Vector Indexes

[vgvoleg]

No description provided.

[Copilot]

Pull Request Overview

This PR introduces support for creating and using a vector index in YDB vector stores, along with end-to-end tests for various search strategies and updated Docker configuration.

• Add index_enabled, index_name, and index configuration parameters in YDBSettings and implement index creation/renaming logic.
• Extend add_texts and search query preparation to leverage the vector index when enabled.
• Add tests covering basic vector-indexed searches across different strategies and update Docker Compose for vector index feature flags.

Reviewed Changes

Copilot reviewed 3 out of 5 changed files in this pull request and generated no comments.

File	Description
tests/test_vectorestore.py	Added tests for indexed searches and a stray drop() call.
langchain_ydb/vectorstores.py	Introduce index config in settings, query builders, and index maintenance.
docker/docker-compose.yml	Bumped YDB image to trunk and enabled vector index feature flag.

Comments suppressed due to low confidence (3)

langchain_ydb/vectorstores.py:92

• The class docstring still says the default table is 'langchain_store'. Update it to match the new default 'ydb_langchain_store'.

    table: str = "ydb_langchain_store"

langchain_ydb/vectorstores.py:73

• The docstring default for index_name ('vectorsearch_index') does not match the code default ('ydb_vector_index'). Please update the docstring.

        index_name (str) : Name of vector index. Default is 'vectorsearch_index'.

tests/test_vectorestore.py:425

• This docsearch.drop() call is at the module level (outside any test) and will raise a NameError. Move it inside the appropriate test or remove it.

    docsearch.drop()