Struct RpcClient 

pub struct RpcClient { /* private fields */ }

One connection to a Vectorizer RPC server.

Implementations

impl RpcClient

pub async fn connect_url(url: &str) -> Result<Self>

Convenience: parse a vectorizer://host[:port] URL and dial.

Accepts every form documented at crate::rpc::endpoint::parse_endpoint:

• vectorizer://host:port → RPC on the given port.
• vectorizer://host → RPC on the default port 15503.
• host:port (no scheme) → RPC.
• http(s)://... → returns RpcClientError::Server with a clear message asking the caller to use the HTTP client instead. The SDK ships the http Cargo feature for that path; an http:// URL is not a transport an RPC client can speak.

pub async fn connect(addr: impl ToSocketAddrs) -> Result<Self>

Open a TCP connection to addr (which must be host:port) and start the background reader task. Does NOT send HELLO — callers MUST call Self::hello before any data-plane command, or the server will reject it.

pub async fn hello(&self, payload: HelloPayload) -> Result<HelloResponse>

Issue the HELLO handshake. Must be the first call on a fresh connection. Returns the server’s capability list and auth flags.

pub async fn ping(&self) -> Result<String>

Health check. The server treats PING as auth-exempt so this works even before HELLO; the typed wrapper still validates the response shape.

pub async fn call( &self, command: impl Into<String>, args: Vec<VectorizerValue>, ) -> Result<VectorizerValue>

Generic call dispatcher. Most callers should use a typed wrapper from crate::rpc::commands instead.

pub fn is_authenticated(&self) -> bool

Returns true once HELLO has succeeded on this connection.

pub fn close(self)

Close the connection. Aborts the reader task; in-flight calls receive ConnectionClosed.

impl RpcClient

pub async fn list_collections(&self) -> Result<Vec<String>>

collections.list — return every collection name visible to the authenticated principal.

pub async fn get_collection_info(&self, name: &str) -> Result<CollectionInfo>

collections.get_info — return metadata for one collection.

pub async fn create_collection( &self, name: &str, config: VectorizerValue, ) -> Result<CreateCollectionResult>

collections.create — create a new collection.

config is a Map with optional keys dimension (Int) and metric (Str: "cosine" | "euclidean" | "dot").

pub async fn delete_collection(&self, name: &str) -> Result<bool>

collections.delete — delete a collection (admin-gated on server).

pub async fn list_empty_collections(&self) -> Result<Vec<String>>

collections.list_empty — list collections that contain zero vectors.

pub async fn cleanup_empty_collections( &self, dry_run: bool, ) -> Result<CleanupEmptyResult>

collections.cleanup_empty — remove empty collections.

Pass dry_run: true to preview which collections would be removed without actually deleting them.

pub async fn force_save_collection(&self, name: &str) -> Result<bool>

collections.force_save — flush a collection’s in-memory state to disk.

impl RpcClient

pub async fn get_vector( &self, collection: &str, vector_id: &str, ) -> Result<VectorizerValue>

vectors.get — fetch one vector by id. Returns the raw VectorizerValue::Map so callers can read whichever fields they care about (id, data, payload, document_id).

pub async fn insert_vector( &self, collection: &str, id: Option<&str>, data: Vec<f32>, payload: Option<VectorizerValue>, ) -> Result<VectorWriteResult>

vectors.insert — insert one pre-computed vector.

data must match the collection’s configured dimension. payload is an optional VectorizerValue::Map of metadata.

pub async fn insert_text_vector( &self, collection: &str, id: Option<&str>, text: &str, payload: Option<VectorizerValue>, ) -> Result<VectorWriteResult>

vectors.insert_text — embed text server-side and insert.

The server auto-creates the collection if it does not exist.

pub async fn update_vector( &self, collection: &str, id: &str, data: Vec<f32>, payload: Option<VectorizerValue>, ) -> Result<VectorWriteResult>

vectors.update — replace a vector’s data and/or payload.

pub async fn delete_vector_rpc( &self, collection: &str, id: &str, ) -> Result<bool>

vectors.delete — delete one vector by id.

pub async fn list_vectors( &self, collection: &str, page: i64, limit: i64, ) -> Result<VectorListResult>

vectors.list — page through vectors in a collection.

page is zero-based; limit is capped at 50 by the server.

pub async fn embed_text( &self, text: &str, model: Option<&str>, ) -> Result<EmbedResult>

vectors.embed — embed text server-side and return the embedding.

pub async fn batch_insert_vectors( &self, collection: &str, items: Vec<VectorizerValue>, ) -> Result<BatchInsertResult>

vectors.batch_insert — insert multiple pre-computed vectors.

