From 0496fbba45a600ed271b71fedd149e4313882d33 Mon Sep 17 00:00:00 2001 From: Tanner Collin Date: Tue, 24 Sep 2019 08:21:26 +0000 Subject: [PATCH] Ignore new Tildes posts and handle deleted ones --- apiserver/feeds/tildes.py | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/apiserver/feeds/tildes.py b/apiserver/feeds/tildes.py index 48286b8..917425b 100644 --- a/apiserver/feeds/tildes.py +++ b/apiserver/feeds/tildes.py @@ -62,14 +62,18 @@ def story(ref): html = api(API_ITEM(ref)) if not html: return False - if 'Topic deleted by author' in html: return False - soup = BeautifulSoup(html, features='html.parser') a = soup.find('article', class_='topic-full') h = a.find('header') + lu = h.find('a', class_='link-user') + + error = a.find('div', class_='text-error') + if error: + if 'deleted' in error.string or 'removed' in error.string: + return False s = {} - s['author'] = str(h.find('a', class_='link-user').string) + s['author'] = str(lu.string if lu else 'unknown user') s['author_link'] = SITE_AUTHOR_LINK(s['author']) s['score'] = int(h.find('span', class_='topic-voting-votes').string) s['date'] = unix(h.find('time')['datetime']) @@ -85,6 +89,9 @@ def story(ref): ch = a.find('header', class_='topic-comments-header') s['num_comments'] = int(ch.h2.string.split(' ')[0]) if ch else 0 + if s['score'] < 8 and s['num_comments'] < 6: + return False + td = a.find('div', class_='topic-full-text') if td: s['text'] = td.encode_contents().decode() @@ -102,7 +109,7 @@ if __name__ == '__main__': #print(self_post) #li_comment = story('gqx') #print(li_comment) - broken = story('h23') + broken = story('hmi') print(broken) # make sure there's no self-reference