Lexhint
Lexhint is a local lexical-evidence engine backed by self-describing SQLite artifacts.
It provides lexical membership, frequency enrichment, compact-string segmentation, stable semantic-domain evidence, and optional rich dictionary entries. It does not speak or normalize text.
- Architecture Documentation
- Introduction and Goals
- Architecture Constraints
- Context and Scope
- Solution Strategy
- Building Block View
- Runtime View
- Deployment View
- Cross-cutting Concepts
- Architecture Decisions
- Quality Requirements
- Risks and Technical Debt
- Glossary
- Changelog
- [v0.4.8] - 2026-09-09
- [v0.4.7] - 2026-09-09
- [v0.4.6] - 2026-09-09
- [v0.4.5] - 2026-09-08
- [v0.4.4] - 2026-09-03
- [v0.4.3] - 2026-09-02
- [v0.4.2] - 2026-08-31
- [v0.4.1] - 2026-08-31
- [v0.4.0] - 2026-08-25
- [0.3.0] - 2026-08-24
- [0.2.1] - 2026-08-23
- [0.2.0] - 2026-08-23
- [0.1.3] - 2026-08-22
- [0.1.2] - 2026-08-21
- [0.1.1] - 2026-08-21
- [0.1.0] - 2026-08-20
See the README for installation, build commands, API examples, and data-source boundaries.
Dataset lifecycle
Install published SQLite datasets explicitly with lexhint dataset download LANGUAGE. The client reads the lexhint-datasets catalog, caches its validated bytes under LEXHINT_CACHE_DIR (or the platform cache directory), and downloads selected assets from immutable GitHub Releases. Online catalog reads use conditional refresh metadata and fall back to a valid cached catalog on transport failure. --offline dataset available and --offline dataset check read the cache without network access. The catalog is an index only; detailed provenance remains in each release’s datasets-v2.json. Exact schema equality is required, and historical releases remain reachable through compatible catalog entries or the compatibility Releases API fallback. The dataset manager stores each immutable artifact by lexical target language, Wiktionary source variant, capability variant, exact schema version, and dataset version side by side, verifies gzip and SQLite metadata before atomic installation, and never downloads from Lexicon or ordinary query commands. Use dataset list, dataset info, dataset validate, and dataset remove for local management; dataset available lists all compatible catalog artifacts, dataset check reports updates for installed source and capability slots, and dataset update updates all installed slots or selected language, source-variant, and capability-variant filters, removing superseded versions after successful installation.