Re: [xml] xmlwriter, attributes and encoding

On Tue, Aug 15, 2006 at 07:06:48PM -0400, Rob Richards wrote:
When using xmlwriter attributes , document encoding that has been set is 
not passed to xmlAttrSerializeTxtContent, so character references are 
written when needed.
The xmlDocPtr passed to xmlAttrSerializeTxtContent is only used to check 
encoding, so I was wondering if the code in the attached patch is fine 
to fix this (it fakes a document so any changes within 
xmlAttrSerializeTxtContent to manipulate the passed document could 
possibly blow up - though i dont see any need/reason to ever do this).

  Hum, that's a bit nasty, because none of the fields in the document
structure are initialized (except encoding), so it internally use an
API based not on the API description but on its implementation. If at
some point someone add something like entity lookup to
xmlAttrSerializeTxtContent or just check the document type, suddently
the writer would crash in an unexpected way. 
  So fields should at least be zero initialized, which is probably
a bit costly if done for every single attribute emitted. So while passing
a document might be a good idea, it's better to create and initialize that
document when getting the xmlTextWriterPtr. Maybe adding a doc in 
struct _xmlTextWriter and making sure it's at least minimally initialized
and properly freed with the xmlTextWriter is a good idea. That sounds
far more clean to me, but a bit more work indeed :-)


Red Hat Virtualization group
Daniel Veillard      | virtualization library
veillard redhat com  | libxml GNOME XML XSLT toolkit | Rpmfind RPM search engine

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]