SAX parseing goes 'all funny' on value [en]

[Fred]

Hi,

I am parsing a small xml document and the parseing goes 'all funny'
when parsing this element: <useragent>Mozi lla/4.61 [en] (WinNT;
I)</useragent>

I've created a subclass of org.xml.sax.hel pers.DefaultHan dler, and an
instance of this subclass is set on my
org.apache.xerc es.parsers.SAXP arser:

SAXParser parser = new SAXParser();
parser.setConte ntHandler(pdh);
parser.setError Handler(pdh);

I've found that the

public void characters(char[] ch, int offset, int length) throws
SAXException

method is called once per element parsed. my debug output confirms
this. e.g. when parsing <useragent>Mobi leExplorer/3.00 (Mozilla/1.22;
compatible; MMEF300; Microsoft; Windows; GenericLarge)</useragent> it
reads:

D: reading characters...(u seragent) length=89, offset=721,
found='MobileEx plorer/3.00 (Mozilla/1.22; compatible; MMEF300;
Microsoft; Windows; GenericLarge)'
D: ending element (useragent) current element value is :
[MobileExplorer/3.00 (Mozilla/1.22; compatible; MMEF300; Microsoft;
Windows; GenericLarge)]
But... when parsing <useragent>Mozi lla/4.61 [en] (WinNT;
I)</useragent>
the debug output reads

D: reading characters...(u seragent) length=16, offset=1097,
found='Mozilla/4.61 [en'
D: reading characters...(u seragent) length=1, offset=0, found=']'
D: reading characters...(u seragent) length=11, offset=1114, found='
(WinNT; I)'
D: ending (useragent) current element value is : [ (WinNT; I)]

It calls the characters method trice?!
Does the [en] bit in the element value have anything to do with this?
Would like to understand what and why.

(As a 'temp fix' I thought to have the DefaultHandlers characters(...)
method concatenate characters read, till the endElement(...) is
invoked; but that seems to break everything.)

Thanks for your input.
Fred.

[Julian Reschke]

Fred wrote:

(As a 'temp fix' I thought to have the DefaultHandlers characters(...)
method concatenate characters read, till the endElement(...) is
invoked; but that seems to break everything.)

I think that's how SAX is supposed to work. There's no guarantee that
you're only getting a single event here.

[Eric Bohlman]

Julian Reschke <ju************ @gmx.de> wrote in
news:3F******** ******@gmx.de:

Fred wrote:

(As a 'temp fix' I thought to have the DefaultHandlers characters(...)
method concatenate characters read, till the endElement(...) is
invoked; but that seems to break everything.)

I think that's how SAX is supposed to work. There's no guarantee that
you're only getting a single event here.

It *is* how SAX is supposed to work. Keep in mind that character data in
XML can be arbitrarily long; if a parser had to deliver character data in a
single chunk, it could find itself constantly allocating and reallocating
buffers. Not imposing such a requirement greatly simplifies buffer
management in a parser; it can use a fixed-size internal buffer and just
call the character handler when everything up to the end of the buffer is
character data, rather than having to shift everything around. That can
greatly speed up parsing.