checkpointpostgres==2.0.5

langgraph-checkpoint-postgres 2.0.5

Summary of Changes

• Added vector search capabilities using pgvector extension with PostgreSQL (#2535)
• Fixed proper handling of search results with dedicated search item processing (#2535, #2547)
• Improved code readability with modern Python syntax for context managers (#2547)

Detailed Changes

langgraph.store.postgres.base.PostgresIndexConfig

• Added new configuration class for vector search support with pgvector (#2535)
• Allows configuring vector dimensions, distance metrics (l2, inner_product, cosine), and vector type (vector, halfvec)

langgraph.store.postgres.base._row_to_search_item

• Added specialized function to handle search results with score information (#2535)
• Ensures search scores are properly returned and converted to float values

langgraph.store.postgres.base.BasePostgresStore

• Enhanced with vector search capabilities (#2535)
• Refactored query methods to support vector similarity search
• Added embedding support for document indexing and retrieval
• Added methods to handle vector-based filtering and scoring

langgraph.store.postgres.aio.AsyncPostgresStore

• Added vector search support with async embedding capabilities (#2535)
• Implemented methods to asynchronously embed documents for vector storage and retrieval

langgraph.checkpoint.postgres.PostgresSaver and langgraph.checkpoint.postgres.aio.AsyncPostgresSaver

• Improved context manager syntax for better readability and maintainability (#2547)