the java source code RDF modeler.
Author: Michele Mostarda ( me@michelemostarda.com )
RDF Coder is a CLI and a flexible library designed to generate RDF models from Java code, both source and bytecode.
RDF Coder is written in Java and can be used to perform multi level code inspection, create code dependency graphs, generate custom documentation.
Currently RDF Coder supports only the Java language.
- Code inspection: navigate large package structures, classes, methods, signatures and dependencies among them.
- Code Refactoring: find package, classe and method relationships, planify refactoring.
- Code issue: troubleshouting large project dependencies, find conflicts.
- Custom Documentation: generate custom documentation, on compiled or incomplete code.
To fully master the key concepts of this tool it is strongly recommended to have a general understaning of RDF and SPARQL.
- [RDF] (https://www.w3.org/RDF) (Resource Description Framework) is a standard model for data interchange on the Web. RDF has features that facilitate data merging even if the underlying schemas differ, and it specifically supports the evolution of schemas over time without requiring all the data consumers to be changed.
- [SPARQL] (https://www.w3.org/TR/sparql11-query/) (SPARQL Protocol and RDF Query Language) is an RDF query language, that is, a semantic query language for databases, able to retrieve and manipulate data stored in Resource Description Framework (RDF) format.
- JDK +1.8
- Apache Ant +1.9.3
Clone or download the latest version on [Github] (https://github.com/michelemostarda/rdfcoder), build and use it.
git clone https://github.com/michelemostarda/rdfcoder.git
cd rdfcoder
ant clean dist
bin/rdfcoder
The RDFCoder executable provides a CLI which works both in batch (via the --command option) and interactive mode (default mode). To get help simply use the -h option.
bin/rdfcoder -h
usage: rdfcoder
-c,--command <arg> Specify command
-d,--dir <arg> Specify working dir
-h,--help Get help
-j,--json-out Specify that all output is in JSON format.
Using the RDFCoder in batch mode allows to execute a list of commands:
$ bin/rdfcoder -c 'pwd;ls lib;loadclasspath arq jar:lib/arq.jar; inspect model'
Initializing JRE data ...
INFO [main] (JavaProfile.java:278) - Objects Table loaded.
JRE data loaded.
/Users/hardest/rdfcoder-github/.
- antlr-2.7.5.jar rw 435563
- arq.jar rw 818802
- bcel-6.0.jar rw 670734
[...]
loading /Users/hardest/repository/RDFCoder/rdfcoder-github/./lib/arq.jar ... done
parsing time (secs):0.0
parsed files: 543
parsed classes: 494
[...]
com.asemantics.rdfcoder.model.java.JavaQueryModelImpl{packages: 31, classes: 494, interfaces: 49, enumerations: 0}
If invoked with no arguments the binary executable runs the Interactive Console.
$ bin/rdfcoder
Initializing JRE data ...
Objects Table loaded.
JRE data loaded.
RDFCoder interactive console [version 0.5]
.~default>
The --json-out option runs the CLI in JSON output mode which means that every command output is espressed with a JSON object. Such modality can be used to interact with the CLI in a programmatic way.
hardest@hardest-mac:~/repository/RDFCoder/rdfcoder-github$ bin/rdfcoder -j
{"type":"out_message","content":"Initializing JRE data ..."}
{"type":"out_message","content":"JRE data loaded."}
{"type":"out_message","content":"RDFCoder interactive console [version 0.5]"}
.~default> pwd
{"operation":"pwd","result":"/Users/hardest/rdfcoder-github/"}
.~default>
Assuming that you have downloaded and compiled RDFCoder successfully, now you can run the Interactive Console
bin/rdfcoder
You'll see a message confirming that the JRE model has been loaded into the default model.
Initializing JRE ...
JRE model loaded.
RDFCoder interactive console [version 0.5]
[version 0.5]
.~default>
A model is a container of RDF statements. RDFCoder provides a default model containing the statements representing the JRE (Java Runtime Environment). Using the CLI you can:
- get help;
- list all the existing models;
- create new models;
- delete existing models;
- load statements into the current model;
- clear the content of the model returning it to the original state;
- inspect the structure of data stored in a model;
- query a model;
- dump the content of a model on the file system.
A command can be run simply specifying its name and parameters when required. It is possible to specify multiple commands per line using the semi-column separator ; example:
.~default> pwd
.~default> ls
.~default> pwd;ls;cd ..
