iOS About Libxml2: charactersFoundSAX called back function

cthesky said:

One of them is charactersFoundSAX function.

Click to expand...

Is this function set as your SAX handler's characters member? I'm going to assume it is. If it's not then none of the following may be true.

cthesky said:

I noticed that this function not only called between start element and end element of a XML file, but also get called before start element and after end element. Is it correct? Is it the actual behaviour of charactersFoundSAX called back function?

Click to expand...

Yes it's correct. I find a pretty much need to build a state machine when using SAX. I'd start in a state were I'd ignore characters until my startElement callback was called for an element where I was interested in the text context.

cthesky said:

I also noticed that between start and end element of an XML file, charactersFoundSAX function mostly will get called one time, but sometimes I noticed that it get called a few times between start and end element. Why? For the number of times of charactersFoundSAX function get called, is it depends on the number or length of characters?

Click to expand...

The number of calls to your characters callback will, as you've seen, be unpredictable. libxml2 does not buffer all the text and then call your characters callback. It calls the callback as soon it is has some text available.

You'll most likely see this happen when parsing XML streamed over a network. Some of the text will arrive in a packet, and libxml2 will immediately call your characters callback. Then as more packets arrive with more text, libxml2 continues to call your characters callback for the text in each packet.

You need to buffer the text if necessary. A technique I use for elements with pure text content is to create a buffer when my startElement callback is called, collect text into the buffer as my characters callback is called, and then only when my endElement callback is called do I process the text in the buffer.

cthesky said:

Jiminaus, I have some questions about state machine as you mentioned in your replied. So, can I said that this state machine can control when the charactersFoundSax will get called? For example, with this state machine, charactersFoundSax can only get called between Start and End element of a XML file or will get called only when the startElement callback function was called for an element where I was interested in the text context. Am I right?

Click to expand...

The state machine would work the opposite way. You can't control when libxml2 calls your characters callback, you can only control what you do in that callback.

Note that I'm using the term state machine in a very abstract way, more as a design device than as an implementation device.

Unless you've got a complex XML structure, you may not need to implement an actual state machine. It should just be as simple as setting a boolean to true in your startElement callback, checking that boolean in your characters callback, and resetting the boolean to false in your endElement callback.