insert comments into elementtree

[Tim Arnold]

Hi, I'm using the TidyHTMLTreeBuilder to generate some elementtrees from
html. One by-product is that I'm losing comments embedded in the html. So
I'm trying to put them back in, but I'm doing something wrong: here's the
code snippet of how I generate the Trees:

from elementtree import ElementTree as ET
from elementtidy import TidyHTMLTreeBuilder
XHTML = "{http://www.w3.org/1999/xhtml}"

htmfile = os.path.join(self.htmloc,filename)
fd = open(htmfile)
tidyTree = TidyHTMLTreeBuilder.TidyHTMLTreeBuilder('utf-8')
tidyTree.feed(fd.read())
fd.close()
try:
tmp = tidyTree.close()
except:
print 'Bad file: %s\nSkipping.' % filename
continue
tree = ET.ElementTree(tmp)

and here's the method I use to put the comments back in:

def addComments(self,tree):
body = tree.find('./%sbody' % XHTML)
for elem in body:
if elem.tag == '%sdiv' % XHTML and elem.get('class'):
if elem.get('class') == 'remapped':
comElem = ET.SubElement(elem,ET.Comment('stopindex'))

self.addComments(tree)
filename = os.path.join(self.deliverloc,name)
self.htmlcontent.write(tree,filename,encoding=self.encoding

when I try this I get errors from the ElementTree _write method:
TypeError: cannot concatenate 'str' and 'instance' objects

thanks for any help!
--Tim Arnold

 

[Stefan Behnel]

Hi, I'm using the TidyHTMLTreeBuilder to generate some elementtrees from
html. One by-product is that I'm losing comments embedded in the html.

That's how the parser in ET works. Use lxml instead, which keeps documents
intact while parsing.

http://codespeak.net/lxml/dev/
http://codespeak.net/lxml/dev/lxmlhtml.html

Stefan

 

[Tim Arnold]

That's how the parser in ET works. Use lxml instead, which keeps documents
intact while parsing.

http://codespeak.net/lxml/dev/
http://codespeak.net/lxml/dev/lxmlhtml.html

Stefan

Thanks Stefan, I certainly would use lxml if I could get everything to
compile on this HPux10.20.
I did manage to get this one solved by inserting the comments back in like
this:
elem.insert(0,ET.Comment('stopindex'))

thanks,
--Tim Arnold