snapshot of https://github.com/go-gitea/gitea.git (main, 2026-09-15; blobs > 131072B omitted)

modules/git/archive.go

@@ -0,0 +1,74 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"os"
+	"path"
+	"path/filepath"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/setting"
+)
+
+// CreateArchive create archive content to the target path
+func CreateArchive(ctx context.Context, repo RepositoryFacade, repoName, format string, target io.Writer, commitID string, paths []string) error {
+	if format == "unknown" {
+		return fmt.Errorf("unknown format: %v", format)
+	}
+
+	cmd := gitcmd.NewCommand("archive")
+	if setting.Repository.PrefixArchiveFiles {
+		cmd.AddOptionFormat("--prefix=%s", strings.ToLower(repoName)+"/")
+	}
+	cmd.AddOptionFormat("--format=%s", format)
+	cmd.AddDynamicArguments(commitID)
+
+	for i := range paths {
+		// although "git archive" already ensures the paths won't go outside the repo, we still clean them here for safety
+		cmd.AddDynamicArguments(path.Clean(paths[i]))
+	}
+	return cmd.WithStdoutCopy(target).WithRepo(repo).RunWithStderr(ctx)
+}
+
+// CreateBundle create bundle content to the target path
+func CreateBundle(ctx context.Context, repo RepositoryFacade, commit string, out io.Writer) error {
+	// TODO: use the following steps instead of creating a temp repo, also need to iterate and clean up outdated refs
+	// the temp ref has to be under refs/heads/*, and a clone only checks out with a HEAD, which needs the temp repo
+	// git update-ref refs/heads/bundle-temp-{timestamp} {commit}
+	// git bundle create - refs/heads/bundle-temp-{timestamp}
+	// git update-ref -d refs/heads/bundle-temp-{timestamp}
+	tmpDir, cleanup, err := setting.AppDataTempDir("git-repo-content").MkdirTempRandom("gitea-bundle")
+	if err != nil {
+		return err
+	}
+	defer cleanup()
+
+	env := append(os.Environ(), "GIT_OBJECT_DIRECTORY="+filepath.Join(gitrepo.RepoLocalPath(repo), "objects"))
+	gitTmpCmd := func() *gitcmd.Command {
+		return gitcmd.NewCommand().WithDir(tmpDir).WithEnv(env)
+	}
+
+	_, _, err = gitTmpCmd().AddArguments("init", "--bare").RunStdString(ctx)
+	if err != nil {
+		return err
+	}
+
+	_, _, err = gitTmpCmd().AddArguments("reset", "--soft").AddDynamicArguments(commit).RunStdString(ctx)
+	if err != nil {
+		return err
+	}
+
+	_, _, err = gitTmpCmd().AddArguments("branch", "-m", "bundle").RunStdString(ctx)
+	if err != nil {
+		return err
+	}
+
+	return gitTmpCmd().AddArguments("bundle", "create", "-", "bundle", "HEAD").WithStdoutCopy(out).RunWithStderr(ctx)
+}

modules/git/archive_test.go

@@ -0,0 +1,29 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+
+	"gitea.dev/modules/setting"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestCreateBundle(t *testing.T) {
+	setting.AppDataPath = t.TempDir()
+
+	buf := &bytes.Buffer{}
+	require.NoError(t, CreateBundle(t.Context(), mockRepository("repo1_bare"), "ce064814f4a0d337b333e646ece456cd39fab612", buf))
+
+	header, _, ok := strings.Cut(buf.String(), "\n\n")
+	require.True(t, ok)
+	assert.Equal(t, "# v2 git bundle", strings.Split(header, "\n")[0])
+	// without a refs/heads/* ref and a HEAD, a clone of the bundle has no branch and no checkout
+	assert.Contains(t, header, "ce064814f4a0d337b333e646ece456cd39fab612 refs/heads/bundle")
+	assert.Contains(t, header, "ce064814f4a0d337b333e646ece456cd39fab612 HEAD")
+}

modules/git/attribute/attribute.go

@@ -0,0 +1,115 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"strings"
+
+	"gitea.dev/modules/optional"
+)
+
+type Attribute string
+
+const (
+	LinguistVendored      = "linguist-vendored"
+	LinguistGenerated     = "linguist-generated"
+	LinguistDocumentation = "linguist-documentation"
+	LinguistDetectable    = "linguist-detectable"
+	LinguistLanguage      = "linguist-language"
+	GitlabLanguage        = "gitlab-language"
+	Lockable              = "lockable"
+	Filter                = "filter"
+	Diff                  = "diff"
+)
+
+var LinguistAttributes = []string{
+	LinguistVendored,
+	LinguistGenerated,
+	LinguistDocumentation,
+	LinguistDetectable,
+	LinguistLanguage,
+	GitlabLanguage,
+}
+
+func (a Attribute) IsUnspecified() bool {
+	return a == "" || a == "unspecified"
+}
+
+func (a Attribute) ToString() optional.Option[string] {
+	if !a.IsUnspecified() {
+		return optional.Some(string(a))
+	}
+	return optional.None[string]()
+}
+
+// ToBool converts the attribute value to optional boolean: true if "set"/"true", false if "unset"/"false", none otherwise
+func (a Attribute) ToBool() optional.Option[bool] {
+	switch a {
+	case "set", "true":
+		return optional.Some(true)
+	case "unset", "false":
+		return optional.Some(false)
+	}
+	return optional.None[bool]()
+}
+
+type Attributes struct {
+	m map[string]Attribute
+}
+
+func NewAttributes() *Attributes {
+	return &Attributes{m: make(map[string]Attribute)}
+}
+
+func (attrs *Attributes) Get(name string) Attribute {
+	if value, has := attrs.m[name]; has {
+		return value
+	}
+	return ""
+}
+
+func (attrs *Attributes) GetVendored() optional.Option[bool] {
+	return attrs.Get(LinguistVendored).ToBool()
+}
+
+func (attrs *Attributes) GetGenerated() optional.Option[bool] {
+	return attrs.Get(LinguistGenerated).ToBool()
+}
+
+func (attrs *Attributes) GetDocumentation() optional.Option[bool] {
+	return attrs.Get(LinguistDocumentation).ToBool()
+}
+
+func (attrs *Attributes) GetDetectable() optional.Option[bool] {
+	return attrs.Get(LinguistDetectable).ToBool()
+}
+
+func (attrs *Attributes) GetLinguistLanguage() optional.Option[string] {
+	return attrs.Get(LinguistLanguage).ToString()
+}
+
+func (attrs *Attributes) GetGitlabLanguage() optional.Option[string] {
+	attrStr := attrs.Get(GitlabLanguage).ToString()
+	if attrStr.Has() {
+		raw := attrStr.Value()
+		// gitlab-language may have additional parameters after the language
+		// ignore them and just use the main language
+		// https://docs.gitlab.com/ee/user/project/highlighting.html#override-syntax-highlighting-for-a-file-type
+		if before, _, ok := strings.Cut(raw, "?"); ok {
+			return optional.Some(before)
+		}
+	}
+	return attrStr
+}
+
+func (attrs *Attributes) GetLanguage() optional.Option[string] {
+	// prefer linguist-language over gitlab-language
+	// if linguist-language is not set, use gitlab-language
+	// if both are not set, return none
+	language := attrs.GetLinguistLanguage()
+	if language.Value() == "" {
+		language = attrs.GetGitlabLanguage()
+	}
+	return language
+}

modules/git/attribute/attribute_test.go

@@ -0,0 +1,37 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func Test_Attribute(t *testing.T) {
+	assert.Empty(t, Attribute("").ToString().Value())
+	assert.Empty(t, Attribute("unspecified").ToString().Value())
+	assert.Equal(t, "python", Attribute("python").ToString().Value())
+	assert.Equal(t, "Java", Attribute("Java").ToString().Value())
+
+	attributes := Attributes{
+		m: map[string]Attribute{
+			LinguistGenerated:     "true",
+			LinguistDocumentation: "false",
+			LinguistDetectable:    "set",
+			LinguistLanguage:      "Python",
+			GitlabLanguage:        "Java",
+			"filter":              "unspecified",
+			"test":                "",
+		},
+	}
+
+	assert.Empty(t, attributes.Get("test").ToString().Value())
+	assert.Empty(t, attributes.Get("filter").ToString().Value())
+	assert.Equal(t, "Python", attributes.Get(LinguistLanguage).ToString().Value())
+	assert.Equal(t, "Java", attributes.Get(GitlabLanguage).ToString().Value())
+	assert.True(t, attributes.Get(LinguistGenerated).ToBool().Value())
+	assert.False(t, attributes.Get(LinguistDocumentation).ToBool().Value())
+	assert.True(t, attributes.Get(LinguistDetectable).ToBool().Value())
+}

modules/git/attribute/batch.go

@@ -0,0 +1,208 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"io"
+	"time"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/log"
+)
+
+// BatchChecker provides a reader for check-attribute content that can be long running
+type BatchChecker struct {
+	attributesNum int
+	repo          *git.Repository
+	stdinWriter   io.WriteCloser
+	stdOut        *nulSeparatedAttributeWriter
+	ctx           context.Context
+	cancel        context.CancelFunc
+	cmd           *gitcmd.Command
+}
+
+// NewBatchChecker creates a check attribute reader for the current repository and provided commit ID
+// If treeish is empty, then it will use current working directory, otherwise it will use the provided treeish on the bare repo
+func NewBatchChecker(ctx context.Context, repo *git.Repository, treeish string, attributes []string) (checker *BatchChecker, returnedErr error) {
+	ctx, cancel := context.WithCancel(ctx)
+	defer func() {
+		if returnedErr != nil {
+			cancel()
+		}
+	}()
+
+	cmd, envs, cleanup, err := checkAttrCommand(ctx, repo, treeish, nil, attributes)
+	if err != nil {
+		return nil, err
+	}
+	defer func() {
+		if returnedErr != nil {
+			cleanup()
+		}
+	}()
+
+	cmd.AddArguments("--stdin")
+
+	checker = &BatchChecker{
+		attributesNum: len(attributes),
+		repo:          repo,
+		ctx:           ctx,
+		cmd:           cmd,
+		cancel: func() {
+			cancel()
+			cleanup()
+		},
+	}
+
+	stdinWriter, stdinWriterClose := cmd.MakeStdinPipe()
+	checker.stdinWriter = stdinWriter
+
+	lw := new(nulSeparatedAttributeWriter)
+	lw.attributes = make(chan attributeTriple, len(attributes))
+	lw.closed = make(chan struct{})
+	checker.stdOut = lw
+
+	cmd.WithEnv(envs).
+		WithRepo(repo).
+		WithStdoutCopy(lw)
+
+	go func() {
+		defer stdinWriterClose()
+		defer checker.cancel()
+		defer lw.Close()
+
+		err := cmd.RunWithStderr(ctx)
+		if err != nil && !gitcmd.IsErrorCanceledOrKilled(err) {
+			log.Error("Attribute checker for commit %s exits with error: %v", treeish, err)
+		}
+	}()
+
+	return checker, nil
+}
+
+// CheckPath check attr for given path
+func (c *BatchChecker) CheckPath(path string) (rs *Attributes, err error) {
+	defer func() {
+		if err != nil && err != c.ctx.Err() {
+			log.Error("Unexpected error when checking path %s in %s, error: %v", path, c.repo.LogString(), err)
+		}
+	}()
+
+	select {
+	case <-c.ctx.Done():
+		return nil, c.ctx.Err()
+	default:
+	}
+
+	if _, err = c.stdinWriter.Write([]byte(path + "\x00")); err != nil {
+		defer c.Close()
+		return nil, err
+	}
+
+	reportTimeout := func() error {
+		stdOutClosed := false
+		select {
+		case <-c.stdOut.closed:
+			stdOutClosed = true
+		default:
+		}
+		debugMsg := fmt.Sprintf("check path %q in repo %q", path, c.repo.LogString())
+		debugMsg += fmt.Sprintf(", stdOut: tmp=%q, pos=%d, closed=%v", string(c.stdOut.tmp), c.stdOut.pos, stdOutClosed)
+		if c.cmd != nil {
+			debugMsg += fmt.Sprintf(", process state: %q", c.cmd.ProcessState())
+		}
+		_ = c.Close()
+		return fmt.Errorf("CheckPath timeout: %s", debugMsg)
+	}
+
+	rs = NewAttributes()
+	for i := 0; i < c.attributesNum; i++ {
+		select {
+		case <-time.After(5 * time.Second):
+			// there is no "hang" problem now. This code is just used to catch other potential problems.
+			return nil, reportTimeout()
+		case attr, ok := <-c.stdOut.ReadAttribute():
+			if !ok {
+				return nil, c.ctx.Err()
+			}
+			rs.m[attr.Attribute] = Attribute(attr.Value)
+		case <-c.ctx.Done():
+			return nil, c.ctx.Err()
+		}
+	}
+	return rs, nil
+}
+
+func (c *BatchChecker) Close() error {
+	c.cancel()
+	err := c.stdinWriter.Close()
+	return err
+}
+
+type attributeTriple struct {
+	Filename  string
+	Attribute string
+	Value     string
+}
+
+type nulSeparatedAttributeWriter struct {
+	tmp        []byte
+	attributes chan attributeTriple
+	closed     chan struct{}
+	working    attributeTriple
+	pos        int
+}
+
+func (wr *nulSeparatedAttributeWriter) Write(p []byte) (n int, err error) {
+	l, read := len(p), 0
+
+	nulIdx := bytes.IndexByte(p, '\x00')
+	for nulIdx >= 0 {
+		wr.tmp = append(wr.tmp, p[:nulIdx]...)
+		switch wr.pos {
+		case 0:
+			wr.working = attributeTriple{
+				Filename: string(wr.tmp),
+			}
+		case 1:
+			wr.working.Attribute = string(wr.tmp)
+		case 2:
+			wr.working.Value = string(wr.tmp)
+		}
+		wr.tmp = wr.tmp[:0]
+		wr.pos++
+		if wr.pos > 2 {
+			wr.attributes <- wr.working
+			wr.pos = 0
+		}
+		read += nulIdx + 1
+		if l > read {
+			p = p[nulIdx+1:]
+			nulIdx = bytes.IndexByte(p, '\x00')
+		} else {
+			return l, nil
+		}
+	}
+	wr.tmp = append(wr.tmp, p...)
+	return l, nil
+}
+
+func (wr *nulSeparatedAttributeWriter) ReadAttribute() <-chan attributeTriple {
+	return wr.attributes
+}
+
+func (wr *nulSeparatedAttributeWriter) Close() error {
+	select {
+	case <-wr.closed:
+		return nil
+	default:
+	}
+	close(wr.attributes)
+	close(wr.closed)
+	return nil
+}

modules/git/attribute/batch_test.go

@@ -0,0 +1,173 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"path/filepath"
+	"testing"
+	"time"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func Test_nulSeparatedAttributeWriter_ReadAttribute(t *testing.T) {
+	wr := &nulSeparatedAttributeWriter{
+		attributes: make(chan attributeTriple, 5),
+	}
+
+	testStr := ".gitignore\"\n\x00linguist-vendored\x00unspecified\x00"
+
+	n, err := wr.Write([]byte(testStr))
+
+	assert.Len(t, testStr, n)
+	assert.NoError(t, err)
+	select {
+	case attr := <-wr.ReadAttribute():
+		assert.Equal(t, ".gitignore\"\n", attr.Filename)
+		assert.Equal(t, LinguistVendored, attr.Attribute)
+		assert.Equal(t, "unspecified", attr.Value)
+	case <-time.After(100 * time.Millisecond):
+		assert.FailNow(t, "took too long to read an attribute from the list")
+	}
+	// Write a second attribute again
+	n, err = wr.Write([]byte(testStr))
+
+	assert.Len(t, testStr, n)
+	assert.NoError(t, err)
+
+	select {
+	case attr := <-wr.ReadAttribute():
+		assert.Equal(t, ".gitignore\"\n", attr.Filename)
+		assert.Equal(t, LinguistVendored, attr.Attribute)
+		assert.Equal(t, "unspecified", attr.Value)
+	case <-time.After(100 * time.Millisecond):
+		assert.FailNow(t, "took too long to read an attribute from the list")
+	}
+
+	// Write a partial attribute
+	_, err = wr.Write([]byte("incomplete-file"))
+	assert.NoError(t, err)
+	_, err = wr.Write([]byte("name\x00"))
+	assert.NoError(t, err)
+
+	select {
+	case <-wr.ReadAttribute():
+		assert.FailNow(t, "There should not be an attribute ready to read")
+	case <-time.After(100 * time.Millisecond):
+	}
+	_, err = wr.Write([]byte("attribute\x00"))
+	assert.NoError(t, err)
+	select {
+	case <-wr.ReadAttribute():
+		assert.FailNow(t, "There should not be an attribute ready to read")
+	case <-time.After(100 * time.Millisecond):
+	}
+
+	_, err = wr.Write([]byte("value\x00"))
+	assert.NoError(t, err)
+
+	attr := <-wr.ReadAttribute()
+	assert.Equal(t, "incomplete-filename", attr.Filename)
+	assert.Equal(t, "attribute", attr.Attribute)
+	assert.Equal(t, "value", attr.Value)
+
+	_, err = wr.Write([]byte("shouldbe.vendor\x00linguist-vendored\x00set\x00shouldbe.vendor\x00linguist-generated\x00unspecified\x00shouldbe.vendor\x00linguist-language\x00unspecified\x00"))
+	assert.NoError(t, err)
+	attr = <-wr.ReadAttribute()
+	assert.NoError(t, err)
+	assert.Equal(t, attributeTriple{
+		Filename:  "shouldbe.vendor",
+		Attribute: LinguistVendored,
+		Value:     "set",
+	}, attr)
+	attr = <-wr.ReadAttribute()
+	assert.NoError(t, err)
+	assert.Equal(t, attributeTriple{
+		Filename:  "shouldbe.vendor",
+		Attribute: LinguistGenerated,
+		Value:     "unspecified",
+	}, attr)
+	attr = <-wr.ReadAttribute()
+	assert.NoError(t, err)
+	assert.Equal(t, attributeTriple{
+		Filename:  "shouldbe.vendor",
+		Attribute: LinguistLanguage,
+		Value:     "unspecified",
+	}, attr)
+}
+
+func expectedAttrs() *Attributes {
+	return &Attributes{
+		m: map[string]Attribute{
+			LinguistGenerated:     "unspecified",
+			LinguistDetectable:    "unspecified",
+			LinguistDocumentation: "unspecified",
+			LinguistVendored:      "unspecified",
+			LinguistLanguage:      "Python",
+			GitlabLanguage:        "unspecified",
+		},
+	}
+}
+
+func Test_BatchChecker(t *testing.T) {
+	setting.AppDataPath = t.TempDir()
+	repoPath := "../tests/repos/language_stats_repo"
+	ctx := t.Context()
+	gitRepo, err := git.OpenRepositoryLocal(ctx, repoPath)
+	require.NoError(t, err)
+	defer gitRepo.Close()
+
+	commitID := "8fee858da5796dfb37704761701bb8e800ad9ef3"
+
+	t.Run("Create index file to run git check-attr", func(t *testing.T) {
+		defer test.MockVariableValue(&git.DefaultFeatures().SupportCheckAttrOnBare, false)()
+		checker, err := NewBatchChecker(ctx, gitRepo, commitID, LinguistAttributes)
+		assert.NoError(t, err)
+		defer checker.Close()
+		attributes, err := checker.CheckPath("i-am-a-python.p")
+		assert.NoError(t, err)
+		assert.Equal(t, expectedAttrs(), attributes)
+	})
+
+	// run git check-attr on work tree
+	t.Run("Run git check-attr on git work tree", func(t *testing.T) {
+		dir := filepath.Join(t.TempDir(), "test-repo")
+		err := git.Clone(t.Context(), repoPath, dir, git.CloneRepoOptions{
+			Shared: true,
+			Branch: "master",
+		})
+		assert.NoError(t, err)
+
+		tempRepo, err := git.OpenRepositoryLocal(ctx, dir)
+		assert.NoError(t, err)
+		defer tempRepo.Close()
+
+		checker, err := NewBatchChecker(t.Context(), tempRepo, "", LinguistAttributes)
+		assert.NoError(t, err)
+		defer checker.Close()
+		attributes, err := checker.CheckPath("i-am-a-python.p")
+		assert.NoError(t, err)
+		assert.Equal(t, expectedAttrs(), attributes)
+	})
+
+	if !git.DefaultFeatures().SupportCheckAttrOnBare {
+		t.Skip("git version 2.40 is required to support run check-attr on bare repo")
+		return
+	}
+
+	t.Run("Run git check-attr in bare repository", func(t *testing.T) {
+		checker, err := NewBatchChecker(ctx, gitRepo, commitID, LinguistAttributes)
+		assert.NoError(t, err)
+		defer checker.Close()
+
+		attributes, err := checker.CheckPath("i-am-a-python.p")
+		assert.NoError(t, err)
+		assert.Equal(t, expectedAttrs(), attributes)
+	})
+}

modules/git/attribute/checker.go

@@ -0,0 +1,97 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"fmt"
+	"os"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/git/gitcmd"
+)
+
+func checkAttrCommand(ctx context.Context, gitRepo *git.Repository, treeish string, filenames, attributes []string) (*gitcmd.Command, []string, func(), error) {
+	cancel := func() {}
+	envs := []string{"GIT_FLUSH=1"}
+	cmd := gitcmd.NewCommand("check-attr", "-z")
+	if len(attributes) == 0 {
+		cmd.AddArguments("--all")
+	}
+
+	// there is treeish, read from bare repo or temp index created by "read-tree"
+	if treeish != "" {
+		if git.DefaultFeatures().SupportCheckAttrOnBare {
+			cmd.AddArguments("--source")
+			cmd.AddDynamicArguments(treeish)
+		} else {
+			indexFilename, worktree, deleteTemporaryFile, err := gitRepo.ReadTreeToTemporaryIndex(ctx, treeish)
+			if err != nil {
+				return nil, nil, nil, err
+			}
+
+			cmd.AddArguments("--cached")
+			envs = append(envs,
+				"GIT_INDEX_FILE="+indexFilename,
+				"GIT_WORK_TREE="+worktree,
+			)
+			cancel = deleteTemporaryFile
+		}
+	} else {
+		// Read from existing index, in cases where the repo is bare and has an index,
+		// or the work tree contains unstaged changes that shouldn't affect the attribute check.
+		// It is caller's responsibility to add changed ".gitattributes" into the index if they want to respect the new changes.
+		cmd.AddArguments("--cached")
+	}
+
+	cmd.AddDynamicArguments(attributes...)
+	if len(filenames) > 0 {
+		cmd.AddDashesAndList(filenames...)
+	}
+	return cmd, envs, cancel, nil
+}
+
+type CheckAttributeOpts struct {
+	Filenames  []string
+	Attributes []string
+}
+
+// CheckAttributes return the attributes of the given filenames and attributes in the given treeish.
+// If treeish is empty, then it will use current working directory, otherwise it will use the provided treeish on the bare repo
+func CheckAttributes(ctx context.Context, gitRepo *git.Repository, treeish string, opts CheckAttributeOpts) (map[string]*Attributes, error) {
+	cmd, envs, cancel, err := checkAttrCommand(ctx, gitRepo, treeish, opts.Filenames, opts.Attributes)
+	if err != nil {
+		return nil, err
+	}
+	defer cancel()
+
+	stdout, _, err := cmd.WithEnv(append(os.Environ(), envs...)).
+		WithRepo(gitRepo).
+		RunStdBytes(ctx)
+	if err != nil {
+		return nil, fmt.Errorf("failed to run check-attr: %w", err)
+	}
+
+	fields := bytes.Split(stdout, []byte{'\000'})
+	if len(fields)%3 != 1 {
+		return nil, errors.New("wrong number of fields in return from check-attr")
+	}
+
+	attributesMap := make(map[string]*Attributes)
+	for i := 0; i < (len(fields) / 3); i++ {
+		filename := string(fields[3*i])
+		attribute := string(fields[3*i+1])
+		info := string(fields[3*i+2])
+		attribute2info, ok := attributesMap[filename]
+		if !ok {
+			attribute2info = NewAttributes()
+			attributesMap[filename] = attribute2info
+		}
+		attribute2info.m[attribute] = Attribute(info)
+	}
+
+	return attributesMap, nil
+}

