Pangor: A JavaScript repository from SALT Lab @ UBC

=======

README

Pangor is a static analyis framework for discovering and analyzing the pervasive bug patterns of JavaScript. For details on Pangor, see our method description and empirical study Mining the Pervasive and Detectable Bug Patterns of JavaScript.

Pangor contains two analysis types: keyword change analysis (a data mining framework) and static change analysis (a flow analysis framework).

Keyword Change Analysis

The keyword change analysis component has two parts. The first builds a data set of keyword changes in bug fixing commits. The second creates a set of clusters from the data set. Clusters contain similar bug patterns.

Data Set Construction

To run Pangor's keyword change analysis from the command line:

java ca.ubc.ece.salt.pangor.learning.LearningAnalysisMain -h

Usage: DataSetMain  [-cc (--complexity) N] [-d (--directory) VAL] [-ds (--dataset) VAL] [-h (--help)] [-r (--repositories) VAL] [-s (--supplement) VAL] [-tr (--threads) N] [-u (--uri) VAL]

 -cc (--complexity) N    : The maximum change complexity of a file to analyze.
 -d (--directory) VAL    : The git directory (e.g., /path/to/project/.git/).
 -ds (--dataset) VAL     : The data set file to read.
 -h (--help)             : Display the help file.
 -r (--repositories) VAL : The path to the file specifying the repositories to analyze.
 -s (--supplement) VAL   : The folder path to place any supplementary files.
 -tr (--threads) N       : The number of threads to be used.
 -u (--uri) VAL          : The uri of the public repository (e.g., https://github.com/qhanam/JSRepairClass.git).

Through Maven:

mvn exec:java -Dexec.mainClass="ca.ubc.ece.salt.pangor.ClassifyAnalysisMain" -Dexec.args="-h"

Cluster Construction

To run Pangor's clustering and keyword metrics program:

java ca.ubc.ece.salt.pangor.learning.LearningDataSetMain -h

Usage: DataSetMain  [-a (--arff-path) VAL] [-c (--clusters)] [-ds (--dataset) VAL] [-f (--filtered) VAL] [-h (--help)] [-m (--metrics)]

 -a (--arff-path) VAL : Folder to write the ARFF files.
 -c (--clusters)      : Print the clusters from the data set.
 -ds (--dataset) VAL  : The data set file to read.
 -f (--filtered) VAL  : The file to write the filtered data set to.
 -h (--help)          : Display the help file.
 -m (--metrics)       : Print the metrics from the data set.

Static Change Analysis

The static change analysis component provides a framework for building static analysis 'checkers' that investigate changes between two versions of source code. For our study we look for instances of bug repairs, but there are many other uses for the framework. The static change analysis component has two main programs. The first builds a data set of bug pattern instance alerts. The second filters out duplicates from the data set and computes metrics (i.e., counts each type of alert).

Analysis

To run Pangor's checker from the command line:

java ca.ubc.ece.salt.pangor.classify.ClassifyAnalysisMain -h

Usage: ClassifyMain  [-ds (--dataset) VAL] [-h (--help)] [-pp (--preprocess)] [-r (--repositories) VAL] [-s (--supplement) VAL] [-tr (--threads) N] [-u (--uri) VAL]

 -ds (--dataset) VAL     : The data set file to write to.
 -h (--help)             : Display the help file.
 -pp (--preprocess)      : Pre-process the AST before running GumTree. Expands ternary operators and short circuits.
 -r (--repositories) VAL : The path to the file specifying the list of repositories to analyze.
 -s (--supplement) VAL   : The folder path to place the supplementary files.
 -tr (--threads) N       : The number of threads to be used.
 -u (--uri) VAL          : The uri of the public repository (e.g., https://github.com/qhanam/JSRepairClass.git).

ClassifyAnalysisMain creates the following artifacts:

dataset.csv: The data set containing the detected instances of the bug patterns.
supplementary/: The buggy and repaired source code files for each instance.

Duplicate Filtering and Metrics

To run Pangor's checker duplicate filtering and metrics program from the command line:

java ca.ubc.ece.salt.pangor.classify.ClassifyDataSetMain -h

Usage: DataSetMain  [-ds (--dataset) VAL] [-f (--filtered) VAL] [-h (--help)] [-m (--metrics)]

 -ds (--dataset) VAL : The data set file to read.
 -f (--filtered) VAL : The file to write the filtered data set to.
 -h (--help)         : Display the help file.
 -m (--metrics)      : Print the metrics from the data set.

ClassifyDataSetMain creates the following artifacts:

filtered.csv: The data set with duplicate alerts removed.

Installation

Pangor is a Maven project.

Clone Pangor.
Build and install the project (mvn clean install).

Optional: Create the Eclipse project files (mvn eclipse:eclipse).

Directory Structure

src\: The source code for the Pangor framework.
test\: Testing files for the Pangor framework.
- src\: The source code for the Pangor framwork tests.
- input\: JavaScript files used in the Pangor framework tests.
tools\: Scripts for inspecting bug pattern clusters.
input\: Contains a repository list for input to LearningAnalysisMain and ClassifyAnalysisMain.

saltlab/Pangor