refactor: benchmarks

Cargo.toml

@@ -46,4 +46,5 @@ expect_used = "warn"
 
 [[bench]]
 name = "benchmark"
+path = "benches/benchmark.rs"
 harness = false

benches/benchmark.rs

@@ -1,145 +1,23 @@
-use anyhow::Result;
+mod common;
+mod scenarios;
+
+use common::setup::init_temp_dir;
 use criterion::{Criterion, criterion_group, criterion_main};
-use csv::Reader;
-use serde::{Deserialize, Deserializer};
-use std::{
-    fs::{File, create_dir_all},
-    path::{Path, PathBuf},
-    process::Command,
-    str::FromStr,
-    u64,
+use scenarios::{
+    basic::benchmark_basic, edge_cases::benchmark_edge_cases,
+    specific::benchmark_specific_scenarios,
 };
-use tempfile::TempDir;
+use std::time::Duration;
 
-const BASE_DIR: &str = env!("CARGO_MANIFEST_DIR");
-
-#[allow(dead_code)]
-#[derive(Debug, Clone, Default, Deserialize)]
-struct Size(u64);
-
-#[allow(dead_code)]
-#[derive(Debug, Clone, Default, Deserialize)]
-struct Modified(u64);
-
-#[allow(dead_code)]
-#[derive(Debug, Clone, Deserialize)]
-struct Permissions(u32);
-
-impl Default for Permissions {
-    fn default() -> Self {
-        Self(644)
-    }
-}
-
-#[derive(Debug, Clone, PartialEq)]
-enum EntryType {
-    Dir,
-    File,
-    Symlink,
-    Other(String),
-}
-
-impl FromStr for EntryType {
-    type Err = String;
-
-    fn from_str(s: &str) -> std::result::Result<Self, Self::Err> {
-        match s {
-            "dir" => Ok(Self::Dir),
-            "file" => Ok(Self::File),
-            "symlink" => Ok(Self::Symlink),
-            other if other.is_empty() => Err("Empty entry type".to_string()),
-            other => Ok(Self::Other(other.into())),
-        }
-    }
-}
-
-impl<'de> Deserialize<'de> for EntryType {
-    fn deserialize<D>(deserializer: D) -> std::result::Result<Self, D::Error>
-    where
-        D: Deserializer<'de>,
-    {
-        let s = String::deserialize(deserializer)?;
-        Self::from_str(&s).map_err(serde::de::Error::custom)
-    }
-}
-
-#[allow(dead_code)]
-#[derive(Debug, Clone, Deserialize)]
-struct FileEntry {
-    #[serde(rename = "type")]
-    entry_type: EntryType,
-    directory: PathBuf,
-    path: PathBuf,
-    #[serde(default)]
-    size: Size,
-    #[serde(default)]
-    modified: Modified,
-    #[serde(default)]
-    permissions: Permissions,
-}
-
-impl FileEntry {
-    fn to_tempfile(&self, base: &Path) -> Result<()> {
-        let full_path = base.join(&self.path);
-        match self.entry_type {
-            EntryType::Dir => create_dir(&full_path),
-            EntryType::File => create_file(&full_path),
-            EntryType::Symlink => Ok(()),
-            EntryType::Other(_) => Ok(()),
-        }
-    }
-}
-
-fn create_file(path: &Path) -> Result<()> {
-    if let Some(parent) = path.parent() {
-        create_dir(parent)?;
-    }
-    File::create(path)?;
-    Ok(())
-}
-
-fn create_dir(path: &Path) -> Result<()> {
-    create_dir_all(path)?;
-    Ok(())
-}
-
-fn process_directory(path: &Path) {
-    let binary_path = PathBuf::from(BASE_DIR).join("target/release/project-finder");
-    Command::new(binary_path)
-        .arg(path)
-        .output()
-        .expect("failed to run binary");
-}
-
-fn setup_entries() -> Result<TempDir> {
-    let temp_dir = TempDir::new()?;
-    println!("Temporary directory: {:?}", temp_dir.path());
-
-    let csv_path = PathBuf::from(BASE_DIR)
-        .join("benches/fixtures")
-        .join("snapshot-2025-03-31_09-20-03.csv");
-    let mut rdr = Reader::from_path(csv_path)?;
-
-    rdr.deserialize::<FileEntry>()
-        .for_each(|entry| match entry {
-            Ok(entry) => {
-                if let Err(e) = entry.to_tempfile(temp_dir.path()) {
-                    // eprintln!("Error processing entry: {}", e);
-                }
-            }
-            Err(e) => eprintln!("Failed to deserialize entry: {}", e),
-        });
-
-    Ok(temp_dir)
-}
-
-fn benchmark_processing(c: &mut Criterion) {
-    let temp_dir = setup_entries().expect("Failed to setup file entries");
-
-    c.bench_function("process_directory", |b| {
-        b.iter(|| process_directory(temp_dir.path()))
-    });
-}
-
-criterion_group!(benches, benchmark_processing);
+criterion_group!(
+    name = benches;
+    config = {
+        let c = Criterion::default()
+            .sample_size(10)
+            .measurement_time(Duration::from_secs(30));
+        init_temp_dir();
+        c
+    };
+    targets = benchmark_basic, benchmark_edge_cases, benchmark_specific_scenarios
+);
 criterion_main!(benches);

