Module encoding

Module encoding 

Source
Expand description

Key encoding for vector storage.

This module provides key encoding for vector embeddings in storage backends. Keys are designed to support efficient prefix-based range scans.

§Key Prefixes

§Embedding spaces (entity-based)

  • 0x10 - Embedding space metadata: [0x10][space_name_hash]
  • 0x11 - Entity embedding: [0x11][space_name_hash][entity_id]
  • 0x12 - Sparse embedding space metadata: [0x12][space_name_hash]
  • 0x13 - Sparse entity embedding: [0x13][space_name_hash][entity_id]
  • 0x14 - Multi-vector space metadata: [0x14][space_name_hash]
  • 0x15 - Multi-vector embedding: [0x15][space_name_hash][entity_id]

§Point collections (Qdrant-style)

  • 0x20 - Collection metadata: [0x20][collection_name_hash]
  • 0x21 - Point payload: [0x21][collection_name_hash][point_id]
  • 0x22 - Dense vector: [0x22][collection_name_hash][point_id][vector_name_hash]
  • 0x23 - Sparse vector: [0x23][collection_name_hash][point_id][vector_name_hash]
  • 0x24 - Multi-vector: [0x24][collection_name_hash][point_id][vector_name_hash]

§Inverted index (sparse vector index)

  • 0x30 - Posting list: [0x30][collection_hash][vector_name_hash][token_id]
  • 0x31 - Index metadata: [0x31][collection_hash][vector_name_hash]
  • 0x32 - Point tokens: [0x32][collection_hash][vector_name_hash][point_id]

§Collection vectors (entity-to-vector mapping)

  • 0x40 - Collection vector: [0x40][collection_id][entity_id][vector_name_hash]

All numeric values are encoded in big-endian format to preserve sort order.

Structs§

CollectionKey
A decoded collection key.
CollectionVectorKey
A decoded collection vector key.
DenseVectorKey
A decoded dense vector key.
EmbeddingKey
A decoded embedding key.
InvertedMetaKey
A decoded inverted meta key.
MultiVectorKey
A decoded multi-vector key.
PointPayloadKey
A decoded point payload key.
PointTokensKey
A decoded point tokens key.
PostingKey
A decoded posting key.
SparseEmbeddingKey
A decoded sparse embedding key.
SparseVectorKey
A decoded sparse vector key.

Constants§

PREFIX_COLLECTION
Key prefix for collection metadata.
PREFIX_COLLECTION_VECTOR
Key prefix for collection vectors.
PREFIX_EMBEDDING
Key prefix for entity embeddings.
PREFIX_EMBEDDING_SPACE
Key prefix for embedding space metadata.
PREFIX_INVERTED_META
Key prefix for inverted index metadata.
PREFIX_MULTI_VECTOR
Key prefix for multi-vector entity embeddings.
PREFIX_MULTI_VECTOR_SPACE
Key prefix for multi-vector embedding space metadata.
PREFIX_POINT_DENSE_VECTOR
Key prefix for dense vectors.
PREFIX_POINT_MULTI_VECTOR
Key prefix for multi-vectors.
PREFIX_POINT_PAYLOAD
Key prefix for point payloads.
PREFIX_POINT_SPARSE_VECTOR
Key prefix for sparse vectors.
PREFIX_POINT_TOKENS
Key prefix for point-to-tokens reverse mapping.
PREFIX_POSTING
Key prefix for posting lists.
PREFIX_SPARSE_EMBEDDING
Key prefix for sparse entity embeddings.
PREFIX_SPARSE_EMBEDDING_SPACE
Key prefix for sparse embedding space metadata.

Functions§

decode_collection_key
Decode a collection key.
decode_collection_vector_entity_id
Decode the entity ID from a collection vector key.
decode_collection_vector_key
Decode a collection vector key.
decode_dense_vector_key
Decode a dense vector key.
decode_embedding_entity_id
Decode an entity ID from an embedding key.
decode_embedding_key
Decode an embedding key.
decode_embedding_space_key
Decode an embedding space name hash from a space key.
decode_inverted_meta_key
Decode an inverted meta key.
decode_multi_vector_key
Decode a multi-vector key.
decode_point_payload_key
Decode a point payload key.
decode_point_payload_point_id
Decode just the point ID from a point payload key.
decode_point_tokens_key
Decode a point tokens key.
decode_posting_key
Decode a posting key.
decode_sparse_embedding_entity_id
Decode an entity ID from a sparse embedding key.
decode_sparse_embedding_key
Decode a sparse embedding key.
decode_sparse_embedding_space_key
Decode a sparse embedding space name hash from a space key.
decode_sparse_vector_key
Decode a sparse vector key.
encode_collection_key
Encode a key for collection metadata.
encode_collection_prefix
Encode a prefix for scanning all collections.
encode_collection_vector_key
Encode a key for a collection vector.
encode_collection_vector_prefix
Encode a prefix for all vectors in a collection.
encode_dense_vector_collection_prefix
Encode a prefix for scanning all dense vectors in a collection.
encode_dense_vector_key
Encode a key for a point’s dense vector.
encode_dense_vector_point_prefix
Encode a prefix for scanning all dense vectors for a point.
encode_embedding_key
Encode a key for an entity’s embedding in a space.
encode_embedding_prefix
Encode a prefix for scanning all embeddings in a space.
encode_embedding_space_key
Encode a key for embedding space metadata.
encode_entity_vector_prefix
Encode a prefix for all vectors of an entity within a collection.
encode_inverted_meta_collection_prefix
Encode a prefix for scanning all inverted indexes in a collection.
encode_inverted_meta_key
Encode a key for inverted index metadata.
encode_multi_vector_collection_prefix
Encode a prefix for scanning all multi-vectors in a collection.
encode_multi_vector_key
Encode a key for a point’s multi-vector.
encode_multi_vector_point_prefix
Encode a prefix for scanning all multi-vectors for a point.
encode_point_payload_key
Encode a key for a point’s payload.
encode_point_payload_prefix
Encode a prefix for scanning all points in a collection.
encode_point_tokens_collection_prefix
Encode a prefix for scanning all point tokens in a collection.
encode_point_tokens_key
Encode a key for point-to-tokens reverse mapping.
encode_point_tokens_prefix
Encode a prefix for scanning all point tokens for a vector.
encode_posting_collection_prefix
Encode a prefix for scanning all posting lists in a collection.
encode_posting_key
Encode a key for a posting list entry.
encode_posting_prefix
Encode a prefix for scanning all posting lists for a vector.
encode_sparse_embedding_key
Encode a key for an entity’s sparse embedding in a space.
encode_sparse_embedding_prefix
Encode a prefix for scanning all sparse embeddings in a space.
encode_sparse_embedding_space_key
Encode a key for sparse embedding space metadata.
encode_sparse_vector_collection_prefix
Encode a prefix for scanning all sparse vectors in a collection.
encode_sparse_vector_key
Encode a key for a point’s sparse vector.
encode_sparse_vector_point_prefix
Encode a prefix for scanning all sparse vectors for a point.
hash_name
Compute a hash for an embedding name.