diff --git a/builder/htmlprocessor/strip_function.py b/builder/htmlprocessor/strip_function.py index 45cef23..d249e62 100644 --- a/builder/htmlprocessor/strip_function.py +++ b/builder/htmlprocessor/strip_function.py @@ -22,7 +22,10 @@ def cleanUp(soup): [s.decompose() for s in content.findAll("div", {"id": "mw-mf-language-section"})] # cut off geo coords as we process them separately in original files [s.decompose() for s in content.findAll("div", {"id": "geoCoord"})] - + # cut off missing images (looks like text File:Image.JPG on pages) + for s in content.findAll("div", {"class": "thumb"}): + if (not s.find("img")): + s.decompose(); # delete empty sections sections = content.findAll("div", {"class": "section"})