Add ability to download subpath archive (#36371)

closes: https://github.com/go-gitea/gitea/issues/4478

---------

Signed-off-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>

modules/git/gitcmd/command.go

@@ -426,9 +426,9 @@ type RunStdError interface {
 }
 
 type runStdError struct {
-	err    error
-	stderr string
-	errMsg string
+	err    error  // usually the low-level error like `*exec.ExitError`
+	stderr string // git command's stderr output
+	errMsg string // the cached error message for Error() method
 }
 
 func (r *runStdError) Error() string {
@@ -448,6 +448,22 @@ func (r *runStdError) Stderr() string {
 	return r.stderr
 }
 
+func ErrorAsStderr(err error) (string, bool) {
+	var runErr RunStdError
+	if errors.As(err, &runErr) {
+		return runErr.Stderr(), true
+	}
+	return "", false
+}
+
+func StderrHasPrefix(err error, prefix string) bool {
+	stderr, ok := ErrorAsStderr(err)
+	if !ok {
+		return false
+	}
+	return strings.HasPrefix(stderr, prefix)
+}
+
 func IsErrorExitCode(err error, code int) bool {
 	var exitError *exec.ExitError
 	if errors.As(err, &exitError) {

modules/git/gitcmd/command_test.go

@@ -12,6 +12,7 @@ import (
 	"code.gitea.io/gitea/modules/tempdir"
 
 	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
 )
 
 func TestMain(m *testing.M) {
@@ -99,3 +100,14 @@ func TestCommandString(t *testing.T) {
 	cmd = NewCommand("url: https://a:b@c/", "/root/dir-a/dir-b")
 	assert.Equal(t, cmd.prog+` "url: https://sanitized-credential@c/" .../dir-a/dir-b`, cmd.LogString())
 }
+
+func TestRunStdError(t *testing.T) {
+	e := &runStdError{stderr: "some error"}
+	var err RunStdError = e
+
+	var asErr RunStdError
+	require.ErrorAs(t, err, &asErr)
+	require.Equal(t, "some error", asErr.Stderr())
+
+	require.ErrorAs(t, fmt.Errorf("wrapped %w", err), &asErr)
+}

modules/git/gitcmd/utils.go

@@ -3,12 +3,18 @@
 
 package gitcmd
 
-import "fmt"
+import (
+	"fmt"
 
-// ConcatenateError concatenats an error with stderr string
+	"code.gitea.io/gitea/modules/util"
+)
+
+// ConcatenateError concatenates an error with stderr string
+// FIXME: use RunStdError instead
 func ConcatenateError(err error, stderr string) error {
 	if len(stderr) == 0 {
 		return err
 	}
-	return fmt.Errorf("%w - %s", err, stderr)
+	errMsg := fmt.Sprintf("%s - %s", err.Error(), stderr)
+	return util.ErrorWrap(&runStdError{err: err, stderr: stderr, errMsg: errMsg}, "%s", errMsg)
 }

modules/gitrepo/archive.go

@@ -8,7 +8,9 @@ import (
 	"fmt"
 	"io"
 	"os"
+	"path"
 	"path/filepath"
+	"slices"
 	"strings"
 
 	"code.gitea.io/gitea/modules/git/gitcmd"
@@ -16,7 +18,7 @@ import (
 )
 
 // CreateArchive create archive content to the target path
-func CreateArchive(ctx context.Context, repo Repository, format string, target io.Writer, usePrefix bool, commitID string) error {
+func CreateArchive(ctx context.Context, repo Repository, format string, target io.Writer, usePrefix bool, commitID string, paths []string) error {
 	if format == "unknown" {
 		return fmt.Errorf("unknown format: %v", format)
 	}
@@ -28,6 +30,13 @@ func CreateArchive(ctx context.Context, repo Repository, format string, target i
 	cmd.AddOptionFormat("--format=%s", format)
 	cmd.AddDynamicArguments(commitID)
 
+	paths = slices.Clone(paths)
+	for i := range paths {
+		// although "git archive" already ensures the paths won't go outside the repo, we still clean them here for safety
+		paths[i] = path.Clean(paths[i])
+	}
+	cmd.AddDynamicArguments(paths...)
+
 	var stderr strings.Builder
 	if err := RunCmd(ctx, repo, cmd.WithStdout(target).WithStderr(&stderr)); err != nil {
 		return gitcmd.ConcatenateError(err, stderr.String())

modules/test/utils.go

@@ -4,6 +4,9 @@
 package test
 
 import (
+	"archive/tar"
+	"compress/gzip"
+	"io"
 	"net/http"
 	"net/http/httptest"
 	"os"
@@ -71,3 +74,31 @@ func SetupGiteaRoot() string {
 	_ = os.Setenv("GITEA_ROOT", giteaRoot)
 	return giteaRoot
 }
+
+func ReadAllTarGzContent(r io.Reader) (map[string]string, error) {
+	gzr, err := gzip.NewReader(r)
+	if err != nil {
+		return nil, err
+	}
+
+	content := make(map[string]string)
+
+	tr := tar.NewReader(gzr)
+	for {
+		hd, err := tr.Next()
+		if err == io.EOF {
+			break
+		}
+		if err != nil {
+			return nil, err
+		}
+
+		buf, err := io.ReadAll(tr)
+		if err != nil {
+			return nil, err
+		}
+
+		content[hd.Name] = string(buf)
+	}
+	return content, nil
+}