|
@@ -765,7 +765,8 @@ class Dataflow_ActivteMQ_extract(Dataflow_extract):
|
|
_soup = BeautifulSoup(_dochtmlcon,"lxml")
|
|
_soup = BeautifulSoup(_dochtmlcon,"lxml")
|
|
if len(_dochtmlcon)>200000:
|
|
if len(_dochtmlcon)>200000:
|
|
_find = _soup.find("div",attrs={"class":"richTextFetch"})
|
|
_find = _soup.find("div",attrs={"class":"richTextFetch"})
|
|
- _find.decompose()
|
|
|
|
|
|
+ if _find is not None:
|
|
|
|
+ _find.decompose()
|
|
else:
|
|
else:
|
|
_soup = article_limit(_soup,50000)
|
|
_soup = article_limit(_soup,50000)
|
|
_dochtmlcon = str(_soup)
|
|
_dochtmlcon = str(_soup)
|