4 releases

0.1.3	Mar 4, 2026
0.1.2	Mar 2, 2026
0.1.1	Feb 23, 2026
0.1.0	Feb 23, 2026

#2 in #ast-analysis

MIT license

320KB
7.5K SLoC

codedash

Code metrics visualization CLI — parse, enrich, and map code structure to visual perceptions.

codedash analyzes source code via tree-sitter AST parsing, enriches it with git history, and transforms structural metrics (lines, complexity, churn, depth) into visual parameters (color, size, opacity) through a declarative Lua DSL pipeline.

Install

cargo install codedash

From source

git clone https://github.com/ynishi/codedash.git
cd codedash
cargo install --path .

Quick Start

# Analyze current directory (Rust)
codedash analyze

# Analyze TypeScript project
codedash analyze src/ -l typescript

# JSON output
codedash analyze -o json

# Top 20 entries
codedash analyze -t 20

# Filter by domain
codedash analyze -d core

# Generate interactive HTML dependency map
codedash view

# With coverage data
cargo llvm-cov --json --output-path coverage.json
codedash analyze --cov-file coverage.json -o json
codedash view --cov-file coverage.json

# Show dependency graph (DOT format)
codedash graph

# Diagnose setup
codedash check-health

Commands

Command	Description
`analyze`	Parse + enrich + evaluate metrics with visual mapping
`parse`	Parse source files (AST only, no enrichment)
`graph`	Show dependency graph (DOT or Mermaid format)
`view`	Generate interactive HTML dependency map
`config-init`	Generate `.codedash.lua` template (auto-detects Cargo workspace)
`list-parsers`	List supported languages and file extensions
`check-health`	Diagnose parser, git, and config status

Configuration

Place .codedash.lua in your project root:

return {
  extends = "recommended",
  domains = {
    { name = "core", patterns = { "router", "route" } },
    { name = "util", patterns = { "utils", "helpers" } },
  },
  exclude = { "index", "vite-env.d" },
}

Generate a template automatically:

codedash config-init   # auto-detects Cargo workspace members

Patterns use substring matching against node name, file path, and short name.

Architecture

Source code
  → tree-sitter AST parsing (Rust)
    → Git enrichment (churn, co-change)
    → Coverage enrichment (--cov-file, llvm-cov JSON)
      → Lua DSL evaluation
        → Visual output (report / JSON / HTML)

Pipeline

Index (what to measure)  →  Normalizer [0,1]  →  Percept (how to perceive)
     ^                                                  |
     |                                                  v
   Node data                                      Visual output

Concept	Description
Node	Immutable AST-derived data record
Index	Extracts a numeric metric from a Node
Percept	Maps normalized [0,1] to a visual range
Binding	Pairs an Index with a Percept
Normalizer	Converts raw values to [0,1] distribution

Built-in Bindings (recommended preset)

Index	Percept	Meaning
churn	hue	High churn → red
lines	size	Many lines → large
params	border	Many params → thick border
depth	opacity	Deep nesting → faded
coverage	clarity	Low coverage → dim

HTML View Encoding (`codedash view`)

Visual Channel	Data Source	Mapping
Circle size	Lines of code	More lines → larger
Circle color	Domain	Auto-assigned palette
Inner ring color	Cyclomatic complexity	Green (low) → Red (high)
Inner ring width	Max cyclomatic	Higher → thicker
Outer ring (amber)	Git churn (30d)	More changes → wider
Badge (shield)	Coverage	Green ≥70%, Yellow ≥40%, Red <40%
Badge (gray ?)	Coverage N/A	No coverage data for this module
Edges	Import dependencies	Arrows between modules

Customizing the DSL

Custom Index

local index = require("codedash.model.index")

local lines = index "lines" { source = "lines" }

local risk = index "risk" {
  compute = function(node) return node.lines * node.cyclomatic end
}

Custom Percept

local percept = require("codedash.model.percept")

local hue  = percept "hue"  { range = { 240, 0 } }       -- blue → red
local size = percept "size" { range = { 0.2, 5.0 } }     -- small → large

Normalizer Override

-- At Binding level
bind { idx.churn, pct.hue, normalize = "rank" }

Normalizer	Strategy	Best for
`percentile`	p10/p90 range, linear interpolation	Normally distributed data
`rank`	Rank-based, average ties	Skewed distributions

CI Integration

codedash integrates with GitHub Actions to automate metrics tracking:

Metrics Dashboard — Deploy an interactive module map, coverage report, and benchmark results to GitHub Pages on every push to main
PR Metrics Comment — Post a summary table (coverage, complexity hotspots, domain breakdown) as a sticky comment on each pull request

See docs/ci-setup.md for step-by-step setup instructions and ready-to-use workflow templates.

Supported Languages

Language	Extensions
Rust	`.rs`
TypeScript	`.ts`, `.tsx`

Requirements

Rust 1.80+

License

MIT

Dependencies

~44MB
~1M SLoC