We have a PHP CMS with a lot of poorly written HTML in the client-contributed content. This kept causing my XSL template system to output XML errors. I got around this problem by:
- wrapping content in CDATA tags
- Checking if the content is valid XML with
xml_parse()in PHP, if not I add a CDATA tag and try again. - Strip out bad characters that may have crept in from Word
- Process the XSL and XML using
xsl:value-of tags with disable-output-escaping="yes"
Using CDATA tags around unpredictable HTML helps prevent problems with the XML parser. Without the final step, the resulting HTML contains the original HTML with HTML entities.
In PHP, mb_convert_encoding($string, 'ASCII') has proven very useful for handling text users paste from applications like Word. PHP has to be compiled with—enable-mbstring for this function to work. It prevents strings with different encodings encodings to your XSL from confusing the XML parser (where the encoding is defined).