/feTruth

Primary LanguageJavaGNU General Public License v3.0GPL-3.0

Table of Contents

General info

feTruth is a tool written in Python that can detect feature envy methods in a Java project.

Requirements

  • Python == 3.6
  • Tensorflow == 2.4.0
  • Keras == 2.4.3
  • Numpy == 1.19.5
  • Gensim == 3.8.0
  • Scikit-learn == 0.24.2
  • Java 11 or newer

You can setup the requirements via any of the following commands:

  1. requirements.txt

    pip install -r requirements.txt

  2. setup.py

    python setup.py or ./setup.py

  3. conda environment

    conda env create -f fetruth.yaml

How to run feTruth

  1. Clone feTruth to your local file system

    git clone https://github.com/lyoubo/feTruth.git

  2. Install dependencies

    cd feTruth/library

    pip install -r requirements.txt

  3. Download the two large files requested by the word2vec model

    Download new_model.bin.trainables.syn1neg.npy and new_model.bin.wv.vectors.npy from Zenodo and put them into folder word2vec. And now, the folder word2vec should contain 3 files (including one copy automatically from this repository).

  4. Run feTruth

    cd feTruth/implementation/

    python feTruth.py -a E:/jsoup

In this case, the identified feature envy methods are available at jsoup.txt in folder output:

Refactoring Type	Source Method	Target Class
Move Method	org.jsoup.helper.DataUtil::mimeBoundary():java.lang.String 	 org.jsoup.internal.StringUtil
Move Method	org.jsoup.helper.HttpConnection.Response::serialiseRequestUrl(org.jsoup.Connection.Request):void 	 org.jsoup.internal.StringUtil
Move Method	org.jsoup.Jsoup::parse(java.io.File, java.lang.String, java.lang.String):org.jsoup.nodes.Document 	 org.jsoup.helper.DataUtil
Move Method	org.jsoup.Jsoup::parse(java.io.File, java.lang.String):org.jsoup.nodes.Document 	 org.jsoup.helper.DataUtil
Move Method	org.jsoup.Jsoup::parse(java.io.InputStream, java.lang.String, java.lang.String):org.jsoup.nodes.Document 	 org.jsoup.helper.DataUtil
Move Method	org.jsoup.Jsoup::parse(java.io.InputStream, java.lang.String, java.lang.String, org.jsoup.parser.Parser):org.jsoup.nodes.Document 	 org.jsoup.helper.DataUtil
Move Method	org.jsoup.Jsoup::parseBodyFragment(java.lang.String):org.jsoup.nodes.Document 	 org.jsoup.parser.Parser
Move Method	org.jsoup.nodes.Attribute::html(java.lang.String, java.lang.String, java.lang.Appendable, org.jsoup.nodes.Document.OutputSettings):void 	 org.jsoup.nodes.Attributes
Move Method	org.jsoup.nodes.Element::appendNormalisedText(java.lang.StringBuilder, org.jsoup.nodes.TextNode):void 	 org.jsoup.internal.StringUtil
Move Method	org.jsoup.nodes.Entities::escape(java.lang.String, org.jsoup.nodes.Document.OutputSettings):java.lang.String 	 org.jsoup.internal.StringUtil
Move Method	org.jsoup.nodes.Entities::escape(java.lang.Appendable, java.lang.String, org.jsoup.nodes.Document.OutputSettings, boolean, boolean, boolean):void 	 org.jsoup.internal.StringUtil
Move Method	org.jsoup.parser.HtmlTreeBuilderState::handleRcData(org.jsoup.parser.Token.StartTag, org.jsoup.parser.HtmlTreeBuilder):void 	 org.jsoup.parser.TreeBuilder
Move Method	org.jsoup.parser.Parser::parseBodyFragment(java.lang.String, java.lang.String):org.jsoup.nodes.Document 	 org.jsoup.nodes.Document
Move Method	org.jsoup.parser.TokeniserState::readEndTag(org.jsoup.parser.Tokeniser, org.jsoup.parser.CharacterReader, org.jsoup.parser.TokeniserState, org.jsoup.parser.TokeniserState):void 	 org.jsoup.parser.CharacterReader
Move Method	org.jsoup.parser.TokenQueue::unescape(java.lang.String):java.lang.String 	 org.jsoup.internal.StringUtil

Dataset

The dataset we used to train feTruth includes 33,419 positive samples and 33,419 negative samples, is available in the following links:

 JSON property descriptions

 id: identity of the moved method
 project_name: project name
 commit_id: Git commit ID
 source_class_name: class name that the moved method belonged to before the move
 method_name: name of the moved method
 source_dist: Jaccard distance between the moved method and the source class
 source_cbmc: coupling between the moved method and the source class
 source_mcmc: message passing coupling between the moved method and the source class
 target_class_name: class name that the moved method belonges to after the move
 target_dist: Jaccard distance between the moved method and the target class
 target_cbmc: coupling between the moved method and the target class
 target_mcmc: message passing coupling between the moved method and the target class
 refactoring_id: identity of the refactoring operation

The source code used by feTruth for data generation is available in the following links: