Re: SAX and stuff ...
Tony Pelton wrote: > > 1. Is this a practical, reasonable and expected approach to doing SAX > parsing for a handful of fairly simple but potentially large XML messages > that we have brewed "in-house" and have not developed DTD's for ? > > The reason that I ask is that on some of the messages ... I might have 10+ > elements that I am looking for events on ... so that I can "do stuff". > > I then find myself thinking about the number of "if's" my code is performing > on every startElement() ... and then I find myself putting the "if's" that I > expect to get more of ... before the one's I expect less of ... I use DTDs; I think you should, too. But, with that in mind, here is how I do it (all examples in C++): o Create an ObjectFactory (in the GoF object sense) that can make objects based on string names. o Make a ParserListener interface that has stuff like this in it: <interface> public: // start creating a child object. virtual ParserListener* createChildObject( const string object ) = 0; // complete a child object. virtual ParserListener* completeChildObject( const string object ) = 0; // get the value of an object attribute. virtual const string getAttribute( const string newKey ) = 0; // set attribute "key" to value "value" for this node. virtual void setAttribute( const string key, const string value ) = 0; protected: // clean up my children after parsetime. virtual void arrangeChildren() = 0; </interface> o Extend the BaseHandler into a class that has a private member that is a ParserListener, and implement a stack of previous ParserListener objects. o StartElement() pushes the current ParserListener onto the stack and sets it to the return of a call to the current ParserListener's createChildObject() method. The createChildObject() method makes a call to the aforementioned ObjectFactory and returns the new object (which, itself, is also a ParserListener). o EndElement() calls the ParserListener's completeChildObject() method and pops the old ParserListener off of the stack (reassigning the current ParserListener to it). The completeChildObject() calls calls the object's arrangeChildren() method to do any final cleanup and reorganization of the tree that was built from the parse. o characters() calls the ParserListener's setAttribute() method. ...this approach, when mixed with objects that implement something like Java's DefaultMutableTreeNode interface, works like a charm to provide a generic and easily extensible framework to parse documents into internal tree structures. I use this technique often for software configuration files and message passing. The beauty of it is that the DTD handles data validation and each object need only know about itself. It makes the abstraction(s) very clean from a traditional OO perspective. Yours, -- Jack Rusher, Senior Engineer | mailto:jar@i... Integratus, Inc. | http://www.integratus.com
PURCHASE STYLUS STUDIO ONLINE TODAY!
Purchasing Stylus Studio from our online shop is Easy, Secure and Value Priced!
Download The World's Best XML IDE!
Accelerate XML development with our award-winning XML IDE - Download a free trial today!
Subscribe in XML format