benches/common/mod.rs

@@ -0,0 +1,2 @@
+pub mod setup;
+pub mod utils;

benches/common/setup.rs

@@ -0,0 +1,188 @@
+use crate::common::utils::BASE_DIR;
+use anyhow;
+use csv::Reader;
+use regex::Regex;
+use serde::{Deserialize, Deserializer};
+use std::{
+    fs::{self, File, create_dir_all},
+    path::{Path, PathBuf},
+    str::FromStr,
+    sync::OnceLock,
+};
+use tempfile::TempDir;
+
+pub static TEMP_DIR: OnceLock<TempDir> = OnceLock::new();
+
+pub fn init_temp_dir() {
+    TEMP_DIR.get_or_init(|| setup_entries().expect("Failed to setup test directory"));
+}
+
+#[derive(Debug, Clone)]
+pub struct BenchParams {
+    pub depth: usize,
+    pub max_results: usize,
+    pub verbose: bool,
+}
+
+#[allow(dead_code)]
+#[derive(Debug, Clone, Deserialize)]
+struct FileEntry {
+    #[serde(rename = "type")]
+    entry_type: EntryType,
+    directory: PathBuf,
+    path: PathBuf,
+    #[serde(default)]
+    size: Size,
+    #[serde(default)]
+    modified: Modified,
+    #[serde(default)]
+    permissions: Permissions,
+}
+
+#[allow(dead_code)]
+#[derive(Debug, Clone, Default, Deserialize)]
+struct Size(#[serde(deserialize_with = "deserialize_u64_from_empty")] u64);
+
+#[allow(dead_code)]
+#[derive(Debug, Clone, Default, Deserialize)]
+struct Modified(#[serde(deserialize_with = "deserialize_u64_from_empty")] u64);
+
+#[allow(dead_code)]
+#[derive(Debug, Clone, Deserialize)]
+struct Permissions(#[serde(deserialize_with = "deserialize_u16_from_empty")] u16);
+
+pub fn setup_entries() -> anyhow::Result<TempDir> {
+    let temp_dir = TempDir::new()?;
+    println!("Temporary directory: {:?}", temp_dir.path());
+
+    let fixtures_dir = PathBuf::from(BASE_DIR).join("benches/fixtures");
+
+    let snapshot_path = last_snaphow_file(&fixtures_dir)?;
+
+    let mut rdr = Reader::from_path(snapshot_path)?;
+
+    rdr.deserialize::<FileEntry>()
+        .for_each(|entry| match entry {
+            Ok(entry) => {
+                if let Err(e) = entry.to_tempfile(temp_dir.path()) {
+                    eprintln!("Error processing entry: {}", e);
+                }
+            }
+            Err(e) => eprintln!("Failed to deserialize entry: {}", e),
+        });
+
+    Ok(temp_dir)
+}
+
+fn last_snaphow_file(dir: &Path) -> anyhow::Result<PathBuf> {
+    let re = Regex::new(r"^snapshot-(\d{4})-(\d{2})-(\d{2})_(\d{2})-(\d{2})-(\d{2})\.csv$")?;
+    let mut snapshots = fs::read_dir(dir)?
+        .filter_map(|entry| {
+            entry.ok().and_then(|entry| {
+                let file_name = entry.file_name();
+
+                if let Some(caps) = re.captures(&file_name.to_string_lossy()) {
+                    let [y, m, d, h, min, s] = (1..=6)
