rsonpath-lib – SIMD-powered JSONPath, as a library 🚀

Library for rsonpath, the JSONPath engine for querying massive streamed datasets.

The main target of this crate is the rsonpath CLI tool. Note that this API is unstable until we reach v1.0.0. This is going to happen (we have a roadmap), but our dev resources are quite limited. Contributions are welcome and appreciated.

Unsafe

The library uses unsafe for SIMD operations, because it has to, at least until portable-simd gets stabilized. Because of this, a compiled library is not portable – if you build on a platform supporting AVX2 and then use the same compiled code on an ARM platform, it will crash. We put special care to not use unsafe code anywhere else – in fact, the crate uses #[forbid(unsafe_code)] when compiled without the default simd feature.

Build & test

The dev workflow utilizes just. Use the included Justfile. It will automatically install Rust for you using the rustup tool if it detects there is no Cargo in your environment.

just build
just test

Architecture diagram

Below is a simplified overview of the module interactions and interfaces, and how data flows from the user's input (query, document) through the pipeline to produce results.

Optional features

The simd feature is enabled by default and is recommended to make use of the performance benefits of the project.

The serde feature is optional and enables the serde dependency, which allows serializing and deserializing the engine after compilation. Note: the binary format of the engine is expected to evolve and so changing it is considered a minor update for semver purposes.

Dependencies

Showing direct dependencies.

cargo tree --package rsonpath-lib --edges normal --depth 1 --target=all --all-features

rsonpath-lib v0.9.4 (/home/mat/src/rsonpath/crates/rsonpath-lib)
├── cfg-if v1.0.0
├── log v0.4.22
├── memmap2 v0.9.5
├── rsonpath-syntax v0.4.0 (/home/mat/src/rsonpath/crates/rsonpath-syntax)
├── serde v1.0.217
├── smallvec v1.13.2
├── static_assertions v1.1.0
├── thiserror v2.0.9
└── vector-map v1.0.1

Justification

• cfg-if – used to support SIMD and no-SIMD versions.
• log – Rust standard logging idiom.
• memmap2 – for fast reading of source files via a memory map instead of buffered copies.
• serde – optional dependency for serialization and deserialization of compiled engines.
• smallvec – crucial for small-stack performance.
• static_assertions – additional reliability by some constant assumptions validated at compile time.
• thiserror – idiomatic Error implementations.
• vector_map – used in the query compiler for measurably better performance.