modules/git/attribute/checker_test.go

@@ -0,0 +1,84 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func Test_Checker(t *testing.T) {
+	setting.AppDataPath = t.TempDir()
+	repoPath := "../tests/repos/language_stats_repo"
+	gitRepo, err := git.OpenRepositoryLocal(t.Context(), repoPath)
+	require.NoError(t, err)
+	defer gitRepo.Close()
+
+	commitID := "8fee858da5796dfb37704761701bb8e800ad9ef3"
+
+	t.Run("Create index file to run git check-attr", func(t *testing.T) {
+		defer test.MockVariableValue(&git.DefaultFeatures().SupportCheckAttrOnBare, false)()
+		attrs, err := CheckAttributes(t.Context(), gitRepo, commitID, CheckAttributeOpts{
+			Filenames:  []string{"i-am-a-python.p"},
+			Attributes: LinguistAttributes,
+		})
+		assert.NoError(t, err)
+		assert.Len(t, attrs, 1)
+		assert.Equal(t, expectedAttrs(), attrs["i-am-a-python.p"])
+	})
+
+	// run git check-attr on work tree
+	t.Run("Run git check-attr on git work tree", func(t *testing.T) {
+		dir := filepath.Join(t.TempDir(), "test-repo")
+		err := git.Clone(t.Context(), repoPath, dir, git.CloneRepoOptions{
+			Shared: true,
+			Branch: "master",
+		})
+		assert.NoError(t, err)
+
+		tempRepo, err := git.OpenRepositoryLocal(t.Context(), dir)
+		assert.NoError(t, err)
+		defer tempRepo.Close()
+
+		attrs, err := CheckAttributes(t.Context(), tempRepo, "", CheckAttributeOpts{
+			Filenames:  []string{"i-am-a-python.p"},
+			Attributes: LinguistAttributes,
+		})
+		assert.NoError(t, err)
+		assert.Len(t, attrs, 1)
+		assert.Equal(t, expectedAttrs(), attrs["i-am-a-python.p"])
+	})
+
+	t.Run("Run git check-attr in bare repository using index", func(t *testing.T) {
+		attrs, err := CheckAttributes(t.Context(), gitRepo, "", CheckAttributeOpts{
+			Filenames:  []string{"i-am-a-python.p"},
+			Attributes: LinguistAttributes,
+		})
+		assert.NoError(t, err)
+		assert.Len(t, attrs, 1)
+		assert.Equal(t, expectedAttrs(), attrs["i-am-a-python.p"])
+	})
+
+	if !git.DefaultFeatures().SupportCheckAttrOnBare {
+		t.Skip("git version 2.40 is required to support run check-attr on bare repo without using index")
+		return
+	}
+
+	t.Run("Run git check-attr in bare repository", func(t *testing.T) {
+		attrs, err := CheckAttributes(t.Context(), gitRepo, commitID, CheckAttributeOpts{
+			Filenames:  []string{"i-am-a-python.p"},
+			Attributes: LinguistAttributes,
+		})
+		assert.NoError(t, err)
+		assert.Len(t, attrs, 1)
+		assert.Equal(t, expectedAttrs(), attrs["i-am-a-python.p"])
+	})
+}

modules/git/attribute/main_test.go

@@ -0,0 +1,39 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package attribute
+
+import (
+	"fmt"
+	"os"
+	"testing"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/setting"
+)
+
+func testRun(m *testing.M) error {
+	gitHomePath, err := os.MkdirTemp(os.TempDir(), "git-home")
+	if err != nil {
+		return fmt.Errorf("unable to create temp dir: %w", err)
+	}
+	defer os.RemoveAll(gitHomePath)
+	setting.Git.HomePath = gitHomePath
+
+	if err = git.InitFull(); err != nil {
+		return fmt.Errorf("failed to call Init: %w", err)
+	}
+
+	exitCode := m.Run()
+	if exitCode != 0 {
+		return fmt.Errorf("run test failed, ExitCode=%d", exitCode)
+	}
+	return nil
+}
+
+func TestMain(m *testing.M) {
+	if err := testRun(m); err != nil {
+		_, _ = fmt.Fprintf(os.Stderr, "Test failed: %v", err)
+		os.Exit(1)
+	}
+}

modules/git/blame.go

@@ -0,0 +1,207 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"bytes"
+	"context"
+	"io"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/setting"
+)
+
+func LineBlame(ctx context.Context, repo RepositoryFacade, revision, file string, line uint) (string, error) {
+	stdout, _, err := gitcmd.NewCommand("blame").WithRepo(repo).
+		AddOptionFormat("-L %d,%d", line, line).
+		AddOptionValues("-p", revision).
+		AddDashesAndList(file).
+		RunStdString(ctx)
+	return stdout, err
+}
+
+// BlamePart represents block of blame - continuous lines with one sha
+type BlamePart struct {
+	Sha          string
+	Lines        []string
+	PreviousSha  string
+	PreviousPath string
+}
+
+// BlameReader returns part of file blame one by one
+type BlameReader struct {
+	bufferedReader *bufio.Reader
+	done           chan error
+	lastSha        *string
+	ignoreRevsFile string
+	objectFormat   ObjectFormat
+	cleanupFuncs   []func()
+}
+
+func (r *BlameReader) UsesIgnoreRevs() bool {
+	return r.ignoreRevsFile != ""
+}
+
+// NextPart returns next part of blame (sequential code lines with the same commit)
+func (r *BlameReader) NextPart() (*BlamePart, error) {
+	var blamePart *BlamePart
+
+	if r.lastSha != nil {
+		blamePart = &BlamePart{
+			Sha:   *r.lastSha,
+			Lines: make([]string, 0),
+		}
+	}
+
+	const previousHeader = "previous "
+	var lineBytes []byte
+	var isPrefix bool
+	var err error
+
+	for err != io.EOF {
+		lineBytes, isPrefix, err = r.bufferedReader.ReadLine()
+		if err != nil && err != io.EOF {
+			return blamePart, err
+		}
+
+		if len(lineBytes) == 0 {
+			// isPrefix will be false
+			continue
+		}
+
+		var objectID string
+		objectFormatLength := r.objectFormat.FullLength()
+
+		if len(lineBytes) > objectFormatLength && lineBytes[objectFormatLength] == ' ' && r.objectFormat.IsValid(string(lineBytes[0:objectFormatLength])) {
+			objectID = string(lineBytes[0:objectFormatLength])
+		}
+		if len(objectID) > 0 {
+			if blamePart == nil {
+				blamePart = &BlamePart{
+					Sha:   objectID,
+					Lines: make([]string, 0),
+				}
+			}
+
+			if blamePart.Sha != objectID {
+				r.lastSha = &objectID
+				// need to munch to end of line...
+				for isPrefix {
+					_, isPrefix, err = r.bufferedReader.ReadLine()
+					if err != nil && err != io.EOF {
+						return blamePart, err
+					}
+				}
+				return blamePart, nil
+			}
+		} else if lineBytes[0] == '\t' {
+			blamePart.Lines = append(blamePart.Lines, string(lineBytes[1:]))
+		} else if bytes.HasPrefix(lineBytes, []byte(previousHeader)) {
+			offset := len(previousHeader) // already includes a space
+			blamePart.PreviousSha = string(lineBytes[offset : offset+objectFormatLength])
+			offset += objectFormatLength + 1 // +1 for space
+			blamePart.PreviousPath = string(lineBytes[offset:])
+		}
+
+		// need to munch to end of line...
+		for isPrefix {
+			_, isPrefix, err = r.bufferedReader.ReadLine()
+			if err != nil && err != io.EOF {
+				return blamePart, err
+			}
+		}
+	}
+
+	r.lastSha = nil
+
+	return blamePart, nil
+}
+
+// Close BlameReader - don't run NextPart after invoking that
+func (r *BlameReader) Close() error {
+	if r.bufferedReader == nil {
+		return nil
+	}
+
+	err := <-r.done
+	r.bufferedReader = nil
+	r.cleanup()
+	return err
+}
+
+func (r *BlameReader) cleanup() {
+	for _, cleanup := range r.cleanupFuncs {
+		cleanup()
+	}
+}
+
+// CreateBlameReader creates reader for given git.RepositoryFacade, commit and file
+func CreateBlameReader(ctx context.Context, objectFormat ObjectFormat, repo RepositoryFacade, gitRepo *Repository, commit *Commit, file string, bypassBlameIgnore bool) (rd *BlameReader, retErr error) {
+	defer func() {
+		if retErr != nil {
+			rd.cleanup()
+		}
+	}()
+
+	rd = &BlameReader{
+		done:         make(chan error, 1),
+		objectFormat: objectFormat,
+	}
+
+	cmd := gitcmd.NewCommand("blame", "--porcelain")
+
+	stdoutReader, stdoutReaderClose := cmd.MakeStdoutPipe()
+	rd.bufferedReader = bufio.NewReader(stdoutReader)
+	rd.cleanupFuncs = append(rd.cleanupFuncs, stdoutReaderClose)
+
+	if !bypassBlameIgnore {
+		ignoreRevsFileName, ignoreRevsFileCleanup, err := tryCreateBlameIgnoreRevsFile(ctx, gitRepo, commit)
+		if err != nil && !IsErrNotExist(err) {
+			return nil, err
+		} else if err == nil {
+			rd.ignoreRevsFile = ignoreRevsFileName
+			rd.cleanupFuncs = append(rd.cleanupFuncs, ignoreRevsFileCleanup)
+			// Possible improvement: use --ignore-revs-file /dev/stdin on unix
+			// There is no equivalent on Windows. May be implemented if Gitea uses an external git backend.
+			cmd.AddOptionValues("--ignore-revs-file", ignoreRevsFileName)
+		}
+	}
+
+	cmd.AddDynamicArguments(commit.ID.String()).AddDashesAndList(file)
+
+	go func() {
+		// TODO: it doesn't work for directories (the directories shouldn't be "blamed"), and the "err" should be returned by "Read" but not by "Close"
+		rd.done <- cmd.WithRepo(repo).RunWithStderr(ctx)
+	}()
+
+	return rd, nil
+}
+
+func tryCreateBlameIgnoreRevsFile(ctx context.Context, gitRepo *Repository, commit *Commit) (string, func(), error) {
+	entry, err := commit.GetTreeEntryByPath(ctx, gitRepo, ".git-blame-ignore-revs")
+	if err != nil {
+		return "", nil, err
+	}
+
+	r, err := entry.Blob(gitRepo).DataAsync(ctx)
+	if err != nil {
+		return "", nil, err
+	}
+	defer r.Close()
+
+	f, cleanup, err := setting.AppDataTempDir("git-repo-content").CreateTempFileRandom("git-blame-ignore-revs")
+	if err != nil {
+		return "", nil, err
+	}
+	filename := f.Name()
+	_, err = io.Copy(f, r)
+	_ = f.Close()
+	if err != nil {
+		cleanup()
+		return "", nil, err
+	}
+
+	return filename, cleanup, nil
+}

modules/git/blame_sha256_test.go

@@ -0,0 +1,155 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"testing"
+
+	"gitea.dev/modules/setting"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestReadingBlameOutputSha256(t *testing.T) {
+	setting.AppDataPath = t.TempDir()
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	if DefaultFeatures().UsingGogit {
+		t.Skip("Skipping test since gogit does not support sha256")
+		return
+	}
+
+	t.Run("Without .git-blame-ignore-revs", func(t *testing.T) {
+		storage := mockRepository("repo5_pulls_sha256")
+		repo, err := OpenRepository(ctx, storage)
+		assert.NoError(t, err)
+		defer repo.Close()
+
+		commit, err := repo.GetCommit(t.Context(), "0b69b7bb649b5d46e14cabb6468685e5dd721290acc7ffe604d37cde57927345")
+		assert.NoError(t, err)
+
+		parts := []*BlamePart{
+			{
+				Sha: "1e35a51dc00fd7de730344c07061acfe80e8117e075ac979b6a29a3a045190ca",
+				Lines: []string{
+					"# test_repo",
+					"Test repository for testing migration from github to gitea",
+				},
+			},
+			{
+				Sha:          "0b69b7bb649b5d46e14cabb6468685e5dd721290acc7ffe604d37cde57927345",
+				Lines:        []string{"", "Do not make any changes to this repo it is used for unit testing"},
+				PreviousSha:  "1e35a51dc00fd7de730344c07061acfe80e8117e075ac979b6a29a3a045190ca",
+				PreviousPath: "README.md",
+			},
+		}
+
+		for _, bypass := range []bool{false, true} {
+			blameReader, err := CreateBlameReader(ctx, Sha256ObjectFormat, storage, repo, commit, "README.md", bypass)
+			assert.NoError(t, err)
+			assert.NotNil(t, blameReader)
+			defer blameReader.Close()
+
+			assert.False(t, blameReader.UsesIgnoreRevs())
+
+			for _, part := range parts {
+				actualPart, err := blameReader.NextPart()
+				assert.NoError(t, err)
+				assert.Equal(t, part, actualPart)
+			}
+
+			// make sure all parts have been read
+			actualPart, err := blameReader.NextPart()
+			assert.Nil(t, actualPart)
+			assert.NoError(t, err)
+		}
+	})
+
+	t.Run("With .git-blame-ignore-revs", func(t *testing.T) {
+		storage := mockRepository("repo6_blame_sha256")
+		repo, err := OpenRepository(ctx, storage)
+		assert.NoError(t, err)
+		defer repo.Close()
+
+		full := []*BlamePart{
+			{
+				Sha:   "ab2b57a4fa476fb2edb74dafa577caf918561abbaa8fba0c8dc63c412e17a7cc",
+				Lines: []string{"line", "line"},
+			},
+			{
+				Sha:          "9347b0198cd1f25017579b79d0938fa89dba34ad2514f0dd92f6bc975ed1a2fe",
+				Lines:        []string{"changed line"},
+				PreviousSha:  "ab2b57a4fa476fb2edb74dafa577caf918561abbaa8fba0c8dc63c412e17a7cc",
+				PreviousPath: "blame.txt",
+			},
+			{
+				Sha:   "ab2b57a4fa476fb2edb74dafa577caf918561abbaa8fba0c8dc63c412e17a7cc",
+				Lines: []string{"line", "line", ""},
+			},
+		}
+
+		cases := []struct {
+			CommitID       string
+			UsesIgnoreRevs bool
+			Bypass         bool
+			Parts          []*BlamePart
+		}{
+			{
+				CommitID:       "e2f5660e15159082902960af0ed74fc144921d2b0c80e069361853b3ece29ba3",
+				UsesIgnoreRevs: true,
+				Bypass:         false,
+				Parts: []*BlamePart{
+					{
+						Sha:   "ab2b57a4fa476fb2edb74dafa577caf918561abbaa8fba0c8dc63c412e17a7cc",
+						Lines: []string{"line", "line", "changed line", "line", "line", ""},
+					},
+				},
+			},
+			{
+				CommitID:       "e2f5660e15159082902960af0ed74fc144921d2b0c80e069361853b3ece29ba3",
+				UsesIgnoreRevs: false,
+				Bypass:         true,
+				Parts:          full,
+			},
+			{
+				CommitID:       "9347b0198cd1f25017579b79d0938fa89dba34ad2514f0dd92f6bc975ed1a2fe",
+				UsesIgnoreRevs: false,
+				Bypass:         false,
+				Parts:          full,
+			},
+			{
+				CommitID:       "9347b0198cd1f25017579b79d0938fa89dba34ad2514f0dd92f6bc975ed1a2fe",
+				UsesIgnoreRevs: false,
+				Bypass:         false,
+				Parts:          full,
+			},
+		}
+
+		objectFormat, err := repo.GetObjectFormat(t.Context())
+		assert.NoError(t, err)
+		for _, c := range cases {
+			commit, err := repo.GetCommit(t.Context(), c.CommitID)
+			assert.NoError(t, err)
+			blameReader, err := CreateBlameReader(ctx, objectFormat, storage, repo, commit, "blame.txt", c.Bypass)
+			assert.NoError(t, err)
+			assert.NotNil(t, blameReader)
+			defer blameReader.Close()
+
+			assert.Equal(t, c.UsesIgnoreRevs, blameReader.UsesIgnoreRevs())
+
+			for _, part := range c.Parts {
+				actualPart, err := blameReader.NextPart()
+				assert.NoError(t, err)
+				assert.Equal(t, part, actualPart)
+			}
+
+			// make sure all parts have been read
+			actualPart, err := blameReader.NextPart()
+			assert.Nil(t, actualPart)
+			assert.NoError(t, err)
+		}
+	})
+}

modules/git/blame_test.go

@@ -0,0 +1,150 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"testing"
+
+	"gitea.dev/modules/setting"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestReadingBlameOutput(t *testing.T) {
+	setting.AppDataPath = t.TempDir()
+	ctx, cancel := context.WithCancel(t.Context())
+	defer cancel()
+
+	t.Run("Without .git-blame-ignore-revs", func(t *testing.T) {
+		storage := mockRepository("repo5_pulls")
+		repo, err := OpenRepository(ctx, storage)
+		assert.NoError(t, err)
+		defer repo.Close()
+		commit, err := repo.GetCommit(t.Context(), "f32b0a9dfd09a60f616f29158f772cedd89942d2")
+		assert.NoError(t, err)
+
+		parts := []*BlamePart{
+			{
+				Sha: "72866af952e98d02a73003501836074b286a78f6",
+				Lines: []string{
+					"# test_repo",
+					"Test repository for testing migration from github to gitea",
+				},
+			},
+			{
+				Sha:          "f32b0a9dfd09a60f616f29158f772cedd89942d2",
+				Lines:        []string{"", "Do not make any changes to this repo it is used for unit testing"},
+				PreviousSha:  "72866af952e98d02a73003501836074b286a78f6",
+				PreviousPath: "README.md",
+			},
+		}
+
+		for _, bypass := range []bool{false, true} {
+			blameReader, err := CreateBlameReader(ctx, Sha1ObjectFormat, storage, repo, commit, "README.md", bypass)
+			assert.NoError(t, err)
+			assert.NotNil(t, blameReader)
+			defer blameReader.Close()
+
+			assert.False(t, blameReader.UsesIgnoreRevs())
+
+			for _, part := range parts {
+				actualPart, err := blameReader.NextPart()
+				assert.NoError(t, err)
+				assert.Equal(t, part, actualPart)
+			}
+
+			// make sure all parts have been read
+			actualPart, err := blameReader.NextPart()
+			assert.Nil(t, actualPart)
+			assert.NoError(t, err)
+		}
+	})
+
+	t.Run("With .git-blame-ignore-revs", func(t *testing.T) {
+		storage := mockRepository("repo6_blame")
+		repo, err := OpenRepository(ctx, storage)
+		assert.NoError(t, err)
+		defer repo.Close()
+
+		full := []*BlamePart{
+			{
+				Sha:   "af7486bd54cfc39eea97207ca666aa69c9d6df93",
+				Lines: []string{"line", "line"},
+			},
+			{
+				Sha:          "45fb6cbc12f970b04eacd5cd4165edd11c8d7376",
+				Lines:        []string{"changed line"},
+				PreviousSha:  "af7486bd54cfc39eea97207ca666aa69c9d6df93",
+				PreviousPath: "blame.txt",
+			},
+			{
+				Sha:   "af7486bd54cfc39eea97207ca666aa69c9d6df93",
+				Lines: []string{"line", "line", ""},
+			},
+		}
+
+		cases := []struct {
+			CommitID       string
+			UsesIgnoreRevs bool
+			Bypass         bool
+			Parts          []*BlamePart
+		}{
+			{
+				CommitID:       "544d8f7a3b15927cddf2299b4b562d6ebd71b6a7",
+				UsesIgnoreRevs: true,
+				Bypass:         false,
+				Parts: []*BlamePart{
+					{
+						Sha:   "af7486bd54cfc39eea97207ca666aa69c9d6df93",
+						Lines: []string{"line", "line", "changed line", "line", "line", ""},
+					},
+				},
+			},
+			{
+				CommitID:       "544d8f7a3b15927cddf2299b4b562d6ebd71b6a7",
+				UsesIgnoreRevs: false,
+				Bypass:         true,
+				Parts:          full,
+			},
+			{
+				CommitID:       "45fb6cbc12f970b04eacd5cd4165edd11c8d7376",
+				UsesIgnoreRevs: false,
+				Bypass:         false,
+				Parts:          full,
+			},
+			{
+				CommitID:       "45fb6cbc12f970b04eacd5cd4165edd11c8d7376",
+				UsesIgnoreRevs: false,
+				Bypass:         false,
+				Parts:          full,
+			},
+		}
+
+		objectFormat, err := repo.GetObjectFormat(t.Context())
+		assert.NoError(t, err)
+		for _, c := range cases {
+			commit, err := repo.GetCommit(t.Context(), c.CommitID)
+			assert.NoError(t, err)
+
+			blameReader, err := CreateBlameReader(ctx, objectFormat, storage, repo, commit, "blame.txt", c.Bypass)
+			assert.NoError(t, err)
+			assert.NotNil(t, blameReader)
+			defer blameReader.Close()
+
+			assert.Equal(t, c.UsesIgnoreRevs, blameReader.UsesIgnoreRevs())
+
+			for _, part := range c.Parts {
+				actualPart, err := blameReader.NextPart()
+				assert.NoError(t, err)
+				assert.Equal(t, part, actualPart)
+			}
+
+			// make sure all parts have been read
+			actualPart, err := blameReader.NextPart()
+			assert.Nil(t, actualPart)
+			assert.NoError(t, err)
+		}
+	})
+}

