[jira] [Commented] (TIKA-2812) NPE when parsing text with write limit set on IBM JDK

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

[jira] [Commented] (TIKA-2812) NPE when parsing text with write limit set on IBM JDK

Hudson (Jira)

    [ https://issues.apache.org/jira/browse/TIKA-2812?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16740315#comment-16740315 ]

Sergiy Shyrkov commented on TIKA-2812:

We went with a workaround, which works for Oracle/ OpenJDK / IBM JDK 8 and JDK 11 for now, by adjusting the {{org.apache.tika.sax.TaggedSAXException.getCause()}} method as follows:
    public SAXException getCause() {
        Throwable theCause = super.getCause();
        if (theCause == null && getException() instanceof SAXException) {
            return (SAXException) getException();
        return (SAXException) theCause;

> NPE when parsing text with write limit set on IBM JDK
> -----------------------------------------------------
>                 Key: TIKA-2812
>                 URL: https://issues.apache.org/jira/browse/TIKA-2812
>             Project: Tika
>          Issue Type: Bug
>          Components: core
>    Affects Versions: 1.20
>         Environment: IBM JDK 8
>            Reporter: Sergiy Shyrkov
>            Priority: Major
>              Labels: regression
>         Attachments: TikaTest.java, output-ibm-jdk-tika-1.14.txt, output-ibm-jdk-tika-1.20.txt, output-open-jdk-tika-1.20.txt, output-oracle-jdk-11-tika-1.20.txt, output-oracle-jdk-tika-1.20.txt, test.txt
> We have updated Tika from version 1.14 to recently released 1.20 and are now experiencing an issue with parsing of texts when write limit is set (we are using {{WriteOutContentHandler}}) on IBM JDK 8.
> Test class [^TikaTest.java] and test file [^test.txt] are attached.
> The issue is present on IBM JDK 8 [^output-ibm-jdk-tika-1.20.txt], but not on Oracle [^output-oracle-jdk-tika-1.20.txt] or Open JDK 8 [^output-open-jdk-tika-1.20.txt].
> With Tika 1.14 we had no this issue [^output-ibm-jdk-tika-1.14.txt].
> Analysis:
>  With the fix in TIKA-2668 ([https://github.com/apache/tika/commit/89a588e4d8d2aa44a9d3c965d514c18c7d3c134d#diff-5a28529cf32968d35a5036172cd8f74fL41)] a line was removed from the constructor of the {{TaggedSAXException}} class:
> {code:java}
> initCause(original); // SAXException has it's own chaining mechanism!
> {code}
> Bringing the line back, solves our issue with JDK 8, but breaks the things on JDK 11 [^output-oracle-jdk-11-tika-1.20.txt].
> Is there any chance the class {{TaggedSAXException}} can be made compatible with JDK 8 and JDK 11 (both Oracle/OpenJDK and IBM one)?
> Thank you in advance!
> Kind regards
>  Sergiy Shyrkov

This message was sent by Atlassian JIRA