Quick fix for storiesonline.net & merging heuristics change.

This commit is contained in:
Jim Miller
2014-02-01 18:54:31 -06:00
2 changed files with 5 additions and 2 deletions
@@ -322,8 +322,8 @@ class StoriesOnlineNetAdapter(BaseSiteAdapter):
div.append(div1)
# removing all the left-over stuff
for a in div.findAll('span'):
a.extract()
# for a in div.findAll('span'):
# a.extract()
for a in div.findAll('h1'):
a.extract()
+3
View File
@@ -137,6 +137,9 @@ def replace_br_with_p(body):
contentLinesSum += lineLen
if lineLen > longestLineLength:
longestLineLength = lineLen
if contentLines == 0:
contentLines = 1
averageLineLength = contentLinesSum/contentLines