caesar,
clean_html,
clean_podcast_url,
+ Config,
date_from_str,
datetime_from_str,
DateRange,
ExtractorError,
find_xpath_attr,
fix_xml_ampersands,
+ format_bytes,
float_or_none,
get_element_by_class,
get_element_by_attribute,
get_elements_by_class,
get_elements_by_attribute,
+ get_element_html_by_class,
+ get_element_html_by_attribute,
+ get_elements_html_by_class,
+ get_elements_html_by_attribute,
+ get_elements_text_and_html_by_attribute,
+ get_element_text_and_html_by_tag,
InAdvancePagedList,
int_or_none,
intlist_to_bytes,
is_html,
js_to_json,
limit_length,
+ locked_file,
merge_dicts,
mimetype2ext,
month_by_name,
compat_chr,
compat_etree_fromstring,
compat_getenv,
+ compat_HTMLParseError,
compat_os_name,
compat_setenv,
)
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'), '--gasdgf')
self.assertEqual(sanitize_filename('--gasdgf', is_id=True), '--gasdgf')
- self.assertEqual(sanitize_filename('.gasdgf'), 'gasdgf')
+ self.assertEqual(sanitize_filename('--gasdgf', is_id=False), '_-gasdgf')
+ self.assertEqual(sanitize_filename('.gasdgf'), '.gasdgf')
self.assertEqual(sanitize_filename('.gasdgf', is_id=True), '.gasdgf')
+ self.assertEqual(sanitize_filename('.gasdgf', is_id=False), 'gasdgf')
forbidden = '"\0\\/'
for fc in forbidden:
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('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('PT1H0.040S'), 3600.04)
self.assertEqual(parse_duration('PT00H03M30SZ'), 210)
self.assertEqual(parse_duration('P0Y0M0DT0H4M20.880S'), 260.88)
+ self.assertEqual(parse_duration('01:02:03:050'), 3723.05)
+ self.assertEqual(parse_duration('103:050'), 103.05)
def test_fix_xml_ampersands(self):
self.assertEqual(
def test_clean_html(self):
self.assertEqual(clean_html('a:\nb'), 'a: b')
- self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
+ self.assertEqual(clean_html('a:\n "b"'), 'a: "b"')
self.assertEqual(clean_html('a<br>\xa0b'), 'a\nb')
def test_intlist_to_bytes(self):
self.assertEqual(parse_count('1000'), 1000)
self.assertEqual(parse_count('1.000'), 1000)
self.assertEqual(parse_count('1.1k'), 1100)
+ self.assertEqual(parse_count('1.1 k'), 1100)
+ self.assertEqual(parse_count('1,1 k'), 1100)
self.assertEqual(parse_count('1.1kk'), 1100000)
self.assertEqual(parse_count('1.1kk '), 1100000)
+ self.assertEqual(parse_count('1,1kk'), 1100000)
+ self.assertEqual(parse_count('100 views'), 100)
+ self.assertEqual(parse_count('1,100 views'), 1100)
self.assertEqual(parse_count('1.1kk views'), 1100000)
+ self.assertEqual(parse_count('10M views'), 10000000)
+ self.assertEqual(parse_count('has 10M views'), 10000000)
def test_parse_resolution(self):
self.assertEqual(parse_resolution(None), {})
self.assertEqual(urshift(3, 1), 1)
self.assertEqual(urshift(-3, 1), 2147483646)
+ GET_ELEMENT_BY_CLASS_TEST_STRING = '''
+ <span class="foo bar">nice</span>
+ '''
+
def test_get_element_by_class(self):
- html = '''
- <span class="foo bar">nice</span>
- '''
+ html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
self.assertEqual(get_element_by_class('foo', html), 'nice')
self.assertEqual(get_element_by_class('no-such-class', html), None)
+ def test_get_element_html_by_class(self):
+ html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
+
+ self.assertEqual(get_element_html_by_class('foo', html), html.strip())
+ self.assertEqual(get_element_by_class('no-such-class', html), None)
+
+ GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING = '''
+ <div itemprop="author" itemscope>foo</div>
+ '''
+
def test_get_element_by_attribute(self):
- html = '''
- <span class="foo bar">nice</span>
- '''
+ html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
self.assertEqual(get_element_by_attribute('class', 'foo bar', html), 'nice')
self.assertEqual(get_element_by_attribute('class', 'foo', html), None)
self.assertEqual(get_element_by_attribute('class', 'no-such-foo', html), None)
- html = '''
- <div itemprop="author" itemscope>foo</div>
- '''
+ html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
self.assertEqual(get_element_by_attribute('itemprop', 'author', html), 'foo')
+ def test_get_element_html_by_attribute(self):
+ html = self.GET_ELEMENT_BY_CLASS_TEST_STRING
+
+ self.assertEqual(get_element_html_by_attribute('class', 'foo bar', html), html.strip())
+ self.assertEqual(get_element_html_by_attribute('class', 'foo', html), None)
+ self.assertEqual(get_element_html_by_attribute('class', 'no-such-foo', html), None)
+
+ html = self.GET_ELEMENT_BY_ATTRIBUTE_TEST_STRING
+
+ self.assertEqual(get_element_html_by_attribute('itemprop', 'author', html), html.strip())
+
+ GET_ELEMENTS_BY_CLASS_TEST_STRING = '''
+ <span class="foo bar">nice</span><span class="foo bar">also nice</span>
+ '''
+ GET_ELEMENTS_BY_CLASS_RES = ['<span class="foo bar">nice</span>', '<span class="foo bar">also nice</span>']
+
def test_get_elements_by_class(self):
- html = '''
- <span class="foo bar">nice</span><span class="foo bar">also nice</span>
- '''
+ html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
self.assertEqual(get_elements_by_class('foo', html), ['nice', 'also nice'])
self.assertEqual(get_elements_by_class('no-such-class', html), [])
+ def test_get_elements_html_by_class(self):
+ html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
+
+ self.assertEqual(get_elements_html_by_class('foo', html), self.GET_ELEMENTS_BY_CLASS_RES)
+ self.assertEqual(get_elements_html_by_class('no-such-class', html), [])
+
def test_get_elements_by_attribute(self):
- html = '''
- <span class="foo bar">nice</span><span class="foo bar">also nice</span>
- '''
+ html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
self.assertEqual(get_elements_by_attribute('class', 'foo bar', html), ['nice', 'also nice'])
self.assertEqual(get_elements_by_attribute('class', 'foo', html), [])
self.assertEqual(get_elements_by_attribute('class', 'no-such-foo', html), [])
+ def test_get_elements_html_by_attribute(self):
+ html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
+
+ self.assertEqual(get_elements_html_by_attribute('class', 'foo bar', html), self.GET_ELEMENTS_BY_CLASS_RES)
+ self.assertEqual(get_elements_html_by_attribute('class', 'foo', html), [])
+ self.assertEqual(get_elements_html_by_attribute('class', 'no-such-foo', html), [])
+
+ def test_get_elements_text_and_html_by_attribute(self):
+ html = self.GET_ELEMENTS_BY_CLASS_TEST_STRING
+
+ self.assertEqual(
+ list(get_elements_text_and_html_by_attribute('class', 'foo bar', html)),
+ list(zip(['nice', 'also nice'], self.GET_ELEMENTS_BY_CLASS_RES)))
+ self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'foo', html)), [])
+ self.assertEqual(list(get_elements_text_and_html_by_attribute('class', 'no-such-foo', html)), [])
+
+ GET_ELEMENT_BY_TAG_TEST_STRING = '''
