Question

I’m using Tika parser to index my files into Solr. I created my own parser (which extends XMLParser). It uses my own mimetype.
I created a jar file which inside looks like this:

src
|-main
    |-some_packages
        |-MyParser.java
    |resources
        |-META-INF
            |-services
                |-org.apache.tika.parser.Parser (which contains a line:some_packages.MyParser.java)
        |_org
            |-apache
                |-tika
                    |-mime
                        |-custom-mimetypes.xml 

In custom-mimetypes I put the definition of new mimetype becouse my xml files have some special tags.

Now where is the problem: I’ve been testing parsing and indexing with Solr on glassfish installed on my local machine. It worked just fine. Then I wanted to install it on some remote server. There is the same version of glassfish installed (3.1.1). I copied-pasted Solr application, it’s home directory with all libraries (including tika jars and the jar with my custom parser). Unfortunately it doesn’t work. After posting files to Solr I can see in content-type field that it detected my custom mime type. But there are no fields that suppose to be there like if MyParser class was never runned. The only fields I get are the ones from Dublin Core. I checked (by simply adding some printlines) that Tika is only using XMLParser.
Have anyone had similar problem? How to handle this?

Answer 1

Problem was that I was using Java 7 to compile my parser but Apache Tika was compiled with Java 5…