+                        .filter_map(|i| caps.get(i)?.as_str().parse().ok())
+                        .collect::<Vec<u32>>()
+                        .try_into()
+                        .ok()?;
+                    return Some(((y, m, d, h, min, s), entry.path()));
+                }
+                None
+            })
+        })
+        .collect::<Vec<_>>();
+
+    snapshots.sort_by_key(|(timestamp, _)| *timestamp);
+
+    snapshots
+        .last()
+        .map(|(_, path)| path.clone())
+        .ok_or_else(|| anyhow::anyhow!("No snapshot files found in directory"))
+}
+
+fn deserialize_u64_from_empty<'de, D>(deserializer: D) -> Result<u64, D::Error>
+where
+    D: Deserializer<'de>,
+{
+    let s = String::deserialize(deserializer)?;
+    if s.trim().is_empty() {
+        return Ok(0);
+    }
+    s.parse().map_err(serde::de::Error::custom)
+}
+
+fn deserialize_u16_from_empty<'de, D>(deserializer: D) -> Result<u16, D::Error>
+where
+    D: Deserializer<'de>,
+{
+    let s = String::deserialize(deserializer)?;
+    if s.trim().is_empty() {
+        return Ok(644);
+    }
+    s.parse().map_err(serde::de::Error::custom)
+}
+
+impl Default for Permissions {
+    fn default() -> Self {
+        Self(644)
+    }
+}
+
+#[derive(Debug, Clone, PartialEq)]
+enum EntryType {
+    Dir,
+    File,
+    Symlink,
+    Other(String),
+}
+
+impl FromStr for EntryType {
+    type Err = String;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        match s {
+            "dir" => Ok(Self::Dir),
+            "file" => Ok(Self::File),
+            "symlink" => Ok(Self::Symlink),
+            other if other.is_empty() => Err("Empty entry type".to_string()),
+            other => Ok(Self::Other(other.into())),
+        }
+    }
+}
+
+impl<'de> Deserialize<'de> for EntryType {
+    fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
+    where
+        D: Deserializer<'de>,
+    {
+        let s = String::deserialize(deserializer)?;
+        Self::from_str(&s).map_err(serde::de::Error::custom)
+    }
+}
+
+impl FileEntry {
+    fn to_tempfile(&self, base: &Path) -> anyhow::Result<()> {
+        let full_path = base.join(&self.path);
+        match self.entry_type {
+            EntryType::Dir => create_dir(&full_path),
+            EntryType::File => create_file(&full_path),
+            EntryType::Symlink => Ok(()),
+            EntryType::Other(_) => Ok(()),
+        }
+    }
+}
+
+fn create_file(path: &Path) -> anyhow::Result<()> {
+    if let Some(parent) = path.parent() {
+        create_dir(parent)?;
+    }
+    File::create(path)?;
+    Ok(())
+}
+
+fn create_dir(path: &Path) -> anyhow::Result<()> {
+    create_dir_all(path)?;
+    Ok(())
+}

