From 908da5744bf36767925e9b43c1430646ca174a16 Mon Sep 17 00:00:00 2001 From: Jim Miller Date: Fri, 6 Mar 2020 14:22:28 -0600 Subject: [PATCH] Update adapter_test1 --- fanficfare/adapters/adapter_test1.py | 20 ++++++++++++++------ 1 file changed, 14 insertions(+), 6 deletions(-) diff --git a/fanficfare/adapters/adapter_test1.py b/fanficfare/adapters/adapter_test1.py index bdee2b33..666234bb 100644 --- a/fanficfare/adapters/adapter_test1.py +++ b/fanficfare/adapters/adapter_test1.py @@ -376,9 +376,6 @@ Some more longer description. "I suck at summaries!" "Better than it sounds!" soup = self.make_soup(data) - if '#' in url: - anchorid = url.split('#')[1] - soup = soup.find('li',id=anchorid) if 'wordpress.com' in url: bq = soup.find('div',{'class':'entry-content'}) @@ -386,15 +383,26 @@ Some more longer description. "I suck at summaries!" "Better than it sounds!" for tag in addiv.find_all_next('div'): tag.extract() addiv.extract() - else: - bq = soup.find('blockquote') - bq.name='div' + elif '#' in url: + anchorid = url.split('#')[1] + if 'spacebattles.com' in url or 'sufficientvelocity.com' in url: + # XF2 + soup = soup.find('article',{'data-content':anchorid}) + bq = soup.find('div',{'class':'bbWrapper'}) + else: + soup = soup.find('li',id=anchorid) + bq = soup.find('blockquote') + bq.name='div' for iframe in bq.find_all('iframe'): iframe.extract() # calibre book reader & editor don't like iframes to youtube. for qdiv in bq.find_all('div',{'class':'quoteExpand'}): qdiv.extract() # Remove
click to expand
+ for qdiv in bq.find_all('div',{'class':re.compile(r'bbCodeBlock-(expand|shrink)Link')}): + qdiv.extract() # Remove
click to expand
+ for tag in bq.find_all('div', class_="bbCodeBlock-expandContent"): + tag.name='blockquote' return self.utf8FromSoup(url[:url.index('/',8)+1],bq)