From 64f1ad80eef6ee27b748089e321ca5f28556ab7f Mon Sep 17 00:00:00 2001
From: Stefan Hengl <stefan@sourcegraph.com>
Date: Fri, 27 Sep 2024 09:38:35 +0200
Subject: [PATCH] archive: use ModTime as proxy for LatestCommitDate (#836)

This is motivated by #832

We use archive index in our e2e tests. In order to test our latest improvements to ranking, archive index needs to set the latest commit date.

Test plan:
- new unit test
- I checked that the tar files downloaded from github have the correct mod time.
---
 internal/archive/archive.go  |  8 +++-
 internal/archive/e2e_test.go | 78 ++++++++++++++++++++++++++++++++++--
 internal/archive/index.go    | 17 ++++++--
 3 files changed, 93 insertions(+), 10 deletions(-)

diff --git a/internal/archive/archive.go b/internal/archive/archive.go
index 2c8cb64fa..2048a25fb 100644
--- a/internal/archive/archive.go
+++ b/internal/archive/archive.go
@@ -11,6 +11,7 @@ import (
 	"net/url"
 	"os"
 	"strings"
+	"time"
 )
 
 type Archive interface {
@@ -20,8 +21,9 @@ type Archive interface {
 
 type File struct {
 	io.ReadCloser
-	Name string
-	Size int64
+	Name    string
+	Size    int64
+	ModTime time.Time
 }
 
 type tarArchive struct {
@@ -45,6 +47,7 @@ func (a *tarArchive) Next() (*File, error) {
 			ReadCloser: io.NopCloser(a.tr),
 			Name:       hdr.Name,
 			Size:       hdr.Size,
+			ModTime:    hdr.ModTime,
 		}, nil
 	}
 }
@@ -71,6 +74,7 @@ func (a *zipArchive) Next() (*File, error) {
 		ReadCloser: r,
 		Name:       f.Name,
 		Size:       int64(f.UncompressedSize64),
+		ModTime:    f.Modified,
 	}, nil
 }
 
diff --git a/internal/archive/e2e_test.go b/internal/archive/e2e_test.go
index 79b3ba538..b861eb09a 100644
--- a/internal/archive/e2e_test.go
+++ b/internal/archive/e2e_test.go
@@ -11,8 +11,12 @@ import (
 	"io"
 	"log"
 	"os"
+	"path/filepath"
 	"strings"
 	"testing"
+	"time"
+
+	"github.com/stretchr/testify/require"
 
 	"github.com/sourcegraph/zoekt"
 	"github.com/sourcegraph/zoekt/build"
@@ -28,11 +32,18 @@ func TestMain(m *testing.M) {
 	os.Exit(m.Run())
 }
 
+var modTime = time.Date(2024, 9, 26, 0, 0, 0, 0, time.UTC)
+
 func writeArchive(w io.Writer, format string, files map[string]string) (err error) {
 	if format == "zip" {
 		zw := zip.NewWriter(w)
 		for name, body := range files {
-			f, err := zw.Create(name)
+			header := &zip.FileHeader{
+				Name:     name,
+				Method:   zip.Deflate,
+				Modified: modTime,
+			}
+			f, err := zw.CreateHeader(header)
 			if err != nil {
 				return err
 			}
@@ -63,9 +74,10 @@ func writeArchive(w io.Writer, format string, files map[string]string) (err erro
 
 	for name, body := range files {
 		hdr := &tar.Header{
-			Name: name,
-			Mode: 0o600,
-			Size: int64(len(body)),
+			Name:    name,
+			Mode:    0o600,
+			Size:    int64(len(body)),
+			ModTime: modTime,
 		}
 		if err := tw.WriteHeader(hdr); err != nil {
 			return err
@@ -189,3 +201,61 @@ func testIndexIncrementally(t *testing.T, format string) {
 		}
 	}
 }
+
+// TestLatestCommitDate tests that the latest commit date is set correctly if
+// the mod time of the files has been set during the archive creation.
+func TestLatestCommitDate(t *testing.T) {
+	for _, format := range []string{"tar", "tgz", "zip"} {
+		t.Run(format, func(t *testing.T) {
+			testLatestCommitDate(t, format)
+		})
+	}
+}
+
+func testLatestCommitDate(t *testing.T, format string) {
+	// Create an archive
+	archive, err := os.CreateTemp("", "TestLatestCommitDate")
+	require.NoError(t, err)
+	defer os.Remove(archive.Name())
+
+	fileSize := 10
+	files := map[string]string{}
+	for i := 0; i < 4; i++ {
+		s := fmt.Sprintf("%d", i)
+		files["F"+s] = strings.Repeat("a", fileSize)
+		files["!F"+s] = strings.Repeat("a", fileSize)
+	}
+
+	err = writeArchive(archive, format, files)
+	if err != nil {
+		t.Fatalf("unable to create archive %v", err)
+	}
+	archive.Close()
+
+	// Index
+	indexDir := t.TempDir()
+	bopts := build.Options{
+		IndexDir: indexDir,
+	}
+	opts := Options{
+		Archive: archive.Name(),
+		Name:    "repo",
+		Branch:  "master",
+		Commit:  "cccccccccccccccccccccccccccccccccccccccc",
+	}
+
+	err = Index(opts, bopts)
+	require.NoError(t, err)
+
+	// Read the metadata of the index we just created and check the latest commit date.
+	f, err := os.Open(indexDir)
+	require.NoError(t, err)
+
+	indexFiles, err := f.Readdirnames(1)
+	require.Len(t, indexFiles, 1)
+
+	repos, _, err := zoekt.ReadMetadataPath(filepath.Join(indexDir, indexFiles[0]))
+	require.NoError(t, err)
+	require.Len(t, repos, 1)
+	require.True(t, repos[0].LatestCommitDate.Equal(modTime))
+}
diff --git a/internal/archive/index.go b/internal/archive/index.go
index 2262c9521..c8836768f 100644
--- a/internal/archive/index.go
+++ b/internal/archive/index.go
@@ -7,6 +7,7 @@ import (
 	"io"
 	"net/url"
 	"strings"
+	"sync"
 
 	"github.com/sourcegraph/zoekt"
 	"github.com/sourcegraph/zoekt/build"
@@ -113,14 +114,22 @@ func Index(opts Options, bopts build.Options) error {
 	defer a.Close()
 
 	bopts.RepositoryDescription.Source = opts.Archive
-	builder, err := build.NewBuilder(bopts)
-	if err != nil {
-		return err
-	}
+	var builder *build.Builder
 
+	once := sync.Once{}
+	var onceErr error
 	add := func(f *File) error {
 		defer f.Close()
 
+		once.Do(func() {
+			// We use the ModTime of the first file as a proxy for the latest commit date.
+			bopts.RepositoryDescription.LatestCommitDate = f.ModTime
+			builder, onceErr = build.NewBuilder(bopts)
+		})
+		if onceErr != nil {
+			return onceErr
+		}
+
 		contents, err := io.ReadAll(f)
 		if err != nil {
 			return err