(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
        self.assertEqual(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')

    def test_sanitize_path(self):
        if sys.platform != 'win32':
            return

        self.assertEqual(sanitize_path('abc'), 'abc')
        self.assertEqual(sanitize_path('abc/def'), 'abc\\def')
        self.assertEqual(sanitize_path('abc\\def'), 'abc\\def')
        self.assertEqual(sanitize_path('abc|def'), 'abc#def')
        self.assertEqual(sanitize_path('<>:"|?*'), '#######')
        self.assertEqual(sanitize_path('C:/abc/def'), 'C:\\abc\\def')
        self.assertEqual(sanitize_path('C?:/abc/def'), 'C##\\abc\\def')

        self.assertEqual(sanitize_path('\\\\?\\UNC\\ComputerName\\abc'), '\\\\?\\UNC\\ComputerName\\abc')
        self.assertEqual(sanitize_path('\\\\?\\UNC/ComputerName/abc'), '\\\\?\\UNC\\ComputerName\\abc')

        self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')
        self.assertEqual(sanitize_path('\\\\?\\C:/abc'), '\\\\?\\C:\\abc')
        self.assertEqual(sanitize_path('\\\\?\\C:\\ab?c\\de:f'), '\\\\?\\C:\\ab#c\\de#f')
        self.assertEqual(sanitize_path('\\\\?\\C:\\abc'), '\\\\?\\C:\\abc')

        self.assertEqual(
            sanitize_path('youtube/%(uploader)s/%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s'),
            'youtube\\%(uploader)s\\%(autonumber)s-%(title)s-%(upload_date)s.%(ext)s')

        self.assertEqual(
            sanitize_path('youtube/TheWreckingYard ./00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part'),
            'youtube\\TheWreckingYard #\\00001-Not bad, Especially for Free! (1987 Yamaha 700)-20141116.mp4.part')
        self.assertEqual(sanitize_path('abc/def...'), 'abc\\def..#')
        self.assertEqual(sanitize_path('abc.../def'), 'abc..#\\def')
        self.assertEqual(sanitize_path('abc.../def...'), 'abc..#\\def..#')

        self.assertEqual(sanitize_path('../abc'), '..\\abc')
        self.assertEqual(sanitize_path('../../abc'), '..\\..\\abc')
        self.assertEqual(sanitize_path('./abc'), 'abc')
        self.assertEqual(sanitize_path('./../abc'), '..\\abc')

    def test_sanitize_url(self):
        self.assertEqual(sanitize_url('//foo.bar'), 'http://foo.bar')
        self.assertEqual(sanitize_url('httpss://foo.bar'), 'https://foo.bar')
        self.assertEqual(sanitize_url('rmtps://foo.bar'), 'rtmps://foo.bar')
        self.assertEqual(sanitize_url('https://foo.bar'), 'https://foo.bar')

    def test_expand_path(self):
        def env(var):
            return '%{0}%'.format(var) if sys.platform == 'win32' else '${0}'.format(var)

        compat_setenv('YOUTUBE_DL_EXPATH_PATH', 'expanded')
        self.assertEqual(expand_path(env('YOUTUBE_DL_EXPATH_PATH')), 'expanded')
        self.assertEqual(expand_path(env('HOME')), compat_getenv('HOME'))
        self.assertEqual(expand_path('~'), compat_getenv('HOME'))
        self.assertEqual(
            expand_path('~/%s' % env('YOUTUBE_DL_EXPATH_PATH')),
            '%s/expanded' % compat_getenv('HOME'))

    def test_prepend_extension(self):
        self.assertEqual(prepend_extension('abc.ext', 'temp'), 'abc.temp.ext')
        self.assertEqual(prepend_extension('abc.ext', 'temp', 'ext'), 'abc.temp.ext')
        self.assertEqual(prepend_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
        self.assertEqual(prepend_extension('abc', 'temp'), 'abc.temp')
        self.assertEqual(prepend_extension('.abc', 'temp'), '.abc.temp')
        self.assertEqual(prepend_extension('.abc.ext', 'temp'), '.abc.temp.ext')

