Class AlloyDBVectorStore (0.2.1)

AlloyDBVectorStore(
    key: object,
    engine: AlloyDBEngine,
    vs: AsyncAlloyDBVectorStore,
    stores_text: bool = True,
    is_embedding_query: bool = True,
)

Google AlloyDB Vector Store class

Properties

client

Get client.

Methods

AlloyDBVectorStore

AlloyDBVectorStore(
    key: object,
    engine: AlloyDBEngine,
    vs: AsyncAlloyDBVectorStore,
    stores_text: bool = True,
    is_embedding_query: bool = True,
)

AlloyDBVectorStore constructor.

Parameters
Name	Description
`key`	`object` Prevent direct constructor usage.
`engine`	`AlloyDBEngine` Connection pool engine for managing connections to Postgres database.
`vs`	`AsyncAlloyDBVectorStore` The async only Vector Store implementation
`stores_text`	`bool` Whether the table stores text. Defaults to "True".
`is_embedding_query`	`bool` Whether the table query can have embeddings. Defaults to "True".

Exceptions
Type	Description
`Exception`	If called directly by user.

aapply_vector_index

aapply_vector_index(
    index: llama_index_alloydb_pg.indexes.BaseIndex,
    name: typing.Optional[str] = None,
    concurrently: bool = False,
) -> None

Create an index on the vector store table.

aclear

aclear() -> None

Asynchronously delete all nodes from the table.

add

add(
    nodes: typing.Sequence[llama_index.core.schema.BaseNode], **add_kwargs: typing.Any
) -> list[str]

Synchronously add nodes to the table.

adelete

adelete(ref_doc_id: str, **delete_kwargs: typing.Any) -> None

Asynchronously delete nodes belonging to provided parent document from the table.

adelete_nodes

adelete_nodes(
    node_ids: typing.Optional[list[str]] = None,
    filters: typing.Optional[
        llama_index.core.vector_stores.types.MetadataFilters
    ] = None,
    **delete_kwargs: typing.Any
) -> None

Asynchronously delete a set of nodes from the table matching the provided nodes and filters.

adrop_vector_index

adrop_vector_index(index_name: typing.Optional[str] = None) -> None

Drop the vector index.

aget_nodes

aget_nodes(
    node_ids: typing.Optional[list[str]] = None,
    filters: typing.Optional[
        llama_index.core.vector_stores.types.MetadataFilters
    ] = None,
) -> list[llama_index.core.schema.BaseNode]

Asynchronously get nodes from the table matching the provided nodes and filters.

ais_valid_index

ais_valid_index(index_name: typing.Optional[str] = None) -> bool

Check if index exists in the table.

apply_vector_index

apply_vector_index(
    index: llama_index_alloydb_pg.indexes.BaseIndex,
    name: typing.Optional[str] = None,
    concurrently: bool = False,
) -> None

Create an index on the vector store table.

aquery

aquery(
    query: llama_index.core.vector_stores.types.VectorStoreQuery, **kwargs: typing.Any
) -> llama_index.core.vector_stores.types.VectorStoreQueryResult

Asynchronously query vector store.

areindex

areindex(index_name: typing.Optional[str] = None) -> None

Re-index the vector store table.

aset_maintenance_work_mem

aset_maintenance_work_mem(num_leaves: int, vector_size: int) -> None

Set database maintenance work memory (for ScaNN index creation).

async_add

async_add(
    nodes: typing.Sequence[llama_index.core.schema.BaseNode], **kwargs: typing.Any
) -> list[str]

Asynchronously add nodes to the table.

class_name

class_name() -> str

Get the class name, used as a unique ID in serialization.

This provides a key that makes serialization robust against actual class name changes.

clear

clear() -> None

Synchronously delete all nodes from the table.

create

create(
    engine: llama_index_alloydb_pg.engine.AlloyDBEngine,
    table_name: str,
    schema_name: str = "public",
    id_column: str = "node_id",
    text_column: str = "text",
    embedding_column: str = "embedding",
    metadata_json_column: str = "li_metadata",
    metadata_columns: list[str] = [],
    ref_doc_id_column: str = "ref_doc_id",
    node_column: str = "node_data",
    stores_text: bool = True,
    is_embedding_query: bool = True,
    distance_strategy: llama_index_alloydb_pg.indexes.DistanceStrategy = DistanceStrategy.COSINE_DISTANCE,
    index_query_options: typing.Optional[
        llama_index_alloydb_pg.indexes.QueryOptions
    ] = None,
) -> llama_index_alloydb_pg.vector_store.AlloyDBVectorStore

Create an AlloyDBVectorStore instance and validates the table schema.

