Skip to content

README.md

Latest commit

 

History

History
110 lines (87 loc) · 4.33 KB

README.md

File metadata and controls

110 lines (87 loc) · 4.33 KB

Java Bytecode Fact Extractor

Usage

From the root folder of this project:

  • Build a jar file of this project with the command mvn package; output jar file is in target/ subfolder
  • Try applying it to itself:
java -jar target/jvm-fact-extractor-0.0.1-SNAPSHOT.jar  target/jvm-fact-extractor-0.0.1-SNAPSHOT.jar

  • Output is in /tmp/data.trig (in TriG syntax); you can also specify output file with -o filename, or use a persistent Jena TDB store with -d dbDir.

  • Run with -s to invoke SPARQL server at http://localhost:3330/

  • Run with -jung to invoke JFrame Jung visualisation of the (manually in-code selected) graph

Interesting queries: http://localhost:3330/dataset?query=PREFIX m: http://model.nuthatchery.org/maven/ SELECT ?x ?p ?y FROM http://db.nuthatchery.org/java/guava-18.0.jar WHERE {?x ?p ?y}

PREFIX nh: <http://model.nuthatchery.org/common/>
PREFIX m: <http://model.nuthatchery.org/maven/project/>
PREFIX j: <http://model.nuthatchery.org/java/>
PREFIX jType: <http://model.nuthatchery.org/java/types/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix owl: <http://www.w3.org/2002/07/owl#>

SELECT ?g ?m ?code  ?op # (group_concat(?code;separator=', ') as ?codes)
WHERE {
  GRAPH ?g {
    ?cls nh:name "org/apache/commons/collections/set/UnmodifiableSet" .
    ?m j:memberOf <this:org/apache/commons/collections/set/UnmodifiableSet> .
    ?m nh:defines ?t .
    ?m j:code  / rdf:rest* / rdf:first ?inv .
    ?inv j:call ?code .
    ?inv (j:varOperand|j:constantOperand|j:memberOperand|j:labelOperand|j:typeOperand) / nh:idName? ?op .
    FILTER (!isBlank(?op)) .
  }
}
#GROUP BY ?g ?cls ?m ?code  ?op
Find all methods that call themselves (directly)
PREFIX nh: <http://model.nuthatchery.org/common/>
PREFIX m: <http://model.nuthatchery.org/maven/project/>
PREFIX j: <http://model.nuthatchery.org/java/>
PREFIX jType: <http://model.nuthatchery.org/java/types/>
PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
prefix owl: <http://www.w3.org/2002/07/owl#>

SELECT ?g ?class ?caller
WHERE {
  GRAPH ?g {
    #?class nh:name "org/apache/commons/collections/set/UnmodifiableSet" .
    ?caller j:memberOf ?class.
    ?caller j:code  / rdf:rest* / rdf:first ?anInstruction .
    
          {?anInstruction j:call j:invokevirtual}
    UNION {?anInstruction j:call j:invokespecial}
    UNION {?anInstruction j:call j:invokeinterface}
    UNION {?anInstruction j:call j:invokedynamic} .

    ?anInstruction j:memberOperand ?callee .
    FILTER (?caller = ?callee) .
  }
}

http://localhost:3330/dataset?query=PREFIX m: http://model.nuthatchery.org/maven/project/ SELECT ?x ?p ?y FROM http://db.nuthatchery.org/java/jvm-fact-extractor-0.0.1-SNAPSHOT.jar WHERE {?x ?p ?y}

http://localhost:3330/dataset?query=PREFIX m: http://model.nuthatchery.org/maven/project/ SELECT ?x ?p ?y FROM http://db.nuthatchery.org/java/jvm-fact-extractor-0.0.1-SNAPSHOT.jar WHERE {?x m:artifactID ?y}

With Graal

Run the application (App.java) in graal-getting-started. It'll load /tmp/data.trig, set up a few inference rules, run a couple of queries, and then expect further queries on stdin.

Ontology conventions

  • To make an ontology model: OntModel m = ModelFactory.createOntologyModel()
  • Name space must be added manually
  • Class names are capitalised CamelCase
    • Make them with m.createClass(NS + uriString)
  • Property names are uncapitalised camelCase
    • Make them with m.createProperty(NS + uriString)

Building models

Create a typed resource:

Resource coord1 = model.createResource(depUri, MavenFacts.MavenCoordinate);
Resource coord2 = model.createResource(depUri, MavenFacts.MavenCoordinate);

Add property:

coord1.addProperty(MavenFacts.dependsOn, coord2);

Inference

If MavenFacts.dependsOn is set up with MavenFacts.MavenCoordinate as rdf:range and rdf:domain, we can also infer that coord1 and coord2 are Maven coordinates.

This also means that anything used with MavenFacts.dependsOn is a Maven coordinate, including something like:

coord1.addProperty(MavenFacts.dependsOn, JAVA_SE_8)

With owl:inverseOf we can also infer something like coord2 mvn:hasDep coord1.