0
0
Fork 0
mirror of https://github.com/yt-dlp/yt-dlp.git synced 2024-11-24 02:35:12 +00:00
yt-dlp/yt_dlp/extractor/commonmistakes.py
2021-05-17 18:59:51 +05:30

50 lines
1.5 KiB
Python

from __future__ import unicode_literals
import sys
from .common import InfoExtractor
from ..utils import ExtractorError
class CommonMistakesIE(InfoExtractor):
IE_DESC = False # Do not list
_VALID_URL = r'''(?x)
(?:url|URL)$
'''
_TESTS = [{
'url': 'url',
'only_matching': True,
}, {
'url': 'URL',
'only_matching': True,
}]
def _real_extract(self, url):
msg = (
'You\'ve asked yt-dlp to download the URL "%s". '
'That doesn\'t make any sense. '
'Simply remove the parameter in your command or configuration.'
) % url
if not self.get_param('verbose'):
msg += ' Add -v to the command line to see what arguments and configuration yt-dlp has'
raise ExtractorError(msg, expected=True)
class UnicodeBOMIE(InfoExtractor):
IE_DESC = False
_VALID_URL = r'(?P<bom>\ufeff)(?P<id>.*)$'
# Disable test for python 3.2 since BOM is broken in re in this version
# (see https://github.com/ytdl-org/youtube-dl/issues/9751)
_TESTS = [] if (3, 0) < sys.version_info <= (3, 3) else [{
'url': '\ufeffhttp://www.youtube.com/watch?v=BaW_jenozKc',
'only_matching': True,
}]
def _real_extract(self, url):
real_url = self._match_id(url)
self.report_warning(
'Your URL starts with a Byte Order Mark (BOM). '
'Removing the BOM and looking for "%s" ...' % real_url)
return self.url_result(real_url)