    def test_replace_extension(self):
        self.assertEqual(replace_extension('abc.ext', 'temp'), 'abc.temp')
        self.assertEqual(replace_extension('abc.ext', 'temp', 'ext'), 'abc.temp')
        self.assertEqual(replace_extension('abc.unexpected_ext', 'temp', 'ext'), 'abc.unexpected_ext.temp')
        self.assertEqual(replace_extension('abc', 'temp'), 'abc.temp')
        self.assertEqual(replace_extension('.abc', 'temp'), '.abc.temp')
        self.assertEqual(replace_extension('.abc.ext', 'temp'), '.abc.temp')

    def test_subtitles_filename(self):
        self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt'), 'abc.en.vtt')
        self.assertEqual(subtitles_filename('abc.ext', 'en', 'vtt', 'ext'), 'abc.en.vtt')
        self.assertEqual(subtitles_filename('abc.unexpected_ext', 'en', 'vtt', 'ext'), 'abc.unexpected_ext.en.vtt')

    def test_remove_start(self):
        self.assertEqual(remove_start(None, 'A - '), None)
        self.assertEqual(remove_start('A - B', 'A - '), 'B')
        self.assertEqual(remove_start('B - A', 'A - '), 'B - A')

    def test_remove_end(self):
        self.assertEqual(remove_end(None, ' - B'), None)
        self.assertEqual(remove_end('A - B', ' - B'), 'A')
        self.assertEqual(remove_end('B - A', ' - B'), 'B - A')

    def test_remove_quotes(self):
        self.assertEqual(remove_quotes(None), None)
        self.assertEqual(remove_quotes('"'), '"')
        self.assertEqual(remove_quotes("'"), "'")
        self.assertEqual(remove_quotes(';'), ';')
        self.assertEqual(remove_quotes('";'), '";')
        self.assertEqual(remove_quotes('""'), '')
        self.assertEqual(remove_quotes('";"'), ';')

    def test_ordered_set(self):
        self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
        self.assertEqual(orderedSet([]), [])
        self.assertEqual(orderedSet([1]), [1])
        # keep the list ordered
        self.assertEqual(orderedSet([135, 1, 1, 1]), [135, 1])

    def test_unescape_html(self):
        self.assertEqual(unescapeHTML('%20;'), '%20;')
        self.assertEqual(unescapeHTML('&#x2F;'), '/')
        self.assertEqual(unescapeHTML('&#47;'), '/')
        self.assertEqual(unescapeHTML('&eacute;'), 'é')
        self.assertEqual(unescapeHTML('&#2013266066;'), '&#2013266066;')
        self.assertEqual(unescapeHTML('&a&quot;'), '&a"')
        # HTML5 entities
        self.assertEqual(unescapeHTML('&period;&apos;'), '.\'')

    def test_date_from_str(self):
        self.assertEqual(date_from_str('yesterday'), date_from_str('now-1day'))
        self.assertEqual(date_from_str('now+7day'), date_from_str('now+1week'))
        self.assertEqual(date_from_str('now+14day'), date_from_str('now+2week'))
        self.assertEqual(date_from_str('now+365day'), date_from_str('now+1year'))
        self.assertEqual(date_from_str('now+30day'), date_from_str('now+1month'))

    def test_daterange(self):
        _20century = DateRange("19000101", "20000101")
        self.assertFalse("17890714" in _20century)
        _ac = DateRange("00010101")
        self.assertTrue("19690721" in _ac)
        _firstmilenium = DateRange(end="10000101")
        self.assertTrue("07110427" in _firstmilenium)

