Fairspec Rust

Updated Oct 28, 2026

This guide will help you get started with Fairspec Rust. If you are new to the core framework's technologies, please take a look at the Fairspec standard and Polars DataFrames documentation.

Platforms

Fairspec Rust is a single fairspec binary — the terminal and the MCP server — and a library of crates. Prebuilt binaries are published for:

Linux x64 and arm64
macOS x64 and arm64
Windows x64

Building from source requires Rust 1.88+.

Installation

Install the binary for your platform from the GitHub releases, checking its checksum:

curl -fsSL https://fairspec.org/install.sh | sh -s -- terminal

Or build it from source:

cargo install --locked --git https://github.com/datisthq/fairspec fairspec

The crates are not on crates.io yet, so a project depends on the library through git:

cargo add --git https://github.com/datisthq/fairspec fairspec
cargo add serde_json

Examples

The examples are bodies of a function returning Result<(), Box<dyn std::error::Error>>; see the library overview for the conventions.

Loading a Dataset from Zenodo merging system Zenodo metadata into a user dataset:

use fairspec::load_dataset;

let dataset = load_dataset("https://zenodo.org/records/10053903")?;

println!("{dataset:?}");
// Some({
//   "id": "https://doi.org/10.5281/zenodo.10053903",
//   ...
// })

Validating an in-memory dataset descriptor:

use fairspec::fairspec_metadata::Descriptor;
use fairspec::{ValidateTableOptions, validate_dataset};
use serde_json::json;

let descriptor: Descriptor = serde_json::from_value(json!({"resources": "bad"}))?;
let report = validate_dataset(descriptor, &ValidateTableOptions::default())?;

println!("{}", report.valid);
// false
println!("{}", serde_json::to_string_pretty(&report.errors)?);
// [
//   {
//     "type": "metadata",
//     "message": "\"bad\" is not of type \"array\"",
//     "jsonPointer": "/resources"
//   }
// ]

Loading a dataset from a remote descriptor, saving it locally as a zip archive, and then using it as a local dataset:

use fairspec::fairspec_dataset::{get_temp_file_path, load_dataset_from_zip, save_dataset_to_zip};
use fairspec::fairspec_metadata::Dataset;
use fairspec::load_dataset;

let archive_path = get_temp_file_path(false, None, Some("zip"))?;
let descriptor = load_dataset(
    "https://raw.githubusercontent.com/fairspec/fairspec-typescript/refs/heads/main/metadata/actions/dataset/fixtures/dataset.json",
)?
.ok_or("unsupported source")?;
let source_dataset: Dataset = serde_json::from_value(descriptor.into())?;

save_dataset_to_zip(&source_dataset, &archive_path, false)?;
let target_dataset = load_dataset_from_zip(&archive_path)?;
println!("{target_dataset:?}");

Reading a CSV table:

use fairspec::fairspec_table::LoadTableOptions;
use fairspec::load_table;
use serde_json::json;

let resource = serde_json::from_value(json!({"data": "data.csv"}))?;
let table = load_table(&resource, &LoadTableOptions::default())?;

// Load with custom format
let resource = serde_json::from_value(json!({
    "data": "data.csv",
    "fileDialect": {"format": "csv", "delimiter": ";", "headerRows": [1]},
}))?;
let table = load_table(&resource, &LoadTableOptions::default())?;

Reference

Note that the fairspec crate re-exports fairspec-library, which re-exports the fairspec-metadata, fairspec-dataset and fairspec-table crates as modules.