19378243882

Committed 14 Nov 2025 09:34PM UTC coverage: 62.201% (+0.06%) from 62.146%

Build # 19378243882

Build Type

push

github

Committed by

vigna

Commit Message

Back to edition 2021

Run Details

5251 of 8442 relevant lines covered (62.2%)

29314157.11 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

64.91

/webgraph/src/transform/simplify.rs

/*
 * SPDX-FileCopyrightText: 2023 Inria
 *
 * SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
 */

use crate::graphs::{
    arc_list_graph, no_selfloops_graph::NoSelfLoopsGraph, union_graph::UnionGraph,
};
use crate::labels::Left;
use crate::traits::{LenderIntoIter, SequentialGraph, SortedIterator, SortedLender, SplitLabeling};
use crate::utils::sort_pairs::{KMergeIters, SortPairs};
use crate::utils::{CodecIter, DefaultBatchCodec, MemoryUsage};
use anyhow::{Context, Result};
use dsi_progress_logger::prelude::*;
use itertools::Itertools;
use lender::*;
use rayon::ThreadPool;
use tempfile::Builder;

use super::transpose;

/// Returns a simplified (i.e., undirected and loopless) version of the provided
/// sorted (both on nodes and successors) graph as a [sequential
/// graph](crate::traits::SequentialGraph).
///
/// This method exploits the fact that the input graph is already sorted,
/// sorting half the number of arcs of [`simplify`].
pub fn simplify_sorted<G: SequentialGraph>(
    graph: G,
    memory_usage: MemoryUsage,
) -> Result<
    NoSelfLoopsGraph<
        UnionGraph<
            G,
            Left<arc_list_graph::ArcListGraph<KMergeIters<CodecIter<DefaultBatchCodec>, ()>>>,
        >,
    >,
>
where
    for<'a> G::Lender<'a>: SortedLender,
    for<'a, 'b> LenderIntoIter<'a, G::Lender<'b>>: SortedIterator,
{
    let transpose = transpose(&graph, memory_usage).context("Could not transpose the graph")?;
    Ok(NoSelfLoopsGraph(UnionGraph(graph, transpose)))
}

/// Returns a simplified (i.e., undirected and loopless) version of the provided
/// graph as a [sequential graph](crate::traits::SequentialGraph).
///
/// Note that if the graph is sorted (both on nodes and successors), it is
/// recommended to use [`simplify_sorted`].
///
/// For the meaning of the additional parameter, see [`SortPairs`].
pub fn simplify(
    graph: &impl SequentialGraph,
    memory_usage: MemoryUsage,
) -> Result<
    Left<
        arc_list_graph::ArcListGraph<
            impl Iterator<Item = ((usize, usize), ())> + Clone + Send + Sync + 'static,
        >,
    >,
> {
    let dir = Builder::new().prefix("simplify_").tempdir()?;
    let mut sorted = SortPairs::new(memory_usage, dir.path())?;

    let mut pl = ProgressLogger::default();
    pl.item_name("node")
        .expected_updates(Some(graph.num_nodes()));
    pl.start("Creating batches...");
    // create batches of sorted edges
    let mut iter = graph.iter();
    while let Some((src, succ)) = iter.next() {
        for dst in succ {
            if src != dst {
                sorted.push(src, dst)?;
                sorted.push(dst, src)?;
            }
        }
        pl.light_update();
    }
    // merge the batches
    let map: fn(((usize, usize), ())) -> (usize, usize) = |(pair, _)| pair;
    let filter: fn(&(usize, usize)) -> bool = |(src, dst)| src != dst;
    let iter = Itertools::dedup(sorted.iter()?.map(map).filter(filter));
    let sorted = arc_list_graph::ArcListGraph::new(graph.num_nodes(), iter);
    pl.done();

