Struct MapReduce 

pub struct MapReduce<G: SwhForwardGraph + SwhBackwardGraph, MR: MapReducer, L> {
    pub map_reducer: MR,
    /* private fields */
}

Associates labels to nodes in the graph using successor nodes’ labels and “bubbling up”

Use swh_graph::views::Subgraph to select the set of nodes to run this on. For example, to avoid content and directory nodes (which are typically much slower to process), use Subgraph::with_node_constraint("rev,rel,snp,ori".parse().unwrap()).

Built from MapReduceBuilder

Example

For example, with this graph:

     - 3
    /
  <-
1 <--- 4 <--+
            +--- 6
2 <--- 5 <--+

We call 1 a successor of 3, consistent with swh-graph’s terminology, even though MapReducer propagates labels in the other direction.

we would:

• compute label of 1
• compute label of 2
• compute label of 2 and merge it with 1’s
• compute label of 4 and merge it with 1’s
• compute label of 5 and merge it with 2’s
• compute label of 6 and merge it with 4’s and 5’s

Fields

map_reducer: MR

Implementations

impl<G: SwhForwardGraph + SwhBackwardGraph, MR: MapReducer, L: Labels<Label = MR::Label>> MapReduce<G, MR, L>

pub fn run_in_topological_order( &mut self, nodes: impl Iterator<Item = NodeId>, ) -> Result<(), MR::Error>

Runs the configured MapReducer sequentially on all nodes in the graph.

nodes must be an iterator of all nodes in topological order, eg. returned by GenerationsReader::iter_nodes with .map(|(_depth, node)| node).

pub fn push_node(&mut self, node: NodeId) -> Result<(), MR::Error>

A single step of Self::run_in_topological_order, in case the caller does not have an iterator of nodes

pub fn labels(&self) -> Result<&L>

Returns every node’s labels, if MapReduceBuilder::keep_labels was set to true.

pub fn take_labels(self) -> Result<L>

Returns every node’s labels, if MapReduceBuilder::keep_labels was set to true.