14130263389

Committed 28 Mar 2025 01:40PM UTC coverage: 49.654% (-0.1%) from 49.798%

Build # 14130263389

Build Type

push

github

Committed by

zommiommy

Commit Message

fixed llp wrong combine types

Run Details

22 of 41 new or added lines in 4 files covered. (53.66%)

1019 existing lines in 41 files now uncovered.

2437 of 4908 relevant lines covered (49.65%)

18919274.03 hits per line

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

49.12

/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 anyhow::{Context, Result};
use dsi_progress_logger::prelude::*;
use itertools::{Dedup, 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).
#[allow(clippy::type_complexity)]
pub fn simplify_sorted<G: SequentialGraph>(
    graph: G,
    batch_size: usize,
) -> 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, batch_size).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).
#[allow(clippy::type_complexity)]
pub fn simplify(
    graph: &impl SequentialGraph,
    batch_size: usize,
) -> Result<
    Left<
        arc_list_graph::ArcListGraph<
            std::iter::Map<
                Dedup<
                    core::iter::Filter<
                        core::iter::Map<
                            KMergeIters<BatchIterator<()>>,
                            fn((usize, usize, ())) -> (usize, usize),
                        >,
                        fn(&(usize, usize)) -> bool,
                    >,
                >,
                fn((usize, usize)) -> (usize, usize, ()),
            >,
        >,
    >,
> {
    let dir = Builder::new().prefix("simplify_").tempdir()?;
    let mut sorted = SortPairs::new(batch_size, 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) = |(src, dst, _)| (src, dst);
    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(Left(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).
#[allow(clippy::type_complexity)]
pub fn simplify_split<S>(
    graph: &S,
    batch_size: usize,
    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(batch_size / num_threads, 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 {} finished", thread_id);
            });
            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 anyhow::{Context, Result};
14	use dsi_progress_logger::prelude::*;
15	use itertools::{Dedup, Itertools};
16	use lender::*;
17	use rayon::ThreadPool;
18	use tempfile::Builder;
19
20	use super::transpose;
21
22	/// Returns a simplified (i.e., undirected and loopless) version of the provided
23	/// sorted (both on nodes and successors) graph as a [sequential
24	/// graph](crate::traits::SequentialGraph).
25	///
26	/// This method exploits the fact that the input graph is already sorted,
27	/// sorting half the number of arcs of
28	/// [`simplify`](crate::transform::simplify::simplify).
29	#[allow(clippy::type_complexity)]
30	pub fn simplify_sorted<G: SequentialGraph>(	1✔
31	graph: G,
32	batch_size: usize,
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, batch_size).context("Could not transpose the graph")?;	2✔
43	Ok(NoSelfLoopsGraph(UnionGraph(graph, transpose)))	×
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	#[allow(clippy::type_complexity)]
UNCOV 55	pub fn simplify(	×
56	graph: &impl SequentialGraph,
57	batch_size: usize,
58	) -> Result<
59	Left<
60	arc_list_graph::ArcListGraph<
61	std::iter::Map<
62	Dedup<
63	core::iter::Filter<
64	core::iter::Map<
65	KMergeIters<BatchIterator<()>>,
66	fn((usize, usize, ())) -> (usize, usize),
67	>,
68	fn(&(usize, usize)) -> bool,
