[
https://issues.apache.org/jira/browse/XERCESJ-970?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12678773#action_12678773
]
Henning Moll commented on XERCESJ-970:
--------------------------------------
I stumbled upon the same problem, and here is a small example to reproduce it.
Please add Xerces2 explicitly to the classpath or make sure that you use a
JRE5. The intregrated version of Xerces in JRE6 does NOT show the problem
(which means, that this bug seems to be fixed there):
import java.io.BufferedInputStream;
import java.io.FileInputStream;
import javax.xml.parsers.SAXParser;
import javax.xml.parsers.SAXParserFactory;
import org.xml.sax.InputSource;
import org.xml.sax.helpers.DefaultHandler;
public class TTT extends DefaultHandler {
public static void main(String[] args) throws Exception {
new TTT().parse(null);
}
public void parse(String[] args) throws Exception {
long before = System.currentTimeMillis();
InputSource source = new InputSource(new BufferedInputStream(new
FileInputStream("garbage.xml")));
SAXParserFactory factory = SAXParserFactory.newInstance();
SAXParser parser = factory.newSAXParser();
parser.parse(source, this);
System.out.println(System.currentTimeMillis() - before);
}
}
create a dummy xml-file "garbage.xml" with the following content:
<?xml version='1.0' encoding='UTF-8' standalone='no'?>
<blubb>
[up to 2Gig of CDATA]
</blubb>
The runtime of the example is very short with this version of the xml file. Now
change the content to
<?xml version='1.0' encoding='UTF-8' standalone='no'?>
<!--blubb>
[up to 2Gig of CDATA]
</blubb-->
Only one big comment. But the runtime increases extremly (on my system about
110ms vs. 50k(!)ms
The interesting part: The bundled Xerces of JRE5 does show this behaviour. The
one from JRE6 does NOT.
> Large comments are extremely slow to parse
> ------------------------------------------
>
> Key: XERCESJ-970
> URL: https://issues.apache.org/jira/browse/XERCESJ-970
> Project: Xerces2-J
> Issue Type: Bug
> Components: XNI
> Affects Versions: 2.2.0, 2.2.1, 2.3.0, 2.4.0, 2.5.0, 2.6.0, 2.6.1, 2.6.2
> Environment: Windows XP running Java 1.4.2
> Reporter: Sean Griffin
> Priority: Minor
>
> Very large comments drastically increase the parsing time for both SAX and
> DOM implementations. Running the sax.Counter and dom.Counter samples with a
> 410KB file where the entire thing is uncommented results in parse times in
> the 100ms to 300ms range. However, if I comment out 95% of the file and run
> the same samples the parse times jump to between 40 and 50 seconds. I ran
> the same samples using the Aelfred parser shipped with Saxon 7.9 and, while
> the file with the large comment was slower than without the comment, it
> jumped by only 100ms or so.
> I briefly compared the code between the two parsers, and they don't look
> significantly different when it comes to handling comments. The only main
> difference I noticed was around low/high byte character checks. I suspect it
> is an inefficiency in the XMLStringBuffer class, but I'm not seeing anything.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