    def test_unified_dates(self):
        self.assertEqual(unified_strdate('December 21, 2010'), '20101221')
        self.assertEqual(unified_strdate('8/7/2009'), '20090708')
        self.assertEqual(unified_strdate('Dec 14, 2012'), '20121214')
        self.assertEqual(unified_strdate('2012/10/11 01:56:38 +0000'), '20121011')
        self.assertEqual(unified_strdate('1968 12 10'), '19681210')
        self.assertEqual(unified_strdate('1968-12-10'), '19681210')
        self.assertEqual(unified_strdate('28/01/2014 21:00:00 +0100'), '20140128')
        self.assertEqual(
            unified_strdate('11/26/2014 11:30:00 AM PST', day_first=False),
            '20141126')
        self.assertEqual(
            unified_strdate('2/2/2015 6:47:40 PM', day_first=False),
            '20150202')
        self.assertEqual(unified_strdate('Feb 14th 2016 5:45PM'), '20160214')
        self.assertEqual(unified_strdate('25-09-2014'), '20140925')
        self.assertEqual(unified_strdate('27.02.2016 17:30'), '20160227')
        self.assertEqual(unified_strdate('UNKNOWN DATE FORMAT'), None)
        self.assertEqual(unified_strdate('Feb 7, 2016 at 6:35 pm'), '20160207')
        self.assertEqual(unified_strdate('July 15th, 2013'), '20130715')
        self.assertEqual(unified_strdate('September 1st, 2013'), '20130901')
        self.assertEqual(unified_strdate('Sep 2nd, 2013'), '20130902')
        self.assertEqual(unified_strdate('November 3rd, 2019'), '20191103')
        self.assertEqual(unified_strdate('October 23rd, 2005'), '20051023')

    def test_unified_timestamps(self):
        self.assertEqual(unified_timestamp('December 21, 2010'), 1292889600)
        self.assertEqual(unified_timestamp('8/7/2009'), 1247011200)
        self.assertEqual(unified_timestamp('Dec 14, 2012'), 1355443200)
        self.assertEqual(unified_timestamp('2012/10/11 01:56:38 +0000'), 1349920598)
        self.assertEqual(unified_timestamp('1968 12 10'), -33436800)
        self.assertEqual(unified_timestamp('1968-12-10'), -33436800)
        self.assertEqual(unified_timestamp('28/01/2014 21:00:00 +0100'), 1390939200)
        self.assertEqual(
            unified_timestamp('11/26/2014 11:30:00 AM PST', day_first=False),
            1417001400)
        self.assertEqual(
            unified_timestamp('2/2/2015 6:47:40 PM', day_first=False),
            1422902860)
        self.assertEqual(unified_timestamp('Feb 14th 2016 5:45PM'), 1455471900)
        self.assertEqual(unified_timestamp('25-09-2014'), 1411603200)
        self.assertEqual(unified_timestamp('27.02.2016 17:30'), 1456594200)
        self.assertEqual(unified_timestamp('UNKNOWN DATE FORMAT'), None)
        self.assertEqual(unified_timestamp('May 16, 2016 11:15 PM'), 1463440500)
        self.assertEqual(unified_timestamp('Feb 7, 2016 at 6:35 pm'), 1454870100)
        self.assertEqual(unified_timestamp('2017-03-30T17:52:41Q'), 1490896361)
        self.assertEqual(unified_timestamp('Sep 11, 2013 | 5:49 AM'), 1378878540)
        self.assertEqual(unified_timestamp('December 15, 2017 at 7:49 am'), 1513324140)
        self.assertEqual(unified_timestamp('2018-03-14T08:32:43.1493874+00:00'), 1521016363)

    def test_determine_ext(self):
        self.assertEqual(determine_ext('http://example.com/foo/bar.mp4/?download'), 'mp4')
        self.assertEqual(determine_ext('http://example.com/foo/bar/?download', None), None)
        self.assertEqual(determine_ext('http://example.com/foo/bar.nonext/?download', None), None)
        self.assertEqual(determine_ext('http://example.com/foo/bar/mp4?download', None), None)
        self.assertEqual(determine_ext('http://example.com/foo/bar.m3u8//?download'), 'm3u8')
        self.assertEqual(determine_ext('foobar', None), None)

    def test_find_xpath_attr(self):
        testxml = '''<root>
            <node/>
            <node x="a"/>
            <node x="a" y="c" />
            <node x="b" y="d" />
            <node x="" />
        </root>'''
        doc = compat_etree_fromstring(testxml)

        self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n'), None)
        self.assertEqual(find_xpath_attr(doc, './/fourohfour', 'n', 'v'), None)
        self.assertEqual(find_xpath_attr(doc, './/node', 'n'), None)
        self.assertEqual(find_xpath_attr(doc, './/node', 'n', 'v'), None)
        self.assertEqual(find_xpath_attr(doc, './/node', 'x'), doc[1])
        self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'a'), doc[1])
        self.assertEqual(find_xpath_attr(doc, './/node', 'x', 'b'), doc[3])
        self.assertEqual(find_xpath_attr(doc, './/node', 'y'), doc[2])
        self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'c'), doc[2])
        self.assertEqual(find_xpath_attr(doc, './/node', 'y', 'd'), doc[3])
        self.assertEqual(find_xpath_attr(doc, './/node', 'x', ''), doc[4])

    def test_xpath_with_ns(self):
        testxml = '''<root xmlns:media="http://example.com/">
            <media:song>
                <media:author>The Author</media:author>
                <url>http://server.com/download.mp3</url>
            </media:song>
        </root>'''
        doc = compat_etree_fromstring(testxml)
        find = lambda p: doc.find(xpath_with_ns(p, {'media': 'http://example.com/'}))
        self.assertTrue(find('media:song') is not None)
        self.assertEqual(find('media:song/media:author').text, 'The Author')
        self.assertEqual(find('media:song/url').text, 'http://server.com/download.mp3')

    def test_xpath_element(self):
        doc = xml.etree.ElementTree.Element('root')
        div = xml.etree.ElementTree.SubElement(doc, 'div')
        p = xml.etree.ElementTree.SubElement(div, 'p')
        p.text = 'Foo'
        self.assertEqual(xpath_element(doc, 'div/p'), p)
        self.assertEqual(xpath_element(doc, ['div/p']), p)
        self.assertEqual(xpath_element(doc, ['div/bar', 'div/p']), p)
        self.assertEqual(xpath_element(doc, 'div/bar', default='default'), 'default')
        self.assertEqual(xpath_element(doc, ['div/bar'], default='default'), 'default')
        self.assertTrue(xpath_element(doc, 'div/bar') is None)
        self.assertTrue(xpath_element(doc, ['div/bar']) is None)
        self.assertTrue(xpath_element(doc, ['div/bar'], 'div/baz') is None)
        self.assertRaises(ExtractorError, xpath_element, doc, 'div/bar', fatal=True)
        self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar'], fatal=True)
        self.assertRaises(ExtractorError, xpath_element, doc, ['div/bar', 'div/baz'], fatal=True)

    def test_xpath_text(self):
        testxml = '''<root>
            <div>
                <p>Foo</p>
            </div>
        </root>'''
        doc = compat_etree_fromstring(testxml)
        self.assertEqual(xpath_text(doc, 'div/p'), 'Foo')
        self.assertEqual(xpath_text(doc, 'div/bar', default='default'), 'default')
        self.assertTrue(xpath_text(doc, 'div/bar') is None)
        self.assertRaises(ExtractorError, xpath_text, doc, 'div/bar', fatal=True)

    def test_xpath_attr(self):
        testxml = '''<root>
            <div>
                <p x="a">Foo</p>
            </div>
        </root>'''
        doc = compat_etree_fromstring(testxml)
        self.assertEqual(xpath_attr(doc, 'div/p', 'x'), 'a')
        self.assertEqual(xpath_attr(doc, 'div/bar', 'x'), None)
        self.assertEqual(xpath_attr(doc, 'div/p', 'y'), None)
        self.assertEqual(xpath_attr(doc, 'div/bar', 'x', default='default'), 'default')
        self.assertEqual(xpath_attr(doc, 'div/p', 'y', default='default'), 'default')
        self.assertRaises(ExtractorError, xpath_attr, doc, 'div/bar', 'x', fatal=True)
        self.assertRaises(ExtractorError, xpath_attr, doc, 'div/p', 'y', fatal=True)

