Merge pull request #635 from Byron/gitoxide-for-traversal

Gitoxide for performance improvements

Author: Byron

Diff for: .github/workflows/ci.yml

@@ -44,6 +44,9 @@ jobs:
       - name: Run clippy
         run: cargo clippy
 
+      - name: Run onefetch
+        run: cargo run
+
   rustfmt:
     name: Rustfmt
     runs-on: ubuntu-latest

Diff for: Cargo.lock

@@ -28,11 +28,12 @@ section = "utility"
 [dependencies]
 anyhow = "1.0"
 askalono = "0.4.4"
-byte-unit = "4.0.14"
+byte-unit = "4.0" # to match git-repository's lower requirement, which it needs for MSRV
 bytecount = "0.6.2"
 clap = {version = "3.1.6", features = ["cargo", "wrap_help"]}
 color_quant = "1.1.0"
 git2 = {version = "0.14.2", default-features = false}
+git-repository = { version = "0.15.0", features = ["max-performance", "unstable", "serde1"] }
 image = "0.24.1"
 owo-colors = "3.3.0"
 regex = "1.5.5"

Diff for: Cargo.toml

@@ -1,3 +1,4 @@
+use git_repository as git;
 use serde::ser::SerializeStruct;
 use serde::Serialize;
 
@@ -10,20 +11,24 @@ pub struct Author {
 
 impl Author {
     pub fn new(
-        name: String,
-        email: Option<String>,
+        name: git::bstr::BString,
+        email: Option<git::bstr::BString>,
         nbr_of_commits: usize,
         total_nbr_of_commits: usize,
     ) -> Self {
         let contribution =
             (nbr_of_commits as f32 * 100. / total_nbr_of_commits as f32).round() as usize;
         Self {
-            name,
-            email,
+            name: name.to_string(),
+            email: email.map(|e| e.to_string()),
             nbr_of_commits,
             contribution,
         }
     }
+
+    pub fn clear_email(&mut self) {
+        self.email = None;
+    }
 }
 
 impl std::fmt::Display for Author {

Diff for: src/info/author.rs

@@ -1,5 +1,6 @@
 use anyhow::Result;
 use std::collections::HashMap;
+use std::path::Path;
 use std::{ffi::OsStr, fs};
 
 pub mod package_manager;
@@ -17,7 +18,7 @@ impl DependencyDetector {
         DependencyDetector { package_managers }
     }
 
-    pub fn get_dependencies(&self, dir: &str) -> Result<String> {
+    pub fn get_dependencies(&self, dir: &Path) -> Result<String> {
         let deps = fs::read_dir(dir)?
             .filter_map(std::result::Result::ok)
             .map(|entry| entry.path())

Diff for: src/info/deps/mod.rs

@@ -1,31 +1,32 @@
-use git2::Oid;
 use serde::ser::SerializeStruct;
 use serde::Serialize;
 
 pub struct HeadRefs {
-    commit: Oid,
+    short_commit_id: String,
     refs: Vec<String>,
 }
 
 impl HeadRefs {
-    pub fn new(commit: Oid, refs: Vec<String>) -> HeadRefs {
-        HeadRefs { commit, refs }
+    pub fn new(short_commit_id: String, refs: Vec<String>) -> HeadRefs {
+        HeadRefs {
+            short_commit_id,
+            refs,
+        }
     }
 }
 
 impl std::fmt::Display for HeadRefs {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
-        let short_commit = self.commit.to_string().chars().take(7).collect::<String>();
         if !self.refs.is_empty() {
             let refs_str = self
                 .refs
                 .iter()
                 .map(|ref_name| ref_name.as_str())
                 .collect::<Vec<&str>>()
                 .join(", ");
-            write!(f, "{} ({})", short_commit, refs_str)
+            write!(f, "{} ({})", self.short_commit_id, refs_str)
         } else {
-            write!(f, "{}", short_commit)
+            write!(f, "{}", self.short_commit_id)
         }
     }
 }
@@ -37,10 +38,7 @@ impl Serialize for HeadRefs {
     {
         let mut state = serializer.serialize_struct("HeadRefs", 2)?;
         state.serialize_field("refs", &self.refs)?;
-        state.serialize_field(
-            "oid",
-            &self.commit.to_string().chars().take(7).collect::<String>(),
-        )?;
+        state.serialize_field("oid", &self.short_commit_id)?;
         state.end()
     }
 }

Diff for: src/info/head_refs.rs

@@ -24,7 +24,7 @@ macro_rules! define_colors {
     ( [ $($bc:ident),+ ] : [ $($c:ident($r:expr, $g:expr, $b:expr)),+ ] ) => { Colors { basic_colors: vec![$(clean_color!($bc)),+], true_colors: Some(vec![$(DynColors::$c($r, $g, $b)),+]) } };
 }
 
-#[derive(PartialEq, EnumString, EnumIter, IntoStaticStr)]
+#[derive(Clone, PartialEq, EnumString, EnumIter, IntoStaticStr)]
 #[strum(serialize_all = "lowercase")]
 pub enum LanguageType {
     Programming,