modules/git/blob.go

@@ -0,0 +1,116 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bytes"
+	"context"
+	"encoding/base64"
+	"errors"
+	"io"
+	"strings"
+
+	"gitea.dev/modules/util"
+)
+
+// This file contains common functions between the gogit and !gogit variants for git Blobs
+
+// Name returns name of the tree entry this blob object was created from (or empty string)
+func (b *Blob) Name() string {
+	return b.name
+}
+
+// GetBlobBytes Gets the limited content of the blob
+func (b *Blob) GetBlobBytes(ctx context.Context, limit int64) ([]byte, error) {
+	if limit <= 0 {
+		return nil, nil
+	}
+	dataRc, err := b.DataAsync(ctx)
+	if err != nil {
+		return nil, err
+	}
+	defer dataRc.Close()
+	return util.ReadWithLimit(dataRc, int(limit))
+}
+
+// GetBlobContent Gets the limited content of the blob as raw text
+func (b *Blob) GetBlobContent(ctx context.Context, limit int64) (string, error) {
+	buf, err := b.GetBlobBytes(ctx, limit)
+	return string(buf), err
+}
+
+// GetBlobLineCount gets line count of the blob.
+// It will also try to write the content to w if it's not nil, then we could pre-fetch the content without reading it again.
+func (b *Blob) GetBlobLineCount(ctx context.Context, w io.Writer) (size int64, count int, _ error) {
+	reader, err := b.DataAsync(ctx)
+	if err != nil {
+		return 0, 0, err
+	}
+	defer reader.Close()
+	return getBlobLineCount(reader, w)
+}
+
+func getBlobLineCount(r io.Reader, w io.Writer) (size int64, count int, _ error) {
+	buf := make([]byte, 32*1024)
+	size, count = 0, 0
+	var lastChar byte
+	lineSep := []byte("\n")
+	for {
+		c, err := r.Read(buf)
+		size += int64(c)
+		if w != nil {
+			if _, err := w.Write(buf[:c]); err != nil {
+				return size, count, err
+			}
+		}
+		if c > 0 {
+			count += bytes.Count(buf[:c], lineSep)
+			lastChar = buf[c-1]
+		}
+		switch {
+		case errors.Is(err, io.EOF):
+			if size > 0 && lastChar != '\n' {
+				// it should match "git diff" hunk line number. "a\nb" => 2 lines, "a\nb\n" => 2 lines
+				count++
+			}
+			return size, count, nil
+		case err != nil:
+			return size, count, err
+		}
+	}
+}
+
+// GetBlobContentBase64 Reads the content of the blob with a base64 encoding and returns the encoded string
+func (b *Blob) GetBlobContentBase64(ctx context.Context, originContent *strings.Builder) (string, error) {
+	dataRc, err := b.DataAsync(ctx)
+	if err != nil {
+		return "", err
+	}
+	defer dataRc.Close()
+
+	base64buf := &strings.Builder{}
+	encoder := base64.NewEncoder(base64.StdEncoding, base64buf)
+	buf := make([]byte, 32*1024)
+loop:
+	for {
+		n, err := dataRc.Read(buf)
+		if n > 0 {
+			if originContent != nil {
+				_, _ = originContent.Write(buf[:n])
+			}
+			if _, err := encoder.Write(buf[:n]); err != nil {
+				return "", err
+			}
+		}
+		switch {
+		case errors.Is(err, io.EOF):
+			break loop
+		case err != nil:
+			return "", err
+		}
+	}
+	_ = encoder.Close()
+	return base64buf.String(), nil
+}

modules/git/blob_gogit.go

@@ -0,0 +1,47 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build gogit
+
+package git
+
+import (
+	"context"
+	"io"
+
+	"gitea.dev/modules/log"
+
+	"github.com/go-git/go-git/v5/plumbing"
+)
+
+// Blob represents a Git object.
+type Blob struct {
+	ID   ObjectID
+	repo *Repository
+	name string
+}
+
+func (b *Blob) gogitEncodedObj() (plumbing.EncodedObject, error) {
+	return b.repo.gogitRepo.Storer.EncodedObject(plumbing.AnyObject, plumbing.Hash(b.ID.RawValue()))
+}
+
+// DataAsync gets a ReadCloser for the contents of a blob without reading it all.
+// Calling the Close function on the result will discard all unread output.
+func (b *Blob) DataAsync(_ context.Context) (io.ReadCloser, error) {
+	obj, err := b.gogitEncodedObj()
+	if err != nil {
+		return nil, err
+	}
+	return obj.Reader()
+}
+
+// Size returns the uncompressed size of the blob
+func (b *Blob) Size(_ context.Context) int64 {
+	obj, err := b.gogitEncodedObj()
+	if err != nil {
+		log.Error("Error getting gogit encoded object for blob %s(%s): %v", b.name, b.ID.String(), err)
+		return 0
+	}
+	return obj.Size()
+}

modules/git/blob_nogogit.go

@@ -0,0 +1,108 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build !gogit
+
+package git
+
+import (
+	"context"
+	"io"
+
+	"gitea.dev/modules/log"
+)
+
+// Blob represents a Git object.
+type Blob struct {
+	ID ObjectID
+
+	gotSize bool
+	size    int64
+	name    string
+	repo    *Repository
+}
+
+// DataAsync gets a ReadCloser for the contents of a blob without reading it all.
+// Calling the Close function on the result will discard all unread output.
+func (b *Blob) DataAsync(ctx context.Context) (_ io.ReadCloser, retErr error) {
+	batch, cancel, err := b.repo.CatFileBatch()
+	if err != nil {
+		return nil, err
+	}
+	defer func() {
+		// if there was an error, cancel the batch right away,
+		// otherwise let the caller close it
+		if retErr != nil {
+			cancel()
+		}
+	}()
+
+	info, contentReader, err := batch.QueryContent(b.ID.String())
+	if err != nil {
+		return nil, err
+	}
+	b.gotSize = true
+	b.size = info.Size
+	return &blobReader{
+		rd:     contentReader,
+		n:      info.Size,
+		cancel: cancel,
+	}, nil
+}
+
+// Size returns the uncompressed size of the blob
+func (b *Blob) Size(ctx context.Context) int64 {
+	if b.gotSize {
+		return b.size
+	}
+
+	batch, cancel, err := b.repo.CatFileBatch()
+	if err != nil {
+		log.Debug("error whilst reading size for %s in %s. Error: %v", b.ID.String(), b.repo.LogString(), err)
+		return 0
+	}
+	defer cancel()
+	info, err := batch.QueryInfo(b.ID.String())
+	if err != nil {
+		log.Debug("error whilst reading size for %s in %s. Error: %v", b.ID.String(), b.repo.LogString(), err)
+		return 0
+	}
+	b.gotSize = true
+	b.size = info.Size
+	return b.size
+}
+
+type blobReader struct {
+	rd     BufferedReader
+	n      int64
+	cancel func()
+}
+
+func (b *blobReader) Read(p []byte) (n int, err error) {
+	if b.n <= 0 {
+		return 0, io.EOF
+	}
+	if int64(len(p)) > b.n {
+		p = p[0:b.n]
+	}
+	n, err = b.rd.Read(p)
+	b.n -= int64(n)
+	return n, err
+}
+
+// Close implements io.Closer
+func (b *blobReader) Close() error {
+	if b.rd == nil {
+		return nil
+	}
+
+	defer b.cancel()
+
+	if err := DiscardFull(b.rd, b.n+1); err != nil {
+		return err
+	}
+
+	b.rd = nil
+
+	return nil
+}

modules/git/blob_test.go

@@ -0,0 +1,76 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"io"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestBlob_Data(t *testing.T) {
+	output := "file2\n"
+	bareRepo1Path := filepath.Join(testReposDir, "repo1_bare")
+	repo, err := OpenRepositoryLocal(t.Context(), bareRepo1Path)
+	require.NoError(t, err)
+	defer repo.Close()
+
+	testBlob, err := repo.GetBlob("6c493ff740f9380390d5c9ddef4af18697ac9375")
+	assert.NoError(t, err)
+
+	r, err := testBlob.DataAsync(t.Context())
+	assert.NoError(t, err)
+	require.NotNil(t, r)
+
+	data, err := io.ReadAll(r)
+	assert.NoError(t, r.Close())
+
+	assert.NoError(t, err)
+	assert.Equal(t, output, string(data))
+}
+
+func Benchmark_Blob_Data(b *testing.B) {
+	bareRepo1Path := filepath.Join(testReposDir, "repo1_bare")
+	repo, err := OpenRepositoryLocal(b.Context(), bareRepo1Path)
+	if err != nil {
+		b.Fatal(err)
+	}
+	defer repo.Close()
+
+	testBlob, err := repo.GetBlob("6c493ff740f9380390d5c9ddef4af18697ac9375")
+	if err != nil {
+		b.Fatal(err)
+	}
+
+	for b.Loop() {
+		r, err := testBlob.DataAsync(b.Context())
+		if err != nil {
+			b.Fatal(err)
+		}
+		io.ReadAll(r)
+		_ = r.Close()
+	}
+}
+
+func TestGetBlobLineCount(t *testing.T) {
+	size, count, err := getBlobLineCount(strings.NewReader(""), nil)
+	assert.NoError(t, err)
+	assert.EqualValues(t, 0, size)
+	assert.Equal(t, 0, count)
+
+	size, count, err = getBlobLineCount(strings.NewReader("\n"), nil)
+	assert.NoError(t, err)
+	assert.EqualValues(t, 1, size)
+	assert.Equal(t, 1, count)
+
+	size, count, err = getBlobLineCount(strings.NewReader("a\nb"), nil)
+	assert.NoError(t, err)
+	assert.EqualValues(t, 3, size)
+	assert.Equal(t, 2, count)
+}

modules/git/branch.go

@@ -0,0 +1,95 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"errors"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+)
+
+// GetBranchesByPath returns a branch by its path
+// if limit = 0 it will not limit
+func GetBranchesByPath(ctx context.Context, repo RepositoryFacade, skip, limit int) ([]string, int, error) {
+	gitRepo, err := OpenRepository(ctx, repo)
+	if err != nil {
+		return nil, 0, err
+	}
+	defer gitRepo.Close()
+
+	return gitRepo.GetBranchNames(ctx, skip, limit)
+}
+
+func GetBranchCommitID(ctx context.Context, repo RepositoryFacade, branch string) (string, error) {
+	gitRepo, err := OpenRepository(ctx, repo)
+	if err != nil {
+		return "", err
+	}
+	defer gitRepo.Close()
+
+	return gitRepo.GetBranchCommitID(ctx, branch)
+}
+
+// SetDefaultBranch sets default branch of repository.
+func SetDefaultBranch(ctx context.Context, repo RepositoryFacade, name string) error {
+	_, _, err := gitcmd.NewCommand("symbolic-ref", "HEAD").
+		AddDynamicArguments(BranchPrefix + name).WithRepo(repo).RunStdString(ctx)
+	return err
+}
+
+// GetDefaultBranch gets default branch of repository.
+func GetDefaultBranch(ctx context.Context, repo RepositoryFacade) (string, error) {
+	stdout, _, err := gitcmd.NewCommand("symbolic-ref", "HEAD").WithRepo(repo).RunStdString(ctx)
+	if err != nil {
+		return "", err
+	}
+	stdout = strings.TrimSpace(stdout)
+	if !strings.HasPrefix(stdout, BranchPrefix) {
+		return "", errors.New("the HEAD is not a branch: " + stdout)
+	}
+	return strings.TrimPrefix(stdout, BranchPrefix), nil
+}
+
+// IsReferenceExist returns true if given reference exists in the repository.
+func IsReferenceExist(ctx context.Context, repo RepositoryFacade, name string) bool {
+	_, _, err := gitcmd.NewCommand("show-ref", "--verify").AddDashesAndList(name).WithRepo(repo).RunStdString(ctx)
+	return err == nil
+}
+
+// IsBranchExist returns true if given branch exists in the repository.
+func IsBranchExist(ctx context.Context, repo RepositoryFacade, name string) bool {
+	return IsReferenceExist(ctx, repo, BranchPrefix+name)
+}
+
+// DeleteBranch delete a branch by name on repository.
+func DeleteBranch(ctx context.Context, repo RepositoryFacade, name string, force bool) error {
+	cmd := gitcmd.NewCommand("branch")
+
+	if force {
+		cmd.AddArguments("-D")
+	} else {
+		cmd.AddArguments("-d")
+	}
+
+	cmd.AddDashesAndList(name)
+	_, _, err := cmd.WithRepo(repo).RunStdString(ctx)
+	return err
+}
+
+// CreateBranch create a new branch
+func CreateBranch(ctx context.Context, repo RepositoryFacade, branch, oldbranchOrCommit string) error {
+	cmd := gitcmd.NewCommand("branch")
+	cmd.AddDashesAndList(branch, oldbranchOrCommit)
+
+	_, _, err := cmd.WithRepo(repo).RunStdString(ctx)
+	return err
+}
+
+// RenameBranch rename a branch
+func RenameBranch(ctx context.Context, repo RepositoryFacade, from, to string) error {
+	_, _, err := gitcmd.NewCommand("branch", "-m").AddDynamicArguments(from, to).WithRepo(repo).RunStdString(ctx)
+	return err
+}

modules/git/catfile_batch.go

@@ -0,0 +1,61 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"io"
+	"os"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/util"
+)
+
+type BufferedReader interface {
+	io.Reader
+	Buffered() int
+	Peek(n int) ([]byte, error)
+	Discard(n int) (int, error)
+	ReadString(sep byte) (string, error)
+	ReadSlice(sep byte) ([]byte, error)
+	ReadBytes(sep byte) ([]byte, error)
+}
+
+type CatFileObject struct {
+	ID   string
+	Type string
+	Size int64
+}
+
+type CatFileBatch interface {
+	// QueryInfo queries the object info from the git repository by its object name using "git cat-file --batch" family commands.
+	// "git cat-file" accepts "<rev>" for the object name, it can be a ref name, object id, etc. https://git-scm.com/docs/gitrevisions
+	// In Gitea, we only use the simple ref name or object id, no other complex rev syntax like "suffix" or "git describe" although they are supported by git.
+	QueryInfo(obj string) (*CatFileObject, error)
+
+	// QueryContent is similar to QueryInfo, it queries the object info and additionally returns a reader for its content.
+	// FIXME: this design still follows the old pattern: the returned BufferedReader is very fragile,
+	// callers should carefully maintain its lifecycle and discard all unread data.
+	// TODO: It needs to be refactored to a fully managed Reader stream in the future, don't let callers manually Close or Discard
+	QueryContent(obj string) (*CatFileObject, BufferedReader, error)
+}
+
+type CatFileBatchCloser interface {
+	CatFileBatch
+	Context() context.Context
+	Close()
+}
+
+// NewBatch creates a "batch object provider (CatFileBatch)" for the given repository path to retrieve object info and content efficiently.
+// The CatFileBatch and the readers create by it should only be used in the same goroutine.
+func NewBatch(ctx context.Context, repo RepositoryFacade) (CatFileBatchCloser, error) {
+	repoPath := gitrepo.RepoLocalPath(repo)
+	if _, err := os.Stat(repoPath); err != nil {
+		return nil, util.NewNotExistErrorf("repo %q doesn't exist", repo.LogString())
+	}
+	if DefaultFeatures().SupportCatFileBatchCommand {
+		return newCatFileBatchCommand(ctx, repo), nil
+	}
+	return newCatFileBatchLegacy(ctx, repo), nil
+}

modules/git/catfile_batch_command.go

@@ -0,0 +1,75 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/util"
+)
+
+// catFileBatchCommand implements the CatFileBatch interface using the "cat-file --batch-command" command
+// for git version >= 2.36
+// ref: https://git-scm.com/docs/git-cat-file#Documentation/git-cat-file.txt---batch-command
+type catFileBatchCommand struct {
+	ctx   context.Context
+	repo  RepositoryFacade
+	batch *catFileBatchCommunicator
+}
+
+var _ CatFileBatch = (*catFileBatchCommand)(nil)
+
+func newCatFileBatchCommand(ctx context.Context, repo RepositoryFacade) *catFileBatchCommand {
+	return &catFileBatchCommand{ctx: ctx, repo: repo}
+}
+
+func (b *catFileBatchCommand) getBatch(callerInfo string) *catFileBatchCommunicator {
+	if b.batch != nil {
+		return b.batch
+	}
+	b.batch = newCatFileBatch(b.ctx, b.repo, gitcmd.NewCommand("cat-file", "--batch-command").WithParentCallerInfo(callerInfo))
+	return b.batch
+}
+
+func (b *catFileBatchCommand) Context() context.Context {
+	return b.ctx
+}
+
+func (b *catFileBatchCommand) QueryContent(obj string) (*CatFileObject, BufferedReader, error) {
+	if strings.Contains(obj, "\n") {
+		setting.PanicInDevOrTesting("invalid object name with newline: %q", obj)
+	}
+	batch := b.getBatch(util.CallerFuncName(1))
+	_, err := batch.reqWriter.Write([]byte("contents " + obj + "\n"))
+	if err != nil {
+		return nil, nil, err
+	}
+	info, err := catFileBatchParseInfoLine(batch.respReader)
+	if err != nil {
+		return nil, nil, err
+	}
+	return info, batch.respReader, nil
+}
+
+func (b *catFileBatchCommand) QueryInfo(obj string) (*CatFileObject, error) {
+	if strings.Contains(obj, "\n") {
+		setting.PanicInDevOrTesting("invalid object name with newline: %q", obj)
+	}
+	batch := b.getBatch(util.CallerFuncName(1))
+	_, err := batch.reqWriter.Write([]byte("info " + obj + "\n"))
+	if err != nil {
+		return nil, err
+	}
+	return catFileBatchParseInfoLine(batch.respReader)
+}
+
+func (b *catFileBatchCommand) Close() {
+	if b.batch != nil {
+		b.batch.Close()
+		b.batch = nil
+	}
+}

modules/git/catfile_batch_legacy.go

@@ -0,0 +1,87 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"io"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/setting"
+)
+
+// catFileBatchLegacy implements the CatFileBatch interface using the "cat-file --batch" command and "cat-file --batch-check" command
+// for git version < 2.36
+// to align with "--batch-command", it creates the two commands for querying object contents and object info separately
+// ref: https://git-scm.com/docs/git-cat-file#Documentation/git-cat-file.txt---batch
+type catFileBatchLegacy struct {
+	ctx          context.Context
+	repo         RepositoryFacade
+	batchContent *catFileBatchCommunicator
+	batchCheck   *catFileBatchCommunicator
+}
+
+var _ CatFileBatchCloser = (*catFileBatchLegacy)(nil)
+
+func newCatFileBatchLegacy(ctx context.Context, repo RepositoryFacade) *catFileBatchLegacy {
+	return &catFileBatchLegacy{ctx: ctx, repo: repo}
+}
+
+func (b *catFileBatchLegacy) getBatchContent() *catFileBatchCommunicator {
+	if b.batchContent != nil {
+		return b.batchContent
+	}
+	b.batchContent = newCatFileBatch(b.ctx, b.repo, gitcmd.NewCommand("cat-file", "--batch"))
+	return b.batchContent
+}
+
+func (b *catFileBatchLegacy) getBatchCheck() *catFileBatchCommunicator {
+	if b.batchCheck != nil {
+		return b.batchCheck
+	}
+	b.batchCheck = newCatFileBatch(b.ctx, b.repo, gitcmd.NewCommand("cat-file", "--batch-check"))
+	return b.batchCheck
+}
+
+func (b *catFileBatchLegacy) Context() context.Context {
+	return b.ctx
+}
+
+func (b *catFileBatchLegacy) QueryContent(obj string) (*CatFileObject, BufferedReader, error) {
+	if strings.Contains(obj, "\n") {
+		setting.PanicInDevOrTesting("invalid object name with newline: %q", obj)
+	}
+	_, err := io.WriteString(b.getBatchContent().reqWriter, obj+"\n")
+	if err != nil {
+		return nil, nil, err
+	}
+	info, err := catFileBatchParseInfoLine(b.getBatchContent().respReader)
+	if err != nil {
+		return nil, nil, err
+	}
+	return info, b.getBatchContent().respReader, nil
+}
+
+func (b *catFileBatchLegacy) QueryInfo(obj string) (*CatFileObject, error) {
+	if strings.Contains(obj, "\n") {
+		setting.PanicInDevOrTesting("invalid object name with newline: %q", obj)
+	}
+	_, err := io.WriteString(b.getBatchCheck().reqWriter, obj+"\n")
+	if err != nil {
+		return nil, err
+	}
+	return catFileBatchParseInfoLine(b.getBatchCheck().respReader)
+}
+
+func (b *catFileBatchLegacy) Close() {
+	if b.batchContent != nil {
+		b.batchContent.Close()
+		b.batchContent = nil
+	}
+	if b.batchCheck != nil {
+		b.batchCheck.Close()
+		b.batchCheck = nil
+	}
+}

modules/git/catfile_batch_reader.go

@@ -0,0 +1,241 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"bytes"
+	"context"
+	"errors"
+	"io"
+	"math"
+	"slices"
+	"strconv"
+	"strings"
+	"sync/atomic"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/util"
+)
+
+type catFileBatchCommunicator struct {
+	closeFunc   atomic.Pointer[func(err error)]
+	reqWriter   io.Writer
+	respReader  *bufio.Reader
+	debugGitCmd *gitcmd.Command
+	closed      chan struct{}
+}
+
+func (b *catFileBatchCommunicator) Close(err ...error) {
+	if fn := b.closeFunc.Swap(nil); fn != nil {
+		(*fn)(util.OptionalArg(err))
+	}
+	// make sure the git process has fully exited before we return from Close()
+	// otherwise, the opened files will block the directory renaming (rename a repo) on Windows
+	<-b.closed
+}
+
+// newCatFileBatch opens git cat-file --batch/--batch-check/--batch-command command and prepares the stdin/stdout pipes for communication.
+func newCatFileBatch(ctx context.Context, repo RepositoryFacade, cmdCatFile *gitcmd.Command) *catFileBatchCommunicator {
+	ctx, ctxCancel := context.WithCancelCause(ctx)
+	stdinWriter, stdoutReader, stdPipeClose := cmdCatFile.MakeStdinStdoutPipe()
+	ret := &catFileBatchCommunicator{
+		debugGitCmd: cmdCatFile,
+		reqWriter:   stdinWriter,
+		respReader:  bufio.NewReaderSize(stdoutReader, 32*1024), // use a buffered reader for rich operations
+		closed:      make(chan struct{}),
+	}
+	ret.closeFunc.Store(new(func(err error) {
+		ctxCancel(err)
+		stdPipeClose()
+	}))
+
+	err := cmdCatFile.WithParentCallerInfo().WithRepo(repo).StartWithStderr(ctx)
+	if err != nil {
+		log.Error("Unable to start git command %v: %v", cmdCatFile.LogString(), err)
+		// ideally here it should return the error, but it would require refactoring all callers
+		// so just return a dummy communicator that does nothing, almost the same behavior as before, not bad
+		close(ret.closed)
+		ret.Close(err)
+		return ret
+	}
+
+	go func() {
+		err := cmdCatFile.WaitWithStderr()
+		if err != nil && !errors.Is(err, context.Canceled) {
+			log.Error("cat-file --batch command failed in repo %s, error: %v", repo.LogString(), err)
+		}
+		close(ret.closed)
+		ret.Close(err)
+	}()
+
+	return ret
+}
+
+func (b *catFileBatchCommunicator) debugKill() (ret struct {
+	beforeClose chan struct{}
+	blockClose  chan struct{}
+	afterClose  chan struct{}
+},
+) {
+	ret.beforeClose = make(chan struct{})
+	ret.blockClose = make(chan struct{})
+	ret.afterClose = make(chan struct{})
+	oldCloseFunc := b.closeFunc.Load()
+	b.closeFunc.Store(new(func(err error) {
+		b.closeFunc.Store(nil)
+		close(ret.beforeClose)
+		<-ret.blockClose
+		(*oldCloseFunc)(err)
+		close(ret.afterClose)
+	}))
+	b.debugGitCmd.DebugKill()
+	return ret
+}
+
+// catFileBatchParseInfoLine reads the header line from cat-file --batch
+// We expect: <oid> SP <type> SP <size> LF
+// then leaving the rest of the stream "<contents> LF" to be read
+func catFileBatchParseInfoLine(rd BufferedReader) (*CatFileObject, error) {
+	typ, err := rd.ReadString('\n')
+	if err != nil {
+		return nil, err
+	}
+	if len(typ) == 1 {
+		typ, err = rd.ReadString('\n')
+		if err != nil {
+			return nil, err
+		}
+	}
+	idx := strings.IndexByte(typ, ' ')
+	if idx < 0 {
+		return nil, ErrNotExist{}
+	}
+	sha := typ[:idx]
+	typ = typ[idx+1:]
+
+	idx = strings.IndexByte(typ, ' ')
+	if idx < 0 {
+		return nil, ErrNotExist{ID: sha}
+	}
+
+	sizeStr := typ[idx+1 : len(typ)-1]
+	typ = typ[:idx]
+
+	size, err := strconv.ParseInt(sizeStr, 10, 64)
+	return &CatFileObject{ID: sha, Type: typ, Size: size}, err
+}
+
+// ReadTagObjectID reads a tag object ID hash from a cat-file --batch stream, throwing away the rest of the stream.
+func ReadTagObjectID(rd BufferedReader, size int64) (string, error) {
+	var id string
+	var n int64
+headerLoop:
+	for {
+		line, err := rd.ReadBytes('\n')
+		if err != nil {
+			return "", err
+		}
+		n += int64(len(line))
+		idx := bytes.Index(line, []byte{' '})
+		if idx < 0 {
+			continue
+		}
+
+		if string(line[:idx]) == "object" {
+			id = string(line[idx+1 : len(line)-1])
+			break headerLoop
+		}
+	}
+
+	// Discard the rest of the tag
+	return id, DiscardFull(rd, size-n+1)
+}
+
+// ReadTreeID reads a tree ID from a cat-file --batch stream, throwing away the rest of the stream.
+func ReadTreeID(rd BufferedReader, size int64) (string, error) {
+	var id string
+	var n int64
+headerLoop:
+	for {
+		line, err := rd.ReadBytes('\n')
+		if err != nil {
+			return "", err
+		}
+		n += int64(len(line))
+		idx := bytes.Index(line, []byte{' '})
+		if idx < 0 {
+			continue
+		}
+
+		if string(line[:idx]) == "tree" {
+			id = string(line[idx+1 : len(line)-1])
+			break headerLoop
+		}
+	}
+
+	// Discard the rest of the commit
+	return id, DiscardFull(rd, size-n+1)
+}
+
+// ParseCatFileTreeLine reads an entry from a tree in a cat-file --batch stream
+// Each entry is composed of:
+// <mode-in-ascii-dropping-initial-zeros> SP <name> NUL <binary-hash>
+func ParseCatFileTreeLine(objectFormat ObjectFormat, rd BufferedReader) (mode EntryMode, name string, objID ObjectID, n int, err error) {
+	// use the in-buffer memory as much as possible to avoid extra allocations
+	bufBytes, err := rd.ReadSlice('\x00')
+	const maxEntryInfoBytes = 1024 * 1024
+	if errors.Is(err, bufio.ErrBufferFull) {
+		bufBytes = slices.Clone(bufBytes)
+		for len(bufBytes) < maxEntryInfoBytes && errors.Is(err, bufio.ErrBufferFull) {
+			var tmp []byte
+			tmp, err = rd.ReadSlice('\x00')
+			bufBytes = append(bufBytes, tmp...)
+		}
+	}
+	if err != nil {
+		return mode, name, objID, len(bufBytes), err
+	}
+
+	idx := bytes.IndexByte(bufBytes, ' ')
+	if idx < 0 {
+		return mode, name, objID, len(bufBytes), errors.New("invalid CatFileTreeLine output")
+	}
+
+	mode = ParseEntryMode(util.UnsafeBytesToString(bufBytes[:idx]))
+	name = string(bufBytes[idx+1 : len(bufBytes)-1]) // trim the NUL terminator, it needs a copy because the bufBytes will be reused by the reader
+	if mode == EntryModeNoEntry {
+		return mode, name, objID, len(bufBytes), errors.New("invalid entry mode: " + string(bufBytes[:idx]))
+	}
+
+	switch objectFormat {
+	case Sha1ObjectFormat:
+		objID = &Sha1Hash{}
+	case Sha256ObjectFormat:
+		objID = &Sha256Hash{}
+	default:
+		panic("unsupported object format: " + objectFormat.Name())
+	}
+	readIDLen, err := io.ReadFull(rd, objID.RawValue())
+	return mode, name, objID, len(bufBytes) + readIDLen, err
+}
+
+func DiscardFull(rd BufferedReader, discard int64) error {
+	if discard > math.MaxInt32 {
+		n, err := rd.Discard(math.MaxInt32)
+		discard -= int64(n)
+		if err != nil {
+			return err
+		}
+	}
+	for discard > 0 {
+		n, err := rd.Discard(int(discard))
+		discard -= int64(n)
+		if err != nil {
+			return err
+		}
+	}
+	return nil
+}

modules/git/catfile_batch_test.go

@@ -0,0 +1,108 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"io"
+	"os"
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestCatFileBatch(t *testing.T) {
+	defer test.MockVariableValue(&DefaultFeatures().SupportCatFileBatchCommand)()
+	DefaultFeatures().SupportCatFileBatchCommand = false
+	t.Run("LegacyCheck", testCatFileBatch)
+	DefaultFeatures().SupportCatFileBatchCommand = true
+	t.Run("BatchCommand", testCatFileBatch)
+}
+
+func testCatFileBatch(t *testing.T) {
+	repo1Path, _ := filepath.Abs(filepath.Join(testReposDir, "repo1_bare"))
+	repo1 := gitrepo.RepositoryUnmanaged(repo1Path)
+	t.Run("CorruptedGitRepo", func(t *testing.T) {
+		tmpDir := t.TempDir()
+		batch, err := NewBatch(t.Context(), gitrepo.RepositoryUnmanaged(tmpDir))
+		// as long as the directory exists, no error, because we can't really know whether the git repo is valid until we run commands
+		require.NoError(t, err)
+		defer batch.Close()
+
+		_, err = batch.QueryInfo("e2129701f1a4d54dc44f03c93bca0a2aec7c5449")
+		require.Error(t, err)
+		_, err = batch.QueryInfo("e2129701f1a4d54dc44f03c93bca0a2aec7c5449")
+		require.Error(t, err)
+	})
+
+	simulateQueryTerminated := func(t *testing.T, errBeforePipeClose, errAfterPipeClose error) {
+		readError := func(t *testing.T, r io.Reader, expectedErr error) {
+			if expectedErr == nil {
+				return // expectedErr == nil means this read should be skipped
+			}
+			n, err := r.Read(make([]byte, 100))
+			assert.Zero(t, n)
+			assert.ErrorIs(t, err, expectedErr)
+		}
+
+		batch, err := NewBatch(t.Context(), repo1)
+		require.NoError(t, err)
+		defer batch.Close()
+		_, err = batch.QueryInfo("e2129701f1a4d54dc44f03c93bca0a2aec7c5449")
+		require.NoError(t, err)
+
+		var c *catFileBatchCommunicator
+		switch b := batch.(type) {
+		case *catFileBatchLegacy:
+			c = b.batchCheck
+			_, _ = c.reqWriter.Write([]byte("in-complete-line-"))
+		case *catFileBatchCommand:
+			c = b.batch
+			_, _ = c.reqWriter.Write([]byte("info"))
+		default:
+			t.FailNow()
+		}
+
+		require.NotEqual(t, errBeforePipeClose == nil, errAfterPipeClose == nil, "must set exactly one of the expected errors")
+
+		inceptor := c.debugKill()
+		<-inceptor.beforeClose                         // wait for the command's Close to be called, the pipe is not closed yet
+		readError(t, c.respReader, errBeforePipeClose) // then caller will read on an open pipe which will be closed soon
+		close(inceptor.blockClose)                     // continue to close the pipe
+		<-inceptor.afterClose                          // wait for the pipe to be closed
+		readError(t, c.respReader, errAfterPipeClose)  // then caller will read on a closed pipe
+	}
+	t.Run("QueryTerminated", func(t *testing.T) {
+		simulateQueryTerminated(t, io.EOF, nil)       // reader is faster
+		simulateQueryTerminated(t, nil, os.ErrClosed) // pipes are closed faster
+	})
+
+	batch, err := NewBatch(t.Context(), repo1)
+	require.NoError(t, err)
+	defer batch.Close()
+
+	t.Run("QueryInfo", func(t *testing.T) {
+		info, err := batch.QueryInfo("e2129701f1a4d54dc44f03c93bca0a2aec7c5449")
+		require.NoError(t, err)
+		assert.Equal(t, "e2129701f1a4d54dc44f03c93bca0a2aec7c5449", info.ID)
+		assert.Equal(t, "blob", info.Type)
+		assert.EqualValues(t, 6, info.Size)
+	})
+
+	t.Run("QueryContent", func(t *testing.T) {
+		info, rd, err := batch.QueryContent("e2129701f1a4d54dc44f03c93bca0a2aec7c5449")
+		require.NoError(t, err)
+		assert.Equal(t, "e2129701f1a4d54dc44f03c93bca0a2aec7c5449", info.ID)
+		assert.Equal(t, "blob", info.Type)
+		assert.EqualValues(t, 6, info.Size)
+
+		content, err := io.ReadAll(io.LimitReader(rd, info.Size))
+		require.NoError(t, err)
+		require.Equal(t, "file1\n", string(content))
+	})
+}

modules/git/clone.go

@@ -0,0 +1,24 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+
+	"gitea.dev/modules/git/gitrepo"
+)
+
+// CloneExternalRepo clones an external repository to the managed repository.
+func CloneExternalRepo(ctx context.Context, fromRemoteURL string, toRepo RepositoryFacade, opts CloneRepoOptions) error {
+	return Clone(ctx, fromRemoteURL, gitrepo.RepoLocalPath(toRepo), opts)
+}
+
+// CloneRepoToLocal clones a managed repository to a local path.
+func CloneRepoToLocal(ctx context.Context, fromRepo RepositoryFacade, toLocalPath string, opts CloneRepoOptions) error {
+	return Clone(ctx, gitrepo.RepoLocalPath(fromRepo), toLocalPath, opts)
+}
+
+func CloneManaged(ctx context.Context, fromRepo, toRepo RepositoryFacade, opts CloneRepoOptions) error {
+	return Clone(ctx, gitrepo.RepoLocalPath(fromRepo), gitrepo.RepoLocalPath(toRepo), opts)
+}

modules/git/cmdverb.go

@@ -0,0 +1,36 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+const (
+	CmdVerbUploadPack      = "git-upload-pack"
+	CmdVerbUploadArchive   = "git-upload-archive"
+	CmdVerbReceivePack     = "git-receive-pack"
+	CmdVerbLfsAuthenticate = "git-lfs-authenticate"
+	CmdVerbLfsTransfer     = "git-lfs-transfer"
+
+	CmdSubVerbLfsUpload   = "upload"
+	CmdSubVerbLfsDownload = "download"
+)
+
+func IsAllowedVerbForServe(verb string) bool {
+	switch verb {
+	case CmdVerbUploadPack,
+		CmdVerbUploadArchive,
+		CmdVerbReceivePack,
+		CmdVerbLfsAuthenticate,
+		CmdVerbLfsTransfer:
+		return true
+	}
+	return false
+}
+
+func IsAllowedVerbForServeLfs(verb string) bool {
+	switch verb {
+	case CmdVerbLfsAuthenticate,
+		CmdVerbLfsTransfer:
+		return true
+	}
+	return false
+}

modules/git/commit.go

@@ -0,0 +1,272 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2018 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"errors"
+	"io"
+	"os/exec"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/util"
+)
+
+// Commit represents a git commit.
+type Commit struct {
+	CommitMessage
+
+	ID        ObjectID
+	TreeID    ObjectID
+	Parents   []ObjectID
+	Author    *Signature // never nil
+	Committer *Signature // never nil
+	Signature *CommitSignature
+
+	submoduleCache *ObjectCache[*SubModule]
+	treeCache      *Tree
+}
+
+// CommitSignature represents a git commit signature part.
+type CommitSignature struct {
+	Signature string
+	Payload   string
+}
+
+// ParentID returns oid of n-th parent (0-based index).
+// It returns nil if no such parent exists.
+func (c *Commit) ParentID(n int) (ObjectID, error) {
+	if n >= len(c.Parents) {
+		return nil, ErrNotExist{"", ""}
+	}
+	return c.Parents[n], nil
+}
+
+// Parent returns n-th parent (0-based index) of the commit.
+func (c *Commit) Parent(ctx context.Context, gitRepo *Repository, n int) (*Commit, error) {
+	id, err := c.ParentID(n)
+	if err != nil {
+		return nil, err
+	}
+	parent, err := gitRepo.getCommit(ctx, id)
+	if err != nil {
+		return nil, err
+	}
+	return parent, nil
+}
+
+// ParentCount returns number of parents of the commit.
+// 0 if this is the root commit,  otherwise 1,2, etc.
+func (c *Commit) ParentCount() int {
+	return len(c.Parents)
+}
+
+// GetCommitByPath return the commit of relative path object.
+func (c *Commit) GetCommitByPath(ctx context.Context, gitRepo *Repository, relpath string) (*Commit, error) {
+	return gitRepo.LastCommitCache.GetCommitByPath(ctx, c.ID, relpath)
+}
+
+func (c *Commit) Tree() *Tree {
+	if c.treeCache == nil {
+		c.treeCache = newTree(c.TreeID)
+	}
+	return c.treeCache
+}
+
+func (c *Commit) GetBlobByPath(ctx context.Context, gitRepo *Repository, relpath string) (*Blob, error) {
+	return c.Tree().GetBlobByPath(ctx, gitRepo, relpath)
+}
+
+func (c *Commit) GetTreeEntryByPath(ctx context.Context, gitRepo *Repository, relpath string) (_ *TreeEntry, err error) {
+	return c.Tree().GetTreeEntryByPath(ctx, gitRepo, relpath)
+}
+
+func (c *Commit) SubTree(ctx context.Context, gitRepo *Repository, relpath string) (*Tree, error) {
+	return c.Tree().SubTree(ctx, gitRepo, relpath)
+}
+
+// CommitsByRange returns the specific page commits before current revision, every page's number default by CommitsRangeSize
+func (c *Commit) CommitsByRange(ctx context.Context, gitRepo *Repository, page, pageSize int, not, since, until string) ([]*Commit, error) {
+	return gitRepo.commitsByRangeWithTime(ctx, c.ID, page, pageSize, not, since, until)
+}
+
+// CommitsBefore returns all the commits before current revision
+func (c *Commit) CommitsBefore(ctx context.Context, gitRepo *Repository) ([]*Commit, error) {
+	return gitRepo.getCommitsBefore(ctx, c.ID)
+}
+
+// HasPreviousCommit returns true if a given commitHash is contained in commit's parents
+func (c *Commit) HasPreviousCommit(ctx context.Context, gitRepo *Repository, objectID ObjectID) (bool, error) {
+	this := c.ID.String()
+	that := objectID.String()
+
+	if this == that {
+		return false, nil
+	}
+
+	_, _, err := gitcmd.NewCommand("merge-base", "--is-ancestor").
+		AddDynamicArguments(that, this).
+		WithRepo(gitRepo).
+		RunStdString(ctx)
+	if err == nil {
+		return true, nil
+	}
+	if exitError, ok := errors.AsType[*exec.ExitError](err); ok {
+		if exitError.ProcessState.ExitCode() == 1 && len(exitError.Stderr) == 0 {
+			return false, nil
+		}
+	}
+	return false, err
+}
+
+// IsForcePush returns true if a push from oldCommitHash to this is a force push
+func (c *Commit) IsForcePush(ctx context.Context, gitRepo *Repository, oldCommitID string) (bool, error) {
+	objectFormat, err := gitRepo.GetObjectFormat(ctx)
+	if err != nil {
+		return false, err
+	}
+	if oldCommitID == objectFormat.EmptyObjectID().String() {
+		return false, nil
+	}
+
+	oldCommit, err := gitRepo.GetCommit(ctx, oldCommitID)
+	if err != nil {
+		return false, err
+	}
+	hasPreviousCommit, err := c.HasPreviousCommit(ctx, gitRepo, oldCommit.ID)
+	return !hasPreviousCommit, err
+}
+
+// CommitsBeforeLimit returns num commits before current revision
+func (c *Commit) CommitsBeforeLimit(ctx context.Context, gitRepo *Repository, num int) ([]*Commit, error) {
+	return gitRepo.getCommitsBeforeLimit(ctx, c.ID, num)
+}
+
+// CommitsBeforeUntil returns the commits in range "[cur, ref)"
+func (c *Commit) CommitsBeforeUntil(ctx context.Context, gitRepo *Repository, ref RefName) ([]*Commit, error) {
+	return gitRepo.CommitsBetween(ctx, c.ID.RefName(), ref, -1)
+}
+
+// SearchCommitsOptions specify the parameters for SearchCommits
+type SearchCommitsOptions struct {
+	Keywords            []string
+	Authors, Committers []string
+	After, Before       string
+	All                 bool
+}
+
+// NewSearchCommitsOptions construct a SearchCommitsOption from a space-delimited search string
+func NewSearchCommitsOptions(searchString string, forAllRefs bool) SearchCommitsOptions {
+	var keywords, authors, committers []string
+	var after, before string
+
+	fields := strings.FieldsSeq(searchString)
+	for k := range fields {
+		switch {
+		case strings.HasPrefix(k, "author:"):
+			authors = append(authors, strings.TrimPrefix(k, "author:"))
+		case strings.HasPrefix(k, "committer:"):
+			committers = append(committers, strings.TrimPrefix(k, "committer:"))
+		case strings.HasPrefix(k, "after:"):
+			after = strings.TrimPrefix(k, "after:")
+		case strings.HasPrefix(k, "before:"):
+			before = strings.TrimPrefix(k, "before:")
+		default:
+			keywords = append(keywords, k)
+		}
+	}
+
+	return SearchCommitsOptions{
+		Keywords:   keywords,
+		Authors:    authors,
+		Committers: committers,
+		After:      after,
+		Before:     before,
+		All:        forAllRefs,
+	}
+}
+
+// SearchCommits returns the commits match the keyword before current revision
+func (c *Commit) SearchCommits(ctx context.Context, gitRepo *Repository, opts SearchCommitsOptions) ([]*Commit, error) {
+	return gitRepo.searchCommits(ctx, c.ID, opts)
+}
+
+// GetFilesChangedSinceCommit get all changed file names between pastCommit to current revision
+func (c *Commit) GetFilesChangedSinceCommit(ctx context.Context, gitRepo *Repository, pastCommit string) ([]string, error) {
+	return gitRepo.GetFilesChangedBetween(ctx, pastCommit, c.ID.String())
+}
+
+// FileChangedSinceCommit Returns true if the file given has changed since the past commit
+// YOU MUST ENSURE THAT pastCommit is a valid commit ID.
+func (c *Commit) FileChangedSinceCommit(ctx context.Context, gitRepo *Repository, filename, pastCommit string) (bool, error) {
+	return gitRepo.FileChangedBetweenCommits(ctx, filename, pastCommit, c.ID.String())
+}
+
+// GetFileContent reads a file content as a string or returns false if this was not possible
+func (c *Commit) GetFileContent(ctx context.Context, gitRepo *Repository, filename string, limit int) (string, error) {
+	entry, err := c.GetTreeEntryByPath(ctx, gitRepo, filename)
+	if err != nil {
+		return "", err
+	}
+
+	r, err := entry.Blob(gitRepo).DataAsync(ctx)
+	if err != nil {
+		return "", err
+	}
+	defer r.Close()
+
+	if limit > 0 {
+		bs := make([]byte, limit)
+		n, err := util.ReadAtMost(r, bs)
+		if err != nil {
+			return "", err
+		}
+		return string(bs[:n]), nil
+	}
+
+	bytes, err := io.ReadAll(r)
+	if err != nil {
+		return "", err
+	}
+	return string(bytes), nil
+}
+
+// GetFullCommitID returns full length (40) of commit ID by given short SHA in a repository.
+func GetFullCommitID(ctx context.Context, repo RepositoryFacade, shortID string) (string, error) {
+	commitID, _, err := gitcmd.NewCommand("rev-parse").
+		AddDynamicArguments(shortID).
+		WithRepo(repo).
+		RunStdString(ctx)
+	if err != nil {
+		if gitcmd.IsErrorExitCode(err, 128) {
+			return "", ErrNotExist{shortID, ""}
+		}
+		return "", err
+	}
+	return strings.TrimSpace(commitID), nil
+}
+
+func IsStringLikelyCommitID(objFmt ObjectFormat, s string, minLength ...int) bool {
+	maxLen := 64 // sha256
+	if objFmt != nil {
+		maxLen = objFmt.FullLength()
+	}
+	minLen := util.OptionalArg(minLength, maxLen)
+	if len(s) < minLen || len(s) > maxLen {
+		return false
+	}
+	return isStringLowerHex(s)
+}
+
+func isStringLowerHex(s string) bool {
+	for _, c := range s {
+		isHex := (c >= '0' && c <= '9') || (c >= 'a' && c <= 'f')
+		if !isHex {
+			return false
+		}
+	}
+	return len(s) > 0 // it accepts odd length because "shorten commit id" can be 7-chars
+}

modules/git/commit2.go

@@ -0,0 +1,97 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"strconv"
+	"strings"
+	"time"
+
+	"gitea.dev/modules/git/gitcmd"
+)
+
+// CommitsCountOptions the options when counting commits
+type CommitsCountOptions struct {
+	Not      string
+	Revision []string
+	RelPath  []string
+	Since    string
+	Until    string
+}
+
+// CommitsCount returns number of total commits of until given revision.
+func CommitsCount(ctx context.Context, repo RepositoryFacade, opts CommitsCountOptions) (int64, error) {
+	cmd := gitcmd.NewCommand("rev-list", "--count")
+
+	cmd.AddDynamicArguments(opts.Revision...)
+
+	if opts.Not != "" {
+		cmd.AddOptionValues("--not", opts.Not)
+	}
+
+	if opts.Since != "" {
+		cmd.AddOptionFormat("--since=%s", opts.Since)
+	}
+
+	if opts.Until != "" {
+		cmd.AddOptionFormat("--until=%s", opts.Until)
+	}
+
+	if len(opts.RelPath) > 0 {
+		cmd.AddDashesAndList(opts.RelPath...)
+	}
+
+	stdout, _, err := cmd.WithRepo(repo).RunStdString(ctx)
+	if err != nil {
+		return 0, err
+	}
+
+	return strconv.ParseInt(strings.TrimSpace(stdout), 10, 64)
+}
+
+// FileCommitsCount return the number of files at a revision
+func FileCommitsCount(ctx context.Context, repo RepositoryFacade, revision, file string) (int64, error) {
+	return CommitsCount(ctx, repo,
+		CommitsCountOptions{
+			Revision: []string{revision},
+			RelPath:  []string{file},
+		})
+}
+
+// CommitsCountOfCommit returns number of total commits of until current revision.
+func CommitsCountOfCommit(ctx context.Context, repo RepositoryFacade, commitID string) (int64, error) {
+	return CommitsCount(ctx, repo, CommitsCountOptions{
+		Revision: []string{commitID},
+	})
+}
+
+// AllCommitsCount returns count of all commits in repository
+func AllCommitsCount(ctx context.Context, repo RepositoryFacade, hidePRRefs bool, files ...string) (int64, error) {
+	cmd := gitcmd.NewCommand("rev-list")
+	if hidePRRefs {
+		cmd.AddArguments("--exclude=" + PullPrefix + "*")
+	}
+	cmd.AddArguments("--all", "--count")
+	if len(files) > 0 {
+		cmd.AddDashesAndList(files...)
+	}
+
+	stdout, _, err := cmd.WithRepo(repo).RunStdString(ctx)
+	if err != nil {
+		return 0, err
+	}
+
+	return strconv.ParseInt(strings.TrimSpace(stdout), 10, 64)
+}
+
+// GetLatestCommitTime returns time for latest commit in repository (across all branches)
+func GetLatestCommitTime(ctx context.Context, repo RepositoryFacade) (time.Time, error) {
+	stdout, _, err := gitcmd.NewCommand("for-each-ref", "--sort=-committerdate", BranchPrefix, "--count", "1", "--format=%(committerdate)").WithRepo(repo).RunStdString(ctx)
+	if err != nil {
+		return time.Time{}, err
+	}
+	commitTime := strings.TrimSpace(stdout)
+	return time.Parse("Mon Jan _2 15:04:05 2006 -0700", commitTime)
+}