+ random text lorem ipsum</p>
+ <div>
+ this should be returned
+ <span>this should also be returned</span>
+ <div>
+ this should also be returned
+ </div>
+ closing tag above should not trick, so this should also be returned
+ </div>
+ but this text should not be returned
+ '''
+ GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[32:276]
+ GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT = GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML[5:-6]
+ GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML = GET_ELEMENT_BY_TAG_TEST_STRING.strip()[78:119]
+ GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT = GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML[6:-7]
+
+ def test_get_element_text_and_html_by_tag(self):
+ html = self.GET_ELEMENT_BY_TAG_TEST_STRING
+
+ self.assertEqual(
+ get_element_text_and_html_by_tag('div', html),
+ (self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_TEXT, self.GET_ELEMENT_BY_TAG_RES_OUTERDIV_HTML))
+ self.assertEqual(
+ get_element_text_and_html_by_tag('span', html),
+ (self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_TEXT, self.GET_ELEMENT_BY_TAG_RES_INNERSPAN_HTML))
+ self.assertRaises(compat_HTMLParseError, get_element_text_and_html_by_tag, 'article', html)
+
def test_iri_to_uri(self):
self.assertEqual(
iri_to_uri('https://www.google.com/search?q=foo&ie=utf-8&oe=utf-8&client=firefox-b'),
ll = reversed(ll)
test(ll, -15, 14, range(15))
+ def test_format_bytes(self):
+ self.assertEqual(format_bytes(0), '0.00B')
+ self.assertEqual(format_bytes(1000), '1000.00B')
+ self.assertEqual(format_bytes(1024), '1.00KiB')
+ self.assertEqual(format_bytes(1024**2), '1.00MiB')
+ self.assertEqual(format_bytes(1024**3), '1.00GiB')
+ self.assertEqual(format_bytes(1024**4), '1.00TiB')
+ self.assertEqual(format_bytes(1024**5), '1.00PiB')
+ self.assertEqual(format_bytes(1024**6), '1.00EiB')
+ self.assertEqual(format_bytes(1024**7), '1.00ZiB')
+ self.assertEqual(format_bytes(1024**8), '1.00YiB')
+ self.assertEqual(format_bytes(1024**9), '1024.00YiB')
+
+ def test_hide_login_info(self):
+ self.assertEqual(Config.hide_login_info(['-u', 'foo', '-p', 'bar']),
+ ['-u', 'PRIVATE', '-p', 'PRIVATE'])
+ self.assertEqual(Config.hide_login_info(['-u']), ['-u'])
+ self.assertEqual(Config.hide_login_info(['-u', 'foo', '-u', 'bar']),
+ ['-u', 'PRIVATE', '-u', 'PRIVATE'])
+ self.assertEqual(Config.hide_login_info(['--username=foo']),
+ ['--username=PRIVATE'])
+
+ def test_locked_file(self):
+ TEXT = 'test_locked_file\n'
+ FILE = 'test_locked_file.ytdl'
+ MODES = 'war' # Order is important
+
+ try:
+ for lock_mode in MODES:
+ with locked_file(FILE, lock_mode, False) as f:
+ if lock_mode == 'r':
+ self.assertEqual(f.read(), TEXT * 2, 'Wrong file content')
+ else:
+ f.write(TEXT)
+ for test_mode in MODES:
+ testing_write = test_mode != 'r'
+ try:
+ with locked_file(FILE, test_mode, False):
+ pass
+ except (BlockingIOError, PermissionError):
+ if not testing_write: # FIXME
+ print(f'Known issue: Exclusive lock ({lock_mode}) blocks read access ({test_mode})')
+ continue
+ self.assertTrue(testing_write, f'{test_mode} is blocked by {lock_mode}')
+ else:
+ self.assertFalse(testing_write, f'{test_mode} is not blocked by {lock_mode}')
+ finally:
+ try:
+ os.remove(FILE)
+ except Exception:
+ pass
+
if __name__ == '__main__':
unittest.main()
projects / yt-dlp.git / blobdiff