[mod] add/modify image fetching for bing_news, qwant and twitter engines

This commit is contained in:
Alexandre Flament 2017-02-12 14:58:49 +01:00 committed by Adam Tauber
parent 4cffd78650
commit f5128c7cb9
4 changed files with 28 additions and 11 deletions

View file

@ -27,6 +27,7 @@ search_url = base_url + 'search?'
# specific xpath variables
results_xpath = '//li[@data-item-type="tweet"]'
avatar_xpath = './/img[contains(@class, "avatar")]/@src'
link_xpath = './/small[@class="time"]//a'
title_xpath = './/span[contains(@class, "username")]'
content_xpath = './/p[contains(@class, "tweet-text")]'
@ -57,6 +58,8 @@ def response(resp):
try:
link = tweet.xpath(link_xpath)[0]
content = extract_text(tweet.xpath(content_xpath)[0])
img_src = tweet.xpath(avatar_xpath)[0]
img_src = img_src.replace('_bigger', '_normal')
except Exception:
continue
@ -71,12 +74,14 @@ def response(resp):
results.append({'url': url,
'title': title,
'content': content,
'img_src': img_src,
'publishedDate': publishedDate})
else:
# append result
results.append({'url': url,
'title': title,
'content': content})
'content': content,
'img_src': img_src})
# return results
return results