modules/git/commit2_test.go

@@ -0,0 +1,76 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestCommitsCount(t *testing.T) {
+	bareRepo1 := mockRepository("repo1_bare")
+
+	commitsCount, err := CommitsCount(t.Context(), bareRepo1,
+		CommitsCountOptions{
+			Revision: []string{"8006ff9adbf0cb94da7dad9e537e53817f9fa5c0"},
+		})
+
+	assert.NoError(t, err)
+	assert.Equal(t, int64(3), commitsCount)
+}
+
+func TestCommitsCountWithSinceUntil(t *testing.T) {
+	bareRepo1 := mockRepository("repo1_bare")
+	revision := []string{"8006ff9adbf0cb94da7dad9e537e53817f9fa5c0"}
+
+	// The three commits on this revision are dated 2018-04-18, 2017-12-19 and 2017-12-19.
+	cases := []struct {
+		name     string
+		since    string
+		until    string
+		expected int64
+	}{
+		{name: "no filter", expected: 3},
+		{name: "since keeps newer commits", since: "2018-01-01", expected: 1},
+		{name: "until keeps older commits", until: "2018-01-01", expected: 2},
+		{name: "since and until bound the range", since: "2017-12-19T22:16:00-08:00", until: "2018-01-01", expected: 1},
+	}
+	for _, tc := range cases {
+		t.Run(tc.name, func(t *testing.T) {
+			commitsCount, err := CommitsCount(t.Context(), bareRepo1,
+				CommitsCountOptions{
+					Revision: revision,
+					Since:    tc.since,
+					Until:    tc.until,
+				})
+
+			assert.NoError(t, err)
+			assert.Equal(t, tc.expected, commitsCount)
+		})
+	}
+}
+
+func TestCommitsCountWithoutBase(t *testing.T) {
+	bareRepo1 := mockRepository("repo1_bare")
+
+	commitsCount, err := CommitsCount(t.Context(), bareRepo1,
+		CommitsCountOptions{
+			Not:      "master",
+			Revision: []string{"branch1"},
+		})
+
+	assert.NoError(t, err)
+	assert.Equal(t, int64(2), commitsCount)
+}
+
+func TestGetLatestCommitTime(t *testing.T) {
+	bareRepo1 := mockRepository("repo1_bare")
+	lct, err := GetLatestCommitTime(t.Context(), bareRepo1)
+	assert.NoError(t, err)
+	// Time is Sun Nov 13 16:40:14 2022 +0100
+	// which is the time of commit
+	// ce064814f4a0d337b333e646ece456cd39fab612 (refs/heads/master)
+	assert.EqualValues(t, 1668354014, lct.Unix())
+}

modules/git/commit_convert_gogit.go

@@ -0,0 +1,75 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2018 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build gogit
+
+package git
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/go-git/go-git/v5/plumbing/object"
+)
+
+func convertPGPSignature(c *object.Commit) *CommitSignature {
+	if c.PGPSignature == "" {
+		return nil
+	}
+
+	var w strings.Builder
+	var err error
+
+	if _, err = fmt.Fprintf(&w, "tree %s\n", c.TreeHash.String()); err != nil {
+		return nil
+	}
+
+	for _, parent := range c.ParentHashes {
+		if _, err = fmt.Fprintf(&w, "parent %s\n", parent.String()); err != nil {
+			return nil
+		}
+	}
+
+	if _, err = fmt.Fprint(&w, "author "); err != nil {
+		return nil
+	}
+
+	if err = c.Author.Encode(&w); err != nil {
+		return nil
+	}
+
+	if _, err = fmt.Fprint(&w, "\ncommitter "); err != nil {
+		return nil
+	}
+
+	if err = c.Committer.Encode(&w); err != nil {
+		return nil
+	}
+
+	if c.Encoding != "" && c.Encoding != "UTF-8" {
+		if _, err = fmt.Fprintf(&w, "\nencoding %s\n", c.Encoding); err != nil {
+			return nil
+		}
+	}
+
+	if _, err = fmt.Fprintf(&w, "\n\n%s", c.Message); err != nil {
+		return nil
+	}
+
+	return &CommitSignature{
+		Signature: c.PGPSignature,
+		Payload:   w.String(),
+	}
+}
+
+func convertCommit(c *object.Commit) *Commit {
+	return &Commit{
+		ID:            ParseGogitHash(c.Hash),
+		CommitMessage: CommitMessage{MessageRaw: c.Message},
+		Committer:     &c.Committer,
+		Author:        &c.Author,
+		Signature:     convertPGPSignature(c),
+		Parents:       ParseGogitHashArray(c.ParentHashes),
+	}
+}

modules/git/commit_count_cache.go

@@ -0,0 +1,27 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+
+	"gitea.dev/modules/cache"
+)
+
+func makeCommitsCountCacheKey(repo RepositoryFacade, ref RefName) string {
+	return cache.SafeCacheKey("git-commits-count:"+repo.GitRepoManagedID(), ref.String())
+}
+
+func RemoveCommitsCountCache(repo RepositoryFacade, ref RefName) {
+	cache.Remove(makeCommitsCountCacheKey(repo, ref))
+}
+
+func GetCommitsCountCache(ctx context.Context, repo RepositoryFacade, ref RefName, commit *Commit) (int64, error) {
+	if commit == nil {
+		return 0, nil
+	}
+	return cache.GetInt64(makeCommitsCountCacheKey(repo, ref), func() (int64, error) {
+		return CommitsCountOfCommit(ctx, repo, commit.ID.String())
+	})
+}

modules/git/commit_file.go

@@ -0,0 +1,88 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"context"
+	"io"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/log"
+)
+
+// CommitFileStatus represents status of files in a commit.
+type CommitFileStatus struct {
+	Added    []string
+	Removed  []string
+	Modified []string
+}
+
+// NewCommitFileStatus creates a CommitFileStatus
+func NewCommitFileStatus() *CommitFileStatus {
+	return &CommitFileStatus{
+		[]string{}, []string{}, []string{},
+	}
+}
+
+func parseCommitFileStatus(fileStatus *CommitFileStatus, stdout io.Reader) {
+	rd := bufio.NewReader(stdout)
+	peek, err := rd.Peek(1)
+	if err != nil {
+		if err != io.EOF {
+			log.Error("Unexpected error whilst reading from git log --name-status. Error: %v", err)
+		}
+		return
+	}
+	if peek[0] == '\n' || peek[0] == '\x00' {
+		_, _ = rd.Discard(1)
+	}
+	for {
+		modifier, err := rd.ReadString('\x00')
+		if err != nil {
+			if err != io.EOF {
+				log.Error("Unexpected error whilst reading from git log --name-status. Error: %v", err)
+			}
+			return
+		}
+		file, err := rd.ReadString('\x00')
+		if err != nil {
+			if err != io.EOF {
+				log.Error("Unexpected error whilst reading from git log --name-status. Error: %v", err)
+			}
+			return
+		}
+		file = file[:len(file)-1]
+		switch modifier[0] {
+		case 'A':
+			fileStatus.Added = append(fileStatus.Added, file)
+		case 'D':
+			fileStatus.Removed = append(fileStatus.Removed, file)
+		case 'M':
+			fileStatus.Modified = append(fileStatus.Modified, file)
+		}
+	}
+}
+
+// GetCommitFileStatus returns file status of commit in given repository.
+func GetCommitFileStatus(ctx context.Context, repo RepositoryFacade, commitID string) (*CommitFileStatus, error) {
+	cmd := gitcmd.NewCommand("log", "--name-status", "-m", "--pretty=format:", "--first-parent", "--no-renames", "-z", "-1")
+	stdout, stdoutClose := cmd.MakeStdoutPipe()
+	defer stdoutClose()
+	done := make(chan struct{})
+	fileStatus := NewCommitFileStatus()
+	go func() {
+		parseCommitFileStatus(fileStatus, stdout)
+		close(done)
+	}()
+	err := cmd.AddDynamicArguments(commitID).
+		WithRepo(repo).
+		RunWithStderr(ctx)
+	if err != nil {
+		return nil, err
+	}
+
+	<-done
+	return fileStatus, nil
+}

modules/git/commit_file_test.go

@@ -0,0 +1,175 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestParseCommitFileStatus(t *testing.T) {
+	type testcase struct {
+		output   string
+		added    []string
+		removed  []string
+		modified []string
+	}
+
+	kases := []testcase{
+		{
+			// Merge commit
+			output: "MM\x00options/locale/locale_en-US.ini\x00",
+			modified: []string{
+				"options/locale/locale_en-US.ini",
+			},
+			added:   []string{},
+			removed: []string{},
+		},
+		{
+			// Spaces commit
+			output: "D\x00b\x00D\x00b b/b\x00A\x00b b/b b/b b/b\x00A\x00b b/b b/b b/b b/b\x00",
+			removed: []string{
+				"b",
+				"b b/b",
+			},
+			modified: []string{},
+			added: []string{
+				"b b/b b/b b/b",
+				"b b/b b/b b/b b/b",
+			},
+		},
+		{
+			// larger commit
+			output: "M\x00go.mod\x00M\x00go.sum\x00M\x00modules/ssh/ssh.go\x00M\x00vendor/github.com/gliderlabs/ssh/circle.yml\x00M\x00vendor/github.com/gliderlabs/ssh/context.go\x00A\x00vendor/github.com/gliderlabs/ssh/go.mod\x00A\x00vendor/github.com/gliderlabs/ssh/go.sum\x00M\x00vendor/github.com/gliderlabs/ssh/server.go\x00M\x00vendor/github.com/gliderlabs/ssh/session.go\x00M\x00vendor/github.com/gliderlabs/ssh/ssh.go\x00M\x00vendor/golang.org/x/sys/unix/mkerrors.sh\x00M\x00vendor/golang.org/x/sys/unix/syscall_darwin.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_darwin_amd64.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_darwin_arm64.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_freebsd_386.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_freebsd_amd64.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_freebsd_arm.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_freebsd_arm64.go\x00M\x00vendor/golang.org/x/sys/unix/zerrors_linux.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_darwin_amd64.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_darwin_arm64.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_dragonfly_amd64.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_freebsd_386.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_freebsd_amd64.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_freebsd_arm.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_freebsd_arm64.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_netbsd_386.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_netbsd_amd64.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_netbsd_arm.go\x00M\x00vendor/golang.org/x/sys/unix/ztypes_netbsd_arm64.go\x00M\x00vendor/modules.txt\x00",
+			modified: []string{
+				"go.mod",
+				"go.sum",
+				"modules/ssh/ssh.go",
+				"vendor/github.com/gliderlabs/ssh/circle.yml",
+				"vendor/github.com/gliderlabs/ssh/context.go",
+				"vendor/github.com/gliderlabs/ssh/server.go",
+				"vendor/github.com/gliderlabs/ssh/session.go",
+				"vendor/github.com/gliderlabs/ssh/ssh.go",
+				"vendor/golang.org/x/sys/unix/mkerrors.sh",
+				"vendor/golang.org/x/sys/unix/syscall_darwin.go",
+				"vendor/golang.org/x/sys/unix/zerrors_darwin_amd64.go",
+				"vendor/golang.org/x/sys/unix/zerrors_darwin_arm64.go",
+				"vendor/golang.org/x/sys/unix/zerrors_freebsd_386.go",
+				"vendor/golang.org/x/sys/unix/zerrors_freebsd_amd64.go",
+				"vendor/golang.org/x/sys/unix/zerrors_freebsd_arm.go",
+				"vendor/golang.org/x/sys/unix/zerrors_freebsd_arm64.go",
+				"vendor/golang.org/x/sys/unix/zerrors_linux.go",
+				"vendor/golang.org/x/sys/unix/ztypes_darwin_amd64.go",
+				"vendor/golang.org/x/sys/unix/ztypes_darwin_arm64.go",
+				"vendor/golang.org/x/sys/unix/ztypes_dragonfly_amd64.go",
+				"vendor/golang.org/x/sys/unix/ztypes_freebsd_386.go",
+				"vendor/golang.org/x/sys/unix/ztypes_freebsd_amd64.go",
+				"vendor/golang.org/x/sys/unix/ztypes_freebsd_arm.go",
+				"vendor/golang.org/x/sys/unix/ztypes_freebsd_arm64.go",
+				"vendor/golang.org/x/sys/unix/ztypes_netbsd_386.go",
+				"vendor/golang.org/x/sys/unix/ztypes_netbsd_amd64.go",
+				"vendor/golang.org/x/sys/unix/ztypes_netbsd_arm.go",
+				"vendor/golang.org/x/sys/unix/ztypes_netbsd_arm64.go",
+				"vendor/modules.txt",
+			},
+			added: []string{
+				"vendor/github.com/gliderlabs/ssh/go.mod",
+				"vendor/github.com/gliderlabs/ssh/go.sum",
+			},
+			removed: []string{},
+		},
+		{
+			// git 1.7.2 adds an unnecessary \x00 on merge commit
+			output: "\x00MM\x00options/locale/locale_en-US.ini\x00",
+			modified: []string{
+				"options/locale/locale_en-US.ini",
+			},
+			added:   []string{},
+			removed: []string{},
+		},
+		{
+			// git 1.7.2 adds an unnecessary \n on normal commit
+			output: "\nD\x00b\x00D\x00b b/b\x00A\x00b b/b b/b b/b\x00A\x00b b/b b/b b/b b/b\x00",
+			removed: []string{
+				"b",
+				"b b/b",
+			},
+			modified: []string{},
+			added: []string{
+				"b b/b b/b b/b",
+				"b b/b b/b b/b b/b",
+			},
+		},
+	}
+
+	for _, kase := range kases {
+		fileStatus := NewCommitFileStatus()
+		parseCommitFileStatus(fileStatus, strings.NewReader(kase.output))
+
+		assert.Equal(t, kase.added, fileStatus.Added)
+		assert.Equal(t, kase.removed, fileStatus.Removed)
+		assert.Equal(t, kase.modified, fileStatus.Modified)
+	}
+}
+
+func TestGetCommitFileStatusMerges(t *testing.T) {
+	bareRepo6 := mockRepository("repo6_merge")
+
+	commitFileStatus, err := GetCommitFileStatus(t.Context(), bareRepo6, "022f4ce6214973e018f02bf363bf8a2e3691f699")
+	assert.NoError(t, err)
+
+	expected := CommitFileStatus{
+		[]string{
+			"add_file.txt",
+		},
+		[]string{
+			"to_remove.txt",
+		},
+		[]string{
+			"to_modify.txt",
+		},
+	}
+
+	assert.Equal(t, expected.Added, commitFileStatus.Added)
+	assert.Equal(t, expected.Removed, commitFileStatus.Removed)
+	assert.Equal(t, expected.Modified, commitFileStatus.Modified)
+}
+
+func TestGetCommitFileStatusMergesSha256(t *testing.T) {
+	bareRepo6Sha256 := mockRepository("repo6_merge_sha256")
+
+	commitFileStatus, err := GetCommitFileStatus(t.Context(), bareRepo6Sha256, "d2e5609f630dd8db500f5298d05d16def282412e3e66ed68cc7d0833b29129a1")
+	assert.NoError(t, err)
+
+	expected := CommitFileStatus{
+		[]string{
+			"add_file.txt",
+		},
+		[]string{},
+		[]string{
+			"to_modify.txt",
+		},
+	}
+
+	assert.Equal(t, expected.Added, commitFileStatus.Added)
+	assert.Equal(t, expected.Removed, commitFileStatus.Removed)
+	assert.Equal(t, expected.Modified, commitFileStatus.Modified)
+
+	expected = CommitFileStatus{
+		[]string{},
+		[]string{
+			"to_remove.txt",
+		},
+		[]string{},
+	}
+
+	commitFileStatus, err = GetCommitFileStatus(t.Context(), bareRepo6Sha256, "da1ded40dc8e5b7c564171f4bf2fc8370487decfb1cb6a99ef28f3ed73d09172")
+	assert.NoError(t, err)
+
+	assert.Equal(t, expected.Added, commitFileStatus.Added)
+	assert.Equal(t, expected.Removed, commitFileStatus.Removed)
+	assert.Equal(t, expected.Modified, commitFileStatus.Modified)
+}

modules/git/commit_info.go

@@ -0,0 +1,115 @@
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"maps"
+	"path"
+	"time"
+
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/util"
+)
+
+// CommitInfo describes the first commit with the provided entry
+type CommitInfo struct {
+	Entry         *TreeEntry
+	Commit        *Commit
+	SubmoduleFile *CommitSubmoduleFile
+}
+
+func GetCommitInfoSubmoduleFile(ctx context.Context, repoLink, fullPath string, gitRepo *Repository, commit *Commit, refCommitID ObjectID) (*CommitSubmoduleFile, error) {
+	submodule, err := commit.GetSubModule(ctx, gitRepo, fullPath)
+	if err != nil {
+		return nil, err
+	}
+	if submodule == nil {
+		// unable to find submodule from ".gitmodules" file
+		return NewCommitSubmoduleFile(repoLink, fullPath, "", refCommitID.String()), nil
+	}
+	return NewCommitSubmoduleFile(repoLink, fullPath, submodule.URL, refCommitID.String()), nil
+}
+
+func getLastCommitForPathsWithTimeout(ctx context.Context, timeout time.Duration, gitRepo *Repository, commit *Commit, treePath string, entryNames []string) (map[string]*Commit, error) {
+	// Actually using "timeout ctx" here is not ideal: it creates a new git cat-file batch process
+	// (the other one uses the request's ctx and is cached by the gitRepo).
+	// Ideally, the underlying functions only need to return in predictable time,
+	// so it should use the shared (cached) git cat-file batcher from the request context and just use "time.After" for deadline.
+	// However, it's difficult to make it right at the moment.
+	ctxInner := ctx
+	if timeout > 0 {
+		var cancel context.CancelFunc
+		ctxInner, cancel = context.WithTimeout(ctx, timeout)
+		defer cancel()
+	}
+	revs, err := GetLastCommitForPaths(ctxInner, gitRepo, commit, treePath, entryNames)
+	if err != nil && ctxInner.Err() == nil {
+		// only return the err if context is not canceled
+		return nil, err
+	}
+	return revs, nil
+}
+
+// GetCommitsInfo gets information of all commits that are corresponding to these entries
+func (tes Entries) GetCommitsInfo(ctx context.Context, timeout time.Duration, repoLink string, gitRepo *Repository, commit *Commit, treePath string) (commitsInfo []CommitInfo, treeLastCommit *Commit, err error) {
+	entryNames := make([]string, 0, len(tes)+1)
+	entryNames = append(entryNames, "") // include the commit for the treePath itself
+	for _, entry := range tes {
+		entryNames = append(entryNames, entry.Name())
+	}
+
+	revs, remainingEntryNames, err := getLastCommitForPathsByCache(ctx, commit.ID.String(), treePath, entryNames, gitRepo.LastCommitCache)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	if len(remainingEntryNames) > 0 {
+		remainingRevs, err := getLastCommitForPathsWithTimeout(ctx, timeout, gitRepo, commit, treePath, remainingEntryNames)
+		if err != nil {
+			return nil, nil, err
+		}
+		maps.Copy(revs, remainingRevs)
+	}
+
+	for _, entry := range tes {
+		ci := CommitInfo{Entry: entry, Commit: revs[entry.Name()]}
+		if ci.Commit == nil {
+			log.Debug("missing commit for %s", entry.Name())
+		}
+		// If the entry is a submodule, add a submodule file for this
+		if entry.IsSubModule() {
+			ci.SubmoduleFile, err = GetCommitInfoSubmoduleFile(ctx, repoLink, path.Join(treePath, entry.Name()), gitRepo, commit, entry.ID)
+			if err != nil {
+				return nil, nil, err
+			}
+		}
+		commitsInfo = append(commitsInfo, ci)
+	}
+
+	// Retrieve the commit for the treePath itself (see above). We basically
+	// get it for free during the tree traversal, and it's used for listing
+	// pages to display information about the newest commit for a given path.
+	treeLastCommit = util.Iif(treePath == "", commit, revs[""])
+	return commitsInfo, treeLastCommit, nil
+}
+
+func getLastCommitForPathsByCache(ctx context.Context, commitID, treePath string, paths []string, cache *LastCommitCache) (map[string]*Commit, []string, error) {
+	var unHitEntryPaths []string
+	results := make(map[string]*Commit)
+	for _, p := range paths {
+		lastCommit, err := cache.Get(ctx, commitID, path.Join(treePath, p))
+		if err != nil {
+			return nil, nil, err
+		}
+		if lastCommit != nil {
+			results[p] = lastCommit
+			continue
+		}
+
+		unHitEntryPaths = append(unHitEntryPaths, p)
+	}
+
+	return results, unHitEntryPaths, nil
+}

modules/git/commit_info_gogit.go

@@ -0,0 +1,207 @@
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build gogit
+
+package git
+
+import (
+	"context"
+	"path"
+
+	"github.com/emirpasic/gods/trees/binaryheap"
+	"github.com/go-git/go-git/v5/plumbing"
+	"github.com/go-git/go-git/v5/plumbing/object"
+	cgobject "github.com/go-git/go-git/v5/plumbing/object/commitgraph"
+)
+
+type commitAndPaths struct {
+	commit cgobject.CommitNode
+	// Paths that are still on the branch represented by commit
+	paths []string
+	// Set of hashes for the paths
+	hashes map[string]plumbing.Hash
+}
+
+func getCommitTree(c cgobject.CommitNode, treePath string) (*object.Tree, error) {
+	tree, err := c.Tree()
+	if err != nil {
+		return nil, err
+	}
+
+	// Optimize deep traversals by focusing only on the specific tree
+	if treePath != "" {
+		tree, err = tree.Tree(treePath)
+		if err != nil {
+			return nil, err
+		}
+	}
+
+	return tree, nil
+}
+
+func getFileHashes(c cgobject.CommitNode, treePath string, paths []string) (map[string]plumbing.Hash, error) {
+	tree, err := getCommitTree(c, treePath)
+	if err == object.ErrDirectoryNotFound {
+		// The whole tree didn't exist, so return empty map
+		return make(map[string]plumbing.Hash), nil
+	}
+	if err != nil {
+		return nil, err
+	}
+
+	hashes := make(map[string]plumbing.Hash)
+	for _, path := range paths {
+		if path != "" {
+			entry, err := tree.FindEntry(path)
+			if err == nil {
+				hashes[path] = entry.Hash
+			}
+		} else {
+			hashes[path] = tree.Hash
+		}
+	}
+
+	return hashes, nil
+}
+
+// GetLastCommitForPaths returns last commit information
+func GetLastCommitForPaths(ctx context.Context, gitRepo *Repository, commit *Commit, treePath string, paths []string) (map[string]*Commit, error) {
+	commitNodeIndex, closer := gitRepo.CommitNodeIndex()
+	defer closer()
+
+	c, err := commitNodeIndex.Get(plumbing.Hash(commit.ID.RawValue()))
+	if err != nil {
+		return nil, err
+	}
+	return getLastCommitForPathsByCommitNode(ctx, gitRepo, c, treePath, paths)
+}
+
+func getLastCommitForPathsByCommitNode(ctx context.Context, gitRepo *Repository, c cgobject.CommitNode, treePath string, paths []string) (map[string]*Commit, error) {
+	refSha := c.ID().String()
+
+	// We do a tree traversal with nodes sorted by commit time
+	heap := binaryheap.NewWith(func(a, b any) int {
+		if a.(*commitAndPaths).commit.CommitTime().Before(b.(*commitAndPaths).commit.CommitTime()) { //nolint:forcetypeassert // this heap only ever holds *commitAndPaths
+			return 1
+		}
+		return -1
+	})
+
+	resultNodes := make(map[string]cgobject.CommitNode)
+	initialHashes, err := getFileHashes(c, treePath, paths)
+	if err != nil {
+		return nil, err
+	}
+
+	// Start search from the root commit and with full set of paths
+	heap.Push(&commitAndPaths{c, paths, initialHashes})
+heaploop:
+	for {
+		select {
+		case <-ctx.Done():
+			if ctx.Err() == context.DeadlineExceeded {
+				break heaploop
+			}
+			return nil, ctx.Err()
+		default:
+		}
+		cIn, ok := heap.Pop()
+		if !ok {
+			break
+		}
+		current := cIn.(*commitAndPaths) //nolint:forcetypeassert // this heap only ever holds *commitAndPaths
+
+		// Load the parent commits for the one we are currently examining
+		numParents := current.commit.NumParents()
+		var parents []cgobject.CommitNode
+		for i := range numParents {
+			parent, err := current.commit.ParentNode(i)
+			if err != nil {
+				break
+			}
+			parents = append(parents, parent)
+		}
+
+		// Examine the current commit and set of interesting paths
+		pathUnchanged := make([]bool, len(current.paths))
+		parentHashes := make([]map[string]plumbing.Hash, len(parents))
+		for j, parent := range parents {
+			parentHashes[j], err = getFileHashes(parent, treePath, current.paths)
+			if err != nil {
+				break
+			}
+
+			for i, path := range current.paths {
+				if parentHashes[j][path] == current.hashes[path] {
+					pathUnchanged[i] = true
+				}
+			}
+		}
+
+		var remainingPaths []string
+		for i, pth := range current.paths {
+			// The results could already contain some newer change for the same path,
+			// so don't override that and bail out on the file early.
+			if resultNodes[pth] == nil {
+				if pathUnchanged[i] {
+					// The path existed with the same hash in at least one parent so it could
+					// not have been changed in this commit directly.
+					remainingPaths = append(remainingPaths, pth)
+				} else {
+					// There are few possible cases how can we get here:
+					// - The path didn't exist in any parent, so it must have been created by
+					//   this commit.
+					// - The path did exist in the parent commit, but the hash of the file has
+					//   changed.
+					// - We are looking at a merge commit and the hash of the file doesn't
+					//   match any of the hashes being merged. This is more common for directories,
+					//   but it can also happen if a file is changed through conflict resolution.
+					resultNodes[pth] = current.commit
+					if err := gitRepo.LastCommitCache.Put(refSha, path.Join(treePath, pth), current.commit.ID().String()); err != nil {
+						return nil, err
+					}
+				}
+			}
+		}
+
+		if len(remainingPaths) > 0 {
+			// Add the parent nodes along with remaining paths to the heap for further
+			// processing.
+			for j, parent := range parents {
+				// Combine remainingPath with paths available on the parent branch
+				// and make union of them
+				remainingPathsForParent := make([]string, 0, len(remainingPaths))
+				newRemainingPaths := make([]string, 0, len(remainingPaths))
+				for _, path := range remainingPaths {
+					if parentHashes[j][path] == current.hashes[path] {
+						remainingPathsForParent = append(remainingPathsForParent, path)
+					} else {
+						newRemainingPaths = append(newRemainingPaths, path)
+					}
+				}
+
+				if remainingPathsForParent != nil {
+					heap.Push(&commitAndPaths{parent, remainingPathsForParent, parentHashes[j]})
+				}
+
+				if len(newRemainingPaths) == 0 {
+					break
+				}
+				remainingPaths = newRemainingPaths
+			}
+		}
+	}
+
+	// Post-processing
+	result := make(map[string]*Commit)
+	for path, commitNode := range resultNodes {
+		commit, err := commitNode.Commit()
+		if err != nil {
+			return nil, err
+		}
+		result[path] = convertCommit(commit)
+	}
+
+	return result, nil
+}

modules/git/commit_info_nogogit.go

@@ -0,0 +1,43 @@
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build !gogit
+
+package git
+
+import (
+	"context"
+)
+
+// GetLastCommitForPaths returns last commit information
+func GetLastCommitForPaths(ctx context.Context, gitRepo *Repository, commit *Commit, treePath string, paths []string) (map[string]*Commit, error) {
+	// We read backwards from the commit to obtain all of the commits
+	revs, err := walkGitLog(ctx, gitRepo, commit, treePath, paths...)
+	if err != nil {
+		return nil, err
+	}
+
+	commitsMap := map[string]*Commit{}
+	commitsMap[commit.ID.String()] = commit
+
+	commitCommits := map[string]*Commit{}
+	for path, commitID := range revs {
+		if len(commitID) == 0 {
+			continue
+		}
+
+		c, ok := commitsMap[commitID]
+		if ok {
+			commitCommits[path] = c
+			continue
+		}
+
+		c, err := gitRepo.GetCommit(ctx, commitID) // Ensure the commit exists in the repository
+		if err != nil {
+			return nil, err
+		}
+		commitCommits[path] = c
+	}
+
+	return commitCommits, nil
+}

modules/git/commit_info_nogogit_test.go

@@ -0,0 +1,57 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build !gogit
+
+package git
+
+import (
+	"context"
+	"path/filepath"
+	"testing"
+	"time"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/test"
+	"gitea.dev/modules/util"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestEntries_GetCommitsInfo_ContextErr(t *testing.T) {
+	repoPath, _ := filepath.Abs(filepath.Join(testReposDir, "repo1_bare"))
+	repo, err := OpenRepository(t.Context(), gitrepo.RepositoryManaged("dummy", repoPath))
+	require.NoError(t, err)
+	defer repo.Close()
+
+	commit, err := repo.GetCommit(t.Context(), "feaf4ba6bc635fec442f46ddd4512416ec43c2c2")
+	require.NoError(t, err)
+	entries, err := commit.Tree().ListEntries(t.Context(), repo)
+	require.NoError(t, err)
+
+	countCommitInfosCommit := func(infos []CommitInfo) (nilCommits, nonNilCommits int) {
+		for _, info := range infos {
+			nilCommits += util.Iif(info.Commit == nil, 1, 0)
+			nonNilCommits += util.Iif(info.Commit != nil, 1, 0)
+		}
+		return nilCommits, nonNilCommits
+	}
+
+	ctx, cancel := context.WithCancel(t.Context())
+	defer test.MockVariableValue(&walkGitLogDebugBeforeNext)()
+
+	walkGitLogDebugBeforeNext = cancel
+	commitInfos, _, err := entries.GetCommitsInfo(ctx, time.Second, "/any/repo-link", repo, commit, "")
+	assert.NoError(t, err)
+	nilCommits, nonNilCommits := countCommitInfosCommit(commitInfos)
+	assert.Equal(t, 0, nonNilCommits) // no commit info due to canceled (or deadline-exceeded) context
+	assert.Equal(t, 3, nilCommits)
+
+	walkGitLogDebugBeforeNext = nil
+	commitInfos, _, err = entries.GetCommitsInfo(t.Context(), time.Second, "/any/repo-link", repo, commit, "")
+	assert.NoError(t, err)
+	nilCommits, nonNilCommits = countCommitInfosCommit(commitInfos)
+	assert.Equal(t, 3, nonNilCommits)
+	assert.Equal(t, 0, nilCommits)
+}

modules/git/commit_info_test.go

@@ -0,0 +1,166 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"path/filepath"
+	"testing"
+	"time"
+
+	"gitea.dev/modules/git/gitrepo"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func cloneRepo(tb testing.TB, url string) (string, error) {
+	repoDir := tb.TempDir()
+	if err := Clone(tb.Context(), url, repoDir, CloneRepoOptions{
+		Mirror:  false,
+		Bare:    false,
+		Quiet:   true,
+		Timeout: 5 * time.Minute,
+	}); err != nil {
+		return "", err
+	}
+	return repoDir, nil
+}
+
+func testGetCommitsInfo(t *testing.T, repo1 *Repository) {
+	type expectedEntryInfo struct {
+		CommitID string
+		Size     int64
+	}
+
+	// these test case are specific to the repo1 test repo
+	testCases := []struct {
+		CommitID           string
+		Path               string
+		ExpectedIDs        map[string]expectedEntryInfo
+		ExpectedTreeCommit string
+	}{
+		{"8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2", "", map[string]expectedEntryInfo{
+			"file1.txt": {
+				CommitID: "95bb4d39648ee7e325106df01a621c530863a653",
+				Size:     6,
+			},
+			"file2.txt": {
+				CommitID: "8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2",
+				Size:     6,
+			},
+		}, "8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2"},
+		{"2839944139e0de9737a044f78b0e4b40d989a9e3", "", map[string]expectedEntryInfo{
+			"file1.txt": {
+				CommitID: "2839944139e0de9737a044f78b0e4b40d989a9e3",
+				Size:     15,
+			},
+			"branch1.txt": {
+				CommitID: "9c9aef8dd84e02bc7ec12641deb4c930a7c30185",
+				Size:     8,
+			},
+		}, "2839944139e0de9737a044f78b0e4b40d989a9e3"},
+		{"5c80b0245c1c6f8343fa418ec374b13b5d4ee658", "branch2", map[string]expectedEntryInfo{
+			"branch2.txt": {
+				CommitID: "5c80b0245c1c6f8343fa418ec374b13b5d4ee658",
+				Size:     8,
+			},
+		}, "5c80b0245c1c6f8343fa418ec374b13b5d4ee658"},
+		{"feaf4ba6bc635fec442f46ddd4512416ec43c2c2", "", map[string]expectedEntryInfo{
+			"file1.txt": {
+				CommitID: "95bb4d39648ee7e325106df01a621c530863a653",
+				Size:     6,
+			},
+			"file2.txt": {
+				CommitID: "8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2",
+				Size:     6,
+			},
+			"foo": {
+				CommitID: "37991dec2c8e592043f47155ce4808d4580f9123",
+				Size:     0,
+			},
+		}, "feaf4ba6bc635fec442f46ddd4512416ec43c2c2"},
+	}
+	for _, testCase := range testCases {
+		commit, err := repo1.GetCommit(t.Context(), testCase.CommitID)
+		if err != nil {
+			assert.NoError(t, err, "Unable to get commit: %s from testcase due to error: %v", testCase.CommitID, err)
+			// no point trying to do anything else for this test.
+			continue
+		}
+		assert.NotNil(t, commit)
+		assert.NotNil(t, commit.TreeID)
+
+		tree, err := commit.SubTree(t.Context(), repo1, testCase.Path)
+		if err != nil {
+			assert.NoError(t, err, "Unable to get subtree: %s of commit: %s from testcase due to error: %v", testCase.Path, testCase.CommitID, err)
+			// no point trying to do anything else for this test.
+			continue
+		}
+
+		assert.NotNil(t, tree, "tree is nil for testCase CommitID %s in Path %s", testCase.CommitID, testCase.Path)
+
+		entries, err := tree.ListEntries(t.Context(), repo1)
+		if err != nil {
+			assert.NoError(t, err, "Unable to get entries of subtree: %s in commit: %s from testcase due to error: %v", testCase.Path, testCase.CommitID, err)
+			// no point trying to do anything else for this test.
+			continue
+		}
+
+		commitsInfo, treeCommit, err := entries.GetCommitsInfo(t.Context(), time.Second, "/any/repo-link", repo1, commit, testCase.Path)
+		assert.NoError(t, err, "Unable to get commit information for entries of subtree: %s in commit: %s from testcase due to error: %v", testCase.Path, testCase.CommitID, err)
+		if err != nil {
+			t.FailNow()
+		}
+		assert.Equal(t, testCase.ExpectedTreeCommit, treeCommit.ID.String())
+		assert.Len(t, commitsInfo, len(testCase.ExpectedIDs))
+		for _, commitInfo := range commitsInfo {
+			entry := commitInfo.Entry
+			commit := commitInfo.Commit
+			expectedInfo, ok := testCase.ExpectedIDs[entry.Name()]
+			if !assert.True(t, ok) {
+				continue
+			}
+			assert.Equal(t, expectedInfo.CommitID, commit.ID.String())
+			assert.Equal(t, expectedInfo.Size, entry.GetSize(t.Context(), repo1), entry.Name())
+		}
+	}
+}
+
+func TestEntries_GetCommitsInfo(t *testing.T) {
+	bareRepo1Path, _ := filepath.Abs(filepath.Join(testReposDir, "repo1_bare"))
+	bareRepo1, err := OpenRepository(t.Context(), gitrepo.RepositoryManaged("repo1_bare", bareRepo1Path))
+	require.NoError(t, err)
+	defer bareRepo1.Close()
+
+	testGetCommitsInfo(t, bareRepo1)
+
+	clonedPath, err := cloneRepo(t, bareRepo1Path)
+	if err != nil {
+		assert.NoError(t, err)
+	}
+	clonedRepo1, err := OpenRepository(t.Context(), gitrepo.RepositoryManaged("repo1_bare-clone", clonedPath))
+	if err != nil {
+		assert.NoError(t, err)
+	}
+	defer clonedRepo1.Close()
+
+	testGetCommitsInfo(t, clonedRepo1)
+
+	t.Run("NonExistingSubmoduleAsNil", func(t *testing.T) {
+		commit, err := bareRepo1.GetCommit(t.Context(), "HEAD")
+		require.NoError(t, err)
+		treeEntry, err := commit.GetTreeEntryByPath(t.Context(), bareRepo1, "file1.txt")
+		require.NoError(t, err)
+		cisf, err := GetCommitInfoSubmoduleFile(t.Context(), "/any/repo-link", "file1.txt", bareRepo1, commit, treeEntry.ID)
+		require.NoError(t, err)
+		assert.Equal(t, &CommitSubmoduleFile{
+			repoLink: "/any/repo-link",
+			fullPath: "file1.txt",
+			refURL:   "",
+			refID:    "e2129701f1a4d54dc44f03c93bca0a2aec7c5449",
+		}, cisf)
+		// since there is no refURL, it means that the submodule info doesn't exist, so it won't have a web link
+		assert.Nil(t, cisf.SubmoduleWebLinkTree(t.Context()))
+	})
+}

modules/git/commit_message.go

@@ -0,0 +1,194 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"net/mail"
+	"regexp"
+	"strings"
+	"sync"
+
+	"gitea.dev/modules/charset"
+	"gitea.dev/modules/util"
+)
+
+// CoAuthoredByTrailer is the canonical token for the `Co-authored-by:` git trailer.
+const CoAuthoredByTrailer = "Co-authored-by"
+
+const (
+	commitIdentityRoleAuthor    = 1
+	commitIdentityRoleCommitter = 2
+	commitIdentityRoleCoAuthor  = 3
+)
+
+type CommitIdentity struct {
+	Name  string
+	Email string
+	role  int
+}
+
+// CommitMessageTrailerValues keys are all in lower-case
+type CommitMessageTrailerValues map[string][]string
+
+type CommitMessage struct {
+	MessageRaw   string
+	messageUTF8  *string
+	messageTitle *string
+	messageBody  *string
+
+	trailerValues CommitMessageTrailerValues
+
+	allAuthors []*CommitIdentity
+}
+
+func (c *CommitMessage) MessageUTF8() string {
+	if c.messageUTF8 == nil {
+		bs := charset.ToUTF8(util.UnsafeStringToBytes(c.MessageRaw), charset.ConvertOpts{ErrorReplacement: []byte{'?'}})
+		c.messageUTF8 = new(util.UnsafeBytesToString(bs))
+	}
+	return *c.messageUTF8
+}
+
+func (c *CommitMessage) MessageTitle() string {
+	if c.messageTitle == nil {
+		s, _, _ := strings.Cut(strings.TrimSpace(c.MessageUTF8()), "\n")
+		c.messageTitle = new(strings.TrimSpace(s))
+	}
+	return *c.messageTitle
+}
+
+func (c *CommitMessage) MessageBody() string {
+	if c.messageBody == nil {
+		_, s, _ := strings.Cut(strings.TrimSpace(c.MessageUTF8()), "\n")
+		c.messageBody = new(strings.TrimSpace(s))
+	}
+	return *c.messageBody
+}
+
+func (c *CommitMessage) MessageTrailer() CommitMessageTrailerValues {
+	if c.trailerValues == nil {
+		_, _, trailer := CommitMessageSplitTrailer(c.MessageUTF8())
+		c.trailerValues = CommitMessageParseTrailer(trailer)
+	}
+	return c.trailerValues
+}
+
+var commitMessageTrailerSplit = sync.OnceValue(func() *regexp.Regexp {
+	// ref: https://git-scm.com/docs/git-interpret-trailers
+	// TODO: the regexp is not able to perfectly parse the all kinds of trailers
+	// It was just copied from legacy code, it is not exactly the same as how Git parses the trailer and not quite right in some cases.
+	// For the key characters: it follows RFC 822 field name syntax (or RFC 2822/RFC 5322): printable ASCII characters between 33 and 126 except the colon (:),
+	// but maybe we don't want to make it that complicated, so here we only support some common "symbol-like" characters.
+	return regexp.MustCompile(`(?s)^(?P<content>.*?)(?P<sep>^|^\n|^-{3,}\n+|\n+-{3,}\n+|\n{2,})(?P<trailer>(?:[A-Za-z0-9][-\w]*:[^\n]*(\n\s+[^\n]*)*\n?)*\n*)$`)
+})
+
+// CommitMessageSplitTrailer tries to split the message by the trailer separator
+// content + sep + trailer will reconstruct the original message
+func CommitMessageSplitTrailer(s string) (content, sep, trailer string) {
+	s = util.NormalizeStringEOL(s)
+	re := commitMessageTrailerSplit()
+	v := re.FindStringSubmatch(s)
+	if v == nil {
+		return s, "", ""
+	}
+	return v[re.SubexpIndex("content")], v[re.SubexpIndex("sep")], v[re.SubexpIndex("trailer")]
+}
+
+// CommitMessageMerge merges two commit messages with their trailers
+func CommitMessageMerge(m1, m2 string) string {
+	c1, s1, t1 := CommitMessageSplitTrailer(m1)
+	c2, s2, t2 := CommitMessageSplitTrailer(m2)
+	c1, t1 = strings.TrimSpace(c1), strings.TrimSpace(t1)
+	c2, t2 = strings.TrimSpace(c2), strings.TrimSpace(t2)
+	out := strings.Builder{}
+	if c1 != "" && c2 != "" {
+		out.WriteString(c1)
+		out.WriteString("\n\n")
+		out.WriteString(c2)
+	} else if c1 != "" {
+		out.WriteString(c1)
+	} else if c2 != "" {
+		out.WriteString(c2)
+	}
+	if t1 != "" || t2 != "" {
+		sep := util.Iif(t1 == "", s2, s1)
+		sep = util.IfZero(sep, "\n\n")
+		if c1 != "" || c2 != "" {
+			out.WriteString(sep)
+		}
+		if t1 != "" {
+			out.WriteString(t1)
+		}
+		if t1 != "" && t2 != "" {
+			out.WriteString("\n")
+		}
+		if t2 != "" {
+			out.WriteString(t2)
+		}
+	}
+	return out.String()
+}
+
+func CommitMessageParseTrailer(s string) CommitMessageTrailerValues {
+	ret := CommitMessageTrailerValues{}
+	for line := range strings.SplitSeq(util.NormalizeStringEOL(s), "\n") {
+		k, v, ok := strings.Cut(line, ":")
+		if !ok {
+			continue
+		}
+		k, v = strings.TrimSpace(k), strings.TrimSpace(v)
+		kLower := strings.ToLower(k)
+		ret[kLower] = append(ret[kLower], v)
+	}
+	return ret
+}
+
+// AllAuthorIdentities returns all the author and co-authors in the commit. Committer is not included:
+// * Author & Co-author: they changed the code (attribution)
+// * Committer: they submitted the commit but didn't change the code (e.g.: maintainer signed a commit)
+func (c *Commit) AllAuthorIdentities() []*CommitIdentity {
+	if c.allAuthors != nil {
+		return c.allAuthors
+	}
+	trailerCoAuthors := c.MessageTrailer()["co-authored-by"]
+	c.allAuthors = make([]*CommitIdentity, 0, 1+len(trailerCoAuthors))
+	exclude := map[string]int{}
+	addAuthor := func(name, email string, role int) {
+		if name == "" && email == "" {
+			return
+		}
+		key := strings.ToLower(email)
+		if key == "" {
+			key = strings.ToLower(name)
+		}
+		if existingRole := exclude[key]; key != "" && existingRole != 0 {
+			return
+		}
+		c.allAuthors = append(c.allAuthors, &CommitIdentity{Name: name, Email: email, role: role})
+		exclude[key] = role
+	}
+
+	addAuthor(c.Author.Name, c.Author.Email, commitIdentityRoleAuthor)
+	for _, coAuthorValue := range trailerCoAuthors {
+		coAuthorName, coAuthorEmail := parseCommitIdentityValue(coAuthorValue)
+		addAuthor(coAuthorName, coAuthorEmail, commitIdentityRoleCoAuthor)
+	}
+	return c.allAuthors
+}
+
+func (c *Commit) CommitterIsAuthor() bool {
+	return c.Committer.Name == c.Author.Name && c.Committer.Email == c.Author.Email
+}
+
+// Git identities are not RFC 5322 addresses: net/mail rejects names like "dependabot[bot]", so fall back to the angle-addr.
+func parseCommitIdentityValue(value string) (name, email string) {
+	if addr, err := mail.ParseAddress(value); err == nil {
+		return addr.Name, addr.Address
+	}
+	begin, end := strings.LastIndex(value, "<"), strings.LastIndex(value, ">")
+	if begin == -1 || end < begin {
+		return value, ""
+	}
+	return strings.TrimSpace(value[:begin]), strings.TrimSpace(value[begin+1 : end])
+}

