quarb

Quarb is a query language for arbors — tree-spanned graphs: graphs with a primary hierarchical backbone enriched by non-hierarchical ("crosslink") relations. File systems with symlinks, HTML with anchor references, JSON/XML with $ref-style links, and abstract syntax trees with name-resolution edges are all arbors.

This crate is the engine: it lexes and parses a query, then evaluates it against an adapter that maps a data source onto the arbor model. Quarb generalizes and unifies XPath and jq, adding first-class graph navigation, regex-style quantifiers over path elements (not just characters), pipelines, register-based breadcrumbs, and subcontexts.

Example

Implement the AstAdapter trait for your own data, or use a ready-made adapter such as quarb-json, quarb-fs, or one of the many others:

use quarb::{run, QueryResult};

// `adapter` maps some data source onto the arbor model.
let result = run("//user[::age > 30]::name", &adapter).unwrap();
match result {
    QueryResult::Nodes(nodes) => { /* matched nodes */ }
    QueryResult::Values(values) => { /* projected scalars */ }
}

The adapter ecosystem

One query language runs over many sources through separate adapter crates: JSON, YAML, TOML, CSV, XML, HTML, Markdown, spreadsheets, archives, filesystems, git, five SQL engines, Neo4j, cloud document stores, mailboxes, source-code ASTs, and more. The qua CLI ties them together. See quarb.org for the full list.

Documentation

• The Quarb Language Specification and User Guide — the authoritative reference and a tutorial introduction.
• Per-tool cookbooks mapping XPath, jq, SQL, Cypher, pandas, and CSS-selector idioms to Quarb.

License

Dual-licensed under either of MIT or Apache-2.0, at your option.