swift-code-metrics

Code metrics analyzer for Swift projects.

Introduction

The goal of this software is to provide an insight of the architectural state of a software written in Swift that consists in several modules. Inspired by the book of Robert C. Martin, Clean Architecture, the software will scan the project to identify the different components in order to assess several common code metrics in the software industry:

• the overall number of concrete classes and interfaces
• the instability and abstractness of the framework
• the distance from the main sequence
• LOC (Lines Of Code)
• NOC (Numbers Of Comments)
• POC (Percentage Of Comments)
• NOM (Number of Methods)
• Number of concretes (Number of classes and structs)
• NOT (Number Of Tests)
• NOI (Number Of Imports)
• Frameworks dependency graph (number of internal and external dependencies)

Requirements

This is a Python 3 script that depends on matplotlib, adjustText, pyfunctional and pygraphviz.

This latest package depends on the Graphviz binary that must be installed before. If you're in a Mac environment, you can install it directly with brew install graphviz.

Usage

The package is available on pip with pip3 install swift-code-metrics.

The syntax is:

swift-code-metrics --source <path-to-swift-project> --artifacts <output-directory> --exclude <excluded-folders> --tests-paths <test-paths> --generate-graphs

• --source is the path to the folder that contains the main Xcode project or Workspace
• --artifacts path to the folder that will contain the generated output.json report
• --exclude (optional) space separated list of path substrings to exclude from analysis (e.g. Tests will ignore all files/folders that contain Tests)
• --tests-paths (default: Test Tests) space separated list of path substrings matching test classes
• --generate-graphs (optional) if passed, it will generate the graphs related to the analysis and save them in the artifacts folder

Development

Please run ./install.sh and ./build_and_test.sh to install dependencies and run the tests.

The repo comes with a predefined setup for VS Code to debug and run tests as well.

Documentation

Please follow the guide with a practical example to get started.

Current limitations

• This tool is designed for medium/large codebases composed by different frameworks. The script will scan the directory and it will identify the frameworks by the name of the 'root' folder, so it's strictly dependent on the file hierarchy (unless a project path override file is specified)

• Libraries built with spm are not supported.

• The framework name is inferred using the directory structure. If the file is in the root dir, the default_framework_name will be used. No inspection of the xcodeproj will be made.

• The list of methods currently doesn't support computed vars

• Inline comments in code (such as struct Data: {} //dummy data) are currently not supported

• Only XCTest test frameworks are currently supported

TODOs

• Code improvements
• Other (open to suggestions)

Contact

Mattia Campolese