refactored the CLI

Author: zommiommy

src/cli/optimize_codes.rs renamed to src/cli/analyze/codes.rs

@@ -13,13 +13,13 @@ use dsi_progress_logger::prelude::*;
 use lender::*;
 use std::path::PathBuf;
 
-pub const COMMAND_NAME: &str = "optimize-codes";
+pub const COMMAND_NAME: &str = "codes";
 
 #[derive(Args, Debug)]
 #[command(about = "Reads a graph and suggests the best codes to use.", long_about = None)]
 pub struct CliArgs {
     /// The basename of the graph.
-    basename: PathBuf,
+    pub src: PathBuf,
 }
 
 pub fn cli(command: Command) -> Command {
@@ -29,7 +29,7 @@ pub fn cli(command: Command) -> Command {
 pub fn main(submatches: &ArgMatches) -> Result<()> {
     let args = CliArgs::from_arg_matches(submatches)?;
 
-    match get_endianness(&args.basename)?.as_str() {
+    match get_endianness(&args.src)?.as_str() {
         #[cfg(any(
             feature = "be_bins",
             not(any(feature = "be_bins", feature = "le_bins"))
@@ -49,7 +49,7 @@ where
     for<'a> BufBitReader<E, MemWordReader<u32, &'a [u32]>>: CodeRead<E> + BitSeek,
 {
     // TODO!: speed it up by using random access graph if possible
-    let graph = BVGraphSeq::with_basename(args.basename)
+    let graph = BVGraphSeq::with_basename(args.src)
         .endianness::<E>()
         .load()?
         .map_factory(StatsDecoderFactory::new);

src/cli/analyze/mod.rs

@@ -0,0 +1,36 @@
+/*
+ * SPDX-FileCopyrightText: 2024 Tommaso Fontana
+ *
+ * SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
+ */
+
+use anyhow::Result;
+use clap::{ArgMatches, Command};
+
+pub mod codes;
+
+pub const COMMAND_NAME: &str = "analyze";
+
+pub fn cli(command: Command) -> Command {
+    let sub_command = Command::new(COMMAND_NAME)
+        .about("Compute statistics on a BVGraph.")
+        .subcommand_required(true)
+        .arg_required_else_help(true)
+        .allow_external_subcommands(true);
+    let sub_command = codes::cli(sub_command);
+    command.subcommand(sub_command)
+}
+
+pub fn main(submatches: &ArgMatches) -> Result<()> {
+    match submatches.subcommand() {
+        Some((codes::COMMAND_NAME, sub_m)) => codes::main(sub_m),
+        Some((command_name, _)) => {
+            eprintln!("Unknown command: {:?}", command_name);
+            std::process::exit(1);
+        }
+        None => {
+            eprintln!("No command given for analyze");
+            std::process::exit(1);
+        }
+    }
+}

src/cli/bench/bf_visit.rs

@@ -21,13 +21,13 @@ pub const COMMAND_NAME: &str = "bf-visit";
 #[command(about = "Breadth-first visits a graph.", long_about = None)]
 struct CliArgs {
     /// The basename of the graph.
-    basename: PathBuf,
+    pub src: PathBuf,
     /// Static dispatch (default BVGraph parameters).
     #[arg(short = 's', long = "static")]
-    _static: bool,
+    pub _static: bool,
     /// Static dispatch (default BVGraph parameters).
     #[arg(short = 'r', long, default_value_t = 1)]
-    repeats: usize,
+    pub repeats: usize,
 }
 
 pub fn cli(command: Command) -> Command {
@@ -37,12 +37,12 @@ pub fn cli(command: Command) -> Command {
 pub fn main(submatches: &ArgMatches) -> Result<()> {
     let args = CliArgs::from_arg_matches(submatches)?;
 
