indymedia
/
epicyon
mirror of https://gitlab.com/bashrc2/epicyon
7
4
Fork 1
Code Issues 14 Releases Wiki Activity

Different rss link style

merge-requests/30/head
Bob Mottram 2022-01-13 22:26:01 +00:00
parent f49863e5d7
commit f8ee7e4da5
1 changed files with 17 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
newswire.py
View File

@ -567,10 +567,20 @@ def get_link_from_rss_item(rss_item: str) -> (str, str):
if '://' in link: if '://' in link:
return link, mime_type return link, mime_type
if '<link>' in rss_item and '</link>' in rss_item:
link = rss_item.split('<link>')[1] link = rss_item.split('<link>')[1]
link = link.split('</link>')[0] link = link.split('</link>')[0]
if '://' not in link: if '://' not in link:
return None, None return None, None
elif '<link ' in rss_item:
link_str = rss_item.split('<link ')[1]
if '>' in link_str:
link_str = link_str.split('>')[0]
if 'href="' in link_str:
link_str = link_str.split('href="')[1]
if '"' in link_str:
link = link_str.split('"')[0]
return link, mime_type return link, mime_type
@ -609,9 +619,7 @@ def _xml2str_to_dict(base_dir: str, domain: str, xml_str: str,
continue continue
if '</title>' not in rss_item: if '</title>' not in rss_item:
continue continue
if '<link>' not in rss_item: if '<link' not in rss_item:
continue
if '</link>' not in rss_item:
continue continue
if '<pubDate>' not in rss_item: if '<pubDate>' not in rss_item:
continue continue
@ -708,9 +716,7 @@ def _xml1str_to_dict(base_dir: str, domain: str, xml_str: str,
continue continue
if '</title>' not in rss_item: if '</title>' not in rss_item:
continue continue
if '<link>' not in rss_item: if '<link' not in rss_item:
continue
if '</link>' not in rss_item:
continue continue
if '<dc:date>' not in rss_item: if '<dc:date>' not in rss_item:
continue continue
@ -793,9 +799,7 @@ def _atom_feed_to_dict(base_dir: str, domain: str, xml_str: str,
continue continue
if '</title>' not in atom_item: if '</title>' not in atom_item:
continue continue
if '<link>' not in atom_item: if '<link' not in atom_item:
continue
if '</link>' not in atom_item:
continue continue
if '<updated>' not in atom_item: if '<updated>' not in atom_item:
continue continue