Parameters
Name	Description
`engine`	`AlloyDBEngine` Alloy DB Engine for managing connections to AlloyDB database.
`table_name`	`str` Name of the existing table or the table to be created.
`schema_name`	`str` Name of the database schema. Defaults to "public".
`id_column`	`str` Column that represents if of a Node. Defaults to "node_id".
`text_column`	`str` Column that represent text content of a Node. Defaults to "text".
`embedding_column`	`str` Column for embedding vectors. The embedding is generated from the content of Node. Defaults to "embedding".
`metadata_json_column`	`str` Column to store metadata as JSON. Defaults to "li_metadata".
`metadata_columns`	`list[str]` Column(s) that represent extracted metadata keys in their own columns.
`ref_doc_id_column`	`str` Column that represents id of a node's parent document. Defaults to "ref_doc_id".
`node_column`	`str` Column that represents the whole JSON node. Defaults to "node_data".
`stores_text`	`bool` Whether the table stores text. Defaults to "True".
`is_embedding_query`	`bool` Whether the table query can have embeddings. Defaults to "True".
`distance_strategy`	`DistanceStrategy` Distance strategy to use for vector similarity search. Defaults to COSINE_DISTANCE.
`index_query_options`	`QueryOptions` Index query option.

Exceptions
Type	Description
`Exception`	If table does not exist or follow the provided structure.

create_sync

create_sync(
    engine: llama_index_alloydb_pg.engine.AlloyDBEngine,
    table_name: str,
    schema_name: str = "public",
    id_column: str = "node_id",
    text_column: str = "text",
    embedding_column: str = "embedding",
    metadata_json_column: str = "li_metadata",
    metadata_columns: list[str] = [],
    ref_doc_id_column: str = "ref_doc_id",
    node_column: str = "node_data",
    stores_text: bool = True,
    is_embedding_query: bool = True,
    distance_strategy: llama_index_alloydb_pg.indexes.DistanceStrategy = DistanceStrategy.COSINE_DISTANCE,
    index_query_options: typing.Optional[
        llama_index_alloydb_pg.indexes.QueryOptions
    ] = None,
) -> llama_index_alloydb_pg.vector_store.AlloyDBVectorStore

Create an AlloyDBVectorStore instance and validates the table schema.

Parameters
Name	Description
`engine`	`AlloyDBEngine` Alloy DB Engine for managing connections to AlloyDB database.
`table_name`	`str` Name of the existing table or the table to be created.
`schema_name`	`str` Name of the database schema. Defaults to "public".
`id_column`	`str` Column that represents if of a Node. Defaults to "node_id".
`text_column`	`str` Column that represent text content of a Node. Defaults to "text".
`embedding_column`	`str` Column for embedding vectors. The embedding is generated from the content of Node. Defaults to "embedding".
`metadata_json_column`	`str` Column to store metadata as JSON. Defaults to "li_metadata".
`metadata_columns`	`list[str]` Column(s) that represent extracted metadata keys in their own columns.
`ref_doc_id_column`	`str` Column that represents id of a node's parent document. Defaults to "ref_doc_id".
`node_column`	`str` Column that represents the whole JSON node. Defaults to "node_data".
`stores_text`	`bool` Whether the table stores text. Defaults to "True".
`is_embedding_query`	`bool` Whether the table query can have embeddings. Defaults to "True".
`distance_strategy`	`DistanceStrategy` Distance strategy to use for vector similarity search. Defaults to COSINE_DISTANCE.
`index_query_options`	`QueryOptions` Index query option.

Exceptions
Type	Description
`Exception`	If table does not exist or follow the provided structure.

delete

delete(ref_doc_id: str, **delete_kwargs: typing.Any) -> None

Synchronously delete nodes belonging to provided parent document from the table.

delete_nodes

delete_nodes(
    node_ids: typing.Optional[list[str]] = None,
    filters: typing.Optional[
        llama_index.core.vector_stores.types.MetadataFilters
    ] = None,
    **delete_kwargs: typing.Any
) -> None

Synchronously delete a set of nodes from the table matching the provided nodes and filters.

drop_vector_index

drop_vector_index(index_name: typing.Optional[str] = None) -> None

Drop the vector index.

get_nodes

get_nodes(
    node_ids: typing.Optional[list[str]] = None,
    filters: typing.Optional[
        llama_index.core.vector_stores.types.MetadataFilters
    ] = None,
) -> list[llama_index.core.schema.BaseNode]

Asynchronously get nodes from the table matching the provided nodes and filters.

is_valid_index

is_valid_index(index_name: typing.Optional[str] = None) -> bool

Check if index exists in the table.

model_post_init

model_post_init(context: Any, /) -> None

This function is meant to behave like a BaseModel method to initialise private attributes.

It takes context as an argument since that's what pydantic-core passes when calling it.

query

query(
    query: llama_index.core.vector_stores.types.VectorStoreQuery, **kwargs: typing.Any
) -> llama_index.core.vector_stores.types.VectorStoreQueryResult

Synchronously query vector store.

reindex

reindex(index_name: typing.Optional[str] = None) -> None

Re-index the vector store table.

set_maintenance_work_mem

set_maintenance_work_mem(num_leaves: int, vector_size: int) -> None

Set database maintenance work memory (for ScaNN index creation).