Vector Store

Vector Store

Vector Store

Value

An R6 object of class vector_store.

Details

In-memory vector store with cosine similarity search and RDS persistence.

Methods

Public methods

• vector_store$new()

• vector_store$add()

• vector_store$search()

• vector_store$remove()

• vector_store$size()

• vector_store$save()

• vector_store$load()

• vector_store$clone()

---

Method new()

Create a new vector store.

Usage

vector_store$new(dims, encryption_key = NULL, audit_log = NULL)

Arguments

dims

Integer, dimensionality of stored vectors.

encryption_key

Raw 32-byte key for AES-256-CBC encryption at rest, or NULL to check the SECURECONTEXT_ENCRYPTION_KEY env var. If neither is set, data is stored unencrypted.

audit_log

Optional path to a JSONL audit log file. If non-NULL, store operations are logged via log_store_event().

---

Method add()

Add vectors to the store.

Usage

vector_store$add(ids, embeddings, metadata = list())

Arguments

ids

Character vector of unique identifiers.

embeddings

Numeric matrix (nrow = length(ids), ncol = dims).

metadata

List of metadata entries (one per id), or empty list.

---

Method search()

Search for nearest neighbors by cosine similarity.

Usage

vector_store$search(query_embedding, k = 5L)

Arguments

query_embedding

Numeric vector or single-row matrix.

k

Number of results to return.

Returns

Data frame with columns id, score, and metadata.

---

Method remove()

Remove vectors by id.

Usage

vector_store$remove(ids)

Arguments

ids

Character vector of ids to remove.

---

Method size()

Number of stored vectors.

Usage

vector_store$size()

Returns

Integer.

---

Method save()

Save store to an RDS file.

Usage

vector_store$save(path)

Arguments

path

File path.

---

Method load()

Load store from an RDS file.

Usage

vector_store$load(path)

Arguments

path

File path.

---

Method clone()

The objects of this class are cloneable with this method.

Usage

vector_store$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.

Examples

vs <- vector_store$new(dims = 3L)
vs$add("id1", matrix(c(1, 0, 0), nrow = 1))
vs$add("id2", matrix(c(0, 1, 0), nrow = 1))
vs$search(c(1, 0, 0), k = 1)
#>    id score
#> 1 id1     1
vs$size()
#> [1] 2