    def test_smuggle_url(self):
        data = {"ö": "ö", "abc": [3]}
        url = 'https://foo.bar/baz?x=y#a'
        smug_url = smuggle_url(url, data)
        unsmug_url, unsmug_data = unsmuggle_url(smug_url)
        self.assertEqual(url, unsmug_url)
        self.assertEqual(data, unsmug_data)

        res_url, res_data = unsmuggle_url(url)
        self.assertEqual(res_url, url)
        self.assertEqual(res_data, None)

        smug_url = smuggle_url(url, {'a': 'b'})
        smug_smug_url = smuggle_url(smug_url, {'c': 'd'})
        res_url, res_data = unsmuggle_url(smug_smug_url)
        self.assertEqual(res_url, url)
        self.assertEqual(res_data, {'a': 'b', 'c': 'd'})

    def test_shell_quote(self):
        args = ['ffmpeg', '-i', encodeFilename('ñ€ß\'.mp4')]
        self.assertEqual(
            shell_quote(args),
            """ffmpeg -i 'ñ€ß'"'"'.mp4'""" if compat_os_name != 'nt' else '''ffmpeg -i "ñ€ß'.mp4"''')

    def test_float_or_none(self):
        self.assertEqual(float_or_none('42.42'), 42.42)
        self.assertEqual(float_or_none('42'), 42.0)
        self.assertEqual(float_or_none(''), None)
        self.assertEqual(float_or_none(None), None)
        self.assertEqual(float_or_none([]), None)
        self.assertEqual(float_or_none(set()), None)

    def test_int_or_none(self):
        self.assertEqual(int_or_none('42'), 42)
        self.assertEqual(int_or_none(''), None)
        self.assertEqual(int_or_none(None), None)
        self.assertEqual(int_or_none([]), None)
        self.assertEqual(int_or_none(set()), None)

    def test_str_to_int(self):
        self.assertEqual(str_to_int('123,456'), 123456)
        self.assertEqual(str_to_int('123.456'), 123456)
        self.assertEqual(str_to_int(523), 523)
        # Python 3 has no long
        if sys.version_info < (3, 0):
            eval('self.assertEqual(str_to_int(123456L), 123456)')
        self.assertEqual(str_to_int('noninteger'), None)
        self.assertEqual(str_to_int([]), None)

    def test_url_basename(self):
        self.assertEqual(url_basename('http://foo.de/'), '')
        self.assertEqual(url_basename('http://foo.de/bar/baz'), 'baz')
        self.assertEqual(url_basename('http://foo.de/bar/baz?x=y'), 'baz')
        self.assertEqual(url_basename('http://foo.de/bar/baz#x=y'), 'baz')
        self.assertEqual(url_basename('http://foo.de/bar/baz/'), 'baz')
        self.assertEqual(
            url_basename('http://media.w3.org/2010/05/sintel/trailer.mp4'),
            'trailer.mp4')

    def test_base_url(self):
        self.assertEqual(base_url('http://foo.de/'), 'http://foo.de/')
        self.assertEqual(base_url('http://foo.de/bar'), 'http://foo.de/')
        self.assertEqual(base_url('http://foo.de/bar/'), 'http://foo.de/bar/')
        self.assertEqual(base_url('http://foo.de/bar/baz'), 'http://foo.de/bar/')
        self.assertEqual(base_url('http://foo.de/bar/baz?x=z/x/c'), 'http://foo.de/bar/')

    def test_urljoin(self):
        self.assertEqual(urljoin('http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin(b'http://foo.de/', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin(b'http://foo.de/', b'/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('//foo.de/', '/a/b/c.txt'), '//foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de/', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de', '/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de', 'a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de/', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de/', '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
        self.assertEqual(urljoin(None, 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin(None, '//foo.de/a/b/c.txt'), '//foo.de/a/b/c.txt')
        self.assertEqual(urljoin('', 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin(['foobar'], 'http://foo.de/a/b/c.txt'), 'http://foo.de/a/b/c.txt')
        self.assertEqual(urljoin('http://foo.de/', None), None)
        self.assertEqual(urljoin('http://foo.de/', ''), None)
        self.assertEqual(urljoin('http://foo.de/', ['foobar']), None)
        self.assertEqual(urljoin('http://foo.de/a/b/c.txt', '.././../d.txt'), 'http://foo.de/d.txt')
        self.assertEqual(urljoin('http://foo.de/a/b/c.txt', 'rtmp://foo.de'), 'rtmp://foo.de')
        self.assertEqual(urljoin(None, 'rtmp://foo.de'), 'rtmp://foo.de')

