gopls/internal/regtest/bench: add flags to reuse benchmark repo dirs

It can be slow and use significant bandwidth to do repeated shallow
clones of e.g. kubernetes while iterating on the benchmark.

Add flags to re-use existing dirs checked out the the correct commit,
assuming the user knows what they are doing.

Change-Id: I2a8584ab26290a317ac23c4dec32f9c13de42d41
Reviewed-on: https://go-review.googlesource.com/c/tools/+/477978
Run-TryBot: Robert Findley <[email protected]>
TryBot-Result: Gopher Robot <[email protected]>
Reviewed-by: Alan Donovan <[email protected]>
Auto-Submit: Robert Findley <[email protected]>

Author: findleyr

gopls/internal/regtest/bench/repo_test.go

@@ -8,6 +8,7 @@ import (
 	"bytes"
 	"context"
 	"errors"
+	"flag"
 	"fmt"
 	"log"
 	"os"
@@ -30,6 +31,7 @@ var repos = map[string]*repo{
 		name:   "istio",
 		url:    "https://github.com/istio/istio",
 		commit: "1.17.0",
+		inDir:  flag.String("istio_dir", "", "if set, reuse this directory as [email protected]"),
 	},
 
 	// Kubernetes is a large repo with many dependencies, and in the past has
@@ -38,13 +40,15 @@ var repos = map[string]*repo{
 		name:   "kubernetes",
 		url:    "https://github.com/kubernetes/kubernetes",
 		commit: "v1.24.0",
+		inDir:  flag.String("kubernetes_dir", "", "if set, reuse this directory as [email protected]"),
 	},
 
 	// A large, industrial application.
 	"kuma": {
 		name:   "kuma",
 		url:    "https://github.com/kumahq/kuma",
 		commit: "2.1.1",
+		inDir:  flag.String("kuma_dir", "", "if set, reuse this directory as [email protected]"),
 	},
 
 	// x/pkgsite is familiar and represents a common use case (a webserver). It
@@ -54,6 +58,7 @@ var repos = map[string]*repo{
 		url:    "https://go.googlesource.com/pkgsite",
 		commit: "81f6f8d4175ad0bf6feaa03543cc433f8b04b19b",
 		short:  true,
+		inDir:  flag.String("pkgsite_dir", "", "if set, reuse this directory as pkgsite@81f6f8d4"),
 	},
 
 	// A tiny self-contained project.
@@ -62,6 +67,7 @@ var repos = map[string]*repo{
 		url:    "https://github.com/google/starlark-go",
 		commit: "3f75dec8e4039385901a30981e3703470d77e027",
 		short:  true,
+		inDir:  flag.String("starlark_dir", "", "if set, reuse this directory as starlark@3f75dec8"),
 	},
 
 	// The current repository, which is medium-small and has very few dependencies.
@@ -70,6 +76,7 @@ var repos = map[string]*repo{
 		url:    "https://go.googlesource.com/tools",
 		commit: "gopls/v0.9.0",
 		short:  true,
+		inDir:  flag.String("tools_dir", "", "if set, reuse this directory as x/[email protected]"),
 	},
 }
 
@@ -94,10 +101,11 @@ func getRepo(tb testing.TB, name string) *repo {
 // codebase.
 type repo struct {
 	// static configuration
-	name   string // must be unique, used for subdirectory
-	url    string // repo url
-	commit string // full commit hash or tag
-	short  bool   // whether this repo runs with -short
+	name   string  // must be unique, used for subdirectory
+	url    string  // repo url
+	commit string  // full commit hash or tag
+	short  bool    // whether this repo runs with -short
+	inDir  *string // if set, use this dir as url@commit, and don't delete
 
 	dirOnce sync.Once
 	dir     string // directory contaning source code checked out to url@commit
@@ -109,14 +117,37 @@ type repo struct {
 	awaiter    *Awaiter
 }
 
+// reusableDir return a reusable directory for benchmarking, or "".
+//
+// If the user specifies a directory, the test will create and populate it
+// on the first run an re-use it on subsequent runs. Otherwise it will
+// create, populate, and delete a temporary directory.
+func (r *repo) reusableDir() string {
+	if r.inDir == nil {
+		return ""
+	}
+	return *r.inDir
+}
+
 // getDir returns directory containing repo source code, creating it if
 // necessary. It is safe for concurrent use.
 func (r *repo) getDir() string {
 	r.dirOnce.Do(func() {
-		r.dir = filepath.Join(getTempDir(), r.name)
-		log.Printf("cloning %s@%s into %s", r.url, r.commit, r.dir)
-		if err := shallowClone(r.dir, r.url, r.commit); err != nil {
+		if r.dir = r.reusableDir(); r.dir == "" {
+			r.dir = filepath.Join(getTempDir(), r.name)
+		}
+
+		_, err := os.Stat(r.dir)
+		switch {
+		case os.IsNotExist(err):
+			log.Printf("cloning %s@%s into %s", r.url, r.commit, r.dir)
+			if err := shallowClone(r.dir, r.url, r.commit); err != nil {
+				log.Fatal(err)
+			}
+		case err != nil:
 			log.Fatal(err)
+		default:
+			log.Printf("reusing %s as %s@%s", r.dir, r.url, r.commit)
 		}
 	})
 	return r.dir
@@ -200,7 +231,7 @@ func (r *repo) Close() error {
 			fmt.Fprintf(&errBuf, "closing sandbox: %v", err)
 		}
 	}
-	if r.dir != "" {
+	if r.dir != "" && r.reusableDir() == "" {
 		if err := os.RemoveAll(r.dir); err != nil {
 			fmt.Fprintf(&errBuf, "cleaning dir: %v", err)
 		}