19076069272

Committed 04 Nov 2025 04:42PM UTC coverage: 61.785% (-0.2%) from 61.976%

Build # 19076069272

Build Type

push

github

Committed by

vigna

Commit Message

Fixed doctests

Run Details

5143 of 8324 relevant lines covered (61.79%)

30278823.63 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::{BatchIterator, KMergeIters, SortPairs};
use crate::utils::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`](crate::transform::simplify::simplify).
pub fn simplify_sorted<G: SequentialGraph>(
    graph: G,
    memory_usage: MemoryUsage,
) -> Result<
    NoSelfLoopsGraph<
        UnionGraph<G, Left<arc_list_graph::ArcListGraph<KMergeIters<BatchIterator<()>, ()>>>>,
    >,
>
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`](crate::transform::simplify::simplify_sorted).
///
/// For the meaning of the additional parameter, see
/// [`SortPairs`](crate::prelude::sort_pairs::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`](crate::prelude::sort_pairs::SortPairs).
pub fn simplify_split<S>(
    graph: &S,
    memory_usage: MemoryUsage,
    threads: &ThreadPool,
) -> Result<Left<arc_list_graph::ArcListGraph<itertools::Dedup<KMergeIters<BatchIterator<()>, ()>>>>>
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<BatchIterator> = 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::{BatchIterator, KMergeIters, SortPairs};
13	use crate::utils::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
29	/// [`simplify`](crate::transform::simplify::simplify).
30	pub fn simplify_sorted<G: SequentialGraph>(	1✔
31	graph: G,
32	memory_usage: MemoryUsage,
33	) -> Result<
34	NoSelfLoopsGraph<
35	UnionGraph<G, Left<arc_list_graph::ArcListGraph<KMergeIters<BatchIterator<()>, ()>>>>,
36	>,
37	>
38	where
39	for<'a> G::Lender<'a>: SortedLender,
40	for<'a, 'b> LenderIntoIter<'a, G::Lender<'b>>: SortedIterator,
41	{
42	let transpose = transpose(&graph, memory_usage).context("Could not transpose the graph")?;	5✔
43	Ok(NoSelfLoopsGraph(UnionGraph(graph, transpose)))	1✔
44	}
45
46	/// Returns a simplified (i.e., undirected and loopless) version of the provided
47	/// graph as a [sequential graph](crate::traits::SequentialGraph).
48	///
49	/// Note that if the graph is sorted (both on nodes and successors), it is
50	/// recommended to use [`simplify_sorted`](crate::transform::simplify::simplify_sorted).
51	///
52	/// For the meaning of the additional parameter, see
53	/// [`SortPairs`](crate::prelude::sort_pairs::SortPairs).
54	pub fn simplify(	×
55	graph: &impl SequentialGraph,
56	memory_usage: MemoryUsage,
57	) -> Result<
58	Left<
59	arc_list_graph::ArcListGraph<
60	impl Iterator<Item = ((usize, usize), ())> + Clone + Send + Sync + 'static,
61	>,
62	>,
63	> {
64	let dir = Builder::new().prefix("simplify_").tempdir()?;	×
65	let mut sorted = SortPairs::new(memory_usage, dir.path())?;	×
66
67	let mut pl = ProgressLogger::default();	×
68	pl.item_name("node")	×
69	.expected_updates(Some(graph.num_nodes()));	×
70	pl.start("Creating batches...");	×
71	// create batches of sorted edges
72	let mut iter = graph.iter();	×
73	while let Some((src, succ)) = iter.next() {	×
74	for dst in succ {	×
75	if src != dst {	×
76	sorted.push(src, dst)?;	×
77	sorted.push(dst, src)?;	×
78	}
79	}
80	pl.light_update();	×
81	}
82	// merge the batches
83	let map: fn(((usize, usize), ())) -> (usize, usize) = \|(pair, _)\| pair;	×
84	let filter: fn(&(usize, usize)) -> bool = \|(src, dst)\| src != dst;	×
85	let iter = Itertools::dedup(sorted.iter()?.map(map).filter(filter));	×
86	let sorted = arc_list_graph::ArcListGraph::new(graph.num_nodes(), iter);	×
87	pl.done();	×
88
89	Ok(sorted)	×
90	}
91
92	/// Returns a simplified (i.e., undirected and loopless) version of the provided
93	/// graph as a [sequential graph](crate::traits::SequentialGraph).
94	///
95	/// This method uses splitting to sort in parallel different parts of the graph.
96	///
97	/// For the meaning of the additional parameter, see
98	/// [`SortPairs`](crate::prelude::sort_pairs::SortPairs).
99	pub fn simplify_split<S>(	4✔
100	graph: &S,
101	memory_usage: MemoryUsage,
102	threads: &ThreadPool,
103	) -> Result<Left<arc_list_graph::ArcListGraph<itertools::Dedup<KMergeIters<BatchIterator<()>, ()>>>>>
104	where
105	S: SequentialGraph + SplitLabeling,
106	{
107	let num_threads = threads.current_num_threads();	12✔
108	let (tx, rx) = std::sync::mpsc::channel();	12✔
109
110	let mut dirs = vec![];	8✔
111
112	threads.in_place_scope(\|scope\| {	12✔
113	let mut thread_id = 0;	8✔
114	#[allow(clippy::explicit_counter_loop)] // enumerate requires some extra bounds here
115	for iter in graph.split_iter(num_threads) {	28✔
116	let tx = tx.clone();	48✔
117	let dir = Builder::new()	48✔
118	.prefix(&format!("simplify_split_{thread_id}_"))	32✔
119	.tempdir()	16✔
120	.expect("Could not create a temporary directory");	32✔
121	let dir_path = dir.path().to_path_buf();	48✔
122	dirs.push(dir);	48✔
123	scope.spawn(move \|_\| {	48✔
124	log::debug!("Spawned thread {thread_id}");	16✔
125	let mut sorted = SortPairs::new(memory_usage, dir_path).unwrap();	80✔
126	for_!( (src, succ) in iter {	2,604,472✔
127	for dst in succ {	27,031,444✔
128	if src != dst {	25,379,448✔
129	sorted.push(src, dst).unwrap();	75,089,040✔
130	sorted.push(dst, src).unwrap();	50,059,360✔
131	}
132	}
133	});
134	let result = sorted.iter().context("Could not read arcs").unwrap();	80✔
135	tx.send(result).expect("Could not send the sorted pairs");	80✔
136	log::debug!("Thread {thread_id} finished");	16✔
137	});
138	thread_id += 1;	16✔
139	}
140	});
141	drop(tx);	8✔
142
143	// get a graph on the sorted data
144	log::debug!("Waiting for threads to finish");	4✔
145	let edges: KMergeIters<BatchIterator> = rx.iter().sum();	20✔
146	let edges = edges.dedup();	12✔
147	log::debug!("All threads finished");	4✔
148	let sorted = arc_list_graph::ArcListGraph::new_labeled(graph.num_nodes(), edges);	20✔
149
150	drop(dirs);	8✔
151	Ok(Left(sorted))	4✔
152	}

vigna / webgraph-rs / 19076069272

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