From 3fe9872e4320217bcc8b353460c8d9bd063e668b Mon Sep 17 00:00:00 2001
From: Jordan Doyle <jordan@doyle.la>
Date: Sun, 13 Mar 2022 17:15:26 +0000
Subject: [PATCH] Sort tree before inserting into packfile

---
 src/git_command_handlers/ls_refs.rs |  2 +-
 src/main.rs                         | 11 ++++++++---
 src/protocol/high_level.rs          |  7 +++++++
 src/protocol/low_level.rs           |  1 +
 4 files changed, 17 insertions(+), 4 deletions(-)

diff --git a/src/git_command_handlers/ls_refs.rs b/src/git_command_handlers/ls_refs.rs
index 4b77740..3c67aa6 100644
--- a/src/git_command_handlers/ls_refs.rs
+++ b/src/git_command_handlers/ls_refs.rs
@@ -22,7 +22,7 @@ pub fn handle<U: UserProvider + PackageProvider + Send + Sync + 'static>(
     let commit_hash = hex::encode(&commit_hash);
 
     handle.write(PktLine::Data(
-        format!("{} HEAD symref-target:refs/heads/master\n", commit_hash).as_bytes(),
+        format!("{} HEAD symref-target:refs/heads/master", commit_hash).as_bytes(),
     ))?;
     handle.write(PktLine::Flush)?;
     handle.flush(session, channel);
diff --git a/src/main.rs b/src/main.rs
index a9d07cd..772f5e3 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -24,6 +24,7 @@ use anyhow::anyhow;
 use bytes::{BufMut, Bytes, BytesMut};
 use clap::Parser;
 use futures::Future;
+use indexmap::IndexMap;
 use parking_lot::RwLock;
 use std::{
     borrow::Cow, collections::HashMap, fmt::Write, net::SocketAddr, net::SocketAddrV6, pin::Pin,
@@ -35,7 +36,7 @@ use thrussh::{
 };
 use thrussh_keys::key::PublicKey;
 use tokio_util::{codec::Decoder, codec::Encoder as CodecEncoder};
-use tracing::{error, info, info_span, Instrument, Span};
+use tracing::{debug, error, info, info_span, Instrument, Span};
 use uuid::Uuid;
 
 const AGENT: &str = concat!(
@@ -110,6 +111,8 @@ type MetadataCache = Arc<RwLock<HashMap<MetadataCacheKey<'static>, Arc<CargoInde
 
 struct Server<U: UserProvider + PackageProvider + Send + Sync + 'static> {
     gitlab: Arc<U>,
+    // todo: we could make our commit hash stable by leaving an update time
+    //  in this cache and using that as our commit time
     metadata_cache: MetadataCache,
 }
 
@@ -184,11 +187,11 @@ impl<U: UserProvider + PackageProvider + Send + Sync + 'static> Handler<U> {
     /// and groups them by crate.
     async fn fetch_releases_by_crate(
         &self,
-    ) -> anyhow::Result<HashMap<(U::CratePath, String), Vec<Release>>> {
+    ) -> anyhow::Result<IndexMap<(U::CratePath, String), Vec<Release>>> {
         let user = self.user()?;
         let group = self.group()?;
 
-        let mut res = HashMap::new();
+        let mut res = IndexMap::new();
 
         for (path, release) in Arc::clone(&self.gitlab)
             .fetch_releases_for_group(group, user)
@@ -297,6 +300,8 @@ impl<U: UserProvider + PackageProvider + Send + Sync + 'static> Handler<U> {
                 let checksum = &release.checksum;
                 let version = &release.version;
 
+                debug!("Fetching metadata for {}-{}", crate_name, version);
+
                 // parses the `cargo metadata` stored in the release, which
                 // should be stored under `metadata.json`.
                 let meta = self
diff --git a/src/protocol/high_level.rs b/src/protocol/high_level.rs
index 917b065..e7dae0d 100644
--- a/src/protocol/high_level.rs
+++ b/src/protocol/high_level.rs
@@ -132,11 +132,13 @@ impl Tree {
             tree.push(match *item {
                 TreeItem::Blob(hash) => LowLevelTreeItem {
                     kind: TreeItemKind::File,
+                    sort_name: name.clone(),
                     name,
                     hash,
                 },
                 TreeItem::Tree(tree) => LowLevelTreeItem {
                     kind: TreeItemKind::Directory,
+                    sort_name: format!("{}/", name),
                     name,
                     // we're essentially working through our tree from the bottom up,
                     // so we can grab the hash of each directory along the way and
@@ -146,6 +148,11 @@ impl Tree {
             });
         }
 
+        // we need to sort our tree alphabetically, otherwise Git will silently
+        // stop parsing the rest of the tree once it comes across a non-sorted
+        // tree entry.
+        tree.sort_unstable_by(|a, b| a.sort_name.cmp(&b.sort_name));
+
         // gets the hash of the tree we've just worked on, and
         // pushes it to the packfile
         let tree = PackFileEntry::Tree(tree);
diff --git a/src/protocol/low_level.rs b/src/protocol/low_level.rs
index 1365b27..b013cb8 100644
--- a/src/protocol/low_level.rs
+++ b/src/protocol/low_level.rs
@@ -143,6 +143,7 @@ pub struct TreeItem {
     pub kind: TreeItemKind,
     pub name: String,
     pub hash: HashOutput,
+    pub sort_name: String,
 }
 
 // `[mode] [name]\0[hash]`
--
libgit2 1.7.2