webgraph_cli/to/
bvgraph.rs1use crate::create_parent_dir;
9use crate::*;
10use anyhow::Result;
11use dsi_bitstream::dispatch::factory::CodesReaderFactoryHelper;
12use dsi_bitstream::prelude::*;
13
14use mmap_rs::MmapFlags;
15use std::path::PathBuf;
16use tempfile::Builder;
17use webgraph::prelude::*;
18
19#[derive(Parser, Debug)]
20#[command(name = "bvgraph", about = "Recompresses a BvGraph, possibly applying a permutation to its node identifiers.", long_about = None)]
21pub struct CliArgs {
22 pub src: PathBuf,
24 pub dst: PathBuf,
26
27 #[clap(flatten)]
28 pub num_threads: NumThreadsArg,
29
30 #[arg(long)]
31 pub permutation: Option<PathBuf>,
33
34 #[clap(flatten)]
35 pub memory_usage: MemoryUsageArg,
36
37 #[clap(flatten)]
38 pub ca: CompressArgs,
39}
40
41pub fn main(global_args: GlobalArgs, args: CliArgs) -> Result<()> {
42 create_parent_dir(&args.dst)?;
43
44 let permutation = if let Some(path) = args.permutation.as_ref() {
45 Some(JavaPermutation::mmap(path, MmapFlags::RANDOM_ACCESS)?)
46 } else {
47 None
48 };
49
50 let target_endianness = args.ca.endianness.clone();
51 match get_endianness(&args.src)?.as_str() {
52 #[cfg(feature = "be_bins")]
53 BE::NAME => compress::<BE>(global_args, args, target_endianness, permutation),
54 #[cfg(feature = "le_bins")]
55 LE::NAME => compress::<LE>(global_args, args, target_endianness, permutation),
56 e => panic!("Unknown endianness: {}", e),
57 }
58}
59
60pub fn compress<E: Endianness>(
61 _global_args: GlobalArgs,
62 args: CliArgs,
63 target_endianness: Option<String>,
64 permutation: Option<JavaPermutation>,
65) -> Result<()>
66where
67 MmapHelper<u32>: CodesReaderFactoryHelper<E>,
68 for<'a> LoadModeCodesReader<'a, E, Mmap>: BitSeek + Send + Sync + Clone,
69{
70 let dir = Builder::new().prefix("to_bvgraph_").tempdir()?;
71
72 let thread_pool = crate::get_thread_pool(args.num_threads.num_threads);
73 let chunk_size = args.ca.chunk_size;
74 let bvgraphz = args.ca.bvgraphz;
75 let mut builder = BvCompConfig::new(&args.dst)
76 .with_comp_flags(args.ca.into())
77 .with_tmp_dir(&dir);
78
79 if bvgraphz {
80 builder = builder.with_chunk_size(chunk_size);
81 }
82
83 if args.src.with_extension(EF_EXTENSION).exists() {
84 let graph = BvGraph::with_basename(&args.src).endianness::<E>().load()?;
85
86 if let Some(permutation) = permutation {
87 let memory_usage = args.memory_usage.memory_usage;
88 thread_pool.install(|| {
89 log::info!("Permuting graph with memory usage {}", memory_usage);
90 let start = std::time::Instant::now();
91 let sorted =
92 webgraph::transform::permute_split(&graph, &permutation, memory_usage)?;
93 log::info!(
94 "Permuted the graph. It took {:.3} seconds",
95 start.elapsed().as_secs_f64()
96 );
97 builder.par_comp_lenders_endianness(
98 &sorted,
99 sorted.num_nodes(),
100 &target_endianness.unwrap_or_else(|| BE::NAME.into()),
101 )
102 })?;
103 } else {
104 thread_pool.install(|| {
105 builder.par_comp_lenders_endianness(
106 &graph,
107 graph.num_nodes(),
108 &target_endianness.unwrap_or_else(|| BE::NAME.into()),
109 )
110 })?;
111 }
112 } else {
113 log::warn!(
114 "The .ef file does not exist. The graph will be read sequentially which will result in slower compression. If you can, run `webgraph build ef` before recompressing."
115 );
116 let seq_graph = BvGraphSeq::with_basename(&args.src)
117 .endianness::<E>()
118 .load()?;
119
120 if let Some(permutation) = permutation {
121 let memory_usage = args.memory_usage.memory_usage;
122
123 log::info!("Permuting graph with memory usage {}", memory_usage);
124 let start = std::time::Instant::now();
125 let permuted = webgraph::transform::permute(&seq_graph, &permutation, memory_usage)?;
126 log::info!(
127 "Permuted the graph. It took {:.3} seconds",
128 start.elapsed().as_secs_f64()
129 );
130
131 thread_pool.install(|| {
132 builder.par_comp_lenders_endianness(
133 &permuted,
134 permuted.num_nodes(),
135 &target_endianness.unwrap_or_else(|| BE::NAME.into()),
136 )
137 })?;
138 } else {
139 thread_pool.install(|| {
140 builder.par_comp_lenders_endianness(
141 &seq_graph,
142 seq_graph.num_nodes(),
143 &target_endianness.unwrap_or_else(|| BE::NAME.into()),
144 )
145 })?;
146 }
147 }
148 Ok(())
149}