-    let config = BVGraph::with_basename(&args.basename)
+    let config = BVGraph::with_basename(&args.src)
         .mode::<Mmap>()
         .flags(MemoryFlags::TRANSPARENT_HUGE_PAGES | MemoryFlags::RANDOM_ACCESS);
 
     for _ in 0..args.repeats {
-        match get_endianness(&args.basename)?.as_str() {
+        match get_endianness(&args.src)?.as_str() {
             #[cfg(any(
                 feature = "be_bins",
                 not(any(feature = "be_bins", feature = "le_bins"))

src/cli/bench/bvgraph.rs

@@ -23,31 +23,31 @@ pub const COMMAND_NAME: &str = "bvgraph";
 #[command(about = "Benchmarks the Rust BVGraph implementation.", long_about = None)]
 struct CliArgs {
     /// The basename of the graph.
-    basename: PathBuf,
+    pub src: PathBuf,
 
     /// Perform a random-access test on this number of randomly chosen nodes.
     #[arg(short, long)]
-    random: Option<usize>,
+    pub random: Option<usize>,
 
     /// The number of repeats.
     #[arg(short = 'R', long, default_value = "10")]
-    repeats: usize,
+    pub repeats: usize,
 
     /// In random-access test, test just access to the first successor.
     #[arg(short = 'f', long)]
-    first: bool,
+    pub first: bool,
 
     /// Static dispatch for speed tests (default BVGraph parameters).
     #[arg(short = 's', long = "static")]
-    _static: bool,
+    pub _static: bool,
 
     /// Test sequential high-speed offset/degree scanning.
     #[arg(short = 'd', long)]
-    degrees: bool,
+    pub degrees: bool,
 
     /// Do not test speed, but check that the sequential and random-access successor lists are the same.
     #[arg(short = 'c', long)]
-    check: bool,
+    pub check: bool,
 }
 
 pub fn cli(command: Command) -> Command {
@@ -57,7 +57,7 @@ pub fn cli(command: Command) -> Command {
 pub fn main(submatches: &ArgMatches) -> Result<()> {
     let args = CliArgs::from_arg_matches(submatches)?;
 
-    match get_endianness(&args.basename)?.as_str() {
+    match get_endianness(&args.src)?.as_str() {
         #[cfg(any(
             feature = "be_bins",
             not(any(feature = "be_bins", feature = "le_bins"))
@@ -137,11 +137,9 @@ where
     for<'a> BufBitReader<E, MemWordReader<u32, &'a [u32]>>: CodeRead<E> + BitSeek,
 {
     if args.check {
-        let graph = BVGraph::with_basename(&args.basename)
-            .endianness::<E>()
-            .load()?;
+        let graph = BVGraph::with_basename(&args.src).endianness::<E>().load()?;
 
-        let seq_graph = BVGraphSeq::with_basename(&args.basename)
+        let seq_graph = BVGraphSeq::with_basename(&args.src)
             .endianness::<E>()
             .load()?;
 
@@ -155,7 +153,7 @@ where
             assert_eq!(succ.collect_vec(), seq_succ.collect_vec());
         }];
     } else if args.degrees {
-        let seq_graph = BVGraphSeq::with_basename(&args.basename)
+        let seq_graph = BVGraphSeq::with_basename(&args.src)
             .endianness::<E>()
             .load()?;
 
@@ -181,7 +179,7 @@ where
         ) {
             (Some(samples), true) => {
                 bench_random(
-                    BVGraph::with_basename(&args.basename)
+                    BVGraph::with_basename(&args.src)
                         .endianness::<E>()
                         .dispatch::<Dynamic>()
                         .mode::<Mmap>()
@@ -194,7 +192,7 @@ where
             }
             (Some(samples), false) => {
                 bench_random(
-                    BVGraph::with_basename(&args.basename)
+                    BVGraph::with_basename(&args.src)
                         .endianness::<E>()
                         .dispatch::<Static>()
                         .mode::<Mmap>()
@@ -207,7 +205,7 @@ where
             }
             (None, true) => {
                 bench_seq(
-                    BVGraphSeq::with_basename(&args.basename)
+                    BVGraphSeq::with_basename(&args.src)
                         .endianness::<E>()
                         .dispatch::<Dynamic>()
                         .mode::<Mmap>()
@@ -218,7 +216,7 @@ where
             }
             (None, false) => {
                 bench_seq(
-                    BVGraphSeq::with_basename(&args.basename)
+                    BVGraphSeq::with_basename(&args.src)
                         .endianness::<E>()
                         .dispatch::<Static>()
                         .mode::<Mmap>()

