Merge pull request #181 from pukkandan/precheck-archive

Check IDs in the archive before downloading webpage
This commit is contained in:
Tom-Oliver Heidel 2020-11-16 22:59:24 +01:00 committed by GitHub
commit 63dcccd07c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -821,12 +821,22 @@ def extract_info(self, url, download=True, ie_key=None, info_dict=None, extra_in
if not ie.suitable(url): if not ie.suitable(url):
continue continue
ie = self.get_info_extractor(ie.ie_key()) ie_key = ie.ie_key()
ie = self.get_info_extractor(ie_key)
if not ie.working(): if not ie.working():
self.report_warning('The program functionality for this site has been marked as broken, ' self.report_warning('The program functionality for this site has been marked as broken, '
'and will probably not work.') 'and will probably not work.')
try: try:
try:
temp_id = ie.extract_id(url) if callable(getattr(ie, 'extract_id', None)) else ie._match_id(url)
except (AssertionError, IndexError):
temp_id = None
if temp_id is not None and self.in_download_archive({'id': temp_id, 'ie_key': ie_key}):
self.to_screen("[%s] %s: has already been recorded in archive" % (
ie_key, temp_id))
break
ie_result = ie.extract(url) ie_result = ie.extract(url)
if ie_result is None: # Finished already (backwards compatibility; listformats and friends should be moved here) if ie_result is None: # Finished already (backwards compatibility; listformats and friends should be moved here)
break break