Allow sink-compress configuration; choose best algorithm

Byron · Byron · commit 29b9c230e35b · 2020-07-28T10:44:41.000+08:00
diff --git a/gitoxide-core/src/pack/explode.rs b/gitoxide-core/src/pack/explode.rs
@@ -116,10 +116,10 @@ impl git_odb::Write for OutputWriter {
 }
 
 impl OutputWriter {
-    fn new(path: Option<impl AsRef<Path>>) -> Self {
+    fn new(path: Option<impl AsRef<Path>>, compress: bool) -> Self {
         match path {
             Some(path) => OutputWriter::Loose(loose::Db::at(path.as_ref())),
-            None => OutputWriter::Sink(git_odb::sink().compress(true)),
+            None => OutputWriter::Sink(git_odb::sink().compress(compress)),
         }
     }
 }
@@ -131,6 +131,7 @@ pub fn pack_or_pack_index<P>(
     thread_limit: Option<usize>,
     progress: Option<P>,
     delete_pack: bool,
+    sink_compress: bool,
 ) -> Result<()>
 where
     P: Progress,
@@ -151,18 +152,28 @@ where
         ));
     }
 
+    let algorithm = object_path
+        .as_ref()
+        .map(|_| {
+            if sink_compress {
+                pack::index::traverse::Algorithm::Lookup
+            } else {
+                pack::index::traverse::Algorithm::DeltaTreeLookup
+            }
+        })
+        .unwrap_or(pack::index::traverse::Algorithm::Lookup);
     let mut progress = bundle.index.traverse(
         &bundle.pack,
         pack::index::traverse::Context {
-            algorithm: pack::index::traverse::Algorithm::Lookup,
+            algorithm,
             thread_limit,
             check: check.into(),
         },
         progress,
         {
             let object_path = object_path.map(|p| p.as_ref().to_owned());
             move || {
-            let out = OutputWriter::new(object_path.clone());
+            let out = OutputWriter::new(object_path.clone(), sink_compress);
             move |object_kind, buf, index_entry, _entry_stats, progress| {
                 let written_id = out
                     .write_buf(object_kind, buf, HashKind::Sha1)
diff --git a/src/plumbing/lean.rs b/src/plumbing/lean.rs
@@ -42,6 +42,14 @@ mod options {
         #[argh(switch, short = 'v')]
         pub verbose: bool,
 
+        /// compress bytes even when using the sink, i.e. no object directory is specified
+        ///
+        /// This helps to determine overhead related to compression. If unset, the sink will
+        /// only create hashes from bytes, which is usually limited by the speed at which input
+        /// can be obtained.
+        #[argh(switch)]
+        pub sink_compress: bool,
+
         /// the amount of checks to run. Defaults to 'all'.
         ///
         /// Allowed values:
@@ -136,6 +144,7 @@ pub fn main() -> Result<()> {
     match cli.subcommand {
         SubCommands::PackExplode(PackExplode {
             pack_path,
+            sink_compress,
             object_path,
             verbose,
             check,
@@ -149,6 +158,7 @@ pub fn main() -> Result<()> {
                 thread_limit,
                 progress,
                 delete_pack,
+                sink_compress,
             )
         }
         SubCommands::PackVerify(PackVerify {
diff --git a/src/plumbing/pretty.rs b/src/plumbing/pretty.rs
@@ -43,6 +43,14 @@ mod options {
             )]
             check: core::pack::explode::SafetyCheck,
 
+            /// Compress bytes even when using the sink, i.e. no object directory is specified
+            ///
+            /// This helps to determine overhead related to compression. If unset, the sink will
+            /// only create hashes from bytes, which is usually limited by the speed at which input
+            /// can be obtained.
+            #[structopt(long)]
+            sink_compress: bool,
+
             /// Display verbose messages and progress information
             #[structopt(long, short = "v")]
             verbose: bool,
@@ -224,6 +232,7 @@ pub fn main() -> Result<()> {
             check,
             progress,
             progress_keep_open,
+            sink_compress,
             delete_pack,
             pack_path,
             object_path,
@@ -240,6 +249,7 @@ pub fn main() -> Result<()> {
                     thread_limit,
                     progress,
                     delete_pack,
+                    sink_compress,
                 )
             },
         ),
diff --git a/tests/stateless-journey.sh b/tests/stateless-journey.sh
@@ -85,7 +85,7 @@ title "CLI ${kind}"
             it "does explode the file" && {
               WITH_SNAPSHOT="$snapshot/plumbing-broken-pack-explode-delete-pack-to-sink-skip-checks-success" \
               expect_run $SUCCESSFULLY "$exe_plumbing" pack-explode --check skip-file-and-object-checksum-and-no-abort-on-decode \
-                                        --delete-pack "${PACK_FILE}.pack" .
+                                        --delete-pack --sink-compress "${PACK_FILE}.pack" .
             }
 
             it "removes the original files" && {

Original file line number	Diff line number	Diff line change
`@@ -85,7 +85,7 @@ title "CLI ${kind}"`
`85`	`85`	`it "does explode the file" && {`
`86`	`86`	`WITH_SNAPSHOT="$snapshot/plumbing-broken-pack-explode-delete-pack-to-sink-skip-checks-success" \`
`87`	`87`	`expect_run $SUCCESSFULLY "$exe_plumbing" pack-explode --check skip-file-and-object-checksum-and-no-abort-on-decode \`
`88`		`- --delete-pack "${PACK_FILE}.pack" .`
	`88`	`+ --delete-pack --sink-compress "${PACK_FILE}.pack" .`
`89`	`89`	`}`
`90`	`90`
`91`	`91`	`it "removes the original files" && {`