69	>,
70	>,
71	fn((usize, usize)) -> (usize, usize, ()),
72	>,
73	>,
74	>,
75	> {
UNCOV 76	let dir = Builder::new().prefix("simplify_").tempdir()?;	×
UNCOV 77	let mut sorted = SortPairs::new(batch_size, dir.path())?;	×
78
79	let mut pl = ProgressLogger::default();	×
80	pl.item_name("node")	×
81	.expected_updates(Some(graph.num_nodes()));	×
82	pl.start("Creating batches...");	×
83	// create batches of sorted edges
84	let mut iter = graph.iter();	×
UNCOV 85	while let Some((src, succ)) = iter.next() {	×
UNCOV 86	for dst in succ {	×
87	if src != dst {	×
UNCOV 88	sorted.push(src, dst)?;	×
UNCOV 89	sorted.push(dst, src)?;	×
90	}
91	}
UNCOV 92	pl.light_update();	×
93	}
94	// merge the batches
UNCOV 95	let map: fn((usize, usize, ())) -> (usize, usize) = \|(src, dst, _)\| (src, dst);	×
UNCOV 96	let filter: fn(&(usize, usize)) -> bool = \|(src, dst)\| src != dst;	×
UNCOV 97	let iter = Itertools::dedup(sorted.iter()?.map(map).filter(filter));	×
98	let sorted = arc_list_graph::ArcListGraph::new(graph.num_nodes(), iter);	×
99	pl.done();	×
100
101	Ok(Left(sorted))	×
102	}
103
104	/// Returns a simplified (i.e., undirected and loopless) version of the provided
105	/// graph as a [sequential graph](crate::traits::SequentialGraph).
106	///
107	/// This method uses splitting to sort in parallel different parts of the graph.
108	///
109	/// For the meaning of the additional parameter, see
110	/// [`SortPairs`](crate::prelude::sort_pairs::SortPairs).
111	#[allow(clippy::type_complexity)]
112	pub fn simplify_split<S>(	2✔
113	graph: &S,
114	batch_size: usize,
115	threads: &ThreadPool,
116	) -> Result<Left<arc_list_graph::ArcListGraph<itertools::Dedup<KMergeIters<BatchIterator<()>, ()>>>>>
117	where
118	S: SequentialGraph + SplitLabeling,
119	{
120	let num_threads = threads.current_num_threads();	2✔
121	let (tx, rx) = std::sync::mpsc::channel();	2✔
122
123	let mut dirs = vec![];	2✔
124
125	threads.in_place_scope(\|scope\| {	4✔
126	let mut thread_id = 0;	2✔
127	#[allow(clippy::explicit_counter_loop)] // enumerate requires some extra bounds here
128	for iter in graph.split_iter(num_threads) {	10✔
129	let tx = tx.clone();	×
130	let dir = Builder::new()	×
131	.prefix(&format!("simplify_split_{}_", thread_id))	×
132	.tempdir()	×
133	.expect("Could not create a temporary directory");	×
134	let dir_path = dir.path().to_path_buf();	×
135	dirs.push(dir);	×
136	scope.spawn(move \|_\| {	8✔
137	log::debug!("Spawned thread {}", thread_id);	8✔
138	let mut sorted = SortPairs::new(batch_size / num_threads, dir_path).unwrap();	8✔
139	for_!( (src, succ) in iter {	651,122✔
140	for dst in succ {	13,515,722✔
141	if src != dst {	6,257,420✔
142	sorted.push(src, dst).unwrap();	6,257,420✔
143	sorted.push(dst, src).unwrap();	6,257,420✔
144	}
145	}
146	});
147	let result = sorted.iter().context("Could not read arcs").unwrap();	8✔
148	tx.send(result).expect("Could not send the sorted pairs");	8✔
149	log::debug!("Thread {} finished", thread_id);	8✔
150	});
151	thread_id += 1;	×
152	}
153	});
154	drop(tx);	2✔
155
156	// get a graph on the sorted data
157	log::debug!("Waiting for threads to finish");	2✔
158	let edges: KMergeIters<BatchIterator> = rx.iter().sum();	2✔
159	let edges = edges.dedup();	2✔
160	log::debug!("All threads finished");	2✔
161	let sorted = arc_list_graph::ArcListGraph::new_labeled(graph.num_nodes(), edges);	2✔
162
163	drop(dirs);	2✔
164	Ok(Left(sorted))	2✔
165	}

vigna / webgraph-rs / 14130263389

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