yt-dlp/yt_dlp/extractor/testurl.py
sepro add96eb9f8
[cleanup] Add more ruff rules (#10149)
Authored by: seproDev

Reviewed-by: bashonly <88596187+bashonly@users.noreply.github.com>
Reviewed-by: Simon Sawicki <contact@grub4k.xyz>
2024-06-12 01:09:58 +02:00

50 lines
2 KiB
Python

import re
from .common import InfoExtractor
from ..utils import ExtractorError
class TestURLIE(InfoExtractor):
""" Allows addressing of the test cases as test:yout.*be_1 """
IE_DESC = False # Do not list
_VALID_URL = r'test(?:url)?:(?P<extractor>.*?)(?:_(?P<num>\d+|all))?$'
def _real_extract(self, url):
from . import gen_extractor_classes
extractor_id, num = self._match_valid_url(url).group('extractor', 'num')
if not extractor_id:
return {'id': ':test', 'title': '', 'url': url}
rex = re.compile(extractor_id, flags=re.IGNORECASE)
matching_extractors = [e for e in gen_extractor_classes() if rex.search(e.IE_NAME)]
if len(matching_extractors) == 0:
raise ExtractorError(f'No extractors matching {extractor_id!r} found', expected=True)
elif len(matching_extractors) > 1:
extractor = next(( # Check for exact match
ie for ie in matching_extractors if ie.IE_NAME.lower() == extractor_id.lower()
), None) or next(( # Check for exact match without plugin suffix
ie for ie in matching_extractors if ie.IE_NAME.split('+')[0].lower() == extractor_id.lower()
), None)
if not extractor:
raise ExtractorError(
'Found multiple matching extractors: {}'.format(' '.join(ie.IE_NAME for ie in matching_extractors)),
expected=True)
else:
extractor = matching_extractors[0]
testcases = tuple(extractor.get_testcases(True))
if num == 'all':
return self.playlist_result(
[self.url_result(tc['url'], extractor) for tc in testcases],
url, f'{extractor.IE_NAME} tests')
try:
tc = testcases[int(num or 0)]
except IndexError:
raise ExtractorError(
f'Test case {num or 0} not found, got only {len(testcases)} tests', expected=True)
self.to_screen(f'Test URL: {tc["url"]}')
return self.url_result(tc['url'], extractor)