Each item in items is a VectorizerValue::Map with at least data (Array<Float>) and optionally id (Str) and payload (Map).

pub async fn batch_insert_texts( &self, collection: &str, items: Vec<VectorizerValue>, ) -> Result<BatchInsertResult>

vectors.batch_insert_texts — embed and insert multiple text items.

Each item in items is a VectorizerValue::Map with at least text (Str) and optionally id (Str) and payload (Map).

pub async fn batch_search( &self, requests: Vec<VectorizerValue>, ) -> Result<Vec<BatchSearchResult>>

vectors.batch_search — run multiple searches in one round-trip.

Each request in requests is a VectorizerValue::Map with collection (Str), query (Str), and optional limit (Int).

pub async fn batch_update_vectors( &self, collection: &str, updates: Vec<VectorizerValue>, ) -> Result<BatchUpdateResult>

vectors.batch_update — update multiple vectors’ data and/or payload.

Each item in updates is a VectorizerValue::Map with id (Str) and optionally data (Array<Float>) and payload (Map).

pub async fn batch_delete_vectors( &self, collection: &str, ids: Vec<String>, ) -> Result<BatchDeleteResult>

vectors.batch_delete — delete multiple vectors by id.

pub async fn move_vectors_rpc( &self, src: &str, dst: &str, ids: Vec<String>, ) -> Result<MoveRpcResult>

vectors.move — move vectors from src to dst collection.

Named move_vectors_rpc to avoid collision with the REST SDK’s move_to_collection.

pub async fn copy_vectors_rpc( &self, src: &str, dst: &str, ids: Vec<String>, ) -> Result<CopyRpcResult>

vectors.copy — copy vectors from src to dst without deleting.

pub async fn delete_by_filter_rpc( &self, collection: &str, filter: VectorizerValue, ) -> Result<DeleteByFilterRpcResult>

vectors.delete_by_filter — delete all vectors matching a Qdrant-style filter predicate.

filter is a VectorizerValue::Map matching the Qdrant filter schema.

pub async fn bulk_update_metadata_rpc( &self, collection: &str, filter: VectorizerValue, patch: VectorizerValue, ) -> Result<BulkUpdateMetadataRpcResult>

vectors.bulk_update_metadata — apply a JSON-merge-patch to all vectors matching filter.

filter selects the target vectors; patch is applied via RFC 7396 merge-patch.

pub async fn set_vector_expiry( &self, collection: &str, id: &str, expires_at: &str, ) -> Result<SetExpiryResult>

vectors.set_expiry — attach a TTL to one vector.

expires_at may be a Unix millisecond timestamp or an RFC3339 string.

impl RpcClient

pub async fn search_basic( &self, collection: &str, query: &str, limit: usize, ) -> Result<Vec<SearchHit>>

search.basic — search collection for query and return up to limit hits sorted by descending similarity.

pub async fn search_intelligent( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

search.intelligent — multi-collection intelligent search.

request is a VectorizerValue::Map with at minimum query (Str) and optionally collections (Array<Str>), max_results (Int), domain_expansion (Bool).

pub async fn search_by_text( &self, collection: &str, query: &str, limit: usize, ) -> Result<Vec<SearchHit>>

search.by_text — search one collection by text query.

pub async fn search_by_file( &self, collection: &str, request: VectorizerValue, ) -> Result<Vec<SearchHit>>

search.by_file — file-content-based search.

request is a VectorizerValue::Map describing the file query. The server currently returns an empty result set (stub surface); this method is provided for forward-compatibility.

pub async fn search_hybrid( &self, collection: &str, request: VectorizerValue, ) -> Result<Vec<SearchHit>>

search.hybrid — RRF / weighted-combination hybrid dense+sparse search.

request is a VectorizerValue::Map with at minimum query (Str) and optional keys alpha, dense_k, sparse_k, final_k, algorithm ("rrf" | "weighted" | "alpha").

pub async fn search_semantic( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

search.semantic — semantic re-ranking search.

request is a VectorizerValue::Map with query (Str), collection (Str), and optional max_results, semantic_reranking, cross_encoder_reranking, similarity_threshold.

pub async fn search_contextual( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

search.contextual — context-filtered semantic search.

