SAX parser splits up character data; I expected Ign. whitesp

Question

Im am working on a XML parser for loading data from some back
end systems into an Oracle 8i database. I am using the SAX
parser for this purpose. After doing some tests with larger
amounts of XML data (> 1M), I found some unexpected behaviour.
The parser sometimes splits up character data into two chunks of
data. The XML looks as follows:
<TAGNAME> this is the character data </TAGNAME>
The parser raises the following events:
1 startElement name = "TAGNAME"
2 characters chbuf = " "
3 characters chbuf = "this is the character data "
4 endElement name = "TAGNAME"
I expected an ignorableWhitespace event at step 2. The XML
document contains repetitive tagnames. The strange thing about
the parse process is that the parser splits up the character
data only sometimes, and I can't determine any kind of logica
for this behaviour. Most occurrences of exactly the same tagname
and character data are parsed correctly (that is, as I
expected).
Am I dealing with correct behaviour here, or is it a bug??
Rolf.
null

Answer 1

Oracle XML Team wrote:
: Rolf van Deursen (guest) wrote:
: : Im am working on a XML parser for loading data from some
back
: : end systems into an Oracle 8i database. I am using the SAX
: : parser for this purpose. After doing some tests with larger
: : amounts of XML data (> 1M), I found some unexpected
behaviour.
: : The parser sometimes splits up character data into two
chunks
: of
: : data. The XML looks as follows:
: : <TAGNAME> this is the character data </TAGNAME>
: : The parser raises the following events:
: : 1 startElement name = "TAGNAME"
: : 2 characters chbuf = " "
: : 3 characters chbuf = "this is the character data "
: : 4 endElement name = "TAGNAME"
: : I expected an ignorableWhitespace event at step 2. The XML
: : document contains repetitive tagnames. The strange thing
about
: : the parse process is that the parser splits up the character
: : data only sometimes, and I can't determine any kind of
logica
: : for this behaviour. Most occurrences of exactly the same
: tagname
: : and character data are parsed correctly (that is, as I
: : expected).
: : Am I dealing with correct behaviour here, or is it a bug??
: : Rolf.
: The behavior is expected and correct. Could you elaborate on
why
: you would expect the parser to treat the whitespace signalled
in
: step 2 as ignorable?
: Oracle XML Team
: http://technet.oracle.com
: Oracle Technology Network
Thank you for your quick response.
In my test XML, there are about 27500 tags containing character
data. All character data starts with a whitespace character.
After parsing the xml, the whitespace of only 5 (!) tags is
treated as a seperate character event (so two character events
are raised in succession). The remaining tags all raise only ONE
character event for the entire character data. I can't explain
the difference in treatment.
null