checkpoint==2.0.7

langgraph-checkpoint 2.0.7

Summary of Changes

• Added vector search capabilities to stores with semantic similarity search (#2547)
• Added embedding utilities for text-to-vector conversion with LangChain integration (#2547)
• Enhanced query filtering with comparison operators ($gt, $lt, etc.) (#2547)
• Improved InMemoryStore with optional NumPy acceleration for vector operations (#2547)
• Added support for accessing nested fields and arrays using path expressions (#2547)

Detailed Changes

langgraph.store.base.SearchItem

• Added new class that extends Item with an additional score field for ranked search results (#2547)

langgraph.store.base.SearchOp

• Added query parameter for natural language search capability (#2547)
• Enhanced filtering with support for comparison operators like $eq, $gt, $lt, etc. (#2547)

langgraph.store.base.PutOp

• Added index parameter to control how items are indexed for vector search (#2547)
• Options include using default indexing, disabling indexing with False, or specifying custom fields (#2547)

langgraph.store.base.IndexConfig

• Added new configuration class for vector search settings (#2547)
• Configurable embedding dimensions and embedding function (#2547)
• Optional field selection for what content to embed (#2547)

langgraph.store.base.NamespacePath

• Renamed from NameSpacePath for consistency (#2547)

langgraph.store.base.BaseStore

• Updated search and asearch methods to return SearchItem instances with similarity scores (#2547)
• Enhanced put and aput methods with indexing control parameter (#2547)
• Improved documentation with detailed examples (#2547)

langgraph.store.base.embed.ensure_embeddings

• Added utility to wrap arbitrary embedding functions into LangChain's Embeddings interface (#2547)
• Handles both synchronous and asynchronous embedding functions (#2547)

langgraph.store.base.embed.EmbeddingsFunc and AEmbeddingsFunc

• Added type definitions for synchronous and asynchronous embedding functions (#2547)

langgraph.store.base.embed.get_text_at_path and tokenize_path

• Added utilities to extract text from nested objects using path expressions (#2547)
• Support for nested fields, array indexing, wildcards, and multi-field selection (#2547)

langgraph.store.memory.InMemoryStore

• Complete rewrite with vector search capabilities (#2547)
• Added optional NumPy acceleration for vector operations (#2547)
• Improved filtering with support for comparison operators (#2547)
• Added efficient batch processing for embedding operations (#2547)