Compiling application with Tika with Java 13 - problems loading modules

I'm trying to migrate a Java application that uses Tika from OracleJDK 1.8 to OPenJDK 13.

My IDE is Eclipse.

I have created the file module-info.java to indicate the required modules for my application.

In order to be able to use Tika classes such as AbstractParser, Detector, etc., I have added requires org.apache.tika.core; in module-info.java.

My code also uses the class org.apache.tika.parser.pdf.PDFParserConfig to extract embedded images:

PDFParserConfig pdfConfig = new PDFParserConfig();
pdfConfig.setExtractInlineImages(true);
context.set(PDFParserConfig.class, pdfConfig);'

I get the compilation error:

PDFParserConfig cannot be resolved to a type

Eclipse suggests to add requires org.apache.tika.parsers; to module-info.java: Eclipse suggestion screenshot.

When I add this module requirement to module-info.java, the application compiles properly.

That is, at this stage we have included in module-info.java:

module myapp {
	/** others ... */ 
	requires org.apache.tika.core;
	requires org.apache.tika.parsers;
}

However, when trying to execute the compiled application, we get the error:

Error occurred during initialization of boot layer
java.lang.module.FindException: Unable to derive module descriptor for C:\Users\Admin\.m2\repository\org\apache\tika\tika-parsers.24\tika-parsers-1.24.jar
Caused by: java.lang.module.InvalidModuleDescriptorException: Provider class org.apache.tika.parser.onenote.OneNoteParser not in module

Inspecting the project Libraries in Eclipse, I can see that tika-core and tika-parsers (v1.24) are both modular: Eclipse Java Build Path

In conclusion: If I don't add org.apache.tika.parsers as a required module, the application won't compile, and if I add it I get the runtime error saying org.apache.tika.parser.onenote.OneNoteParser is not in the module.

I have inspected the JAR files for these packages to see the dependencies they have. The core packages seems to be right:

$ jar --file=tika-core-1.24.jar --describe-module

No module descriptor found. Derived automatic module.

org.apache.tika.core@1.24 automatic
requires java.base mandated
contains org.apache.tika
contains org.apache.tika.concurrent
contains org.apache.tika.config
contains org.apache.tika.detect
contains org.apache.tika.embedder
contains org.apache.tika.exception
contains org.apache.tika.extractor
contains org.apache.tika.fork
contains org.apache.tika.io
contains org.apache.tika.language
contains org.apache.tika.language.detect
contains org.apache.tika.language.translate
contains org.apache.tika.metadata
contains org.apache.tika.mime
contains org.apache.tika.parser
contains org.apache.tika.parser.digest
contains org.apache.tika.parser.external
contains org.apache.tika.sax
contains org.apache.tika.sax.xpath
contains org.apache.tika.utils

...but the 'parsers' jar gives an error:

$ jar --file=tika-parsers-1.24.jar --describe-module

Unable to derive module descriptor for: tika-parsers-1.24.jar
Provider class org.apache.tika.parser.onenote.OneNoteParser not in module

Does this mean the jar package for parsers is not well formed?
Is there any workaround for this?

Thank you.

EDIT:
If I try with version 1.24.1, I get the execution error:

Error occurred during initialization of boot layer
java.lang.module.FindException: Unable to derive module descriptor for C:\Users\Admin\.m2\repository\org\apache\tika\tika-parsers.24.1\tika-parsers-1.24.1.jar
Caused by: java.lang.module.InvalidModuleDescriptorException: Provider class org.apache.tika.parser.external.CompositeExternalParser not in module

That is: the failing class is CompositeExternalParser instead of OneNoreParser.

Inspecting META-INF/services/org.apache.tika.parser.Parser of tika-parsers-1.42.1.jarI can see the entryorg.apache.tika.parser.external.CompositeExternalParser` but the package does not contain this class.

So, it seems to be an error in this META-INF file. Id this due to an error when compiling the package and submitting it to Maven Central?

I've found a JIRA issue, TIKA-2929, where they say "Apache Tika needs to be on the Java Classpath, not the module path". I've tried this, but, as explained before, I get a compilation error if I don't add it to the module path and set requires org.apache.tika.parsers;.

This is a hard puzzle...

Ran into the same issues.<br>
Also found the faulty entries in<br>
org.apache.tika.parser.Parser (and also org.apache.tika.parser.Detector) in META-INF/services/ <br>

A quick fix is to ...

• Unpack those files
• delete the lines that seem to reference non existing classes
• pack them back into the jar <br>

My project compiled after that.<br>
For sure no longterm solution, but since even older versions i tried ran into that problem, it might help out some people.