Struct CsrGraph 

pub struct CsrGraph { /* private fields */ }

A directed, weighted graph in Compressed Sparse Row format.

Built once from the edge set and reused by every graph algorithm. Both forward (out-edges) and reverse (in-edges) CSR arrays are stored: PageRank and Dijkstra consume out-edges; community detection and similarity use the undirected view (both directions).

Implementations

impl CsrGraph

pub fn from_edges(node_ids: &[String], edges: &[GraphEdge]) -> Self

Build a CSR graph from an explicit node-id list and edge list.

Edges whose source or target is absent from node_ids are dropped (foreign-reference integrity). Duplicate node IDs collapse to the first occurrence. The construction is O(V + E log E).

pub fn build_csr(conn: &Connection) -> Result<Self>

Build a CSR snapshot from a SQLite connection.

Loads all node IDs (including edge-less dangling nodes) and a node-count-proportional edge cap.

pub fn node_count(&self) -> usize

Number of nodes.

pub fn edge_count(&self) -> usize

Number of resolved edges (orphans excluded).

pub fn node_index(&self, id: &str) -> Option<u32>

Resolve a node ID to its index, or None if absent.

pub fn node_id(&self, idx: u32) -> &str

Resolve an index back to its node ID. Panics on out-of-range index.

pub fn out_neighbors(&self, src: u32) -> impl Iterator<Item = (u32, f64)> + '_

Iterate (target_idx, weight) over the out-edges of src.

pub fn in_neighbors(&self, dst: u32) -> impl Iterator<Item = (u32, f64)> + '_

Iterate (source_idx, weight) over the in-edges of dst.

pub fn out_degree(&self, idx: u32) -> u32

Out-degree (number of out-edges) of idx.

pub fn out_weight_sum(&self, idx: u32) -> f64

Sum of out-edge weights for idx. <= 0.0 marks a dangling node.

pub fn is_dangling(&self, idx: u32) -> bool

Whether idx is a dangling node — no out-edges, or all-zero weights.

PageRank redistributes a dangling node’s rank mass globally each iteration to prevent rank leakage and convergence failure.

pub fn node_indices(&self) -> Range<u32>

Iterate all node indices in [0, node_count).