    Ok(sorted)
}

/// Returns a simplified (i.e., undirected and loopless) version of the provided
/// graph as a [sequential graph](crate::traits::SequentialGraph).
///
/// This method uses splitting to sort in parallel different parts of the graph.
///
/// For the meaning of the additional parameter, see [`SortPairs`].
pub fn simplify_split<S>(
    graph: &S,
    memory_usage: MemoryUsage,
    threads: &ThreadPool,
) -> Result<
    Left<
        arc_list_graph::ArcListGraph<
            itertools::Dedup<KMergeIters<CodecIter<DefaultBatchCodec>, ()>>,
        >,
    >,
>
where
    S: SequentialGraph + SplitLabeling,
{
    let num_threads = threads.current_num_threads();
    let (tx, rx) = std::sync::mpsc::channel();

    let mut dirs = vec![];

    threads.in_place_scope(|scope| {
        let mut thread_id = 0;
        #[allow(clippy::explicit_counter_loop)] // enumerate requires some extra bounds here
        for iter in graph.split_iter(num_threads) {
            let tx = tx.clone();
            let dir = Builder::new()
                .prefix(&format!("simplify_split_{thread_id}_"))
                .tempdir()
                .expect("Could not create a temporary directory");
            let dir_path = dir.path().to_path_buf();
            dirs.push(dir);
            scope.spawn(move |_| {
                log::debug!("Spawned thread {thread_id}");
                let mut sorted = SortPairs::new(memory_usage, dir_path).unwrap();
                for_!( (src, succ) in iter {
                    for dst in succ {
                        if src != dst {
                            sorted.push(src, dst).unwrap();
                            sorted.push(dst, src).unwrap();
                        }
                    }
                });
                let result = sorted.iter().context("Could not read arcs").unwrap();
                tx.send(result).expect("Could not send the sorted pairs");
                log::debug!("Thread {thread_id} finished");
            });
            thread_id += 1;
        }
    });
    drop(tx);

    // get a graph on the sorted data
    log::debug!("Waiting for threads to finish");
    let edges: KMergeIters<CodecIter<DefaultBatchCodec>> = rx.iter().sum();
    let edges = edges.dedup();
    log::debug!("All threads finished");
    let sorted = arc_list_graph::ArcListGraph::new_labeled(graph.num_nodes(), edges);

