# bugfinder-framework > Framework used to research bug correlations to quantifications of software artefacts and automatically find bugs with machine learning methods Latest version **10.1.0** (published 2021-11-27) · ISC license · 0 weekly downloads ## Install ```sh npm install bugfinder-framework pnpm add bugfinder-framework yarn add bugfinder-framework bun add bugfinder-framework ``` ## Health **Score 20/100 (F)** — status: abandoned. Positive: has types; no vulnerabilities. Warnings: low downloads; no esm support. Negative: abandoned; low maintenance score. ## Facts | | | |---|---| | Version | 10.1.0 | | Published | 2021-11-27 | | First published | 2021-09-03 | | Weekly downloads | 0 | | License | ISC | | TypeScript types | bundled | | Module format | CommonJS | | Dependencies | 3 | | Unpacked size | 95 KB | | Known vulnerabilities | 0 | | Install scripts | no | | GitHub stars | 0 | | Author | Robert Klinger | | Maintainers | penguinsarefunny | ## Links - npm: https://www.npmjs.com/package/bugfinder-framework - Repository: https://github.com/penguinsAreFunny/bugFinder-framework - Homepage: https://github.com/penguinsAreFunny/bugFinder-framework#readme - Issues: https://github.com/penguinsAreFunny/bugFinder-framework/issues - npm.io page: https://npm.io/package/bugfinder-framework ## Dependencies (3) - [ts-log](https://npm.io/package/ts-log.md) ^2.2.3 - [@types/node](https://npm.io/package/@types/node.md) ^15.12.2 - [bugfinder-framework-defaultcontainer](https://npm.io/package/bugfinder-framework-defaultcontainer.md) ^1.4.0 ## Recent versions - 10.1.0 (latest) — 2021-11-27 - 10.0.2 — 2021-11-27 - 10.0.1 — 2021-11-25 - 10.0.0 — 2021-11-03 - 9.13.2 — 2021-10-26 - 9.13.1 — 2021-10-26 - 9.13.0 — 2021-10-26 - 9.12.0 — 2021-10-25 - 9.11.0 — 2021-10-25 - 9.10.1 — 2021-10-25 - 9.10.0 — 2021-10-24 - 9.9.0 — 2021-10-24 - 9.8.3 — 2021-10-15 - 9.8.2 — 2021-10-15 - 9.8.1 — 2021-10-15 - … 41 more at https://npm.io/package/bugfinder-framework/versions ## README # BugFinder-Framework This is a framework dedicated to automating finding bugs in source artefacts based on measurements and machine learning algorithms. This framework´s main use case is researching correlations between measurements and bug occurrence. It addresses reusability, modifiability and portability (adaptability, installibility and replaceability) of components needed to research correlations between source artefact measurements and bug occurrence. Different researchers should be able to exchange components of the bugfinder program and therefore be able to distribute research among researchers world wide. With all needed components given you should be able to automatically analyse code and find localities with high bug probability. # Table of Contents - [BugFinder-Framework](#bugfinder-framework) - [Table of Contents](#table-of-contents) - [Quick start](#quick-start) - [Introduction](#introduction) * [Recording by example](#recording-by-example) * [Preprocessing by example](#preprocessing-by-example) * [Training](#training) - [Component realisations available](#component-realisations-available) * [Shared](#shared) + [DB](#db) * [Recording](#recording) + [LocalityRecorder (and therefore localities)](#localityrecorder--and-therefore-localities-) + [LocalityPreprocessors](#localitypreprocessors) + [Quantifier](#quantifier) + [Annotator](#annotator) * [Preprocessor](#preprocessor) * [Machine Learning](#machine-learning) # Quick start The best way to start is to clone the [bugFinder repository](https://github.com/penguinsAreFunny/bugFinder) and make changes there. You can change components for locality-recording, locality-preprocessing, quantification, annotation and preprocessing. # Introduction The whole process beginning with recording until classification of localities is designed with a pipeline: ![Machine_Learning_Pipeline](./doc/Pipeline.svg) This project supports the learning phase. This framework defines all interfaces for each step of the pipeline. Researchers want to develop or use others´ implementation of the components needed to realise this machine learning pipeline. Please see the main repository: [bugFinder repository](https://github.com/penguinsAreFunny/bugFinder) for whole documentation. ## Recording by example All used examples in the following documentations are already implemented and ready to use. Record all features. F.e. record all Lines Of Code for each file in each Commit. 