This repository contains:
- Darcy, a tool designed to automatically detect and repair architectural inconsistencies that may arise in developing Java applications using Java Platform Module System
- The input data needed for running this tool
- The materials used in evaluating it
The DOI for this repository is
In the following sections, we explain all the materials contained in this repository.
Darcy’s main goal is to detect and repair architectural inconsistencies that may arise within Java applications using Java Platform Module System (JPMS), introduced in the 9th version of Java. In the paper, we formally specify eight types of inconsistent dependencies that may arise when using JPMS. Darcy leverages these definitions and static analyses to automatically (1) detect the specified inconsistent dependencies within Java applications and (2) repair those identified inconsistencies.Darcy’s implementation contains several components:
- Class Dependency Analysis: To identify actual dependencies of an input Java application, Darcy relies on an static analysis tool, Classycle [1] , which provides a complete report of all dependencies in source code of a Java application at both the class and package levels. (
TraverseDir.py) - Parsing the XML files generated by Classycle, implemented in Python (
ParseXML.py) - Java Inconsistency Analysis: To identify all types of inconsistency scenarios, for each directive in a module-info.java file, Java Inconsistency Analysis explores actual and specified dependencies, stored in their respective database components. It identifies any occurrence of an inconsistent dependency previously defined and reports the identified architectural inconsistency, the module affected,and the specific directive involved. This component is implemented in Java (
Check_Consistency.jar). - Java Reflection Analysis: It leverages a custom static analysis, which we have implemented using the Soot framework [2], to identify usage of reflection in the input application. This component is implemented in Java (
java_reflection_analysis.jar) - ServiceLoader Analysis: It leverages a custom static analysis using the Soot framework to identify usage of java.util.ServiceLoader in the input application. This component is implemented in Java (
uses_provides_analysis.jar) - Module-Info Transformer: This component leverages ANTLR [3] and deletes or modifies the explicit dependencies defined in the module-info files to repair the identified inconsistencies within Java applications. This component is implemented in Java (
ModuleInfoTransformer.jar).
The runAll.sh script, located in the Darcy directory, runs all these components in the explained order. After running this script, it asks for a Java project path as an input, runs Darcy’s components, and reports its output. The output contains all the reports regarding each component. The identified inconsistencies are reported as "FOUND INCONSISTENCIES", and also saved in a text file excessDirectives.txt in the Darcy directory. The identified inconsistencies of type opens are reported in the Java reflection analysis output and also saved in excessOpenDirectives.txt in the Darcy directory. The inconsistencies of type uses are reported as "Excess Uses", and also saved in excessUsesDirectives.txt in the Darcy directory. Finally, in the output of the repair section, the modified module-info files are printed after Darcy’s repair. The original module-info files within the target Java project is also modified and saved.
In this repository, all the source files and the runAll.sh script are located in Darcy directoy.
The Darcy directory also contains the input data for executing Darcy; this data were used in evaluating Darcy as reported in the paper. This input data contains open-source Java-9 applications which are located in a folder named dataset_projects with the following address:
Darcy/dataset_projects
You can use any of the Java projects within this folder as an input to the runAll.sh script; however, for more convenience, you can use the following example input:
- First go to
Darcydirectory and run the script:
cd Darcy
./runAll.sh
- When it asks for "Project Path", enter the absolute path for the following direcotry:
Darcy/dataset_projects/sense-nine-start-point
A copy of the output for this specific input is provided in exampleOutput.txt at the root directory of this repository. Note that the statistics regarding the execution time at the end of the output may slightly vary.
We have provided all the materials used in the evaluation process located at a folder named evaluation in the root directory of the GitHub repository. There is an excel file results.xlsx in the evaluation folder that contains all the evaluation results presented in the paper (Table 2-6) along with their calculations. Each table’s result is represented in a separate sheet:
- Identified Inconsistencies: This sheet contains information about all 38 Java-9 application in our dataset, along with their identified inconsistencies, separated by their types, and whether they were successfully compiled after the repair. It also shows the test passing rate for those project containing test suite, after Darcy’s repair.
- Attack Surface Reduction: This sheet includes information about Darcy’s ability to reduce the attack surface of Java-9 applications and its calculations using the number of exposed packages.
- Encapsulation Improvement: This sheet contains information about the encapsulation improvement within Java projects of our dataset after Darcy's repair, along with the calculations of the two metrics used in the paper (RoC and NCD).
- Software Bloat Reduction: This sheet includes information about the runtime memory reduction in Java-9 applications after Darcy’s repair.
- Execution Time: This sheet includes the execution time of each component of Darcy for all Java-9 applications within our dataset.
[1] F.-J. Elmer, "Classycle: Analysing Tools for Java Class and Package Dependencies," How Classycle works, 2012.
[2] R.Valle ́e-Rai,P.Co,E.Gagnon, L.Hendren, P.Lam, and V.Sundaresan, "Soot: A Java bytecode optimization framework," in CASCON First Decade High Impact Papers. IBM Corp., 2010, pp. 214–224.
[3] "ANTLR," http://www.antlr.org, 2018.