> One of the reason we discussed with Bernd during ApacheCon was that I > needed a 'smart' XML parser, and Vysper obviously has one. The problem > I had was that all the Java XML parsers I know of aren't able to stop > parsing in the middle of a tag and start again without complaining > with a hell lot of Exceptions. > > For instance, if you receive "<ta" when the full tag is "<tag>", then > the parser will complain. I would expect the parser to simply returns, > storing the current state somwhere, and restart smoothly when the next > few bytes are received. Maybe this 'smart' statefull parser exist > however...
> So let's suppose it does not exist : we need to extend some existing > parser in order to accept the inherently fragmented nature of TCP > messages. Otherwise, you have to do what Vysper does : pre-parse, > heading for the closing '>', expecting that the user didn't > encapsulate any comments in its XML file ! My code example has this flaw, that it was 2 stage parsing, first collect the complete XML and then use any parser Jibx or any XML Data binding framework to get the POJO something similar to Vysper. > > Is this too much to ask for? Is there a java XML parser already > wirking the way I want it to work ? If so, please tell me ! Otherwise, > we might want to extend xerces, or something like XPP3 > (http://www.extreme.indiana.edu/xgws/xsoap/xpp/mxp1/index.html) > > wdyt ? StAX parser can be a possible solution, though my experience is limited wth StAX parsers. http://en.wikipedia.org/wiki/StAX Either way's, its very much doable and interesting :-) What many people would like is use of a Data binding framework along with, so the decoder has to be smart. May be with a hook for easy customization. -- thanks ashish Blog: http://www.ashishpaliwal.com/blog My Photo Galleries: http://www.pbase.com/ashishpaliwal
