Crate invars 

Invars

Invars is a declarative data validation engine for Rust.

It allows you to define invariants (validation rules) and evaluate them against a dataset using an execution engine (currently Polars).

---

Core Concepts

• Invariant: A validation rule (e.g. “column must be unique”).
• Scope: Where the rule applies (Dataset or Column).
• Engine: Executes invariants against data.
• Violation: Returned when an invariant fails.

Execution model:

Invariant
   ↓
plan() -> Metric Expression
   ↓
Engine execution (Polars LazyFrame)
   ↓
map() -> Option<Violation>

---

Quick Example (Polars)

#[cfg(feature = “polars”)]

fn main() -> Result<(), Box> {

use invars::prelude::; use polars::prelude::;

let df = df! { “age” => [18, 25, 130], }?;

let invariant = Invariant::new( “age_between”.to_string().try_into()?, PolarsKind::ValueBetween, Scope::Column { name: “age”.into() }, ) .with_param(“min”, “0”) .with_param(“max”, “120”);

let engine = EnginePolarsDataFrame::default(); let violations = engine.evaluate(&df, vec![invariant])?;

Ok(())

}

#[cfg(not(feature = “polars”))]

fn main() {}

---

Feature Flags

• polars — Enables the Polars execution engine.

Design Goals

• Declarative validation rules
• Engine-agnostic core domain
• Lazy execution support
• Explicit metric-to-violation mapping
• Fully testable invariant units

Invars is designed to be predictable, extensible, and AI-friendly.

Modules

engine

error

infrastructure

interface

invariant

prelude

report

scope

severity

spec

use_cases

violation