Skip to content

Latest commit

 

History

History
47 lines (42 loc) · 3.19 KB

README.md

File metadata and controls

47 lines (42 loc) · 3.19 KB
Raw

Clones Detection

This project features support for performing clones detection on MPS models.

Motivation

When models (both on language development and use sides) are developed over several years, people tend to copy/paste and adapt existing code. This leads to redundancies and hard to maintain code. Furthermore, there are situations when copy-paste is the only way to reuse some code, and this can reveal lack of adequate abstractions which would enable reuse.

Building the mps.qa.clones plugin

  1. rename "gradle.properties.example" in "gradle.properties"
  2. set the required macros (mpsqa.home and mpsHomeDir)
  3. open a terminal and call gradlew.bat

Usage

  1. create a model and import the language org.mpsqa.clones.config
  2. create a clones detection config root node and specify
  • how big should be the clones (minimmum number of sibling nodes, how deep can a node be)
  • desired scope (global scope with all models in the repo, project scope)
  • modules which should be considered (white list)
  • modules which should be ignored (black list)
  • models which should be ignored (black list)
  • names of concepts which should be ignored (black list)
  • strategy to postprocess the identified clones - e.g.
    • save them in a file as baseline
    • load previously found clones from file
    • filter newly found clones w.r.t. a previously saved baseline
  1. trigger the clones detection via Tools -> Detect Clones from Configuration
  • usually is very fast and takes ca. 30s for big scopes
  1. after the algorithm runs, a tool-view with the detected clones will open
  • double click on an entry to display details about a clone (which nodes it contains)
  • double click on a table entry from the detailed view to open the node in the editor
  • open "tree-map" view to gain an overview over the clones
    • black squares are models without clones, red squares are models with clones
    • right-click on a red node, and filter clones contained in that model
  1. if manual investigation of the clones reveals too many false positives, then you can refine the configuration (Step 2) and re-run the detection

IMAGE ALT TEXT HERE

Internal implementation

  • the core clones detection algorithm is contained in org.mpsqa.clones.core
    • computes hash-codes of sibling nodes and saves them in a hash-map
    • if the hash-map has multiple entries for a certain hash-code, then we identified cloned code (we detect Type 2 clones as defined in Comparison and evaluation of clone detection tools - pdf
    • our algorithm is inspired by Index-Based Code Clone Detection: Incremental, Distributed, Scalable - pdf
  • besides the core algorithm, we have the following components
    • UI consisting of tables and tree-map visualization
    • post-processors of detected clones (saving, loading, filtering)
    • definition of the clones detection configuration