pub struct Dfs<N, VM> {
pub stack: Vec<N>,
pub discovered: VM,
}
Expand description
Visit nodes of a graph in a depth-first-search (DFS) emitting nodes in preorder (when they are first discovered).
The traversal starts at a given node and only traverses nodes reachable from it.
Dfs
is not recursive.
Dfs
does not itself borrow the graph, and because of this you can run
a traversal over a graph while still retaining mutable access to it, if you
use it like the following example:
use petgraph::Graph;
use petgraph::visit::Dfs;
let mut graph = Graph::<_,()>::new();
let a = graph.add_node(0);
let mut dfs = Dfs::new(&graph, a);
while let Some(nx) = dfs.next(&graph) {
// we can access `graph` mutably here still
graph[nx] += 1;
}
assert_eq!(graph[a], 1);
Note: The algorithm may not behave correctly if nodes are removed during iteration. It may not necessarily visit added nodes or edges.
Fields§
§stack: Vec<N>
The stack of nodes to visit
discovered: VM
The map of discovered nodes
Implementations§
Source§impl<N, VM> Dfs<N, VM>
impl<N, VM> Dfs<N, VM>
Sourcepub fn new<G>(graph: G, start: N) -> Self
pub fn new<G>(graph: G, start: N) -> Self
Create a new Dfs, using the graph’s visitor map, and put start in the stack of nodes to visit.
Sourcepub fn from_parts(stack: Vec<N>, discovered: VM) -> Self
pub fn from_parts(stack: Vec<N>, discovered: VM) -> Self
Create a Dfs
from a vector and a visit map
Sourcepub fn empty<G>(graph: G) -> Self
pub fn empty<G>(graph: G) -> Self
Create a new Dfs using the graph’s visitor map, and no stack.
Sourcepub fn move_to(&mut self, start: N)
pub fn move_to(&mut self, start: N)
Keep the discovered map, but clear the visit stack and restart the dfs from a particular node.
Sourcepub fn next<G>(&mut self, graph: G) -> Option<N>where
G: IntoNeighbors<NodeId = N>,
pub fn next<G>(&mut self, graph: G) -> Option<N>where
G: IntoNeighbors<NodeId = N>,
Return the next node in the dfs, or None if the traversal is done.