src/cli/build/dcf.rs

@@ -23,7 +23,7 @@ pub const COMMAND_NAME: &str = "dcf";
 #[command(about = "Builds an Elias–Fano representation of the degree cumulative function of a graph.", long_about = None)]
 pub struct CliArgs {
     /// The basename of the graph.
-    pub basename: PathBuf,
+    pub src: PathBuf,
 }
 
 pub fn cli(command: Command) -> Command {
@@ -33,7 +33,7 @@ pub fn cli(command: Command) -> Command {
 pub fn main(submatches: &ArgMatches) -> Result<()> {
     let args = CliArgs::from_arg_matches(submatches)?;
 
-    match get_endianness(&args.basename)?.as_str() {
+    match get_endianness(&args.src)?.as_str() {
         #[cfg(any(
             feature = "be_bins",
             not(any(feature = "be_bins", feature = "le_bins"))
@@ -52,7 +52,7 @@ pub fn build_dcf<E: Endianness + 'static>(args: CliArgs) -> Result<()>
 where
     for<'a> BufBitReader<E, MemWordReader<u32, &'a [u32]>>: CodeRead<E> + BitSeek,
 {
-    let basename = args.basename;
+    let basename = args.src;
     let properties_path = basename.with_extension(PROPERTIES_EXTENSION);
     let f = File::open(&properties_path).with_context(|| {
         format!(

src/cli/build/ef.rs

@@ -23,7 +23,7 @@ pub const COMMAND_NAME: &str = "ef";
 #[command(about = "Builds the .ef file for a graph.", long_about = None)]
 pub struct CliArgs {
     /// The basename of the graph.
-    pub basename: PathBuf,
+    pub src: PathBuf,
     /// The number of elements to be inserted in the Elias-Fano
     /// starting from a label offset file. It is usually one more than
     /// the number of nodes in the graph.
@@ -37,7 +37,7 @@ pub fn cli(command: Command) -> Command {
 pub fn main(submatches: &ArgMatches) -> Result<()> {
     let args = CliArgs::from_arg_matches(submatches)?;
 
-    match get_endianness(&args.basename)?.as_str() {
+    match get_endianness(&args.src)?.as_str() {
         #[cfg(any(
             feature = "be_bins",
             not(any(feature = "be_bins", feature = "le_bins"))
@@ -56,7 +56,7 @@ pub fn build_eliasfano<E: Endianness + 'static>(args: CliArgs) -> Result<()>
 where
     for<'a> BufBitReader<E, MemWordReader<u32, &'a [u32]>>: CodeRead<E> + BitSeek,
 {
-    let basename = args.basename;
+    let basename = args.src;
     if let Some(num_nodes) = args.n {
         // Horribly temporary duplicated code for the case of label offsets.
         let of_file_path = basename.with_extension(LABELOFFSETS_EXTENSION);

src/cli/build/offsets.rs

@@ -23,7 +23,7 @@ pub const COMMAND_NAME: &str = "offsets";
 
 pub struct CliArgs {
     /// The basename of the graph.
-    pub basename: PathBuf,
+    pub src: PathBuf,
 }
 
 pub fn cli(command: Command) -> Command {
@@ -33,7 +33,7 @@ pub fn cli(command: Command) -> Command {
 pub fn main(submatches: &ArgMatches) -> Result<()> {
     let args = CliArgs::from_arg_matches(submatches)?;
 