    def test_url_or_none(self):
        self.assertEqual(url_or_none(None), None)
        self.assertEqual(url_or_none(''), None)
        self.assertEqual(url_or_none('foo'), None)
        self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
        self.assertEqual(url_or_none('https://foo.de'), 'https://foo.de')
        self.assertEqual(url_or_none('http$://foo.de'), None)
        self.assertEqual(url_or_none('http://foo.de'), 'http://foo.de')
        self.assertEqual(url_or_none('//foo.de'), '//foo.de')

    def test_parse_age_limit(self):
        self.assertEqual(parse_age_limit(None), None)
        self.assertEqual(parse_age_limit(False), None)
        self.assertEqual(parse_age_limit('invalid'), None)
        self.assertEqual(parse_age_limit(0), 0)
        self.assertEqual(parse_age_limit(18), 18)
        self.assertEqual(parse_age_limit(21), 21)
        self.assertEqual(parse_age_limit(22), None)
        self.assertEqual(parse_age_limit('18'), 18)
        self.assertEqual(parse_age_limit('18+'), 18)
        self.assertEqual(parse_age_limit('PG-13'), 13)
        self.assertEqual(parse_age_limit('TV-14'), 14)
        self.assertEqual(parse_age_limit('TV-MA'), 17)
        self.assertEqual(parse_age_limit('TV14'), 14)
        self.assertEqual(parse_age_limit('TV_G'), 0)

    def test_parse_duration(self):
        self.assertEqual(parse_duration(None), None)
        self.assertEqual(parse_duration(False), None)
        self.assertEqual(parse_duration('invalid'), None)
        self.assertEqual(parse_duration('1'), 1)
        self.assertEqual(parse_duration('1337:12'), 80232)
        self.assertEqual(parse_duration('9:12:43'), 33163)
        self.assertEqual(parse_duration('12:00'), 720)
        self.assertEqual(parse_duration('00:01:01'), 61)
        self.assertEqual(parse_duration('x:y'), None)
        self.assertEqual(parse_duration('3h11m53s'), 11513)
        self.assertEqual(parse_duration('3h 11m 53s'), 11513)
        self.assertEqual(parse_duration('3 hours 11 minutes 53 seconds'), 11513)
        self.assertEqual(parse_duration('3 hours 11 mins 53 secs'), 11513)
        self.assertEqual(parse_duration('62m45s'), 3765)
        self.assertEqual(parse_duration('6m59s'), 419)
        self.assertEqual(parse_duration('49s'), 49)
        self.assertEqual(parse_duration('0h0m0s'), 0)
        self.assertEqual(parse_duration('0m0s'), 0)
        self.assertEqual(parse_duration('0s'), 0)
        self.assertEqual(parse_duration('01:02:03.05'), 3723.05)
        self.assertEqual(parse_duration('T30M38S'), 1838)
        self.assertEqual(parse_duration('5 s'), 5)
        self.assertEqual(parse_duration('3 min'), 180)
        self.assertEqual(parse_duration('2.5 hours'), 9000)
        self.assertEqual(parse_duration('02:03:04'), 7384)
        self.assertEqual(parse_duration('01:02:03:04'), 93784)
        self.assertEqual(parse_duration('1 hour 3 minutes'), 3780)
        self.assertEqual(parse_duration('87 Min.'), 5220)
        self.assertEqual(parse_duration('PT1H0.040S'), 3600.04)
        self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
        self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)

    def test_fix_xml_ampersands(self):