anda_db

anda_db is the embedded database core of the AndaDB workspace. It gives Rust applications schema-aware collections, object-store-backed persistence, and three built-in retrieval modes: B-Tree for exact and range filters, BM25 for full-text search, and HNSW for vector similarity search.

What This Crate Provides

• Database lifecycle management through AndaDB
• Schema-aware collections and document CRUD
• B-Tree, BM25, and HNSW index integration
• Hybrid search with reciprocal-rank fusion reranking
• Object-store-backed persistence with checkpoints, flushing, and recovery
• Lightweight metadata extensions at database and collection scope

When to Use It

Use anda_db when you want:

• an embedded database inside a Rust service
• durable AI memory without operating an external database service
• structured, lexical, and semantic retrieval in one collection model
• local development on filesystem or in-memory storage, with optional cloud object storage in production

Getting Started

Add the crate to your project:

[dependencies]
anda_db = { version = "0.7", features = ["full"] }
object_store = { version = "0.13", features = ["fs"] }
tokio = { version = "1", features = ["full"] }
serde = { version = "1", features = ["derive"] }

For a complete runnable example, see:

• examples/db_demo.rs
• cargo run -p anda_db --example db_demo --features full

For a product-level introduction, see the workspace root README.

Technical Reference

Deep technical documentation for this crate lives in:

• docs/anda_db.md
• docs/anda_db_schema.md
• docs/anda_db_btree.md
• docs/anda_db_tfs.md
• docs/anda_db_hnsw.md
• docs/anda_object_store.md

Related Crates

• anda_db_schema for field types, schemas, and documents
• anda_db_derive for AndaDBSchema and FieldTyped
• anda_db_btree for exact and range indexing
• anda_db_tfs for BM25 full-text search
• anda_db_hnsw for vector search
• anda_object_store for metadata and encryption wrappers over object_store

License

MIT. See LICENSE.