MoonCardinalityKit

MoonCardinalityKit is a MoonBit foundation library for approximate distinct-count analysis over streams.

It is useful when a project needs to estimate UV, active device count, distinct trace IDs, unique error signatures, or unique keys in data-quality checks without storing every raw value forever.

What It Solves

• exact distinct counting for small sets
• approximate cardinality estimation for larger streams
• mergeable sketches for sharded logs or distributed jobs
• stable basis-point error reporting
• sketch diagnostics for saturation, duplicate ratio, and rebucket signals
• JSON summaries for CLI, dashboard, and regression snapshots

Why It Is Not A Duplicate

The project is not a hash library, not a bitmap index, not a database, and not a charting tool. It focuses on a reusable analytics primitive: compact distinct-count sketches that can sit under monitoring systems, log analyzers, event pipelines, and data-quality tools.

Core API

• SketchConfig::new(buckets, exact_limit)
• CardinalitySketch::new(config)
• add(sketch, value)
• add_many(sketch, values)
• estimate(sketch)
• merge(left, right)
• evaluate(values, sketch)
• sketch.diagnostics()
• Estimate::to_json()
• AccuracyReport::to_json()

Quick Start

let values = demo_values()
let sketch = CardinalitySketch::new(default_config()).add_many(values)
let estimate = sketch.estimate()
println(estimate.to_json())

Run:

moon check --target all
moon test --target wasm
moon test --target wasm-gc
moon run cmd/main

Boundary

MoonCardinalityKit does not collect logs, store events, render dashboards, or guarantee cryptographic hashing. It provides deterministic MoonBit data structures and algorithms for distinct-count estimation.

License

Apache-2.0.