add tree-info subcommand to more easily test actual tree-traversal performance (#301)

Author: Byron

Diff for: gitoxide-core/src/repository/tree.rs

@@ -6,6 +6,7 @@ use std::path::PathBuf;
 use crate::OutputFormat;
 use git_repository as git;
 use git_repository::prelude::ObjectIdExt;
+use git_repository::Tree;
 
 mod entries {
     use git_repository as git;
@@ -17,28 +18,32 @@ mod entries {
     use git::traverse::tree::visit::Action;
     use git_repository::bstr::{ByteSlice, ByteVec};
 
-    pub struct Traverse<'a> {
+    #[cfg_attr(feature = "serde1", derive(serde::Serialize))]
+    #[derive(Default)]
+    pub struct Statistics {
         pub num_trees: usize,
         pub num_links: usize,
         pub num_blobs: usize,
         pub num_blobs_exec: usize,
         pub num_submodules: usize,
+        #[cfg_attr(feature = "serde1", serde(skip_serializing_if = "Option::is_none"))]
+        pub bytes: Option<u64>,
+        #[cfg_attr(feature = "serde1", serde(skip))]
         pub num_bytes: u64,
-        repo: Option<git::Repository>,
-        out: &'a mut dyn std::io::Write,
+    }
+
+    pub struct Traverse<'repo, 'a> {
+        pub stats: Statistics,
+        repo: Option<&'repo git::Repository>,
+        out: Option<&'a mut dyn std::io::Write>,
         path: BString,
         path_deque: VecDeque<BString>,
     }
 
-    impl<'a> Traverse<'a> {
-        pub fn new(repo: Option<git::Repository>, out: &'a mut dyn std::io::Write) -> Self {
+    impl<'repo, 'a> Traverse<'repo, 'a> {
+        pub fn new(repo: Option<&'repo git::Repository>, out: Option<&'a mut dyn std::io::Write>) -> Self {
             Traverse {
-                num_trees: 0,
-                num_links: 0,
-                num_blobs: 0,
-                num_blobs_exec: 0,
-                num_submodules: 0,
-                num_bytes: 0,
+                stats: Default::default(),
                 repo,
                 out,
                 path: BString::default(),
@@ -62,7 +67,7 @@ mod entries {
         }
     }
 
-    impl<'a> git::traverse::tree::Visit for Traverse<'a> {
+    impl<'repo, 'a> git::traverse::tree::Visit for Traverse<'repo, 'a> {
         fn pop_front_tracked_path_and_set_current(&mut self) {
             self.path = self.path_deque.pop_front().expect("every parent is set only once");
         }
@@ -81,59 +86,78 @@ mod entries {
         }
 
         fn visit_tree(&mut self, _entry: &EntryRef<'_>) -> Action {
-            self.num_trees += 1;
+            self.stats.num_trees += 1;
             Action::Continue
         }
 
         fn visit_nontree(&mut self, entry: &EntryRef<'_>) -> Action {
             use git::objs::tree::EntryMode::*;
             let size = self
                 .repo
-                .as_ref()
                 .and_then(|repo| repo.find_object(entry.oid).map(|o| o.data.len()).ok());
-            format_entry(&mut *self.out, entry, self.path.as_bstr(), size).ok();
+            if let Some(out) = &mut self.out {
+                format_entry(out, entry, self.path.as_bstr(), size).ok();
+            }
             if let Some(size) = size {
-                self.num_bytes += size as u64;
+                self.stats.num_bytes += size as u64;
             }
 
             match entry.mode {
-                Commit => self.num_submodules += 1,
-                Blob => self.num_blobs += 1,
-                BlobExecutable => self.num_blobs_exec += 1,
-                Link => self.num_links += 1,
+                Commit => self.stats.num_submodules += 1,
+                Blob => self.stats.num_blobs += 1,
+                BlobExecutable => self.stats.num_blobs_exec += 1,
+                Link => self.stats.num_links += 1,
                 Tree => unreachable!("BUG"),
             }
             Action::Continue
         }
     }
 }
 