    drop(dirs);
    Ok(Left(sorted))
}

1	/*
2	* SPDX-FileCopyrightText: 2023 Inria
3	*
4	* SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
5	*/
6
7	use crate::graphs::{
8	arc_list_graph, no_selfloops_graph::NoSelfLoopsGraph, union_graph::UnionGraph,
9	};
10	use crate::labels::Left;
11	use crate::traits::{LenderIntoIter, SequentialGraph, SortedIterator, SortedLender, SplitLabeling};
12	use crate::utils::sort_pairs::{KMergeIters, SortPairs};
13	use crate::utils::{CodecIter, DefaultBatchCodec, MemoryUsage};
14	use anyhow::{Context, Result};
15	use dsi_progress_logger::prelude::*;
16	use itertools::Itertools;
17	use lender::*;
18	use rayon::ThreadPool;
19	use tempfile::Builder;
20
21	use super::transpose;
22
23	/// Returns a simplified (i.e., undirected and loopless) version of the provided
24	/// sorted (both on nodes and successors) graph as a [sequential
25	/// graph](crate::traits::SequentialGraph).
26	///
27	/// This method exploits the fact that the input graph is already sorted,
28	/// sorting half the number of arcs of [`simplify`].
29	pub fn simplify_sorted<G: SequentialGraph>(	1✔
30	graph: G,
31	memory_usage: MemoryUsage,
32	) -> Result<
33	NoSelfLoopsGraph<
34	UnionGraph<
35	G,
36	Left<arc_list_graph::ArcListGraph<KMergeIters<CodecIter<DefaultBatchCodec>, ()>>>,
37	>,
38	>,
39	>
40	where
41	for<'a> G::Lender<'a>: SortedLender,
42	for<'a, 'b> LenderIntoIter<'a, G::Lender<'b>>: SortedIterator,
43	{
44	let transpose = transpose(&graph, memory_usage).context("Could not transpose the graph")?;	5✔
45	Ok(NoSelfLoopsGraph(UnionGraph(graph, transpose)))	1✔
46	}
47
48	/// Returns a simplified (i.e., undirected and loopless) version of the provided
49	/// graph as a [sequential graph](crate::traits::SequentialGraph).
50	///
51	/// Note that if the graph is sorted (both on nodes and successors), it is
52	/// recommended to use [`simplify_sorted`].
53	///
54	/// For the meaning of the additional parameter, see [`SortPairs`].
55	pub fn simplify(	×
56	graph: &impl SequentialGraph,
57	memory_usage: MemoryUsage,
58	) -> Result<
59	Left<
60	arc_list_graph::ArcListGraph<
61	impl Iterator<Item = ((usize, usize), ())> + Clone + Send + Sync + 'static,
62	>,
63	>,
64	> {
65	let dir = Builder::new().prefix("simplify_").tempdir()?;	×
66	let mut sorted = SortPairs::new(memory_usage, dir.path())?;	×
67
68	let mut pl = ProgressLogger::default();	×
69	pl.item_name("node")	×
70	.expected_updates(Some(graph.num_nodes()));	×
71	pl.start("Creating batches...");	×
72	// create batches of sorted edges
73	let mut iter = graph.iter();	×
74	while let Some((src, succ)) = iter.next() {	×
75	for dst in succ {	×
76	if src != dst {	×
77	sorted.push(src, dst)?;	×
78	sorted.push(dst, src)?;	×
79	}
80	}
81	pl.light_update();	×
82	}
83	// merge the batches
84	let map: fn(((usize, usize), ())) -> (usize, usize) = \|(pair, _)\| pair;	×
85	let filter: fn(&(usize, usize)) -> bool = \|(src, dst)\| src != dst;	×
86	let iter = Itertools::dedup(sorted.iter()?.map(map).filter(filter));	×
87	let sorted = arc_list_graph::ArcListGraph::new(graph.num_nodes(), iter);	×
88	pl.done();	×
89
90	Ok(sorted)	×
91	}
92
93	/// Returns a simplified (i.e., undirected and loopless) version of the provided
94	/// graph as a [sequential graph](crate::traits::SequentialGraph).
95	///
96	/// This method uses splitting to sort in parallel different parts of the graph.
97	///
98	/// For the meaning of the additional parameter, see [`SortPairs`].
99	pub fn simplify_split<S>(	4✔
100	graph: &S,
101	memory_usage: MemoryUsage,
102	threads: &ThreadPool,
103	) -> Result<
104	Left<
105	arc_list_graph::ArcListGraph<
106	itertools::Dedup<KMergeIters<CodecIter<DefaultBatchCodec>, ()>>,
107	>,
108	>,
109	>
110	where
111	S: SequentialGraph + SplitLabeling,
112	{
113	let num_threads = threads.current_num_threads();	12✔
114	let (tx, rx) = std::sync::mpsc::channel();	12✔
115
116	let mut dirs = vec![];	8✔
117
118	threads.in_place_scope(\|scope\| {	12✔
119	let mut thread_id = 0;	8✔
120	#[allow(clippy::explicit_counter_loop)] // enumerate requires some extra bounds here
121	for iter in graph.split_iter(num_threads) {	28✔
122	let tx = tx.clone();	48✔
123	let dir = Builder::new()	48✔
124	.prefix(&format!("simplify_split_{thread_id}_"))	32✔
125	.tempdir()	16✔
126	.expect("Could not create a temporary directory");	32✔
127	let dir_path = dir.path().to_path_buf();	48✔
128	dirs.push(dir);	48✔
129	scope.spawn(move \|_\| {	48✔
130	log::debug!("Spawned thread {thread_id}");	16✔
131	let mut sorted = SortPairs::new(memory_usage, dir_path).unwrap();	80✔
132	for_!( (src, succ) in iter {	2,604,472✔
133	for dst in succ {	27,031,444✔
134	if src != dst {	25,379,448✔
135	sorted.push(src, dst).unwrap();	75,089,040✔
136	sorted.push(dst, src).unwrap();	50,059,360✔
137	}
138	}
139	});
140	let result = sorted.iter().context("Could not read arcs").unwrap();	80✔
141	tx.send(result).expect("Could not send the sorted pairs");	80✔
142	log::debug!("Thread {thread_id} finished");	16✔
143	});
144	thread_id += 1;	16✔
145	}
146	});
147	drop(tx);	8✔
148
149	// get a graph on the sorted data
150	log::debug!("Waiting for threads to finish");	4✔
151	let edges: KMergeIters<CodecIter<DefaultBatchCodec>> = rx.iter().sum();	20✔
152	let edges = edges.dedup();	12✔
153	log::debug!("All threads finished");	4✔
154	let sorted = arc_list_graph::ArcListGraph::new_labeled(graph.num_nodes(), edges);	20✔
155
156	drop(dirs);	8✔
157	Ok(Left(sorted))	4✔
158	}

vigna / webgraph-rs / 19378243882

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous