git: implement git on top of go-git

Code is simpler, faster, and we can get rid of ref detection (could previously already have done that, just need to pull HEAD).
hound-search · Aug 28, 2024 · 8172ac9 · 8172ac9
1 parent 69ff895
commit 8172ac9
Show file tree

Hide file tree

Showing 3 changed files with 37 additions and 222 deletions.
diff --git a/vcs/git.go b/vcs/git.go
@@ -4,35 +4,23 @@ import (
 	"bytes"
 	"encoding/json"
 	"fmt"
-	"io"
 	"log"
 	"os/exec"
-	"path/filepath"
-	"regexp"
-	"strings"
+
+	gogit "github.com/go-git/go-git/v5"
+	"github.com/go-git/go-git/v5/config"
 )
 
-const defaultRef = "master"
+const indexRef = "localindex"
 const autoGeneratedAttribute = "linguist-generated"
 
-var headBranchRegexp = regexp.MustCompile(`HEAD branch: (?P<branch>.+)`)
-
 func init() {
 	Register(newGit, "git")
 }
 
 type GitDriver struct {
-	DetectRef     bool   `json:"detect-ref"`
-	Ref           string `json:"ref"`
-	Bare          bool   `json:"bare"`
-	refDetetector refDetetector
-}
-
-type refDetetector interface {
-	detectRef(dir string) string
-}
-
-type headBranchDetector struct {
+	Ref  string `json:"ref"`
+	Bare bool   `json:"bare"`
 }
 
 func newGit(b []byte) (Driver, error) {
@@ -44,115 +32,56 @@ func newGit(b []byte) (Driver, error) {
 		}
 	}
 
-	d.refDetetector = &headBranchDetector{}
-
 	return &d, nil
 }
 
-func (g *GitDriver) HeadRev(dir string) (string, error) {
-	targetRef := "HEAD"
-	if g.Bare {
-		targetRef = fmt.Sprintf("origin/%s", g.targetRef(dir))
-	}
-
-	cmd := exec.Command(
-		"git",
-		"rev-parse",
-		targetRef)
-	cmd.Dir = dir
-	r, err := cmd.StdoutPipe()
+func (g *GitDriver) Pull(dir string) (string, error) {
+	repo, err := gogit.PlainOpen(dir)
 	if err != nil {
 		return "", err
 	}
-	defer r.Close()
 
-	if err := cmd.Start(); err != nil {
-		return "", err
+	fetchTarget := "HEAD"
+	if g.Ref != "" {
+		fetchTarget = g.Ref
 	}
 
-	var buf bytes.Buffer
-
-	if _, err := io.Copy(&buf, r); err != nil {
+	if err = repo.Fetch(&gogit.FetchOptions{
+		Depth: 1,
+		Prune: true,
+		Tags:  gogit.NoTags,
+		RefSpecs: []config.RefSpec{
+			config.RefSpec(fmt.Sprintf("+%s:refs/heads/%s", fetchTarget, indexRef)),
+		},
+	}); err != nil {
 		return "", err
 	}
 
-	return strings.TrimSpace(buf.String()), cmd.Wait()
-}
-
-func run(desc, dir, cmd string, args ...string) (string, error) {
-	c := exec.Command(cmd, args...)
-	c.Dir = dir
-	out, err := c.CombinedOutput()
+	newHead, err := repo.ResolveRevision(indexRef)
 	if err != nil {
-		log.Printf(
-			"Failed to %s %v at %q, see output below\n%s: %+v\nContinuing...",
-			desc,
-			c.Args, c.Dir,
-			out, err)
-	}
-
-	return string(out), nil
-}
-
-func (g *GitDriver) Pull(dir string) (string, error) {
-	targetRef := g.targetRef(dir)
-
-	if _, err := run("git fetch", dir,
-		"git",
-		"fetch",
-		"--prune",
-		"--no-tags",
-		"--depth", "1",
-		"origin",
-		fmt.Sprintf("+%s:remotes/origin/%s", targetRef, targetRef)); err != nil {
 		return "", err
 	}
 
 	if !g.Bare {
-		// XXX(tvdw) Check if this works without fetch?
-		if _, err := run("git reset", dir,
-			"git",
-			"reset",
-			"--hard",
-			fmt.Sprintf("origin/%s", targetRef)); err != nil {
+		worktree, err := repo.Worktree()
+		if err != nil {
 			return "", err
 		}
+		worktree.Reset(&gogit.ResetOptions{
+			Mode:   gogit.HardReset,
+			Commit: *newHead,
+		})
 	}
 
-	return g.HeadRev(dir)
-}
-
-func (g *GitDriver) targetRef(dir string) string {
-	var targetRef string
-	if g.Ref != "" {
-		targetRef = g.Ref
-	} else if g.DetectRef {
-		targetRef = g.refDetetector.detectRef(dir)
-	}
-
-	if targetRef == "" {
-		targetRef = defaultRef
-	}
-
-	return targetRef
+	return newHead.String(), nil
 }
 
 func (g *GitDriver) Clone(dir, url string) (string, error) {
-	par, rep := filepath.Split(dir)
-
-	cmdArgs := []string{
-		"clone",
-		"--depth", "1",
-	}
-	if g.Bare {
-		cmdArgs = append(cmdArgs, "--bare")
-	}
-	cmdArgs = append(cmdArgs, url, rep)
-	cmd := exec.Command("git", cmdArgs...)
-	cmd.Dir = par
-	out, err := cmd.CombinedOutput()
+	_, err := gogit.PlainClone(dir, g.Bare, &gogit.CloneOptions{
+		Depth: 1,
+		URL:   url,
+	})
 	if err != nil {
-		log.Printf("Failed to clone %s, see output below\n%sContinuing...", url, out)
 		return "", err
 	}
 
@@ -168,6 +97,11 @@ func (g *GitDriver) SpecialFiles() []string {
 func (g *GitDriver) AutoGeneratedFiles(dir string) []string {
 	var files []string
 
+	// XXX(tvdw): broken under Bare
+	if g.Bare {
+		return nil
+	}
+
 	filesCmd := exec.Command("git", "ls-files", "-z")
 	filesCmd.Dir = dir
 	pipe, err := filesCmd.StdoutPipe()
@@ -210,40 +144,9 @@ func (g *GitDriver) AutoGeneratedFiles(dir string) []string {
 	return files
 }
 
-func (d *headBranchDetector) detectRef(dir string) string {
-	output, err := run("git show remote info", dir,
-		"git",
-		"remote",
-		"show",
-		"origin",
-	)
-
-	if err != nil {
-		log.Printf(
-			"error occured when fetching info to determine target ref in %s: %s. Will fall back to default ref %s",
-			dir,
-			err,
-			defaultRef,
-		)
-		return ""
-	}
-
-	matches := headBranchRegexp.FindStringSubmatch(output)
-	if len(matches) != 2 {
-		log.Printf(
-			"could not determine target ref in %s. Will fall back to default ref %s",
-			dir,
-			defaultRef,
-		)
-		return ""
-	}
-
-	return matches[1]
-}
-
 func (g *GitDriver) FileSystem(dir string) (FileSystem, error) {
 	if g.Bare {
-		return NewGitFilesystem(dir, fmt.Sprintf("origin/%s", g.targetRef(dir)))
+		return NewGitFilesystem(dir, indexRef)
 	} else {
 		return NewDirFilesystem(dir)
 	}

diff --git a/vcs/git_test.go b/vcs/git_test.go
diff --git a/vcs/vcs.go b/vcs/vcs.go
@@ -39,9 +39,6 @@ type Driver interface {
 	// Pull new changes from the server and update the working directory.
 	Pull(dir string) (string, error)
 
-	// Return the revision at the head of the vcs directory.
-	HeadRev(dir string) (string, error)
-
 	// Return a list of special filenames that should not be indexed.
 	SpecialFiles() []string