23365769388

Committed 20 Mar 2026 10:52PM UTC coverage: 68.228% (-3.0%) from 71.245%

Build # 23365769388

Build Type

push

github

Committed by

vigna

Commit Message

No le_bins,be_bins for webgraph

Coverage Stats

6655 of 9754 relevant lines covered (68.23%)

46582760.24 hits per line

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

82.76

/cli/src/build/offsets.rs

/*
 * SPDX-FileCopyrightText: 2023 Inria
 * SPDX-FileCopyrightText: 2023 Tommaso Fontana
 * SPDX-FileCopyrightText: 2026 Sebastiano Vigna
 *
 * SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
 */

use crate::LogIntervalArg;
use anyhow::{Context, Result};
use clap::Parser;
use dsi_bitstream::{dispatch::factory::CodesReaderFactoryHelper, prelude::*};
use dsi_progress_logger::prelude::*;
use std::path::PathBuf;
use webgraph::prelude::*;

#[derive(Parser, Debug)]
#[command(name = "offsets", about = "Builds the offsets file of a graph.", long_about = None, next_line_help = true)]
pub struct CliArgs {
    /// The basename of the graph.​
    pub basename: PathBuf,

    #[clap(flatten)]
    pub log_interval: LogIntervalArg,
}

pub fn main(args: CliArgs) -> Result<()> {
    match get_endianness(&args.basename)?.as_str() {
        #[cfg(feature = "be_bins")]
        BE::NAME => build_offsets::<BE>(args),
        #[cfg(feature = "le_bins")]
        LE::NAME => build_offsets::<LE>(args),
        e => panic!("Unknown endianness: {}", e),
    }
}

pub fn build_offsets<E: Endianness + 'static>(args: CliArgs) -> Result<()>
where
    MmapHelper<u32>: CodesReaderFactoryHelper<E>,
    for<'a> LoadModeCodesReader<'a, E, Mmap>: BitSeek,
{
    // Creates the sequential iterator over the graph
    let seq_graph = BvGraphSeq::with_basename(&args.basename)
        .endianness::<E>()
        .load()?;
    let offsets = args.basename.with_extension(OFFSETS_EXTENSION);
    // create a bit writer on the file
    let mut writer = buf_bit_writer::from_path::<BE, usize>(&offsets)
        .with_context(|| format!("Could not create {}", offsets.display()))?;
    // progress bar
    let mut pl = progress_logger![
        display_memory = true,
        item_name = "node",
        expected_updates = Some(seq_graph.num_nodes()),
    ];
    if let Some(duration) = args.log_interval.log_interval {
        pl.log_interval(duration);
    }
    pl.start("Computing offsets...");
    // read the graph a write the offsets
    let mut offset = 0;
    let mut degs_iter = seq_graph.offset_deg_iter();
    for (new_offset, _degree) in &mut degs_iter {
        // write where
        writer
            .write_gamma((new_offset - offset) as _)
            .context("Could not write gamma")?;
        offset = new_offset;
        // decode the next nodes so we know where the next node_id starts
        pl.light_update();
    }
    // write the last offset, this is done to avoid decoding the last node
    writer
        .write_gamma((degs_iter.get_pos() - offset) as _)
        .context("Could not write final gamma")?;
    pl.light_update();
    pl.done();
    Ok(())
}

1	/*
2	* SPDX-FileCopyrightText: 2023 Inria
3	* SPDX-FileCopyrightText: 2023 Tommaso Fontana
4	* SPDX-FileCopyrightText: 2026 Sebastiano Vigna
5	*
6	* SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
7	*/
8
9	use crate::LogIntervalArg;
10	use anyhow::{Context, Result};
11	use clap::Parser;
12	use dsi_bitstream::{dispatch::factory::CodesReaderFactoryHelper, prelude::*};
13	use dsi_progress_logger::prelude::*;
14	use std::path::PathBuf;
15	use webgraph::prelude::*;
16
17	#[derive(Parser, Debug)]
18	#[command(name = "offsets", about = "Builds the offsets file of a graph.", long_about = None, next_line_help = true)]
19	pub struct CliArgs {
20	/// The basename of the graph.
21	pub basename: PathBuf,
22
23	#[clap(flatten)]
24	pub log_interval: LogIntervalArg,
25	}
26
27	pub fn main(args: CliArgs) -> Result<()> {	15✔
28	match get_endianness(&args.basename)?.as_str() {	45✔
29	#[cfg(feature = "be_bins")]
30	BE::NAME => build_offsets::<BE>(args),	45✔
31	#[cfg(feature = "le_bins")]
32	LE::NAME => build_offsets::<LE>(args),	×
33	e => panic!("Unknown endianness: {}", e),	×
34	}
35	}
36
37	pub fn build_offsets<E: Endianness + 'static>(args: CliArgs) -> Result<()>	15✔
38	where
39	MmapHelper<u32>: CodesReaderFactoryHelper<E>,
40	for<'a> LoadModeCodesReader<'a, E, Mmap>: BitSeek,
41	{
42	// Creates the sequential iterator over the graph
43	let seq_graph = BvGraphSeq::with_basename(&args.basename)	45✔
44	.endianness::<E>()
45	.load()?;
46	let offsets = args.basename.with_extension(OFFSETS_EXTENSION);	30✔
47	// create a bit writer on the file
48	let mut writer = buf_bit_writer::from_path::<BE, usize>(&offsets)	45✔
49	.with_context(\|\| format!("Could not create {}", offsets.display()))?;	15✔
50	// progress bar
51	let mut pl = progress_logger![	30✔
52	display_memory = true,	×
53	item_name = "node",	×
54	expected_updates = Some(seq_graph.num_nodes()),	15✔
55	];
56	if let Some(duration) = args.log_interval.log_interval {	15✔
57	pl.log_interval(duration);	×
58	}
59	pl.start("Computing offsets...");	30✔
60	// read the graph a write the offsets
61	let mut offset = 0;	30✔
62	let mut degs_iter = seq_graph.offset_deg_iter();	45✔
63	for (new_offset, _degree) in &mut degs_iter {	9,766,725✔
64	// write where
65	writer	4,883,355✔
66	.write_gamma((new_offset - offset) as _)	9,766,710✔
67	.context("Could not write gamma")?;
68	offset = new_offset;	4,883,355✔
69	// decode the next nodes so we know where the next node_id starts
70	pl.light_update();	9,766,710✔
71	}
72	// write the last offset, this is done to avoid decoding the last node
73	writer	15✔
74	.write_gamma((degs_iter.get_pos() - offset) as _)	45✔
75	.context("Could not write final gamma")?;
76	pl.light_update();	30✔
77	pl.done();	30✔
78	Ok(())	15✔
79	}

vigna / webgraph-rs / 23365769388

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