Struct GraphHandle 

pub struct GraphHandle { /* private fields */ }

Implementations

impl GraphHandle

pub fn new( path: &Path, durability: Durability, tuning: SqliteTuning, _lru_cache_size: NonZeroUsize, read_pool_size: usize, ) -> Result<Self>

pub fn get_entity(&self, name: &str) -> Result<Option<Entity>>

pub fn create_entities(&self, entities: &[EntityInput]) -> Result<Vec<Entity>>

pub fn upsert_entities(&self, entities: &[EntityInput]) -> Result<Vec<Entity>>

pub fn delete_entities(&self, names: &[String]) -> Result<()>

pub fn create_relations( &self, relations: &[RelationInput], ) -> Result<Vec<Relation>>

pub fn delete_relations(&self, relations: &[Relation]) -> Result<()>

pub fn add_relation_observations( &self, from: &str, to: &str, relation_type: &str, contents: &[ObservationInput], ) -> Result<Vec<Observation>>

pub fn delete_relation_observations( &self, from: &str, to: &str, relation_type: &str, observations: &[ObservationInput], ) -> Result<()>

pub fn set_attributes(&self, targets: &[AttributeSet]) -> Result<()>

pub fn delete_attributes(&self, targets: &[AttributeDelete]) -> Result<()>

pub fn add_observations( &self, entity_name: &str, contents: &[ObservationInput], ) -> Result<Vec<Observation>>

pub fn delete_observations( &self, entity_name: &str, observations: &[ObservationInput], ) -> Result<()>

pub fn merge_entities(&self, source: &str, target: &str) -> Result<Entity>

pub fn rename_entity(&self, old_name: &str, new_name: &str) -> Result<Entity>

pub fn code_purge_file(&self, rel_path: &str) -> Result<usize>

Delete a code file and all its defined symbols in the same transaction.

pub fn search_nodes_filtered( &self, query: &str, filter_type: Option<&str>, offset: usize, limit: usize, ) -> Vec<Entity>

pub fn search_nodes_lite_json( &self, query: &str, filter_type: Option<&str>, offset: usize, limit: usize, ) -> (String, usize, bool)

The viewer’s search payload: a JSON array of {name, entityType, obsCount} (no observation bodies — see [batch_entity_lite_by_ids]), in the same order and with the same post-filter offset semantics as Self::search_nodes_filtered. Returns (entities_json, returned, has_more); has_more is detected by fetching one extra match past the page. Builds the JSON directly to avoid a Vec<Entity> round-trip.

pub fn read_graph_filtered( &self, filter_type: Option<&str>, offset: usize, limit: usize, ) -> Result<String>

pub fn read_graph_filtered_lite( &self, filter_type: Option<&str>, offset: usize, limit: usize, ) -> Result<(String, usize)>

Observation-free page for the browser viewer: entities carry obsCount (the denormalised count) instead of the observation bodies, which the canvas never renders in bulk — the inspector lazy-loads them for the one selected node via GET /ui/node. Returns (json, returned) so the caller can build the pagination cursor without re-parsing the payload.

pub fn open_nodes(&self, names: &[String]) -> String

pub fn entities_exist(&self, names: &[String]) -> Result<Vec<bool>>

pub fn degree(&self, name: &str, direction: Direction) -> Result<usize>

pub fn get_entity_count(&self) -> Result<usize>

pub fn get_relation_count(&self) -> Result<usize>

pub fn search_relations( &self, from: Option<&str>, to: Option<&str>, rtype: Option<&str>, query: Option<&str>, limit: Option<usize>, ) -> Result<Vec<RelationDetail>>

pub fn find_path(&self, from: &str, to: &str) -> Result<Option<Vec<String>>>

pub fn compact(&self) -> Result<()>

pub fn neighbors( &self, name: &str, direction: Direction, rtype: Option<&str>, depth: u32, ) -> Result<String>

pub fn extract_subgraph(&self, names: &[String], depth: u32) -> Result<String>

pub fn describe_entity(&self, name: &str) -> Result<EntityDescription>

pub fn entity_type_counts(&self) -> Vec<(String, usize)>

pub fn entity_type_catalog(&self) -> Vec<(String, usize, Option<String>)>

(name, count, desc) for every registered entity type: one with members, or one whose description registers it before first use.

pub fn ui_meta(&self) -> (Vec<(String, usize)>, usize, usize)

The viewer’s shared page metadata — entity-type legend and the graph-wide entity/relation totals — gathered on a single reader connection. The /ui/graph and /ui/search handlers used to take three or four separate reader-pool acquisitions per request (entity_type_counts + get_entity_count + get_relation_count); folding them into one lock acquisition cuts pool churn and shortens the reader hold, which is what bounds concurrent read throughput.

pub fn relation_type_counts(&self) -> Vec<(String, usize)>

pub fn relation_type_catalog(&self) -> Vec<(String, usize, Option<String>)>

(name, count, desc) for every registered relation type: one with members, or one whose description registers it before first use.

pub fn set_type_description( &self, kind: i64, name: &str, description: Option<&str>, ) -> Result<()>

Set or clear the description of one entity type (kind 0) or relation type (kind 1) by exact name. The type row is created when it does not exist yet — a description can thus register a type before its first member (count stays 0). None clears the stored description. This is registry metadata, not a graph mutation: no change event is emitted and the taxonomy index revision is untouched.

pub fn entity_type_exists(&self, name: &str) -> bool

Whether an entity type with the given name exists. Read-only: a missing type stays absent and no row is inserted.

pub fn relation_type_exists(&self, name: &str) -> bool

Whether a relation type with the given name exists. Read-only: a missing type stays absent and no row is inserted.

pub fn batch_get_entities(&self, names: &[String]) -> Vec<Option<Entity>>

pub fn find_all_paths( &self, from: &str, to: &str, max_depth: usize, max_paths: usize, ) -> Result<Vec<Vec<String>>>

pub fn export(&self, _format: &str, max_rows: i64) -> Result<String>

Export the whole graph as a JSON string. max_rows caps both the entity and relation arrays so a pathologically large graph cannot be coerced into an unbounded in-memory string (DoS guard); callers pass a generous constant. A negative value means “no limit”.

pub fn wipe(&self) -> Result<()>

pub fn run_maintenance(&self) -> Result<()>

Periodic database maintenance: WAL checkpoint, query planner analysis, and FTS index optimization. Call from a background timer.

pub fn checkpoint_passive(&self) -> Result<()>

Run a non-blocking wal_checkpoint(PASSIVE) to fsync committed WAL frames without stalling readers or writers. Call from a short-interval timer to bound the durability window in async mode.