On 31/10/2010 05:35, Roger Smith wrote:

We are working on a project to integrate Drools with Apache Hadoop, http://hadoop.apache.org/, and run into some road blocks. We would very much appreciate any suggestions/help from this list. 

We have an app where we call drools rule engine inside the reducer task of a Hadoop map reduce job. It throws a NullPointerException when we the rule package resource to the knowledge builder. The same code works fine when run as part of a stand alone app. 

Code:

    private static Map<String, StatefulKnowledgeSession> sessions =
            new HashMap<String, StatefulKnowledgeSession>();
    private static final String RULE_PACK_DIR = "file:///home/roger/Projects/gridx/";
    private static final String RULE_PACK_EXT = ".drl";
   

    public int process(String rulePackage, String dateTime, String type) throws TException {
        int rate = 0;
        StatefulKnowledgeSession session = sessions.get(rulePackage);
        if (null == session){
            KnowledgeBuilder kbuilder = KnowledgeBuilderFactory.newKnowledgeBuilder();
            String rulePackPath = RULE_PACK_DIR + rulePackage + RULE_PACK_EXT;
            kbuilder.add( ResourceFactory.newFileResource(rulePackPath ), ResourceType.DRL);
            if ( kbuilder.hasErrors() ) {
                System.err.println( kbuilder.getErrors().toString() );
            }
            KnowledgeBase kbase = KnowledgeBaseFactory.newKnowledgeBase();
            kbase.addKnowledgePackages(kbuilder.getKnowledgePackages());

            session = kbase.newStatefulKnowledgeSession();
            sessions.put(rulePackage, session);
        }

        ContractRule contractRule = new ContractRule();
        contractRule.prepare(dateTime, type);;
        FactHandle ruleHandle = session.insert(contractRule);
        session.fireAllRules();
        System.out.println("" + contractRule);
        rate = contractRule.getRate();
        session.retract(ruleHandle);

        return rate;
    }

This line throws the exception:
kbuilder.add( ResourceFactory.newFileResource(rulePackPath ), ResourceType.DRL);

ResourceFactory.newFileResource just creates a file object but cleaning up the paths.
    public FileSystemResource(File file) {
        if ( file == null ) {
            throw new IllegalArgumentException( "File must not be null" );
        }
        this.file = new File( StringUtils.cleanPath(file.getPath()) );
    }
The code is borrowed from Spring and uses the sme cleanPath logic. Once the file is created it just creates the stream as so:
    public InputStream getInputStream() throws IOException {
        this.lastRead = getLastModified();
        return new FileInputStream(this.file);
    }

So it's pretty basic, the stack trace with line numbers may help. But I suspect you'll need to do a debug and step into the drools code at the point where you get the nullpointer and see what's going on.

Mark


It works fine as a stand alone app, outside hadoop

Roger Smith

 

_______________________________________________
rules-users mailing list
rules-users@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-users