request is a VectorizerValue::Map with query (Str), collection (Str), and optional context_filters (Map), max_results, context_weight, context_reranking.

pub async fn search_multi_collection( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

search.multi_collection — fan-out search across multiple collections.

request is a VectorizerValue::Map with query (Str), collections (Array<Str>), and optional max_per_collection, max_total_results, cross_collection_reranking.

pub async fn search_explain( &self, collection: &str, request: VectorizerValue, ) -> Result<SearchExplainResult>

search.explain — run a vector search and return HNSW traversal trace.

collection is the target collection; request must contain vector (Array<Float>) and optionally k (Int).

impl RpcClient

pub async fn discover(&self, request: VectorizerValue) -> Result<DiscoverResult>

discovery.discover — full discovery pipeline: embed → search → compress → build plan → render prompt.

request must contain query (Str) and optionally include_collections, exclude_collections (Array<Str>), max_bullets (Int).

pub async fn filter_collections( &self, request: VectorizerValue, ) -> Result<Vec<String>>

discovery.filter_collections — filter collection list by query relevance.

request must contain query (Str) and optionally include / exclude (Array<Str>).

pub async fn score_collections( &self, request: VectorizerValue, ) -> Result<Vec<ScoredCollection>>

discovery.score_collections — score all collections for a query.

request must contain query (Str).

pub async fn expand_queries( &self, request: VectorizerValue, ) -> Result<ExpandQueriesResult>

discovery.expand_queries — generate query variants via baseline expansion.

request must contain query (Str) and optionally max_expansions (Int), include_definition, include_features, include_architecture (Bool).

pub async fn broad_discovery( &self, request: VectorizerValue, ) -> Result<Vec<DiscoveryChunk>>

discovery.broad_discovery — multi-query broad search across all collections.

request must contain queries (Array<Str>) and optionally k (Int).

pub async fn semantic_focus( &self, request: VectorizerValue, ) -> Result<Vec<DiscoveryChunk>>

discovery.semantic_focus — deep semantic search within one collection.

request must contain collection (Str), queries (Array<Str>), and optionally k (Int).

pub async fn promote_readme( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

discovery.promote_readme — promote README chunks to the top of a chunk set.

request must contain chunks (Array<Map>) where each map has collection, doc_id, content, score, file_path, chunk_index, file_extension.

pub async fn compress_evidence( &self, request: VectorizerValue, ) -> Result<Vec<CompressBullet>>

discovery.compress_evidence — compress a chunk set into ranked bullets.

request must contain chunks (Array<Map>) and optionally max_bullets (Int), max_per_doc (Int).

pub async fn build_answer_plan( &self, request: VectorizerValue, ) -> Result<AnswerPlanResult>

discovery.build_answer_plan — organise bullets into a structured answer plan.

request must contain bullets (Array<Map>), each with text, source_id, score, category.

pub async fn render_llm_prompt( &self, request: VectorizerValue, ) -> Result<RenderPromptResult>

discovery.render_llm_prompt — render an answer plan into an LLM prompt string.

request must contain plan (Map) with sections (Array<Map>) and optionally total_bullets, sources.

impl RpcClient

pub async fn file_content( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.content — retrieve raw file content stored in a collection.

request must contain collection (Str) and file_path (Str), and optionally max_size_kb (Int).

pub async fn file_list( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.list — list files indexed in a collection.

request must contain collection (Str) and optionally filter_by_type (Array<Str>), min_chunks (Int), max_results (Int), sort_by (Str).

pub async fn file_summary( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.summary — extractive or structural summary of one file.

request must contain collection (Str), file_path (Str), and optionally summary_type ("extractive" | "structural" | "both"), max_sentences (Int).

pub async fn file_chunks( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.chunks — retrieve ordered chunks for one file.

request must contain collection (Str), file_path (Str), and optionally start_chunk (Int), limit (Int), include_context (Bool).

pub async fn file_outline( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.outline — directory-tree outline of a collection’s files.

request must contain collection (Str) and optionally max_depth (Int), include_summaries (Bool), highlight_key_files (Bool).

pub async fn file_related( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.related — find files semantically related to a given file.

request must contain collection (Str), file_path (Str), and optionally limit (Int), similarity_threshold (Float), include_reason (Bool).

pub async fn file_search_by_type( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

file.search_by_type — search within files of specific extension types.

request must contain collection (Str), query (Str), file_types (Array<Str>), and optionally limit (Int), return_full_files (Bool).

impl RpcClient

pub async fn graph_list_nodes( &self, collection: &str, ) -> Result<VectorizerValue>

graph.list_nodes — list all graph nodes in a collection.

pub async fn graph_neighbors( &self, collection: &str, node_id: &str, ) -> Result<VectorizerValue>

graph.neighbors — fetch direct neighbors of a graph node.

pub async fn graph_find_related( &self, collection: &str, node_id: &str, max_hops: i64, ) -> Result<VectorizerValue>

graph.find_related — find nodes reachable within max_hops of a node.

pub async fn graph_find_path( &self, collection: &str, from: &str, to: &str, ) -> Result<VectorizerValue>

graph.find_path — shortest path between two graph nodes.

pub async fn graph_create_edge( &self, collection: &str, edge: VectorizerValue, ) -> Result<VectorizerValue>

graph.create_edge — create a directed edge between two nodes.

edge is a VectorizerValue::Map with source (Str), target (Str), relationship_type (Str), and optionally weight (Float).

pub async fn graph_delete_edge( &self, collection: &str, edge_id: &str, ) -> Result<VectorizerValue>

graph.delete_edge — remove an edge by its id.

pub async fn graph_list_edges( &self, collection: &str, ) -> Result<VectorizerValue>

graph.list_edges — list all edges in a collection’s graph.

pub async fn graph_discover_edges( &self, collection: &str, request: VectorizerValue, ) -> Result<DiscoverEdgesResult>

graph.discover_edges — auto-discover edges by vector similarity across the whole collection.

request is an optional VectorizerValue::Map with similarity_threshold (Float) and max_per_node (Int).

pub async fn graph_discover_edges_for_node( &self, collection: &str, node_id: &str, request: VectorizerValue, ) -> Result<DiscoverEdgesForNodeResult>

graph.discover_edges_for_node — auto-discover edges for one node.

request is an optional VectorizerValue::Map with similarity_threshold (Float) and max_per_node (Int).

pub async fn graph_discovery_status( &self, collection: &str, ) -> Result<GraphDiscoveryStatus>

graph.discovery_status — percentage of nodes that have edges.

impl RpcClient

pub async fn admin_stats(&self) -> Result<AdminStats>

admin.stats — aggregate vector/collection counts.

pub async fn admin_status(&self) -> Result<AdminStatus>

admin.status — readiness probe and basic counts.

pub async fn admin_logs( &self, request: Option<VectorizerValue>, ) -> Result<VectorizerValue>

admin.logs — in-process log entries (the server currently returns empty; use the REST /logs endpoint for live streaming).

pub async fn admin_indexing_progress(&self) -> Result<VectorizerValue>

admin.indexing_progress — how many collections have been indexed.

pub async fn admin_config_get(&self) -> Result<VectorizerValue>

admin.config_get — read the server’s config.yml.

pub async fn admin_config_update(&self, patch: VectorizerValue) -> Result<bool>

admin.config_update — write a patch map to config.yml (admin).

patch is a VectorizerValue::Map of config keys to new values.

pub async fn admin_backups_list(&self) -> Result<VectorizerValue>

admin.backups_list — list available backup files.

pub async fn admin_backups_create( &self, request: VectorizerValue, ) -> Result<String>

admin.backups_create — create a backup (admin).

request must contain name (Str) and optionally collections (Array<Str>).

pub async fn admin_backups_restore( &self, request: VectorizerValue, ) -> Result<bool>

admin.backups_restore — restore a backup by id (admin).

request must contain backup_id (Str).

pub async fn admin_workspaces_list(&self) -> Result<VectorizerValue>

admin.workspaces_list — list configured workspaces.

pub async fn admin_workspace_get(&self) -> Result<VectorizerValue>

admin.workspace_get — read workspace.yml.

pub async fn admin_workspace_add( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

admin.workspace_add — register a new workspace directory (admin).

request must contain path (Str) and collection_name (Str).

pub async fn admin_workspace_remove(&self, name: &str) -> Result<bool>

admin.workspace_remove — remove a workspace by name (admin).

pub async fn admin_restart(&self) -> Result<bool>

admin.restart — schedule a server restart (admin).

pub async fn admin_slow_queries_list(&self) -> Result<VectorizerValue>

admin.slow_queries_list — retrieve the slow-query ring buffer.

pub async fn admin_slow_queries_config( &self, config: VectorizerValue, ) -> Result<SlowQueryConfigResult>

admin.slow_queries_config — configure slow-query threshold and ring-buffer capacity.

config must contain threshold_ms (Int) and optionally capacity (Int).

impl RpcClient

pub async fn auth_me(&self) -> Result<AuthMeResult>

auth.me — return the authenticated principal’s identity.

pub async fn auth_logout(&self, token: &str) -> Result<VectorizerValue>

auth.logout — blacklist the supplied JWT so it cannot be reused.

pub async fn auth_refresh_token( &self, token: &str, ) -> Result<RefreshTokenResult>

auth.refresh_token — exchange a valid JWT for a fresh one.

pub async fn auth_validate_password( &self, password: &str, ) -> Result<ValidatePasswordResult>

auth.validate_password — check a plaintext password against the server’s password policy.

pub async fn auth_api_keys_create( &self, request: VectorizerValue, ) -> Result<ApiKeyCreated>

auth.api_keys_create — create a new API key.

request must contain name (Str) and optionally expires_in (Int, seconds) and permissions (Array<Str>).

pub async fn auth_api_keys_list(&self) -> Result<VectorizerValue>

auth.api_keys_list — list API keys for the current principal.

pub async fn auth_api_keys_revoke(&self, key_id: &str) -> Result<bool>

auth.api_keys_revoke — permanently revoke an API key by id.

pub async fn rotate_api_key_rpc(&self, key_id: &str) -> Result<RotatedApiKey>

auth.api_keys_rotate — rotate an API key (5-minute grace period).

Named rotate_api_key_rpc to avoid collision with the REST SDK’s rotate_api_key.

pub async fn auth_api_keys_create_scoped( &self, request: VectorizerValue, ) -> Result<ApiKeyCreated>

auth.api_keys_create_scoped — create a collection-scoped API key.

request must contain name (Str) and optionally expires_in (Int), permissions (Array<Str>), scopes (Array<Map>).

pub async fn auth_users_create( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

auth.users_create — create a user (admin; returns server error in v1 — RpcState does not carry AuthHandlerState; use the REST endpoint).

pub async fn auth_users_list(&self) -> Result<VectorizerValue>

auth.users_list — list users (admin; returns server error in v1).

pub async fn auth_users_delete( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

auth.users_delete — delete a user (admin; returns server error in v1).

pub async fn auth_users_change_password( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

auth.users_change_password — change a user’s password (returns server error in v1 — use REST).

pub async fn auth_introspect(&self, token: &str) -> Result<VectorizerValue>

auth.introspect — inspect a token’s claims and blacklist status.

pub async fn auth_audit( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

auth.audit — query the auth audit log.

request is an optional VectorizerValue::Map with from (Str), to (Str), actor (Str), action (Str), limit (Int).

impl RpcClient

pub async fn replication_status(&self) -> Result<VectorizerValue>

replication.status — current replication role and replica list.

pub async fn replication_configure( &self, config: VectorizerValue, ) -> Result<ReplicationConfigureResult>

replication.configure — set the replication role for this node.

config must contain role (Str: "master" | "replica" | "standalone") and optionally bind_address, master_address (Str). A server restart is required for the change to take effect.

pub async fn replication_stats(&self) -> Result<VectorizerValue>

replication.stats — replication throughput and lag statistics.

pub async fn replication_replicas_list(&self) -> Result<VectorizerValue>

replication.replicas_list — list connected replicas (master only).

impl RpcClient

pub async fn cluster_failover( &self, replica_id: &str, ) -> Result<VectorizerValue>

cluster.failover — promote a replica to master (admin).

pub async fn cluster_replica_resync( &self, replica_id: &str, ) -> Result<VectorizerValue>

cluster.replica_resync — force a replica to resync from master (admin).

pub async fn cluster_peer_add( &self, request: VectorizerValue, ) -> Result<VectorizerValue>

cluster.peer_add — add a new peer to the cluster (admin).

request must contain address (Str) and optionally role (Str: "member" | "observer").

pub async fn cluster_rebalance(&self) -> Result<VectorizerValue>

cluster.rebalance — trigger a shard rebalance across peers (admin).

pub async fn cluster_rebalance_status(&self) -> Result<RebalanceStatus>

cluster.rebalance_status — check the status of an in-progress rebalance (or confirm idle).

Trait Implementations

impl Drop for RpcClient

fn drop(&mut self)

Executes the destructor for this type.