cupido

Explore your codebase with graph view.

Goal & Motivation

It efficiently analyzes the entire commit history of a repository, combining information such as files, issue references, authors, etc., to generate a graph for the entire repository, in seconds.

All analyses can be performed flexibly and quickly on the graph.

graph TD
;
    File1("src/main.rs") === Commit1("abcd1234");
    File2("src/utils.rs") === Commit1;
    Commit1 === Issue1("#123");
    Issue1 === File1;
    Dir("src/") --- File1;
    Dir --- File2;
    Author1("Williamfzc") --- Commit1;
    File3("src/config.rs") === Commit2("efgh5678");
    File4("README.md") === Commit2;
    Commit2 === Issue2("#456");
    Issue2 === File4;
    Dir --- File3;
    Author2("JaneDoe") --- Commit2;

Loading

Concept

graph TD
;
    File === Commit;
    Commit === Issue;
    Issue === File;
    Dir --- File;
    Author --- Commit;

Loading

Conceptually, the entire graph consists of three core node types:

• File Node: logic unit.
• Commit Node: developer unit.
• Issue Node: story unit.

These nodes are interconnected and support bidirectional fast searching.

In addition to these, there are also some additional nodes to support more extensive retrieval and analysis:

• Author Node
• Dir Node
• ...

Usage

We primarily offer three usage modes:

• Rust library
• Local server mode (similar to LSP)
• CLI

Rust library

See examples/mini.rs

use cupido::collector::config::{get_collector, Collect, Config};

fn main() {
    let collector = get_collector();
    let mut conf = Config::default();
    conf.repo_path = String::from(".");
    let graph = collector.walk(Config::default());

    // 1. search from files to issues
    let file_name = String::from("src/server/app.rs");
    let issues = graph.file_related_issues(&file_name).unwrap();

    // src/server/app.rs related to ["#1"]
    println!("1. {} related to {:?}", file_name, issues);

    // 2. search from issues to commits
    let issue_label = issues.get(0).unwrap();
    let commits = graph.issue_related_commits(issue_label).unwrap();

    // #1 related to ["b7574411fbf685a777d1929bff26b3ad4ebd84f2"]
    println!("2. {} related to {:?}", issue_label, commits);

    // 3. search from commits to files
    let commit = commits.get(0).unwrap();
    let files = graph.commit_related_files(commit).unwrap();

    // b7574411fbf685a777d1929bff26b3ad4ebd84f2 related to ["src/server/mod.rs", "src/server/handler.rs", "src/server/config.rs", "src/server/app.rs", "src/server.rs", "src/main.rs"]
    println!("3. {} related to {:?}", commit, files);

    // Also, you can do it vice versa.
}

Local server mode

You can find the corresponding binary files for your system on the release page:

https://github.com/williamfzc/cupido/releases/

You can start the service using the following command:

./cupido up --repo-path ~/workspace/github/axios

Upon successful startup, you should see logs similar to the following:

2024-02-08T13:46:02.932406Z  INFO cupido: relation creating ...
2024-02-08T13:46:02.932754Z  INFO cupido: config: UpCommand { issue_regex: None, repo_path: Some("/Users/bytedance/workspace/github/axios"), path_specs: None, multi_parents: None }
2024-02-08T13:46:03.177632Z  INFO cupido: relation ready in 244.838094ms: GraphSize { file_size: 321, commit_size: 1136, issue_size: 753 }
2024-02-08T13:46:03.178575Z  INFO cupido: server up: http://127.0.0.1:9410

The service is exposed on port 9410, and you can access the HTTP API through it. You can use our client or other HTTP tools to interact with it.

➜ curl http://127.0.0.1:9410/size
{"file_size":10486,"commit_size":6983,"issue_size":1403}

You can find our client and API documentation here: node client

CLI

Please see cupido --help.

Performance

cupido can also work with bare repo. At the most time, the analysis should finish in seconds.

Repository	Time Taken	File Size	Commit Size	Issue Size
https://github.com/microsoft/pyright	8.046621521s	10486	6983	1403
https://github.com/axios/axios	244.838094ms	321	1136	753

Contribution

Issues and PRs are always welcome. :)

Currently, we are working on API v1.

License

Apache 2.0