[
https://issues.apache.org/jira/browse/ARTEMIS-747?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15515794#comment-15515794
]
ASF GitHub Bot commented on ARTEMIS-747:
----------------------------------------
Github user TomasHofman commented on a diff in the pull request:
https://github.com/apache/activemq-artemis/pull/791#discussion_r80201908
--- Diff:
artemis-cli/src/main/java/org/apache/activemq/artemis/cli/commands/tools/XmlDataImporter.java
---
@@ -444,33 +444,59 @@ private void processMessageBody(Message message)
throws XMLStreamException, IOEx
}
}
reader.next();
+ ActiveMQServerLogger.LOGGER.debug("XMLStreamReader impl: " + reader);
if (isLarge) {
tempFileName = UUID.randomUUID().toString() + ".tmp";
ActiveMQServerLogger.LOGGER.debug("Creating temp file " +
tempFileName + " for large message.");
try (OutputStream out = new FileOutputStream(tempFileName)) {
- while (reader.hasNext()) {
- if (reader.getEventType() ==
XMLStreamConstants.END_ELEMENT) {
- break;
- }
- else {
- String characters = new
String(reader.getTextCharacters(), reader.getTextStart(),
reader.getTextLength());
- String trimmedCharacters = characters.trim();
- if (trimmedCharacters.length() > 0) { // this will skip
"indentation" characters
- byte[] data = decode(trimmedCharacters);
- out.write(data);
- }
- }
- reader.next();
- }
+ getMessageBodyBytes(new MessageBodyBytesProcessor() {
+ @Override
+ public void processBodyBytes(byte[] bytes) throws
IOException {
+ out.write(bytes);
+ }
+ });
}
FileInputStream fileInputStream = new
FileInputStream(tempFileName);
BufferedInputStream bufferedInput = new
BufferedInputStream(fileInputStream);
((ClientMessage) message).setBodyInputStream(bufferedInput);
}
else {
- reader.next(); // step past the "indentation" characters to get
to the CDATA with the message body
- String characters = new String(reader.getTextCharacters(),
reader.getTextStart(), reader.getTextLength());
- message.getBodyBuffer().writeBytes(decode(characters.trim()));
+ getMessageBodyBytes(new MessageBodyBytesProcessor() {
+ @Override
+ public void processBodyBytes(byte[] bytes) throws IOException {
+ message.getBodyBuffer().writeBytes(bytes);
+ }
+ });
+ }
+ }
+
+ /**
+ * Message bodies are written to XML as Base64 encoded CDATA elements.
Some parser implementations won't read the
+ * entire CDATA element at once (e.g. Woodstox) so it's possible for
multiple CDATA events to be combined into a
+ * single Base64 encoded string. You can't decode bits and pieces of
each CDATA. Each CDATA has to be decoded in
+ * its entirety.
+ *
+ * @param processor used to deal with the decoded CDATA elements
+ * @throws IOException
+ * @throws XMLStreamException
+ */
+ private void getMessageBodyBytes(MessageBodyBytesProcessor processor)
throws IOException, XMLStreamException {
+ int currentEventType;
+ StringBuilder cdata = new StringBuilder();
+ while (reader.hasNext()) {
+ currentEventType = reader.getEventType();
+ if (currentEventType == XMLStreamConstants.END_ELEMENT) {
+ break;
+ }
+ // when we hit a CHARACTERS event we know that the entire CDATA
is complete so decode and pass back to the processor
+ else if (currentEventType == XMLStreamConstants.CHARACTERS &&
cdata.length() > 0) {
--- End diff --
JDK's STAX implementation reports CDATA sections also as
XMLStreamConstants.CHARACTERS (instead of XMLStreamConstants.CDATA like
Woodstox do) by default. So this condition needs to check that the text
contains white spaces only. Probably reader.isWhiteSpace() should do.
> Multiple CDATA events during import fails
> -----------------------------------------
>
> Key: ARTEMIS-747
> URL: https://issues.apache.org/jira/browse/ARTEMIS-747
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Affects Versions: 1.4.0
> Reporter: Justin Bertram
> Assignee: Justin Bertram
>
> Message bodies are written to XML as Base64 encoded CDATA elements. Some
> parser implementations won't read the entire CDATA element at once (e.g.
> Woodstox) so it's possible for multiple CDATA events to be combined into a
> single Base64 encoded string. You can't decode bits and pieces of each
> CDATA. Each CDATA has to be decoded in its entirety. The current importer
> doesn't deal with this properly.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
