A new Rust-based tool called jsongrep claims to be faster than established JSON query tools like jq, jmespath, jsonpath-rust, and jql. The tool uses a deterministic finite automaton (DFA) approach to search JSON documents, compiling queries into a state machine that processes documents in a single pass. This contrasts with traditional tools that interpret path expressions recursively.

Inspired by Andrew Gallant's ripgrep, jsongrep treats JSON documents as trees and queries as regular expressions over keys and indices. The query language supports field selection, wildcards, alternation, and recursive descent, but lacks jq's transformation capabilities. The tool is available via crates.io and binaries are provided for cross-platform use.

The performance advantage comes from compiling queries into DFAs before processing JSON, enabling O(1) work per input symbol with no backtracking. Benchmarks show jsongrep significantly outperforms competitors on large datasets, with end-to-end search times that demonstrate the practical benefits of the DFA approach. While less feature-rich than jq, jsongrep excels at its core purpose: fast, single-pass JSON value searching.