flash1293/kibana_analyze

Kibana Source Code Analysis

This is a small weekend project analyzing the source code of kibana on a file/LOC basis and visualizing the results in kibana.

Run it

• Check out this repo: git clone [email protected]:flash1293/kibana_analyze.git
• Check out kibana besides this repo: [email protected]:elastic/kibana.git
• Start a local elasticsearch / kibana instance (you can use docker-compose up)
• Install dependencies (yarn install)
• Collect data node index.js (this may take a while)
• Import the dashboards and visualizations from dashboards.json
• Look at the data in your kibana instance

Current State (Latest Tag)

History

React Implementations of Directives

Data

The script checks out the last 80 tags of the repo and collects for each tag the following values for each file:

• Lines of code (separated by comment, code, mixed, ...) as provided by the sloc lib
• Filename
• Path of directories (dir0, dir1, ... dirN)
• File extension
• Whether it is a unit test file
• Whether it references the AngularJs ui module
• Whether it defines an angular directive, controller or service
• Whether it uses ngReact
• Git tag

Running index.js prunes the index and collects everything again.

Improvement TODOs

• Incremental index build
• Collect for each commit (or each n commits of master)
• Separate collection for code changes instead of files at a given time (the current data doesn't reflect whether existing LOCs has been changed)
• More robust React/Angular detection
• More robust unit test detection
• Scripted fields for ratio code metrics like todos per total LOC
• Fix alphabetical tag ordering bug (maybe just skip alpha and beta versions?)
• Index since the beginning of time
• Collect additional metrics (# of classes in TS/JS, style rules in CSS, ...)
• Collect amount of services/directives/controller for each file
• Merge with information about code coverage / linting errors
• Merge with information about build sizes