1. You first need to record all localities, CommitPaths, of a repository. A CommitPath is a path in a commit. 2. After that you can preprocess your CommitPaths. For example filtering all the paths you do not want to quantify, like .txt-, .md-files and only keep CommitPaths with paths in src ending with .ts. 3. After that you want to quantify your CommitPaths, f.e. with the SonarQubeQuantifier. Therefore getting a map of CommitPaths to SonarQubeMeasurements. 4. You might also want to annotate your localities so that you have labels needed for suppervised learning algorithms. You can label your CommitPaths with analyzing the Commit-message for semantic key words like "bug, fix, error, fail" indicating a correction of a bug. Now you have a database of localities, quantifications (features) and annotations (classes). ## Preprocessing by example Preprocess your data from the recording phase. F.e.: Find the last 3 changes of a CommitPath with a bug indicating commit message. Those changes are localities by them self: CommitPaths. Now you can analyse the changes made before a CommitPath got a bug. F.e. the max. added Lines Of Code in a change. In the training phase you might find statistical correlation between the amount of lines of codes added in one single change and the likelyhood of a bug beeing fixed later in that file. Therefore indicating a bug added in that file. ## Training Training is give by a template project: @see [bugFinder-machineLearning](https://github.com/penguinsAreFunny/bugFinder-machineLearning) You want to train a model to be able to predict bugs using the dataset generated by the previous steps. # Component realisations available You can find different components realisations open source on github and npm. Search for bugfinder-* ## Shared ### DB - [Commit-MongoDB](https://www.npmjs.com/package/bugfinder-commit-db-mongodb) - [CommitPath-MongoDB](https://www.npmjs.com/package/bugfinder-commitpath-db-mongodb) ## Recording ### LocalityRecorder (and therefore localities) npm search: bugfinder-localityrecorder-* - [CommitPath](https://www.npmjs.com/package/bugfinder-localityrecorder-commitpath) - [Commit](https://www.npmjs.com/package/bugfinder-localityrecorder-commit) ### LocalityPreprocessors npm search: bugfinder-$LOCALITY_CLASS-localityPreprocessor-* - [Predecessors](https://www.npmjs.com/package/bugfinder-commitpath-localitypreprocessor-commitsubset) ### Quantifier npm search: bugfinder-$LOCALITY_CLASS-quantifier-* - [SonarQube](https://github.com/penguinsAreFunny/bugFinder-commitPath-quantifier-sonarqube) - [SonarQubePredecessors](https://github.com/penguinsAreFunny/bugFinder-commitPath-quantifier-sonarqubePredecessors) ### Annotator npm search: bugfinder-$LOCALITY_CLASS-annotator-* - [CommitMsg](https://www.npmjs.com/package/bugfinder-commitpath-annotator-commitmsg) - [CommitMsgPredecessors](https://github.com/penguinsAreFunny/bugFinder-commitPath-annotator-commitMsgPredecessors) ## Preprocessor npm search: bugfinder-$LOCALITY_CLASS-$ANNOTATION_TYPE-$QUANTIFICATION_TYPE-preprocessor-* - [featureSelection](https://github.com/penguinsAreFunny/bugFinder-commitpath-number-sonarqube-preprocessor-featureSelection) ## Machine Learning - [bugFinder-machineLearning](https://github.com/penguinsAreFunny/bugFinder-machineLearning) --- _Source: https://npm.io/package/bugfinder-framework · Machine-readable twin of the npm.io package page. Health data is recomputed on every publish._