[
https://issues.apache.org/jira/browse/CAMEL-27304?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Work on CAMEL-27304 started by Claus Ibsen.
-------------------------------------------
> camel-tika - parse with tikaParseOutputFormat=html fails on a document
> without a title on JDK 21+
> -------------------------------------------------------------------------------------------------
>
> Key: CAMEL-27304
> URL: https://issues.apache.org/jira/browse/CAMEL-27304
> Project: Camel
> Issue Type: Bug
> Components: camel-tika
> Reporter: Claus Ibsen
> Assignee: Claus Ibsen
> Priority: Minor
> Fix For: 4.24.0
>
>
> With {{tikaParseOutputFormat=html}}, {{TikaProducer}} (line 192) wraps the
> JDK {{TransformerHandler}} in Tika's {{ExpandedTitleContentHandler}}. For an
> empty {{<title>}}, that handler calls {{characters(new char[0], 0, 1)}} and
> catches the {{ArrayIndexOutOfBoundsException}} it expects from the serializer
> (the TIKA-725 workaround). The JDK serializer ({{ToStream$CharacterBuffer}})
> does not throw: it buffers one NUL character and rejects it when the element
> ends. So any document without a title fails:
> {noformat}
> TIKA-237: Illegal SAXException ... An invalid XML character (Unicode: 0x0)
> was found in the node's character data content
> {noformat}
> Reproduced on Camel 4.24.0-SNAPSHOT (Tika 4.1.0) with JDK 21 and JDK 25:
> * plain text, or HTML without a {{<title>}}, to {{html}}: fails;
> * the same to {{text}} or {{xml}}: works ({{xml}} writes {{<title/>}});
> * HTML with a {{<title>}} to {{html}}: works.
> A standalone program with only tika-core 4.1.0 and the JDK serializer shows
> the same: it fails on JDK 21 and 25, and on JDK 17 it writes the invalid
> {{<title>�</title>}}. Tika's own
> {{org.apache.tika.sax.ToHTMLContentHandler}} writes {{<title></title>}}
> correctly.
> Found through the {{tika-extract-text-action}} Kamelet: extracting text and
> then HTML in one route fails at the second call, because the extracted text
> has no title. The existing tests ({{TikaParseTest}},
> {{TikaMetadataHeaderFilterTest}}) always use documents with a title.
> Expected: {{html}} works for documents without a title, for example by using
> {{ToHTMLContentHandler}} or a wrapper that does not depend on the serializer
> throwing, plus a test with a document that has no title.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)