Diff for: src/info/langs/language.rs

@@ -2,6 +2,7 @@ use anyhow::{Context, Result};
 use language::{Language, LanguageType};
 use regex::Regex;
 use std::collections::HashMap;
+use std::path::Path;
 use strum::IntoEnumIterator;
 
 pub mod language;
@@ -11,7 +12,7 @@ pub fn get_dominant_language(languages_stat_vec: &[(Language, f64)]) -> Language
 }
 
 pub fn get_language_statistics(
-    dir: &str,
+    dir: &Path,
     ignored_directories: &[String],
     language_types: &[LanguageType],
     include_hidden: bool,
@@ -71,7 +72,7 @@ fn get_total_loc(languages: &tokei::Languages) -> usize {
 }
 
 fn get_statistics(
-    dir: &str,
+    dir: &Path,
     ignored_directories: &[String],
     language_types: &[LanguageType],
     include_hidden: bool,

Diff for: src/info/langs/mod.rs

@@ -1,5 +1,6 @@
 use anyhow::{bail, Result};
 use askalono::{Store, TextData};
+use std::path::Path;
 use std::{ffi::OsStr, fs};
 
 const LICENSE_FILES: [&str; 3] = ["LICENSE", "LICENCE", "COPYING"];
@@ -24,7 +25,7 @@ impl Detector {
         }
     }
 
-    pub fn get_license(&self, dir: &str) -> Result<String> {
+    pub fn get_license(&self, dir: &Path) -> Result<String> {
         fn is_license_file<S: AsRef<str>>(file_name: S) -> bool {
             LICENSE_FILES
                 .iter()

Diff for: src/info/license.rs

@@ -1,7 +1,8 @@
 use crate::cli::{self, Config};
+use crate::repo::Commits;
 use crate::ui::get_ascii_colors;
 use crate::ui::text_colors::TextColors;
-use anyhow::Result;
+use anyhow::{Context, Result};
 use author::Author;
 use deps::DependencyDetector;
 use git2::Repository;
@@ -159,29 +160,61 @@ impl Info {
     pub fn new(config: Config) -> Result<Self> {
         let git_version = cli::get_git_version();
         let repo = Repository::discover(&config.repo_path)?;
-        let internal_repo = Repo::new(&repo, config.no_merges, &config.bot_regex_pattern)?;
-        let (repo_name, repo_url) = internal_repo.get_name_and_url()?;
-        let head_refs = internal_repo.get_head_refs()?;
-        let pending_changes = internal_repo.get_pending_changes()?;
-        let version = internal_repo.get_version()?;
-        let git_username = internal_repo.get_git_username()?;
-        let number_of_tags = internal_repo.get_number_of_tags()?;
-        let number_of_branches = internal_repo.get_number_of_branches()?;
-        let creation_date = internal_repo.get_creation_date(config.iso_time)?;
-        let number_of_commits = internal_repo.get_number_of_commits();
-        let (authors, contributors) =
-            internal_repo.get_authors(config.number_of_authors, config.show_email)?;
-        let last_change = internal_repo.get_date_of_last_commit(config.iso_time);
-        let (repo_size, file_count) = internal_repo.get_repo_size();
-        let workdir = internal_repo.get_work_dir()?;
+        let workdir = repo.workdir().expect("non-bare repo").to_owned();
+
+        let pending_changes = std::thread::spawn({
+            let git_dir = repo.path().to_owned();
+            move || {
+                let repo = git2::Repository::open(git_dir)?;
+                repo::get_pending_changes(&repo)
+            }
+        });
+        let languages_handle = std::thread::spawn({
+            let ignored_directories = config.ignored_directories.clone();
+            let language_types = config.language_types.clone();
+            let include_hidden = config.include_hidden;
+            let workdir = workdir.clone();
+            move || {
+                langs::get_language_statistics(
+                    &workdir,
+                    &ignored_directories,
+                    &language_types,
+                    include_hidden,
+                )
+            }
+        });
+
+        let repo = Repo::new(repo)?;
+        let mut commits = Commits::new(
+            repo.gitoxide(),
+            config.no_merges,
+            &config.bot_regex_pattern,
+            config.number_of_authors,
+        )?;
+        let (repo_name, repo_url) = repo.get_name_and_url()?;
+        let head_refs = repo.get_head_refs()?;
+        let version = repo.get_version()?;
+        let git_username = repo.get_git_username()?;
+        let number_of_tags = repo.get_number_of_tags()?;
+        let number_of_branches = repo.get_number_of_branches()?;
+        let (repo_size, file_count) = repo.get_repo_size();
         let license = Detector::new()?.get_license(&workdir)?;
         let dependencies = DependencyDetector::new().get_dependencies(&workdir)?;
-        let (languages, lines_of_code) = langs::get_language_statistics(
-            &workdir,
-            &config.ignored_directories,
-            &config.language_types,
-            config.include_hidden,
-        )?;
+
+        let creation_date = commits.get_creation_date(config.iso_time);
+        let number_of_commits = commits.count();
+        let (authors, contributors) = commits.take_authors(config.show_email);
+        let last_change = commits.get_date_of_last_commit(config.iso_time);
+
+        let pending_changes = pending_changes
+            .join()
+            .ok()
+            .context("BUG: panic in pending-changes thread")??;
+
+        let (languages, lines_of_code) = languages_handle
+            .join()
+            .ok()
+            .context("BUG: panic in language statistics thread")??;
         let dominant_language = langs::get_dominant_language(&languages);
         let ascii_colors = get_ascii_colors(
             &config.ascii_language,