The problem:
Today I am adding Solr Cell(Tika) to our Solr application, during test, it throws the following exception:
Caused by: java.lang.NoSuchMethodError: org.apache.tika.mime.MediaType.set([Lorg/apache/tika/mime/MediaType;)Ljava/util/Set;
at org.apache.tika.parser.crypto.Pkcs7Parser.getSupportedTypes(Pkcs7Parser.java:52)
at org.apache.tika.parser.CompositeParser.getParsers(CompositeParser.java:81)
This looks like there is a conflicting tika-core jar.
I used the java decompiler JD-GUI to check the jars I added: solr\contrib\extraction\lib\tika-core-1.3.jar, the class MediaType does contain this method: set(MediaType[] types).
Then it seems there are some other jars containing the MediaType class, I checked solr.war\WEB-INF\lib, but no obvious hint.
Using Eclipse Display View to Check which jar Contains the Class
I enabled the remote debug, added a breakpoint at Pkcs7Parser.getSupportedTypes(Pkcs7Parser.java:52), reran the Solr Cell request, it hit and stops at Pkcs7Parser.getSupportedTypes(Pkcs7Parser.java:52).
(java.net.URL) file:omitted/webapps/server/WEB-INF/lib/crawler4j-dependency.jar
Check crawler4j-dependency.ja, so now the root cause is obvious.
The culprit is that some one added crawler4j into the Solr application and put its all dependencies into crawler4j-dependency.jar. It uses tika-core-1.0.jar, the MediaType classes doesn't contain the method: set(MediaType[] types).
We can use following code to return all methods in MediaType:
The Problem - Breakpoint doesn't work
When we step through classes from some jars in Eclipse, we may find that the code doesn't match or breakpoint doesn't work at all.
This usually means there are multiple versions of same class or library in your application, the one java uses is not same as the Eclipse loads to debug.
you can check what java is using by - this.getClass()/(XClass.class).getProtectionDomain().getCodeSource().getLocation()
You can check what jar Eclipse is loading in package view - if "Link with Editor" is enabled.
References
Uploading Data with Solr Cell using Apache Tika
Solr ExtractingRequestHandler
Today I am adding Solr Cell(Tika) to our Solr application, during test, it throws the following exception:
Caused by: java.lang.NoSuchMethodError: org.apache.tika.mime.MediaType.set([Lorg/apache/tika/mime/MediaType;)Ljava/util/Set;
at org.apache.tika.parser.crypto.Pkcs7Parser.getSupportedTypes(Pkcs7Parser.java:52)
at org.apache.tika.parser.CompositeParser.getParsers(CompositeParser.java:81)
This looks like there is a conflicting tika-core jar.
I used the java decompiler JD-GUI to check the jars I added: solr\contrib\extraction\lib\tika-core-1.3.jar, the class MediaType does contain this method: set(MediaType[] types).
Then it seems there are some other jars containing the MediaType class, I checked solr.war\WEB-INF\lib, but no obvious hint.
Using Eclipse Display View to Check which jar Contains the Class
I enabled the remote debug, added a breakpoint at Pkcs7Parser.getSupportedTypes(Pkcs7Parser.java:52), reran the Solr Cell request, it hit and stops at Pkcs7Parser.getSupportedTypes(Pkcs7Parser.java:52).
java.security.CodeSource src = org.apache.tika.mime.MediaType.class.getProtectionDomain().getCodeSource(); return src.getLocation();The output:
(java.net.URL) file:omitted/webapps/server/WEB-INF/lib/crawler4j-dependency.jar
Check crawler4j-dependency.ja, so now the root cause is obvious.
The culprit is that some one added crawler4j into the Solr application and put its all dependencies into crawler4j-dependency.jar. It uses tika-core-1.0.jar, the MediaType classes doesn't contain the method: set(MediaType[] types).
We can use following code to return all methods in MediaType:
java.lang.reflect.Method[] methods = org.apache.tika.mime.MediaType.class.getMethods(); return methods;
The Problem - Breakpoint doesn't work
When we step through classes from some jars in Eclipse, we may find that the code doesn't match or breakpoint doesn't work at all.
This usually means there are multiple versions of same class or library in your application, the one java uses is not same as the Eclipse loads to debug.
you can check what java is using by - this.getClass()/(XClass.class).getProtectionDomain().getCodeSource().getLocation()
You can check what jar Eclipse is loading in package view - if "Link with Editor" is enabled.
References
Uploading Data with Solr Cell using Apache Tika
Solr ExtractingRequestHandler