benches/common/utils.rs

@@ -0,0 +1,62 @@
+use std::{
+    path::{Path, PathBuf},
+    process::Command,
+};
+
+pub const BASE_DIR: &str = env!("CARGO_MANIFEST_DIR");
+
+pub fn run_binary_with_args(
+    path: &Path,
+    depth: usize,
+    max_results: usize,
+    verbose: bool,
+) -> anyhow::Result<()> {
+    let binary_path = PathBuf::from(BASE_DIR).join("target/release/project-finder");
+
+    if !binary_path.exists() {
+        return Err(anyhow::anyhow!(
+            "Binary not found at {}. Did you run `cargo build --release`?",
+            binary_path.display()
+        ));
+    }
+
+    let mut cmd = Command::new(&binary_path);
+
+    // Add the path to search
+    cmd.arg(path);
+
+    // Add depth parameter
+    cmd.arg("--depth").arg(depth.to_string());
+
+    // Add max_results parameter if not zero
+    if max_results > 0 {
+        cmd.arg("--max-results").arg(max_results.to_string());
+    }
+
+    // Add verbose flag if true
+    if verbose {
+        cmd.arg("--verbose");
+    }
+
+    let output = cmd.output().map_err(|e| {
+        anyhow::anyhow!("Failed to execute binary {}: {}", binary_path.display(), e)
+    })?;
+
+    if !output.status.success() {
+        let stderr = String::from_utf8_lossy(&output.stderr);
+        return Err(anyhow::anyhow!(
+            "Process failed with status: {}\nStderr: {}",
+            output.status,
+            stderr
+        ));
+    }
+
+    Ok(())
+}
+pub fn create_deep_directory(base: &Path, depth: usize) -> anyhow::Result<()> {
+    todo!()
+}
+
+pub fn create_wide_directory(base: &Path, width: usize) -> anyhow::Result<()> {
+    todo!()
+}

benches/scenarios/basic.rs

@@ -0,0 +1,60 @@
+use crate::common::{
+    setup::{BenchParams, TEMP_DIR, init_temp_dir, setup_entries},
+    utils::{BASE_DIR, run_binary_with_args},
+};
+use criterion::{BenchmarkId, Criterion};
+use std::{
+    path::{Path, PathBuf},
+    process::Command,
+};
+
+fn process_directory(path: &Path) {
+    let binary_path = PathBuf::from(BASE_DIR).join("target/release/project-finder");
+    Command::new(binary_path)
+        .arg(path)
+        .output()
+        .expect("failed to run binary");
+}
+
+pub fn benchmark_basic(c: &mut Criterion) {
+    init_temp_dir();
+    let temp_dir = TEMP_DIR.get().unwrap().path();
+
+    let params = vec![
+        BenchParams {
+            depth: 1,
+            max_results: 0,
+            verbose: false,
+        },
+        BenchParams {
+            depth: 5,
+            max_results: 0,
+            verbose: false,
+        },
+    ];
+
+    let mut group = c.benchmark_group("basic_scenarios");
+
+    group.bench_function("process_directory", |b| {
+        b.iter(|| process_directory(temp_dir))
+    });
+
+    for param in params {
+        let id = BenchmarkId::new(
+            format!(
+                "depth{}_max{}_verbose{}",
+                param.depth, param.max_results, param.verbose
+            ),
+            param.depth,
+        );
+
+        group.bench_with_input(id, &param, |b, param| {
+            b.iter(|| {
+                run_binary_with_args(temp_dir, param.depth, param.max_results, param.verbose)
+                    .expect("Failed to run binary")
+            })
+        });
+    }
+
+    group.finish();
+}

benches/scenarios/edge_cases.rs

@@ -0,0 +1,6 @@
+use criterion::Criterion;
+
+pub fn benchmark_edge_cases(c: &mut Criterion) {
+    let mut group = c.benchmark_group("edge_cases");
+    group.finish();
+}

benches/scenarios/mod.rs

@@ -0,0 +1,3 @@
+pub mod basic;
+pub mod edge_cases;
+pub mod specific;

benches/scenarios/specific.rs

@@ -0,0 +1,6 @@
+use criterion::Criterion;
+
+pub fn benchmark_specific_scenarios(c: &mut Criterion) {
+    let mut group = c.benchmark_group("specific_scenarios");
+    group.finish();
+}