-    match get_endianness(&args.basename)?.as_str() {
+    match get_endianness(&args.src)?.as_str() {
         #[cfg(any(
             feature = "be_bins",
             not(any(feature = "be_bins", feature = "le_bins"))
@@ -54,10 +54,10 @@ where
     for<'a> BufBitReader<E, WordAdapter<u32, BufReader<File>>>: CodeRead<E> + BitSeek,
 {
     // Create the sequential iterator over the graph
-    let seq_graph = BVGraphSeq::with_basename(&args.basename)
+    let seq_graph = BVGraphSeq::with_basename(&args.src)
         .endianness::<E>()
         .load()?;
-    let offsets = args.basename.with_extension(OFFSETS_EXTENSION);
+    let offsets = args.src.with_extension(OFFSETS_EXTENSION);
     let file = std::fs::File::create(&offsets)
         .with_context(|| format!("Could not create {}", offsets.display()))?;
     // create a bit writer on the file

src/cli/check_ef.rs renamed to src/cli/check/ef.rs

@@ -26,7 +26,7 @@ pub const COMMAND_NAME: &str = "check-ef";
 #[command(about = "Check that the '.ef' file (and `.offsets` if present) is coherent with the graph.", long_about = None)]
 pub struct CliArgs {
     /// The basename of the graph.
-    basename: PathBuf,
+    pub src: PathBuf,
 }
 
 pub fn cli(command: Command) -> Command {
@@ -38,7 +38,7 @@ pub fn main(submatches: &ArgMatches) -> Result<()> {
 }
 
 pub fn check_ef(args: CliArgs) -> Result<()> {
-    let properties_path = args.basename.with_extension(PROPERTIES_EXTENSION);
+    let properties_path = args.src.with_extension(PROPERTIES_EXTENSION);
     let f = File::open(&properties_path).with_context(|| {
         format!(
             "Could not load properties file: {}",
@@ -49,9 +49,9 @@ pub fn check_ef(args: CliArgs) -> Result<()> {
     let num_nodes = map.get("nodes").unwrap().parse::<usize>()?;
 
     // Create the offsets file
-    let of_file_path = args.basename.with_extension(OFFSETS_EXTENSION);
+    let of_file_path = args.src.with_extension(OFFSETS_EXTENSION);
 
-    let ef = EF::mmap(args.basename.with_extension(EF_EXTENSION), Flags::default())?;
+    let ef = EF::mmap(args.src.with_extension(EF_EXTENSION), Flags::default())?;
 
     let mut pl = ProgressLogger::default();
     pl.display_memory(true)
@@ -85,7 +85,7 @@ pub fn check_ef(args: CliArgs) -> Result<()> {
         .expected_updates(Some(num_nodes));
 
     info!("The offsets file does not exists, reading the graph to build Elias-Fano");
-    let seq_graph = crate::graphs::bvgraph::sequential::BVGraphSeq::with_basename(&args.basename)
+    let seq_graph = crate::graphs::bvgraph::sequential::BVGraphSeq::with_basename(&args.src)
         .endianness::<BE>()
         .load()?;
     // otherwise directly read the graph

src/cli/check/mod.rs

@@ -0,0 +1,36 @@
+/*
+ * SPDX-FileCopyrightText: 2024 Tommaso Fontana
+ *
+ * SPDX-License-Identifier: Apache-2.0 OR LGPL-2.1-or-later
+ */
+
+use anyhow::Result;
+use clap::{ArgMatches, Command};
+
+pub mod ef;
+
+pub const COMMAND_NAME: &str = "check";
+
+pub fn cli(command: Command) -> Command {
+    let sub_command = Command::new(COMMAND_NAME)
+        .about("Check coherence of files.")
+        .subcommand_required(true)
+        .arg_required_else_help(true)
+        .allow_external_subcommands(true);
+    let sub_command = ef::cli(sub_command);
+    command.subcommand(sub_command)
+}
+
+pub fn main(submatches: &ArgMatches) -> Result<()> {
+    match submatches.subcommand() {
+        Some((ef::COMMAND_NAME, sub_m)) => ef::main(sub_m),
+        Some((command_name, _)) => {
+            eprintln!("Unknown command: {:?}", command_name);
+            std::process::exit(1);
+        }
+        None => {
+            eprintln!("No command given for check");
+            std::process::exit(1);
+        }
+    }
+}