Vector

Milvus implementation of vector similarity search capability.

This module provides a Milvus implementation of the VectorCapability protocol for vector-based semantic similarity search operations.

MilvusVectorCapability(collection_name, client, em_invoker, dimension, distance_metric='L2', vector_field='dense_vector', id_max_length=100, content_max_length=65535)

Milvus implementation of VectorCapability protocol.

This class provides vector similarity search operations using Milvus.

Attributes:

Name	Type	Description
collection_name	str	The name of the Milvus collection.
client	AsyncMilvusClient	Async Milvus client instance.
em_invoker	BaseEMInvoker	Embedding model invoker for vectorization.
dimension	int	Vector dimension.
distance_metric	str	Distance metric ("L2", "IP", "COSINE").
vector_field	str	Field name for dense vectors.
id_max_length	int	Maximum length for ID field.
content_max_length	int	Maximum length for content field.

Initialize the Milvus vector capability.

Parameters:

Name	Type	Description	Default
collection_name	str	The name of the Milvus collection.	required
client	AsyncMilvusClient	The async Milvus client instance.	required
em_invoker	BaseEMInvoker	The embedding model invoker.	required
dimension	int	Vector dimension.	required
distance_metric	str	Distance metric. Defaults to "L2". Supported: "L2", "IP", "COSINE".	'L2'
vector_field	str	Field name for dense vectors. Defaults to "dense_vector".	'dense_vector'
id_max_length	int	Maximum length for ID field. Defaults to 100.	100
content_max_length	int	Maximum length for content field. Defaults to 65535.	65535

em_invoker property

Returns the EM Invoker instance.

Returns:

Name	Type	Description
BaseEMInvoker	BaseEMInvoker	The EM Invoker instance.

clear() async

Clear all records from the datastore.

create(data, **kwargs) async

Add chunks to the vector store with automatic embedding generation.

Parameters:

Name	Type	Description	Default
data	Chunk | list[Chunk]	Single chunk or list of chunks to add.	required
**kwargs	Any	Backend-specific parameters (e.g., partition_name).	{}

Raises:

Type	Description
ValueError	If vector dimension mismatch occurs.

create_from_vector(chunk_vectors, **kwargs) async

Add pre-computed vectors directly.

Parameters:

Name	Type	Description	Default
chunk_vectors	list[tuple[Chunk, Vector]]	List of tuples containing chunks and their corresponding vectors.	required
**kwargs	Any	Backend-specific parameters (e.g., partition_name).	{}

Raises:

Type	Description
ValueError	If vector dimension mismatch occurs.

delete(filters=None, **kwargs) async

Delete records from the datastore.

Parameters:

Name	Type	Description	Default
filters	FilterClause | QueryFilter | None	Filters to select records to delete. FilterClause objects are automatically converted to QueryFilter internally. Defaults to None, in which case no operation is performed (no-op).	None
**kwargs	Any	Backend-specific parameters.	{}

Note

If filters is None, no operation is performed (no-op).

ensure_index(index_type='IVF_FLAT', index_params=None, query_field='content', **kwargs) async

Ensure collection and vector index exist, creating them if necessary.

This method is idempotent - if the collection and index already exist, it will skip creation and return early. Uses a lock to prevent race conditions when called concurrently from multiple coroutines.

Parameters:

Name	Type	Description	Default
index_type	str	Index type. Defaults to "IVF_FLAT". Supported: "IVF_FLAT", "HNSW".	'IVF_FLAT'
index_params	dict[str, Any] | None	Index-specific parameters. Defaults to None, in which case default parameters are used.	None
query_field	str	Field name for text content. Defaults to "content".	'content'
**kwargs	Any	Additional parameters.	{}

Raises:

Type	Description
RuntimeError	If collection or index creation fails.

retrieve(query, filters=None, options=None, **kwargs) async

Read records from the datastore using text-based similarity search with optional filtering.

Parameters:

Name	Type	Description	Default
query	str	Input text to embed and search with.	required
filters	FilterClause | QueryFilter | None	Query filters to apply. FilterClause objects are automatically converted to QueryFilter internally. Defaults to None.	None
options	QueryOptions | None	Query options like limit and sorting. Defaults to None.	None
**kwargs	Any	Backend-specific parameters (e.g., search_params).	{}

Returns:

Type	Description
list[Chunk]	list[Chunk]: Query results ordered by similarity score.

retrieve_by_vector(vector, filters=None, options=None, search_params=None, **kwargs) async

Direct vector similarity search.

Parameters:

Name	Type	Description	Default
vector	Vector	Query embedding vector.	required
filters	FilterClause | QueryFilter | None	Query filters to apply. FilterClause objects are automatically converted to QueryFilter internally. Defaults to None.	None
options	QueryOptions | None	Query options like limit and sorting. Defaults to None.	None
search_params	dict[str, Any] | None	Search parameters for Milvus. If None, default search parameters based on distance metric will be used. Defaults to None.	None
**kwargs	Any	Backend-specific parameters.	{}

Returns:

Type	Description
list[Chunk]	list[Chunk]: List of chunks ordered by similarity score.

update(update_values, filters=None, **kwargs) async

Update existing records in the datastore.

Parameters:

Name	Type	Description	Default
update_values	dict[str, Any]	Values to update. Supports content for updating document content and metadata for updating metadata. If content is updated, vectors are re-embedded.	required
filters	FilterClause | QueryFilter | None	Filters to select records to update. FilterClause objects are automatically converted to QueryFilter internally. Defaults to None.	None
**kwargs	Any	Backend-specific parameters (e.g., partition_name).	{}

Note

If filters is None, no operation is performed (no-op).