Indexes API

C-SPANN index configuration.

langchain_cockroachdb.indexes.CSPANNIndex

CockroachDB C-SPANN vector index configuration.

C-SPANN (CockroachDB SPANN) is CockroachDB's vector indexing algorithm for real-time indexing of billions of vectors in distributed systems. Based on Microsoft's SPANN (Space Partition with Approximate Nearest Neighbor).

__init__(distance_strategy=DistanceStrategy.COSINE, min_partition_size=None, max_partition_size=None, name=None)

Initialize C-SPANN index configuration.

Parameters:

Name	Type	Description	Default
distance_strategy	DistanceStrategy	Distance metric to use	COSINE
min_partition_size	int | None	Minimum partition size for C-SPANN	None
max_partition_size	int | None	Maximum partition size for C-SPANN	None
name	str | None	Optional custom index name	None

get_create_index_sql(table_name, column_name, *, schema='public', prefix_columns=None)

Generate CREATE VECTOR INDEX SQL.

Parameters:

Name	Type	Description	Default
table_name	str	Target table name	required
column_name	str	Vector column name	required
schema	str	Database schema	'public'
prefix_columns	list[str] | None	Optional prefix columns for multi-tenant indexes	None

Returns:

Type	Description
str	SQL statement to create the index

get_drop_index_sql(table_name, column_name, *, schema='public')

Generate DROP INDEX SQL.

Parameters:

Name	Type	Description	Default
table_name	str	Target table name	required
column_name	str	Vector column name	required
schema	str	Database schema	'public'

Returns:

Type	Description
str	SQL statement to drop the index

langchain_cockroachdb.indexes.DistanceStrategy

Bases: str, Enum

Distance strategies for vector similarity.

get_opclass()

Get index operator class.

get_operator()

Get SQL operator for distance calculation.

Distance Strategies

Strategy	Operator	Best For
COSINE	<=>	Normalized vectors (most embeddings)
EUCLIDEAN	<->	Spatial data, L2 distance
INNER_PRODUCT	<#>	Pre-normalized vectors

Examples

See Vector Indexes Guide for complete examples.