Skip to content

CLAUDE.md

Latest commit

 

History

History
71 lines (49 loc) · 2.58 KB

CLAUDE.md

File metadata and controls

71 lines (49 loc) · 2.58 KB

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Build Commands

# Build
cargo build

# Run tests (tests must run serially due to shared global state)
cargo test -- --test-threads=1

# Run a specific test
cargo test <test_name> -- --test-threads=1

# Run clippy
cargo clippy

# Install locally
cargo install --path .

Project Overview

rust-diagnostics is a CLI utility that embeds Clippy diagnostic messages as inline comments in Rust source code. It also analyzes git commit history to track how warnings are manually fixed between revisions.

Architecture

The project is a single-binary CLI tool with two main modules:

  • src/main.rs: Core logic including:

    • CLI argument parsing via structopt (Args struct at line 19)
    • Clippy invocation and diagnostic message parsing via cargo_metadata
    • Warning markup insertion into source files (markup function at line 84)
    • Git diff analysis using git2-rs to find patches that fix warnings
    • Function-level context extraction using tree-sitter queries (splitup function at line 137)
    • Hunk processing and output formatting (lines 754-802)

  • src/language.rs: Tree-sitter language binding for Rust parsing

Key Data Structures

  • Warning: Represents a clippy diagnostic with byte offsets, line numbers, and suggestion text
  • Hunk: Represents a git diff hunk with old/new text, surrounding function context, and associated warnings
  • LineRange: Used to index function items by their source location

Output

The tool creates a diagnostics/ folder containing:

  • Marked-up source files with embedded warning comments
  • diagnostics.json: Cached clippy output
  • diagnostics.log: Summary and patch analysis results

CLI Options

Key flags for the --patch workflow:

  • --patch <commit_id>: Compare current revision against specified commit
  • --confirm: Only show hunks that actually fix warnings
  • --pair: Output before/after code separated by === 19a3477889393ea2cdd0edcb5e6ab30c ===
  • --function / -W: Include surrounding function context
  • --location: Insert warning markers in the function context
  • --mixed: Combine context view with patch format
  • --single: Only show hunks that fix exactly one warning
  • --fix: Use machine-applicable clippy rules only

Testing Notes

Tests use serial_test crate because they share global state (ARGS static mutex). Each test creates a temporary directory (e.g., tmp_<uuid>) and performs git operations to set up test scenarios.

Snapshot tests use insta crate for comparing expected outputs.