@taladb/react-native

Provides native document persistence in React Native via JSI (JavaScript Interface) HostObject bindings that expose a native database layer without requiring network calls. Documents are stored locally on the device with structured schema support, enabling offline-first applications to maintain full CRUD operations on document collections without cloud synchronization overhead.

Stores vector embeddings alongside documents and provides semantic similarity search via vector distance calculations (likely cosine or Euclidean metrics). The system indexes embeddings for efficient retrieval, enabling RAG (Retrieval-Augmented Generation) patterns where documents are ranked by semantic relevance rather than keyword matching.

Encrypts documents stored on the device using device-level encryption keys, protecting data if the device is lost or stolen. Encryption is transparent to the application — documents are encrypted on write and decrypted on read without explicit key management in JavaScript code.

Enforces document structure through schema definitions that validate incoming documents before storage, providing type safety and preventing malformed data from corrupting the database. Schemas define required fields, data types, and constraints that are checked at write time, with validation errors returned to the application layer.

Exposes synchronous create, read, update, and delete operations on documents through JSI HostObject methods, allowing React Native code to perform database operations without async/await overhead. Operations return results immediately from the native layer, enabling responsive UI updates without promise chains or callback hell.

Stores all data locally on the device with no required network connectivity, supporting eventual consistency patterns where local changes are persisted immediately and synchronized to remote systems when connectivity is available. The database tracks local modifications independently of sync state, enabling applications to function fully offline.

Supports querying documents using filter predicates (equality, comparison, range, logical operators) to retrieve subsets of the document collection matching specified conditions. Queries execute in native code via JSI, returning filtered result sets without loading the entire collection into memory.

Automatically or manually creates indexes on frequently-queried document fields to accelerate retrieval operations. Indexes are maintained in native code and used transparently during query execution to reduce search time from O(n) to O(log n) or better, depending on index type and query selectivity.

Supports inserting, updating, or deleting multiple documents in a single operation, reducing JSI bridge overhead compared to individual CRUD calls. Batch operations execute atomically in native code, ensuring consistency across multiple document modifications without interleaving from concurrent operations.

Tracks modifications to documents (create, update, delete) and emits events that React Native code can subscribe to, enabling reactive UI updates when data changes. Change events include document deltas or full snapshots, allowing listeners to respond to specific modifications without polling.

Provides transaction semantics for multi-step operations, ensuring that either all changes commit atomically or all roll back if any step fails. Transactions prevent partial updates and maintain database consistency even if the application crashes mid-transaction, using write-ahead logging or similar durability mechanisms.