modules/git/commit_message_test.go

@@ -0,0 +1,151 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestCommitMessageSanitizesInvalidUTF8(t *testing.T) {
+	commit := &Commit{
+		CommitMessage: CommitMessage{MessageRaw: "title \xff\n\n\n\nbody \xff\n\n\n"},
+	}
+	assert.Equal(t, "title ÿ", commit.MessageTitle())
+	assert.Equal(t, "body ÿ", commit.MessageBody())
+	assert.Equal(t, "title ÿ\n\n\n\nbody ÿ\n\n\n", commit.MessageUTF8())
+}
+
+func TestCommitMessageTrailer(t *testing.T) {
+	cases := []struct {
+		msg, body, sep, trailer string
+	}{
+		{"", "", "", ""},
+		{"a", "a", "", ""},
+		{"a\n\nk", "a\n\nk", "", ""},
+		{"a\n\nk:v", "a", "\n\n", "k:v"},
+		{"a\n\nk:v\n  next-line", "a", "\n\n", "k:v\n  next-line"},
+		{"a\n\nk:v\n  next-line\nother: v", "a", "\n\n", "k:v\n  next-line\nother: v"},
+		{"a\n\nk:v\n\n", "a", "\n\n", "k:v\n\n"},
+		{"a\n--\nk:v", "a\n--\nk:v", "", ""},
+		{"a\n---\nk:v", "a", "\n---\n", "k:v"}, // TODO: should we support such case? No empty line between "---" and the trailer
+		{"a\n\n---\n\nk:v", "a", "\n\n---\n\n", "k:v"},
+
+		{"k: v", "", "", "k: v"},
+		{"\nk:v", "", "\n", "k:v"},
+		{"\n\nk:v", "", "\n\n", "k:v"},
+
+		{"---\nk:v", "", "---\n", "k:v"},
+		{"\n---\nk:v", "", "\n---\n", "k:v"},
+		{"a:b\n---\nk:v", "a:b", "\n---\n", "k:v"},
+	}
+	for _, c := range cases {
+		body, sep, trailer := CommitMessageSplitTrailer(c.msg)
+		assert.Equal(t, c.body, body, "input=%q", c.msg)
+		assert.Equal(t, c.sep, sep, "input=%q", c.msg)
+		assert.Equal(t, c.trailer, trailer, "input=%q", c.msg)
+	}
+}
+
+func TestCommitMessageParticipants(t *testing.T) {
+	sig := func(n, e string) *Signature { return &Signature{Name: n, Email: e} }
+	idt := func(n, e string, r int) *CommitIdentity { return &CommitIdentity{n, e, r} }
+	roleAuthor, _, roleCoAuthor := commitIdentityRoleAuthor, commitIdentityRoleCommitter, commitIdentityRoleCoAuthor
+	type testCase struct {
+		name       string
+		commit     *Commit
+		identities []*CommitIdentity
+	}
+
+	t.Run("AllAuthors", func(t *testing.T) {
+		cases := []testCase{
+			{
+				"CommitterExcluded",
+				&Commit{
+					Author: sig("a", "a@m.com"), Committer: sig("c", "c@m.com"),
+					CommitMessage: CommitMessage{MessageRaw: "CO-Authored-BY: Full Name <x@m.com>"},
+				},
+				[]*CommitIdentity{idt("a", "a@m.com", roleAuthor), idt("Full Name", "x@m.com", roleCoAuthor)},
+			},
+			{
+				"AuthorIsCoAuthor",
+				&Commit{
+					Author: sig("a", "a@m.com"), Committer: sig("c", "c@m.com"),
+					CommitMessage: CommitMessage{MessageRaw: "CO-Authored-BY: other-name <a@m.com>"},
+				},
+				[]*CommitIdentity{idt("a", "a@m.com", roleAuthor)},
+			},
+			{
+				"EmptyAuthor", // synthesized commits (push feed) may have no author signature at all
+				&Commit{
+					Author: sig("", ""), Committer: sig("", ""),
+					CommitMessage: CommitMessage{MessageRaw: "Co-authored-by: c <c@m.com>"},
+				},
+				// but if the commit message contains co-authors, the co-authors are still parsed for "all authors"
+				// if it is a problem, the caller should fix the problem (provide correct "author")
+				[]*CommitIdentity{idt("c", "c@m.com", roleCoAuthor)},
+			},
+			{
+				"CoAuthorIsCommitter",
+				&Commit{
+					Author: sig("a", "a@m.com"), Committer: sig("c", "c@m.com"),
+					CommitMessage: CommitMessage{MessageRaw: "Co-authored-by: c <c@m.com>"},
+				},
+				[]*CommitIdentity{idt("a", "a@m.com", roleAuthor), idt("c", "c@m.com", roleCoAuthor)},
+			},
+			{
+				"CoAuthorNameOnlyAndDuplicate",
+				&Commit{
+					Author: sig("a", "a@m.com"), Committer: sig("c", "c@m.com"),
+					CommitMessage: CommitMessage{MessageRaw: "Co-authored-by: b\nCo-authored-by: b\nCo-authored-by: c"},
+				},
+				[]*CommitIdentity{idt("a", "a@m.com", roleAuthor), idt("b", "", roleCoAuthor), idt("c", "", roleCoAuthor)},
+			},
+			{
+				"CoAuthorNameNotAnEmailAddress", // names net/mail rejects, e.g. bots and names with a comma
+				&Commit{
+					Author: sig("a", "a@m.com"), Committer: sig("c", "c@m.com"),
+					CommitMessage: CommitMessage{MessageRaw: "Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>\nCo-authored-by: Smith, John <j@m.com>"},
+				},
+				[]*CommitIdentity{
+					idt("a", "a@m.com", roleAuthor),
+					idt("dependabot[bot]", "49699333+dependabot[bot]@users.noreply.github.com", roleCoAuthor),
+					idt("Smith, John", "j@m.com", roleCoAuthor),
+				},
+			},
+		}
+		for _, c := range cases {
+			assert.Equal(t, c.identities, c.commit.AllAuthorIdentities(), "case: %s", c.name)
+		}
+	})
+}
+
+func TestCommitMessageMerge(t *testing.T) {
+	cases := []struct {
+		m1, m2 string
+		out    string
+	}{
+		{"", "", ""},
+		{"msg1", "", "msg1"},
+		{"", "msg2", "msg2"},
+		{"msg1", "msg2", "msg1\n\nmsg2"},
+		{"k1: a", "", "k1: a"},
+		{"", "k2: b", "k2: b"},
+		{"k1: a", "k2: b", "k1: a\nk2: b"},
+		{"msg1", "k2: b", "msg1\n\nk2: b"},
+		{"k1: a", "msg2", "msg2\n\nk1: a"},
+		{"msg1\n\nk1: a", "msg2", "msg1\n\nmsg2\n\nk1: a"},
+		{"msg1\n----\nk1: a", "msg2", "msg1\n\nmsg2\n----\nk1: a"},
+		{"msg1\n\n----\n\nk1: a", "msg2", "msg1\n\nmsg2\n\n----\n\nk1: a"},
+		{"msg1", "msg2\n----\nk2: b", "msg1\n\nmsg2\n----\nk2: b"},
+		{"msg1", "msg2\n\nk2: b", "msg1\n\nmsg2\n\nk2: b"},
+		{"msg1\n\nk1: a", "msg2\n\nk2: b", "msg1\n\nmsg2\n\nk1: a\nk2: b"},
+	}
+
+	for i, c := range cases {
+		out := CommitMessageMerge(c.m1, c.m2)
+		assert.Equal(t, c.out, out, "idx=%d, m1=%q m2=%q", i, c.m1, c.m2)
+	}
+}

modules/git/commit_reader.go

@@ -0,0 +1,100 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"bytes"
+	"fmt"
+	"io"
+)
+
+const (
+	commitHeaderGpgsig       = "gpgsig"
+	commitHeaderGpgsigSha256 = "gpgsig-sha256"
+)
+
+func assignCommitFields(commit *Commit, headerKey string, headerValue []byte) error {
+	if len(headerValue) > 0 && headerValue[len(headerValue)-1] == '\n' {
+		headerValue = headerValue[:len(headerValue)-1] // remove trailing newline
+	}
+	switch headerKey {
+	case "tree":
+		objID, err := NewIDFromString(string(headerValue))
+		if err != nil {
+			return fmt.Errorf("invalid tree ID %q: %w", string(headerValue), err)
+		}
+		commit.TreeID = objID
+	case "parent":
+		objID, err := NewIDFromString(string(headerValue))
+		if err != nil {
+			return fmt.Errorf("invalid parent ID %q: %w", string(headerValue), err)
+		}
+		commit.Parents = append(commit.Parents, objID)
+	case "author":
+		commit.Author.Decode(headerValue)
+	case "committer":
+		commit.Committer.Decode(headerValue)
+	case commitHeaderGpgsig, commitHeaderGpgsigSha256:
+		// if there are duplicate "gpgsig" and "gpgsig-sha256" headers, then the signature must have already been invalid
+		// so we don't need to handle duplicate headers here
+		commit.Signature = &CommitSignature{Signature: string(headerValue)}
+	}
+	return nil
+}
+
+// CommitFromReader will generate a Commit from a provided reader
+// We need this to interpret commits from cat-file or cat-file --batch
+//
+// If used as part of a cat-file --batch stream you need to limit the reader to the correct size
+func CommitFromReader(objectID ObjectID, reader io.Reader) (*Commit, error) {
+	commit := &Commit{
+		ID:        objectID,
+		Author:    &Signature{},
+		Committer: &Signature{},
+	}
+
+	bufReader := bufio.NewReader(reader)
+	inHeader := true
+	var payloadSB, messageSB bytes.Buffer
+	var headerKey string
+	var headerValue []byte
+	for {
+		line, err := bufReader.ReadBytes('\n')
+		if err != nil && err != io.EOF {
+			return nil, fmt.Errorf("unable to read commit %q: %w", objectID.String(), err)
+		}
+		if len(line) == 0 {
+			break
+		}
+
+		if inHeader {
+			inHeader = !(len(line) == 1 && line[0] == '\n') // still in header if line is not just a newline
+			k, v, _ := bytes.Cut(line, []byte{' '})
+			if len(k) != 0 || !inHeader {
+				if headerKey != "" {
+					if err = assignCommitFields(commit, headerKey, headerValue); err != nil {
+						return nil, fmt.Errorf("unable to parse commit %q: %w", objectID.String(), err)
+					}
+				}
+				headerKey = string(k) // it also resets the headerValue to empty string if not inHeader
+				headerValue = v
+			} else {
+				headerValue = append(headerValue, v...)
+			}
+			if headerKey != commitHeaderGpgsig && headerKey != commitHeaderGpgsigSha256 {
+				_, _ = payloadSB.Write(line)
+			}
+		} else {
+			_, _ = messageSB.Write(line)
+			_, _ = payloadSB.Write(line)
+		}
+	}
+
+	commit.MessageRaw = messageSB.String()
+	if commit.Signature != nil {
+		commit.Signature.Payload = payloadSB.String()
+	}
+	return commit, nil
+}

modules/git/commit_sha256_test.go

@@ -0,0 +1,128 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build !gogit
+
+package git
+
+import (
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestGetFullCommitIDSha256(t *testing.T) {
+	id, err := GetFullCommitID(t.Context(), mockRepository("repo1_bare_sha256"), "f004f4")
+	assert.NoError(t, err)
+	assert.Equal(t, "f004f41359117d319dedd0eaab8c5259ee2263da839dcba33637997458627fdc", id)
+}
+
+func TestGetFullCommitIDErrorSha256(t *testing.T) {
+	id, err := GetFullCommitID(t.Context(), mockRepository("repo1_bare_sha256"), "unknown")
+	assert.Empty(t, id)
+	if assert.Error(t, err) {
+		assert.EqualError(t, err, "object does not exist [id: unknown, rel_path: ]")
+	}
+}
+
+func TestCommitFromReaderSha256(t *testing.T) {
+	commitString := `tree e7f9e96dd79c09b078cac8b303a7d3b9d65ff9b734e86060a4d20409fd379f9e
+parent 26e9ccc29fad747e9c5d9f4c9ddeb7eff61cc45ef6a8dc258cbeb181afc055e8
+author Adam Majer <amajer@suse.de> 1698676906 +0100
+committer Adam Majer <amajer@suse.de> 1698676906 +0100
+gpgsig-sha256 -----BEGIN PGP SIGNATURE-----
+` + " " + `
+ iQIrBAABCgAtFiEES+fB08xlgTrzSdQvhkUIsBsmec8FAmU/wKoPHGFtYWplckBz
+ dXNlLmRlAAoJEIZFCLAbJnnP4s4PQIJATa++WPzR6/H4etT7bsOGoMyguEJYyWOd
+ aTybplzT7QAL7h2to0QszGabtzMJPIA39xSFZNYNN30voK5YyyYibXluPKgjemfK
+ WNXwF+gkwgZI38gSvKf+vlqI+EYyIFe19wOhiju0m8SIlB5NEPiWHa17q2mqmqqx
+ 1FWa2JdqLPYjAtSLFXeSZegrY5V1FxdemyMUONkg8YO9OSIMZiE0GsnnOXQ3xcT4
+ JTCnmlUxIKw689UiEY80JopUIq+Wl7+qq9507IYYSUCyB6JazL42AKMzVCbD+qBP
+ oOzh/hafYgk9H9qCQXaLbmvs17zXRpicig1bAzqgAy1FDelvpERyRTydEajSLIG6
+ U1cRCkgXCZ0NfsYNPPmBa8b3+rnstypXYTbyMwTln7FfUAaGo6o9JYiPMkzxlmsy
+ zfp/tcaY8+LlBL9aOJjtv+a0p+HrpCGd6CCa4ARfphTLq8QRSSh8uzlB9N+6HnRI
+ VAEUo6ecdDxSpyt2naeg9pKus/BRi7P6g4B1hkk/zZstUX/QP4IQuAJbXjkvsC+X
+ HKRr3NlRM/DygzTyj0gN74uoa0goCIbyAQhiT42nm0cuhM7uN/W0ayrlZjGF1cbR
+ 8NCJUL2Nwj0ywKIavC99Ipkb8AsFwpVT6U6effs6
+ =xybZ
+ -----END PGP SIGNATURE-----
+
+signed commit`
+
+	sha := &Sha256Hash{
+		0x94, 0x33, 0xb2, 0xa6, 0x2b, 0x96, 0x4c, 0x17, 0xa4, 0x48, 0x5a, 0xe1, 0x80, 0xf4, 0x5f, 0x59,
+		0x5d, 0x3e, 0x69, 0xd3, 0x1b, 0x78, 0x60, 0x87, 0x77, 0x5e, 0x28, 0xc6, 0xb6, 0x39, 0x9d, 0xf0,
+	}
+	gitRepo, err := OpenRepositoryLocal(t.Context(), filepath.Join(testReposDir, "repo1_bare_sha256"))
+	assert.NoError(t, err)
+	assert.NotNil(t, gitRepo)
+	defer gitRepo.Close()
+
+	commitFromReader, err := CommitFromReader(sha, strings.NewReader(commitString))
+	assert.NoError(t, err)
+	require.NotNil(t, commitFromReader)
+	assert.EqualValues(t, sha, commitFromReader.ID)
+	assert.Equal(t, `-----BEGIN PGP SIGNATURE-----
+
+iQIrBAABCgAtFiEES+fB08xlgTrzSdQvhkUIsBsmec8FAmU/wKoPHGFtYWplckBz
+dXNlLmRlAAoJEIZFCLAbJnnP4s4PQIJATa++WPzR6/H4etT7bsOGoMyguEJYyWOd
+aTybplzT7QAL7h2to0QszGabtzMJPIA39xSFZNYNN30voK5YyyYibXluPKgjemfK
+WNXwF+gkwgZI38gSvKf+vlqI+EYyIFe19wOhiju0m8SIlB5NEPiWHa17q2mqmqqx
+1FWa2JdqLPYjAtSLFXeSZegrY5V1FxdemyMUONkg8YO9OSIMZiE0GsnnOXQ3xcT4
+JTCnmlUxIKw689UiEY80JopUIq+Wl7+qq9507IYYSUCyB6JazL42AKMzVCbD+qBP
+oOzh/hafYgk9H9qCQXaLbmvs17zXRpicig1bAzqgAy1FDelvpERyRTydEajSLIG6
+U1cRCkgXCZ0NfsYNPPmBa8b3+rnstypXYTbyMwTln7FfUAaGo6o9JYiPMkzxlmsy
+zfp/tcaY8+LlBL9aOJjtv+a0p+HrpCGd6CCa4ARfphTLq8QRSSh8uzlB9N+6HnRI
+VAEUo6ecdDxSpyt2naeg9pKus/BRi7P6g4B1hkk/zZstUX/QP4IQuAJbXjkvsC+X
+HKRr3NlRM/DygzTyj0gN74uoa0goCIbyAQhiT42nm0cuhM7uN/W0ayrlZjGF1cbR
+8NCJUL2Nwj0ywKIavC99Ipkb8AsFwpVT6U6effs6
+=xybZ
+-----END PGP SIGNATURE-----`, commitFromReader.Signature.Signature)
+	assert.Equal(t, `tree e7f9e96dd79c09b078cac8b303a7d3b9d65ff9b734e86060a4d20409fd379f9e
+parent 26e9ccc29fad747e9c5d9f4c9ddeb7eff61cc45ef6a8dc258cbeb181afc055e8
+author Adam Majer <amajer@suse.de> 1698676906 +0100
+committer Adam Majer <amajer@suse.de> 1698676906 +0100
+
+signed commit`, commitFromReader.Signature.Payload)
+	assert.Equal(t, "Adam Majer <amajer@suse.de>", commitFromReader.Author.String())
+
+	commitFromReader2, err := CommitFromReader(sha, strings.NewReader(commitString+"\n\n"))
+	assert.NoError(t, err)
+	commitFromReader.CommitMessage.MessageRaw += "\n\n"
+	commitFromReader.Signature.Payload += "\n\n"
+	assert.Equal(t, commitFromReader, commitFromReader2)
+}
+
+func TestHasPreviousCommitSha256(t *testing.T) {
+	bareRepo1Path := filepath.Join(testReposDir, "repo1_bare_sha256")
+
+	repo, err := OpenRepositoryLocal(t.Context(), bareRepo1Path)
+	assert.NoError(t, err)
+	defer repo.Close()
+
+	commit, err := repo.GetCommit(t.Context(), "f004f41359117d319dedd0eaab8c5259ee2263da839dcba33637997458627fdc")
+	assert.NoError(t, err)
+
+	objectFormat, err := repo.GetObjectFormat(t.Context())
+	assert.NoError(t, err)
+
+	parentSHA := MustIDFromString("b0ec7af4547047f12d5093e37ef8f1b3b5415ed8ee17894d43a34d7d34212e9c")
+	notParentSHA := MustIDFromString("42e334efd04cd36eea6da0599913333c26116e1a537ca76e5b6e4af4dda00236")
+	assert.Equal(t, objectFormat, parentSHA.Type())
+	assert.Equal(t, "sha256", objectFormat.Name())
+
+	haz, err := commit.HasPreviousCommit(t.Context(), repo, parentSHA)
+	assert.NoError(t, err)
+	assert.True(t, haz)
+
+	hazNot, err := commit.HasPreviousCommit(t.Context(), repo, notParentSHA)
+	assert.NoError(t, err)
+	assert.False(t, hazNot)
+
+	selfNot, err := commit.HasPreviousCommit(t.Context(), repo, commit.ID)
+	assert.NoError(t, err)
+	assert.False(t, selfNot)
+}

modules/git/commit_submodule.go

@@ -0,0 +1,54 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import "context"
+
+type SubmoduleWebLink struct {
+	RepoWebLink, CommitWebLink string
+}
+
+// GetSubModules get all the submodules of current revision git tree
+func (c *Commit) GetSubModules(ctx context.Context, gitRepo *Repository) (*ObjectCache[*SubModule], error) {
+	if c.submoduleCache != nil {
+		return c.submoduleCache, nil
+	}
+
+	entry, err := c.GetTreeEntryByPath(ctx, gitRepo, ".gitmodules")
+	if err != nil {
+		if _, ok := err.(ErrNotExist); ok {
+			return nil, nil //nolint:nilnil // return nil to indicate that the submodule does not exist
+		}
+		return nil, err
+	}
+
+	rd, err := entry.Blob(gitRepo).DataAsync(ctx)
+	if err != nil {
+		return nil, err
+	}
+	defer rd.Close()
+
+	// at the moment we do not strictly limit the size of the .gitmodules file because some users would have huge .gitmodules files (>1MB)
+	c.submoduleCache, err = configParseSubModules(rd)
+	if err != nil {
+		return nil, err
+	}
+	return c.submoduleCache, nil
+}
+
+// GetSubModule gets the submodule by the entry name.
+// It returns "nil, nil" if the submodule does not exist, caller should always remember to check the "nil"
+func (c *Commit) GetSubModule(ctx context.Context, gitRepo *Repository, entryName string) (*SubModule, error) {
+	modules, err := c.GetSubModules(ctx, gitRepo)
+	if err != nil {
+		return nil, err
+	}
+
+	if modules != nil {
+		if module, has := modules.Get(entryName); has {
+			return module, nil
+		}
+	}
+	return nil, nil //nolint:nilnil // return nil to indicate that the submodule does not exist
+}

modules/git/commit_submodule_file.go

@@ -0,0 +1,69 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"path"
+	"strings"
+
+	giturl "gitea.dev/modules/git/url"
+	"gitea.dev/modules/util"
+)
+
+// CommitSubmoduleFile represents a file with submodule type.
+type CommitSubmoduleFile struct {
+	repoLink string
+	fullPath string
+	refURL   string
+	refID    string
+
+	parsed           bool
+	parsedTargetLink string
+}
+
+// NewCommitSubmoduleFile create a new submodule file
+func NewCommitSubmoduleFile(repoLink, fullPath, refURL, refID string) *CommitSubmoduleFile {
+	return &CommitSubmoduleFile{repoLink: repoLink, fullPath: fullPath, refURL: refURL, refID: refID}
+}
+
+// RefID returns the commit ID of the submodule, it returns empty string for nil receiver
+func (sf *CommitSubmoduleFile) RefID() string {
+	if sf == nil {
+		return ""
+	}
+	return sf.refID
+}
+
+func (sf *CommitSubmoduleFile) getWebLinkInTargetRepo(ctx context.Context, moreLinkPath string) *SubmoduleWebLink {
+	if sf == nil || sf.refURL == "" {
+		return nil
+	}
+	if strings.HasPrefix(sf.refURL, "../") {
+		targetLink := path.Join(sf.repoLink, sf.refURL)
+		return &SubmoduleWebLink{RepoWebLink: targetLink, CommitWebLink: targetLink + moreLinkPath}
+	}
+	if !sf.parsed {
+		sf.parsed = true
+		parsedURL, err := giturl.ParseRepositoryURL(ctx, sf.refURL)
+		if err != nil {
+			return nil
+		}
+		sf.parsedTargetLink = giturl.MakeRepositoryWebLink(parsedURL)
+	}
+	return &SubmoduleWebLink{RepoWebLink: sf.parsedTargetLink, CommitWebLink: sf.parsedTargetLink + moreLinkPath}
+}
+
+// SubmoduleWebLinkTree tries to make the submodule's tree link in its own repo, it also works on "nil" receiver
+// It returns nil if the submodule does not have a valid URL or is nil
+func (sf *CommitSubmoduleFile) SubmoduleWebLinkTree(ctx context.Context, optCommitID ...string) *SubmoduleWebLink {
+	return sf.getWebLinkInTargetRepo(ctx, "/tree/"+util.OptionalArg(optCommitID, sf.RefID()))
+}
+
+// SubmoduleWebLinkCompare tries to make the submodule's compare link in its own repo, it also works on "nil" receiver
+// It returns nil if the submodule does not have a valid URL or is nil
+func (sf *CommitSubmoduleFile) SubmoduleWebLinkCompare(ctx context.Context, commitID1, commitID2 string) *SubmoduleWebLink {
+	return sf.getWebLinkInTargetRepo(ctx, "/compare/"+commitID1+"..."+commitID2)
+}