+pub fn info(
+    repository: PathBuf,
+    treeish: Option<&str>,
+    extended: bool,
+    format: OutputFormat,
+    out: &mut dyn io::Write,
+    err: &mut dyn io::Write,
+) -> anyhow::Result<()> {
+    if format == OutputFormat::Human {
+        writeln!(err, "Only JSON is implemented - using that instead")?;
+    }
+
+    let repo = git::open(repository)?.apply_environment();
+    let tree = treeish_to_tree(treeish, &repo)?;
+
+    let mut delegate = entries::Traverse::new(extended.then(|| &repo), None);
+    tree.traverse().breadthfirst(&mut delegate)?;
+
+    #[cfg(feature = "serde1")]
+    {
+        delegate.stats.bytes = extended.then(|| delegate.stats.num_bytes);
+        serde_json::to_writer_pretty(out, &delegate.stats)?;
+    }
+
+    Ok(())
+}
+
 pub fn entries(
     repository: PathBuf,
     treeish: Option<&str>,
     recursive: bool,
     extended: bool,
     format: OutputFormat,
     out: &mut dyn io::Write,
-    _err: &mut dyn io::Write,
 ) -> anyhow::Result<()> {
     if format == OutputFormat::Json {
         bail!("Only human output format is supported at the moment");
     }
 
-    let tree_repo = git::open(repository)?;
-    let repo = tree_repo.clone().apply_environment();
-
-    let tree = match treeish {
-        Some(hex) => git::hash::ObjectId::from_hex(hex.as_bytes())
-            .map(|id| id.attach(&repo))?
-            .object()?
-            .try_into_tree()?,
-        None => repo.head()?.peel_to_commit_in_place()?.tree()?,
-    };
+    let repo = git::open(repository)?.apply_environment();
+    let tree = treeish_to_tree(treeish, &repo)?;
 
     if recursive {
-        let mut delegate = entries::Traverse::new(extended.then(|| tree_repo), out);
+        let mut delegate = entries::Traverse::new(extended.then(|| &repo), out.into());
         tree.traverse().breadthfirst(&mut delegate)?;
     } else {
         for entry in tree.iter() {
@@ -152,6 +176,16 @@ pub fn entries(
     Ok(())
 }
 
+fn treeish_to_tree<'repo>(treeish: Option<&str>, repo: &'repo git::Repository) -> anyhow::Result<Tree<'repo>> {
+    Ok(match treeish {
+        Some(hex) => git::hash::ObjectId::from_hex(hex.as_bytes())
+            .map(|id| id.attach(&repo))?
+            .object()?
+            .try_into_tree()?,
+        None => repo.head()?.peel_to_commit_in_place()?.tree()?,
+    })
+}
+
 fn format_entry(
     mut out: impl io::Write,
     entry: &git::objs::tree::EntryRef<'_>,

Diff for: src/plumbing/main.rs

@@ -154,16 +154,20 @@ pub fn main() -> Result<()> {
                 progress,
                 progress_keep_open,
                 None,
+                move |_progress, out, _err| {
+                    core::repository::tree::entries(repository, treeish.as_deref(), recursive, extended, format, out)
+                },
+            ),
+            repo::Subcommands::Tree {
+                cmd: repo::tree::Subcommands::Info { treeish, extended },
+            } => prepare_and_run(
+                "repository-tree-info",
+                verbose,
+                progress,
+                progress_keep_open,
+                None,
                 move |_progress, out, err| {
-                    core::repository::tree::entries(
-                        repository,
-                        treeish.as_deref(),
-                        recursive,
-                        extended,
-                        format,
-                        out,
-                        err,
-                    )
+                    core::repository::tree::info(repository, treeish.as_deref(), extended, format, out, err)
                 },
             ),
             repo::Subcommands::Verify {

Diff for: src/plumbing/options.rs

@@ -351,18 +351,27 @@ pub mod repo {
     pub mod tree {
         #[derive(Debug, clap::Subcommand)]
         pub enum Subcommands {
+            /// Print entries in a given tree
             Entries {
                 /// Traverse the entire tree and its subtrees respectively, not only this tree.
                 #[clap(long, short = 'r')]
                 recursive: bool,
 
-                /// Provide the files size as well. This is expensive as the object is decoded entirely.
+                /// Provide files size as well. This is expensive as the object is decoded entirely.
                 #[clap(long, short = 'e')]
                 extended: bool,
 
                 /// The tree to traverse, or the tree at `HEAD` if unspecified.
                 treeish: Option<String>,
             },
+            /// Provide information about a tree.
+            Info {
+                /// Provide files size as well. This is expensive as the object is decoded entirely.
+                #[clap(long, short = 'e')]
+                extended: bool,
+                /// The tree to traverse, or the tree at `HEAD` if unspecified.
+                treeish: Option<String>,
+            },
         }
     }
 }