Search code examples
runtime-errorstatic-analysisinria-spoon

How do we deal with "The type xxx is already defined" in spoon's source code analysis for a big project's source code?

When we use spoon to analyze the source code of a big project like Hadoop, it occurs frequently the problem "The type xxx is already defined", because there may exist Java classes with the same class name and the same package directory in different sub projects.

I got the following error when I run "java -cp xx spoon.Launcher -i ~/hadoop-0.23.3-src/ -p myspoon.CatchProcessor".

Exception in thread "main" spoon.compiler.ModelBuildingException: The type JobInProgress is already defined
    at spoon.support.compiler.jdt.JDTBasedSpoonCompiler.reportProblem(JDTBasedSpoonCompiler.java:550)
    at spoon.support.compiler.jdt.TreeBuilderRequestor.acceptResult(TreeBuilderRequestor.java:37)
    at spoon.support.compiler.jdt.TreeBuilderCompiler.buildUnits(TreeBuilderCompiler.java:73)
    at spoon.support.compiler.jdt.JDTBatchCompiler.getUnits(JDTBatchCompiler.java:120)
    at spoon.support.compiler.jdt.JDTBasedSpoonCompiler.buildUnits(JDTBasedSpoonCompiler.java:410)
    at spoon.support.compiler.jdt.JDTBasedSpoonCompiler.buildUnitsAndModel(JDTBasedSpoonCompiler.java:372)
    at spoon.support.compiler.jdt.JDTBasedSpoonCompiler.buildSources(JDTBasedSpoonCompiler.java:348)
    at spoon.support.compiler.jdt.JDTBasedSpoonCompiler.build(JDTBasedSpoonCompiler.java:119)
    at spoon.support.compiler.jdt.JDTBasedSpoonCompiler.build(JDTBasedSpoonCompiler.java:102)
    at spoon.Launcher.buildModel(Launcher.java:700)
    at spoon.Launcher.run(Launcher.java:651)
    at spoon.Launcher.run(Launcher.java:106)
    at spoon.Launcher.main(Launcher.java:99)

And I find there are two JobInProgress.java in the same sub-project "hadoop-mapreduce-project" 

./hadoop-mapreduce-project/hadoop-mapreduce-client/hadoop-mapreduce-client-core/src/main/java/org/apache/hadoop/mapred/JobInProgress.java
./hadoop-mapreduce-project/src/java/org/apache/hadoop/mapred/JobInProgress.java

So how to fix this problem ????? I already found that it seems we couldn't move away one of both to run spoon analysis, because it will cause another problem of "missing files"

Exception in thread "main" spoon.compiler.ModelBuildingException: The import org.apache.hadoop.conf cannot be resolved at xxx

Any comments for the problem? Thanks.

Solution

  • Projects like Hadoop are defined with several sub-modules compiling independently: then you cannot consider the whole source code as one big project.

    Spoon is not designed to support maven project with submodules, so you have to submit each submodule to Spoon independently, but you have to respect the module hierarchy in doing so, and more important, you have to use the right classpath: for example, if module A is dependent to module B, then don't forget to pass B.jar when you analyze A.