To start let's invoke the help command:
.~default> help
Usage: <command> <parameters>
available commands:
cd Change the current directory
debug Show/set the debug flag
setmodel Set a new model
removemodel Remove the current model
ls List content of current directory
loadjre Load the JRE data into the active model
clearmodel Cleanup the current model
loadclasspath Load a classpath on the active model
list List the loaded models
savemodel Save the current model on file system
help Print this help
to obtain more information about a command type: help <command>
newmodel Create a new model
loadmodel Load a model from file system
inspect Inspect the active model
describe Describe the object referenced by the given path
pwd Print the current directory
querymodel Query the active model
A description of all available CLI commands can be found here.
Let's start creating a new model called test:
.~default> newmodel test
Model 'test' created.
Now let's check that test model has been created:
.~default> list
Models:
default[X]
test
We see two models, default (the active one is marked with [X]) and test.
Let's change now the active model to test:
.~default> setmodel test
Model set to 'test'
.~test>
we can see the new model into the prompt.
It's time now to load statements into the active model. RDFCoder supports two sources of data:
- JAR files;
- source code directories. To parse data from any source we always use the command loadclasspath.
The loadclasspath command accepts both relative and absolue paths, to use relative path the RDF CLI must be avare of a current directory.
You can check your current directory with pwd:
.~test> pwd
/Users/hardest/repository/RDFCoder/rdfcoder-github/.
You can also list the content of the current directory with ls:
.~test> ls
/Users/hardest/repository/RDFCoder/rdfcoder-github/.
d bin rwed 170
- build.xml rwed 9938
d classes rwed 102
d dist rwed 102
d doc rwed 102
d javadoc rwed 612
d lib rwed 782
[...]
For this tutorial we can use the content of the lib directory:
.~test> ls lib
/Users/hardest/repository/RDFCoder/rdfcoder-github/./lib
- antlr-2.7.5.jar rwed 435563
- arq.jar rwed 818802
- bcel-6.0.jar rwed 670734
- commons-logging.jar rwed 38015
- concurrent.jar rwed 126474
[...]
.~test>
Let's choose the arq.jar for the test.
.~test> loadclasspath arq jar:lib/arq.jar
loading /Users/hardest/repository/RDFCoder/rdfcoder-github/lib/arq.jar ... done
parsing time (secs):0.0
parsed files: 543
parsed classes: 494
parsed interfaces: 49
parsed attributes: 1853
parsed constructors: 768
parsed methods: 3748
parsed enumerations: 0
javadoc entries: 0
classes javadoc: 0
fields javadoc: 0
constructors javadoc: 0
methods javadoc: 0
generated temporary identifiers: 0
replaced entries: 0
unresolved [0] {
}
parse errors[0] {
}
The command output shows the data extraction statistics in term of detected entities.
The loadclasspath command accepts an identifier for the library and classpath resolutor, expressed with syntax <format>:<path>.
Read the command documentation with:
.~test> help loadclasspath
Load a classpath on the active model
syntax: loadclasspath <library_name> <library_location> [<library_name> <library_location>]+
where <library_location> can be expressed as
a jar file: jar:/path/to/jarfile.jar
a source dir: src:/path/to/src
a javadoc dir: javadoc:/path/to/src
a class dir: class:/path/to/class
Performs a parsing of the given set of resources
and loads extracted data within the current model
Now the model is ready to be queried. It is possible to query a model in two ways: using the querymodel command or using the inspect command.
The inspect command allows to perform predefined queries about the structure of packages, classes, methods and relations among them.
The querymodel instead allows to run generic [SPARQL] (https://www.w3.org/TR/sparql11-query/) (SPARQL Protocol and RDF Query Language) queries but require a specific degree of knowledge about such technology. For further reading about SPARQL and RDF please refer section [#background].
To inspect is used in combination with describe command. The describe command shows the structure of the specified element, while the inspect command allows to retrieve the content of the specified element.
These two commands accept an accessor expression with format obj1.obj2... or obj1[i].obj2[j]... or combinations of these expression.
The root element for the active model inspection is the model object.
.~test> describe model
class com.asemantics.rdfcoder.model.java.JavaQueryModelImpl
- asset: com.asemantics.rdfcoder.model.Asset
- classes: com.asemantics.rdfcoder.model.java.JClass[0]
- enumerations: com.asemantics.rdfcoder.model.java.JEnumeration[0]
- interfaces: com.asemantics.rdfcoder.model.java.JInterface[0]
- libraries: java.lang.String[0]
- packages: com.asemantics.rdfcoder.model.java.JPackage[0]
The command describe lists all the available accessors for the current object and the relative data type. For example, to access all the declared classes in model you must use the classes accessor which returns a list of com.asemantics.rdfcoder.model.java.JClass objects. To see all the available accessors of a com.asemantics.rdfcoder.model.java.JClass object simply run:
.~test> describe model.classes[0]
class com.asemantics.rdfcoder.model.java.JClass
- attributes: com.asemantics.rdfcoder.model.java.JAttribute[1]
- enumerations: com.asemantics.rdfcoder.model.java.JEnumeration[0]
- identifier: com.asemantics.rdfcoder.model.Identifier
- innerClass: boolean
- innerClasses: com.asemantics.rdfcoder.model.java.JClass[0]
- methods: com.asemantics.rdfcoder.model.java.JMethod[4]
- modifiers: com.asemantics.rdfcoder.model.java.JavaCodeModel$JModifier[1]
- name: java.lang.String
- parent: com.asemantics.rdfcoder.model.java.JContainer
- parentClass: com.asemantics.rdfcoder.model.java.JClass
- parentPackage: com.asemantics.rdfcoder.model.java.JPackage
- path: com.asemantics.rdfcoder.model.java.JContainer[8]
- queryModel: com.asemantics.rdfcoder.model.java.JavaQueryModel
- visibility: com.asemantics.rdfcoder.model.java.JavaCodeModel$JVisibility
To get the content of an accessor use the inspect command:
.~test> inspect model.classes[0]
Identifier<http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmd.jclass:QExec>
.~test>
with any depth:
.~default> inspect model.classes[0].parentPackage
Identifier<http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmd>
A simple way to obtain all interfaces declared within the active model is using the following query:
.~test> querymodel "select ?a where {?a <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.rdfcoder.org/2007/1.0#JInterface>}"
-------------------------------------------------------------------------------------------------------------------
| a |
===================================================================================================================
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.util.jinterface:Callback> |
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.lang.sparql.jinterface:SPARQLParserConstants> |
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.lang.rdql.jinterface:NodeValue> |
[...]
similarly to get all classes
.~test> querymodel "select ?a where {?a <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.rdfcoder.org/2007/1.0#JClass>}"
---------------------------------------------------------------------------------------------------------------------------
| a |
===========================================================================================================================
| <http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmd.jclass:QExec> |
| <http://www.rdfcoder.org/2007/1.0#jpackage:arq.jclass:rdql> |
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.serializer.jclass:FmtExprAbstract> |
[...]
and all methods
.~test> querymodel "select ?a where {?a <http://www.w3.org/2000/01/rdf-schema#subClassOf> <http://www.rdfcoder.org/2007/1.0#JMethod>}"
-----------------------------------------------------------------------------------------------------------------------------------------------
| a |
=============================================================================================================================================================
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.resultset.jclass:OutputBase.jmethod:class$> |
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.core.jclass:DataSourceImpl.jmethod:replaceNamedModel> |
| <http://www.rdfcoder.org/2007/1.0#jpackage:com.hp.hpl.jena.query.engine1.jclass:BindingImmutable.jmethod:equals> |
Get details about a specific method
.~test> querymodel "select * where {<http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmdline.jclass:CmdLineArgs.jmethod:getValue> ?p ?o}"
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| p | o |
==============================================================================================================================================================================
| <http://www.rdfcoder.org/2007/1.0/contains_signature> | <http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmdline.jclass:CmdLineArgs.jmethod:getValue.jsignature:_-378625919> |
| <http://www.rdfcoder.org/2007/1.0/contains_signature> | <http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmdline.jclass:CmdLineArgs.jmethod:getValue.jsignature:_-450679132> |
| <http://www.rdfcoder.org/2007/1.0/has_visibility> | "public" |
| <http://www.rdfcoder.org/2007/1.0/has_modifiers> | "0" |
| <http://www.w3.org/2000/01/rdf-schema#subClassOf> | <http://www.rdfcoder.org/2007/1.0#JMethod> |
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
.~test> querymodel "select * where {http://www.rdfcoder.org/2007/1.0#jpackage:arq.cmdline.jclass:CmdLineArgs.jmethod:getValue ?p ?o}"
Let's end this tutorial describing the model persistence. All models are persisted in memory. To store them on file system you can use the command savemodel. This command accepts a driver (fs or db) and additional attributes depending on the selected driver to perform the operation. To store the data on a file, you must specify the fs driver and the filename for the output file:
.~test> savemodel fs filename=test.rdf
Model saved.
this will create a file named test.rdf in the current path.
RDF Coder is mantained on [Github] (https://github.com/michelemostarda/rdfcoder)
Please refer to the internal [issue tracker] (https://github.com/michelemostarda/rdfcoder/projects/1).
I'm always looking for feedback and collaboration. Please feel free to write [me] (me@michelemostarda.com) and fork this project to improve it.
RDF Coder is released under the Apache 2 License the Apache License 2.0.