snap_preview = doc.find_class('snap_preview')[0]
# lastnode = snap_preview[-1]
# if lastnode.tag == 'div' and lastnode.get('class') == 'possibly-related':
# snap_preview.remove(lastnode)
post.Contents = html.tostring(snap_preview)
# commentstring = doc.get_element_by_id('comments').text.split()[0]
# if commentstring == "No":
# post.Comments = 0
# else:
# post.Comments = int(commentstring)
return post
if __name__ == "__main__":
blog = Blog(filepath="dorait.blog", createnew=False)
for ui in sitemap.parse_sitemap("http://krishashok.wordpress.com/sitemap.xml"):
if not '/20' in ui.loc:
#guards against non-post pages. Needs to be more robust
#But okay for now!
continue
if not blog.has_key(ui.loc):
p = post_from_page(ui.loc)
p.PostedAt = ui.lastmod
blog.persist(ui.loc, p)
print "Done %s" % ui.loc
else:
print "Skipped %s" % ui.loc
snap_preview = doc.find_class('snap_preview')[0]
lastnode = snap_preview[-1]
if lastnode.tag == 'div' and lastnode.get('class') == 'possibly-related':
snap_preview.remove(lastnode)
post.Contents = html.tostring(snap_preview)
commentstring = doc.get_element_by_id('comments').text.split()[0]
if commentstring == "No":
post.Comments = 0
else:
post.Comments = int(commentstring)
return post
if __name__ == "__main__":
blog = Blog(filepath="dorait.blog", createnew=False)
for ui in sitemap.parse_sitemap("http://dorai.wordpress.com/sitemap.xml"):
if not '/20' in ui.loc:
#guards against non-post pages. Needs to be more robust
#But okay for now!
continue
if not blog.has_key(ui.loc):
p = post_from_page(ui.loc)
p.PostedAt = ui.lastmod
blog.persist(ui.loc, p)
print "Done %s" % ui.loc
else:
print "Skipped %s" % ui.loc
if categories:
post.Categories = [c.text for c in categories]
snap_preview = doc.find_class('snap_preview')[0]
post.Contents = html.tostring(snap_preview)
# commentstring = doc.get_element_by_id('comments').text.split()[0]
# if commentstring == "No":
# post.Comments = 0
# else:
# post.Comments = int(commentstring)
return post
if __name__ == "__main__":
blog = Blog(filepath="mona.blog", createnew=False)
for ui in sitemap.parse_sitemap("http://pixelbits.wordpress.com/sitemap.xml"):
if not '/20' in ui.loc:
#guards against non-post pages. Needs to be more robust
#But okay for now!
continue
if not blog.has_key(ui.loc):
p = post_from_page(ui.loc)
p.PostedAt = ui.lastmod
blog.persist(ui.loc, p)
print "Done %s" % ui.loc
else:
print "Skipped %s" % ui.loc
