DDG: attempt at simplify code

From: b62249c557
This commit is contained in:
James Lu 2014-12-14 21:55:15 -08:00 committed by GLolol
parent be3112197b
commit 9298b71732

View File

@ -71,35 +71,22 @@ class DDG(callbacks.Plugin):
data = re.sub('\t|\r|\n', '', data)
data = re.sub('\s{2,}', ' ', data)
soup = BeautifulSoup(data)
# DuckDuckGo lite uses tables for everything. Each WEB result is made
# up of 3 <tr> tags:
tables = soup.find_all('table')
# Sometimes there is an extra table for page navigation
try:
webresults = tables[2].find_all('tr')
except IndexError:
webresults = tables[1].find_all('tr')
if webresults:
try:
while 'result-sponsored' in webresults[0]["class"]:
self.log.debug("DDG: stripping 1 sponsored/ad result.")
webresults = webresults[4:]
except KeyError: pass
# 1) The link and title.
link = webresults[0].find('a').get('href')
# 2) A result snippet.
snippet = webresults[1].find("td", class_="result-snippet")
tds = soup.find_all('td')
for t in tds:
if "1." in t.text:
res = t.next_sibling.next_sibling
break
try:
# 1) Fetch the result link.
link = res.a.get('href')
# 2) Get a result snippet.
snippet = res.parent.next_sibling.next_sibling.find("td",
class_="result-snippet")
snippet = snippet.text.strip()
except AttributeError:
snippet = webresults[1].td.text.strip()
# 3) The link-text; essentially the same as the link in 1), but with the
# URI (http(s)://) removed. We do not need this section.
s = format("%s - %u", snippet, link)
irc.reply(s)
else:
except AttributeError:
irc.error("No results found.")
search = wrap(search, ['text'])