Understanding Your Codebase Using ctags

Luke Jaggernauth - August 9th, 2023

In the past few weeks, we have greatly improved the performance of Sweep and I would like to share a significant piece of our journey: getting to know your codebase using universal-ctags.

Background

In the realm of coding, understanding the structure and semantics of a codebase is key. It’s vital for pinpointing definitions, implementations, and usages of functions, classes, and variables. Traditionally, this would require manual navigation and significant familiarity with the codebase - a time-consuming process. We wanted to automate this and make it more efficient.

File Tree

Previously, Sweep could only see the file tree structure of the repository as well as some snippets from relevant scripts. The file tree looks similar to this:

[README.md](http://readme.md/)
chatRoom/
|- [README.md](http://readme.md/)
|- demo_connect.gif
|- demo_send_msg.gif
|- include/
|   |- Practical.cpp

However, we found that this did not provide enough adequate information about the codebase. So, we decided to include ctags as context.

New File Tree

Universal-ctags allow us to look through the repository with a better understanding of the code, like the location of methods, variables, and classes. With this context, Sweep can now provide intelligent context-aware assistance. When you query about a particular function, Sweep can quickly identify it and provide relevant information about its definition and usage, enabling more accurate responses. The file tree has been updated to include the ctags:

.gitignore
jp-app/
|- App.xaml
|- App.xaml.cs
|   |- namespace jp_app
|   |- class App
|   |- method App (ISystemLanguageService systemLanguageService)
|   |- method OnStart ()

Conclusion

We wanted to give a thank you to Paul Gauthier! His blog (opens in a new tab) was helpful in learning about ctags, and we're excited to further build on his approach! We’re incredibly excited about how universal-ctags is boosting Sweep’s understanding of your codebase and can’t wait to share more updates! For anyone eager to learn more, check out our GitHub repo: https://github.com/sweepai/sweep (opens in a new tab) 🌟