modules/git/commit_submodule_file_test.go

@@ -0,0 +1,40 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestCommitSubmoduleLink(t *testing.T) {
+	assert.Nil(t, (*CommitSubmoduleFile)(nil).SubmoduleWebLinkTree(t.Context()))
+	assert.Nil(t, (*CommitSubmoduleFile)(nil).SubmoduleWebLinkCompare(t.Context(), "", ""))
+	assert.Nil(t, (&CommitSubmoduleFile{}).SubmoduleWebLinkTree(t.Context()))
+	assert.Nil(t, (&CommitSubmoduleFile{}).SubmoduleWebLinkCompare(t.Context(), "", ""))
+
+	t.Run("GitHubRepo", func(t *testing.T) {
+		sf := NewCommitSubmoduleFile("/any/repo-link", "full-path", "git@github.com:user/repo.git", "aaaa")
+		wl := sf.SubmoduleWebLinkTree(t.Context())
+		assert.Equal(t, "https://github.com/user/repo", wl.RepoWebLink)
+		assert.Equal(t, "https://github.com/user/repo/tree/aaaa", wl.CommitWebLink)
+
+		wl = sf.SubmoduleWebLinkCompare(t.Context(), "1111", "2222")
+		assert.Equal(t, "https://github.com/user/repo", wl.RepoWebLink)
+		assert.Equal(t, "https://github.com/user/repo/compare/1111...2222", wl.CommitWebLink)
+	})
+
+	t.Run("RelativePath", func(t *testing.T) {
+		sf := NewCommitSubmoduleFile("/subpath/any/repo-home-link", "full-path", "../../user/repo", "aaaa")
+		wl := sf.SubmoduleWebLinkTree(t.Context())
+		assert.Equal(t, "/subpath/user/repo", wl.RepoWebLink)
+		assert.Equal(t, "/subpath/user/repo/tree/aaaa", wl.CommitWebLink)
+
+		sf = NewCommitSubmoduleFile("/subpath/any/repo-home-link", "dir/submodule", "../../user/repo", "aaaa")
+		wl = sf.SubmoduleWebLinkCompare(t.Context(), "1111", "2222")
+		assert.Equal(t, "/subpath/user/repo", wl.RepoWebLink)
+		assert.Equal(t, "/subpath/user/repo/compare/1111...2222", wl.CommitWebLink)
+	})
+}

modules/git/commit_test.go

@@ -0,0 +1,204 @@
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestGetFullCommitID(t *testing.T) {
+	id, err := GetFullCommitID(t.Context(), mockRepository("repo1_bare"), "8006ff9a")
+	assert.NoError(t, err)
+	assert.Equal(t, "8006ff9adbf0cb94da7dad9e537e53817f9fa5c0", id)
+}
+
+func TestGetFullCommitIDError(t *testing.T) {
+	id, err := GetFullCommitID(t.Context(), mockRepository("repo1_bare"), "unknown")
+	assert.Empty(t, id)
+	if assert.Error(t, err) {
+		assert.EqualError(t, err, "object does not exist [id: unknown, rel_path: ]")
+	}
+}
+
+func TestCommitFromReader(t *testing.T) {
+	commitString := `tree f1a6cb52b2d16773290cefe49ad0684b50a4f930
+parent 37991dec2c8e592043f47155ce4808d4580f9123
+author silverwind <me@silverwind.io> 1563741793 +0200
+committer silverwind <me@silverwind.io> 1563741793 +0200
+gpgsig -----BEGIN PGP SIGNATURE-----
+` + " " + `
+ iQIzBAABCAAdFiEEWPb2jX6FS2mqyJRQLmK0HJOGlEMFAl00zmEACgkQLmK0HJOG
+ lEMDFBAAhQKKqLD1VICygJMEB8t1gBmNLgvziOLfpX4KPWdPtBk3v/QJ7OrfMrVK
+ xlC4ZZyx6yMm1Q7GzmuWykmZQJ9HMaHJ49KAbh5MMjjV/+OoQw9coIdo8nagRUld
+ vX8QHzNZ6Agx77xHuDJZgdHKpQK3TrMDsxzoYYMvlqoLJIDXE1Sp7KYNy12nhdRg
+ R6NXNmW8oMZuxglkmUwayMiPS+N4zNYqv0CXYzlEqCOgq9MJUcAMHt+KpiST+sm6
+ FWkJ9D+biNPyQ9QKf1AE4BdZia4lHfPYU/C/DEL/a5xQuuop/zMQZoGaIA4p2zGQ
+ /maqYxEIM/yRBQpT1jlODKPJrMEgx7SgY2hRU47YZ4fj6350fb6fNBtiiMAfJbjL
+ S3Gh85E9fm3hJaNSPKAaJFYL1Ya2svuWfgHj677C56UcmYis7fhiiy1aJuYdHnSm
+ sD53z/f0J+We4VZjY+pidvA9BGZPFVdR3wd3xGs8/oH6UWaLJAMGkLG6dDb3qDLm
+ 1LFZwsX8sdD32i1SiWanYQYSYMyFWr0awi4xdoMtYCL7uKBYtwtPyvq3cj4IrJlb
+ mfeFhT57UbE4qukTDIQ0Y0WM40UYRTakRaDY7ubhXgLgx09Cnp9XTVMsHgT6j9/i
+ 1pxsB104XLWjQHTjr1JtiaBQEwFh9r2OKTcpvaLcbNtYpo7CzOs=
+ =FRsO
+ -----END PGP SIGNATURE-----
+
+empty commit`
+
+	sha := &Sha1Hash{0xfe, 0xaf, 0x4b, 0xa6, 0xbc, 0x63, 0x5f, 0xec, 0x44, 0x2f, 0x46, 0xdd, 0xd4, 0x51, 0x24, 0x16, 0xec, 0x43, 0xc2, 0xc2}
+	gitRepo, err := OpenRepositoryLocal(t.Context(), filepath.Join(testReposDir, "repo1_bare"))
+	assert.NoError(t, err)
+	assert.NotNil(t, gitRepo)
+	defer gitRepo.Close()
+
+	commitFromReader, err := CommitFromReader(sha, strings.NewReader(commitString))
+	assert.NoError(t, err)
+	require.NotNil(t, commitFromReader)
+	assert.EqualValues(t, sha, commitFromReader.ID)
+	assert.Equal(t, `-----BEGIN PGP SIGNATURE-----
+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+=FRsO
+-----END PGP SIGNATURE-----`, commitFromReader.Signature.Signature)
+	assert.Equal(t, `tree f1a6cb52b2d16773290cefe49ad0684b50a4f930
+parent 37991dec2c8e592043f47155ce4808d4580f9123
+author silverwind <me@silverwind.io> 1563741793 +0200
+committer silverwind <me@silverwind.io> 1563741793 +0200
+
+empty commit`, commitFromReader.Signature.Payload)
+	assert.Equal(t, "silverwind <me@silverwind.io>", commitFromReader.Author.String())
+
+	commitFromReader2, err := CommitFromReader(sha, strings.NewReader(commitString+"\n\n"))
+	assert.NoError(t, err)
+	commitFromReader.CommitMessage.MessageRaw += "\n\n"
+	commitFromReader.Signature.Payload += "\n\n"
+	assert.Equal(t, commitFromReader, commitFromReader2)
+}
+
+func TestCommitWithEncodingFromReader(t *testing.T) {
+	commitString := `tree ca3fad42080dd1a6d291b75acdfc46e5b9b307e5
+parent 47b24e7ab977ed31c5a39989d570847d6d0052af
+author KN4CK3R <admin@oldschoolhack.me> 1711702962 +0100
+committer KN4CK3R <admin@oldschoolhack.me> 1711702962 +0100
+encoding ISO-8859-1
+gpgsig -----BEGIN PGP SIGNATURE-----
+<SPACE>
+ iQGzBAABCgAdFiEE9HRrbqvYxPT8PXbefPSEkrowAa8FAmYGg7IACgkQfPSEkrow
+ Aa9olwv+P0HhtCM6CRvlUmPaqswRsDPNR4i66xyXGiSxdI9V5oJL7HLiQIM7KrFR
+ gizKa2COiGtugv8fE+TKqXKaJx6uJUJEjaBd8E9Af9PrAzjWj+A84lU6/PgPS8hq
+ zOfZraLOEWRH4tZcS+u2yFLu3ez2Wqh1xW5LNy7xqEedMXEFD1HwSJ0+pjacNkzr
+ frp6Asyt7xRI6YmgFJZJoRsS3Ktr6rtKeRL2IErSQQyorOqj6gKrglhrhfG/114j
+ FKB1v4or0WZ1DE8iP2SJZ3n+/K1IuWAINh7MVdb7PndfBPEa+IL+ucNk5uzEE8Jd
+ G8smGxXUeFEt2cP1dj2W8EgAxuA9sTnH9dqI5aRqy5ifDjuya7Emm8sdOUvtGdmn
+ SONRzusmu5n3DgV956REL7x62h7JuqmBz/12HZkr0z0zgXkcZ04q08pSJATX5N1F
+ yN+tWxTsWg+zhDk96d5Esdo9JMjcFvPv0eioo30GAERaz1hoD7zCMT4jgUFTQwgz
+ jw4YcO5u
+ =r3UU
+ -----END PGP SIGNATURE-----
+
+ISO-8859-1`
+	commitString = strings.ReplaceAll(commitString, "<SPACE>", " ")
+	sha := &Sha1Hash{0xfe, 0xaf, 0x4b, 0xa6, 0xbc, 0x63, 0x5f, 0xec, 0x44, 0x2f, 0x46, 0xdd, 0xd4, 0x51, 0x24, 0x16, 0xec, 0x43, 0xc2, 0xc2}
+	gitRepo, err := OpenRepositoryLocal(t.Context(), filepath.Join(testReposDir, "repo1_bare"))
+	assert.NoError(t, err)
+	assert.NotNil(t, gitRepo)
+	defer gitRepo.Close()
+
+	commitFromReader, err := CommitFromReader(sha, strings.NewReader(commitString))
+	assert.NoError(t, err)
+	require.NotNil(t, commitFromReader)
+	assert.EqualValues(t, sha, commitFromReader.ID)
+	assert.Equal(t, `-----BEGIN PGP SIGNATURE-----
+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+=r3UU
+-----END PGP SIGNATURE-----`, commitFromReader.Signature.Signature)
+	assert.Equal(t, `tree ca3fad42080dd1a6d291b75acdfc46e5b9b307e5
+parent 47b24e7ab977ed31c5a39989d570847d6d0052af
+author KN4CK3R <admin@oldschoolhack.me> 1711702962 +0100
+committer KN4CK3R <admin@oldschoolhack.me> 1711702962 +0100
+encoding ISO-8859-1
+
+ISO-8859-1`, commitFromReader.Signature.Payload)
+	assert.Equal(t, "KN4CK3R <admin@oldschoolhack.me>", commitFromReader.Author.String())
+
+	commitFromReader2, err := CommitFromReader(sha, strings.NewReader(commitString+"\n\n"))
+	assert.NoError(t, err)
+	commitFromReader.CommitMessage.MessageRaw += "\n\n"
+	commitFromReader.Signature.Payload += "\n\n"
+	assert.Equal(t, commitFromReader, commitFromReader2)
+}
+
+func TestHasPreviousCommit(t *testing.T) {
+	bareRepo1Path := filepath.Join(testReposDir, "repo1_bare")
+
+	repo, err := OpenRepositoryLocal(t.Context(), bareRepo1Path)
+	assert.NoError(t, err)
+	defer repo.Close()
+
+	commit, err := repo.GetCommit(t.Context(), "8006ff9adbf0cb94da7dad9e537e53817f9fa5c0")
+	assert.NoError(t, err)
+
+	parentSHA := MustIDFromString("8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2")
+	notParentSHA := MustIDFromString("2839944139e0de9737a044f78b0e4b40d989a9e3")
+
+	haz, err := commit.HasPreviousCommit(t.Context(), repo, parentSHA)
+	assert.NoError(t, err)
+	assert.True(t, haz)
+
+	hazNot, err := commit.HasPreviousCommit(t.Context(), repo, notParentSHA)
+	assert.NoError(t, err)
+	assert.False(t, hazNot)
+
+	selfNot, err := commit.HasPreviousCommit(t.Context(), repo, commit.ID)
+	assert.NoError(t, err)
+	assert.False(t, selfNot)
+}
+
+func Test_GetCommitBranchStart(t *testing.T) {
+	bareRepo1Path := filepath.Join(testReposDir, "repo1_bare")
+	repo, err := OpenRepositoryLocal(t.Context(), bareRepo1Path)
+	assert.NoError(t, err)
+	defer repo.Close()
+	commit, err := repo.GetBranchCommit(t.Context(), "branch1")
+	assert.NoError(t, err)
+	assert.Equal(t, "2839944139e0de9737a044f78b0e4b40d989a9e3", commit.ID.String())
+
+	startCommitID, err := repo.GetCommitBranchStart(t.Context(), os.Environ(), "branch1", commit.ID.String())
+	assert.NoError(t, err)
+	assert.NotEmpty(t, startCommitID)
+	assert.Equal(t, "95bb4d39648ee7e325106df01a621c530863a653", startCommitID)
+}
+
+func TestIsStringLikelyCommitID(t *testing.T) {
+	assert.True(t, IsStringLikelyCommitID(nil, "abc", 3))
+	assert.False(t, IsStringLikelyCommitID(nil, "abc", 4))
+	assert.True(t, IsStringLikelyCommitID(nil, strings.Repeat("a", 64), 4))
+	assert.False(t, IsStringLikelyCommitID(nil, strings.Repeat("a", 65), 4))
+}

modules/git/compare.go

@@ -0,0 +1,70 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"strconv"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+)
+
+// DivergeObject represents commit count diverging commits
+type DivergeObject struct {
+	Ahead  int
+	Behind int
+}
+
+// GetDivergingCommits returns the number of commits a targetBranch is ahead or behind a baseBranch
+func GetDivergingCommits(ctx context.Context, repo RepositoryFacade, baseBranch, targetBranch string) (*DivergeObject, error) {
+	cmd := gitcmd.NewCommand("rev-list", "--count", "--left-right").
+		AddDynamicArguments(baseBranch + "..." + targetBranch).AddArguments("--")
+	stdout, _, err1 := cmd.WithRepo(repo).RunStdString(ctx)
+	if err1 != nil {
+		return nil, err1
+	}
+
+	left, right, found := strings.Cut(strings.Trim(stdout, "\n"), "\t")
+	if !found {
+		return nil, fmt.Errorf("git rev-list output is missing a tab: %q", stdout)
+	}
+
+	behind, err := strconv.Atoi(left)
+	if err != nil {
+		return nil, err
+	}
+	ahead, err := strconv.Atoi(right)
+	if err != nil {
+		return nil, err
+	}
+	return &DivergeObject{Ahead: ahead, Behind: behind}, nil
+}
+
+// GetCommitIDsBetweenReverse returns the last commit IDs between two commits in reverse order (from old to new) with limit.
+// If the result exceeds the limit, the old commits IDs will be ignored
+func GetCommitIDsBetweenReverse(ctx context.Context, repo RepositoryFacade, startRef, endRef, notRef string, limit int) ([]string, error) {
+	genCmd := func(reversions ...string) *gitcmd.Command {
+		cmd := gitcmd.NewCommand("rev-list", "--reverse").
+			AddArguments("-n").AddDynamicArguments(strconv.Itoa(limit)).
+			AddDynamicArguments(reversions...)
+		if notRef != "" { // --not should be kept as the last parameter of git command, otherwise the result will be wrong
+			cmd.AddOptionValues("--not", notRef)
+		}
+		return cmd
+	}
+	stdout, _, err := genCmd(startRef + ".." + endRef).WithRepo(repo).RunStdString(ctx)
+	if gitcmd.IsStderr(err, gitcmd.StderrNoMergeBase) {
+		// if the start and end are not related (no merge base), just get all commits pushed by "end ref"
+		// previously it would return the results of git rev-list before last so let's try that...
+		stdout, _, err = genCmd(endRef).WithRepo(repo).RunStdString(ctx)
+	}
+	if err != nil {
+		return nil, err
+	}
+
+	commitIDs := strings.Fields(strings.TrimSpace(stdout))
+	return commitIDs, nil
+}

modules/git/compare_test.go

@@ -0,0 +1,138 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"path/filepath"
+	"strings"
+	"testing"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/util"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestMergeBaseNoCommonHistory(t *testing.T) {
+	repoDir := filepath.Join(t.TempDir(), "repo.git")
+	require.NoError(t, gitcmd.NewCommand("init").AddDynamicArguments(repoDir).Run(t.Context()))
+	_, _, runErr := gitcmd.NewCommand("fast-import").WithDir(repoDir).WithStdinBytes([]byte(strings.TrimSpace(`
+commit refs/heads/branch1
+committer User <user@example.com> 1714310400 +0000
+data 12
+First commit
+M 100644 inline file1.txt
+data 12
+Hello from 1
+
+commit refs/heads/branch2
+committer User <user@example.com> 1714310400 +0000
+data 13
+Second commit
+M 100644 inline file2.txt
+data 12
+Hello from 2
+`))).RunStdString(t.Context())
+	require.NoError(t, runErr)
+	mergeBase, err := MergeBase(t.Context(), mockRepository(repoDir), "branch1", "branch2")
+	assert.Empty(t, mergeBase)
+	assert.ErrorIs(t, err, util.ErrNotExist)
+}
+
+func TestRepoGetDivergingCommits(t *testing.T) {
+	repo := mockRepository("repo1_bare")
+	do, err := GetDivergingCommits(t.Context(), repo, "master", "branch2")
+	assert.NoError(t, err)
+	assert.Equal(t, &DivergeObject{
+		Ahead:  1,
+		Behind: 5,
+	}, do)
+
+	do, err = GetDivergingCommits(t.Context(), repo, "master", "master")
+	assert.NoError(t, err)
+	assert.Equal(t, &DivergeObject{
+		Ahead:  0,
+		Behind: 0,
+	}, do)
+
+	do, err = GetDivergingCommits(t.Context(), repo, "master", "test")
+	assert.NoError(t, err)
+	assert.Equal(t, &DivergeObject{
+		Ahead:  0,
+		Behind: 2,
+	}, do)
+}
+
+func TestGetCommitIDsBetweenReverse(t *testing.T) {
+	repo := mockRepository("repo1_bare")
+
+	// tests raw commit IDs
+	commitIDs, err := GetCommitIDsBetweenReverse(t.Context(), repo,
+		"8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+		"",
+		100,
+	)
+	assert.NoError(t, err)
+	assert.Equal(t, []string{
+		"8006ff9adbf0cb94da7dad9e537e53817f9fa5c0",
+		"6fbd69e9823458e6c4a2fc5c0f6bc022b2f2acd1",
+		"37991dec2c8e592043f47155ce4808d4580f9123",
+		"feaf4ba6bc635fec442f46ddd4512416ec43c2c2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+	}, commitIDs)
+
+	commitIDs, err = GetCommitIDsBetweenReverse(t.Context(), repo,
+		"8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+		"6fbd69e9823458e6c4a2fc5c0f6bc022b2f2acd1",
+		100,
+	)
+	assert.NoError(t, err)
+	assert.Equal(t, []string{
+		"37991dec2c8e592043f47155ce4808d4580f9123",
+		"feaf4ba6bc635fec442f46ddd4512416ec43c2c2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+	}, commitIDs)
+
+	commitIDs, err = GetCommitIDsBetweenReverse(t.Context(), repo,
+		"8d92fc957a4d7cfd98bc375f0b7bb189a0d6c9f2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+		"",
+		3,
+	)
+	assert.NoError(t, err)
+	assert.Equal(t, []string{
+		"37991dec2c8e592043f47155ce4808d4580f9123",
+		"feaf4ba6bc635fec442f46ddd4512416ec43c2c2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+	}, commitIDs)
+
+	// test branch names instead of raw commit IDs.
+	commitIDs, err = GetCommitIDsBetweenReverse(t.Context(), repo,
+		"test",
+		"master",
+		"",
+		100,
+	)
+	assert.NoError(t, err)
+	assert.Equal(t, []string{
+		"feaf4ba6bc635fec442f46ddd4512416ec43c2c2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+	}, commitIDs)
+
+	// add notref to exclude test
+	commitIDs, err = GetCommitIDsBetweenReverse(t.Context(), repo,
+		"test",
+		"master",
+		"test",
+		100,
+	)
+	assert.NoError(t, err)
+	assert.Equal(t, []string{
+		"feaf4ba6bc635fec442f46ddd4512416ec43c2c2",
+		"ce064814f4a0d337b333e646ece456cd39fab612",
+	}, commitIDs)
+}

modules/git/config.go

@@ -0,0 +1,205 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"regexp"
+	"runtime"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/setting"
+)
+
+// syncGitConfig only modifies gitconfig, won't change global variables (otherwise there will be data-race problem)
+func syncGitConfig(ctx context.Context) (err error) {
+	if err = os.MkdirAll(gitcmd.HomeDir(), os.ModePerm); err != nil {
+		return fmt.Errorf("unable to prepare git home directory %s, err: %w", gitcmd.HomeDir(), err)
+	}
+
+	// Git requires setting user.name and user.email in order to commit changes - old comment: "if they're not set just add some defaults"
+	// TODO: need to confirm whether users really need to change these values manually. It seems that these values are dummy only and not really used.
+	// If these values are not really used, then they can be set (overwritten) directly without considering about existence.
+	for configKey, defaultValue := range map[string]string{
+		"user.name":  "Gitea",
+		"user.email": "gitea@fake.local",
+	} {
+		if err := configSetNonExist(ctx, configKey, defaultValue); err != nil {
+			return err
+		}
+	}
+
+	// Set git some configurations - these must be set to these values for gitea to work correctly
+	if err := configSet(ctx, "core.quotePath", "false"); err != nil {
+		return err
+	}
+
+	if err := configSet(ctx, "receive.advertisePushOptions", "true"); err != nil {
+		return err
+	}
+
+	if err := configSet(ctx, "core.commitGraph", "true"); err != nil {
+		return err
+	}
+
+	if err := configSet(ctx, "gc.writeCommitGraph", "true"); err != nil {
+		return err
+	}
+
+	if err := configSet(ctx, "fetch.writeCommitGraph", "true"); err != nil {
+		return err
+	}
+
+	if DefaultFeatures().SupportProcReceive {
+		// set support for AGit flow
+		if err := configAddNonExist(ctx, "receive.procReceiveRefs", "refs/for"); err != nil {
+			return err
+		}
+	} else {
+		if err := configUnsetAll(ctx, "receive.procReceiveRefs", "refs/for"); err != nil {
+			return err
+		}
+	}
+
+	// Due to CVE-2022-24765, git now denies access to git directories which are not owned by current user.
+	// However, some docker users and samba users find it difficult to configure their systems correctly,
+	// so that Gitea's git repositories are owned by the Gitea user.
+	// (Possibly Windows Service users - but ownership in this case should really be set correctly on the filesystem.)
+	// See issue: https://github.com/go-gitea/gitea/issues/19455
+	// As Gitea now always use its internal git config file, and access to the git repositories is managed through Gitea,
+	// it is now safe to set "safe.directory=*" for internal usage only.
+	// Although this setting is only supported by some new git versions, it is also tolerated by earlier versions
+	if err := configAddNonExist(ctx, "safe.directory", "*"); err != nil {
+		return err
+	}
+
+	if runtime.GOOS == "windows" {
+		if err := configSet(ctx, "core.longpaths", "true"); err != nil {
+			return err
+		}
+		if setting.Git.DisableCoreProtectNTFS {
+			err = configSet(ctx, "core.protectNTFS", "false")
+		} else {
+			err = configUnsetAll(ctx, "core.protectNTFS", "false")
+		}
+		if err != nil {
+			return err
+		}
+	}
+
+	if !setting.Git.DisablePartialClone {
+		if err = configSet(ctx, "uploadpack.allowfilter", "true"); err != nil {
+			return err
+		}
+		err = configSet(ctx, "uploadpack.allowAnySHA1InWant", "true")
+	} else {
+		if err = configUnsetAll(ctx, "uploadpack.allowfilter", "true"); err != nil {
+			return err
+		}
+		err = configUnsetAll(ctx, "uploadpack.allowAnySHA1InWant", "true")
+	}
+	if err != nil {
+		return err
+	}
+
+	// Apply user's git config options last so they take precedence over builtin defaults
+	for k, v := range setting.GitConfig.Options {
+		if err = configSet(ctx, strings.ToLower(k), v); err != nil {
+			return err
+		}
+	}
+
+	GlobalConfig = &GlobalConfigStruct{}
+	// HINT: GIT-DIFF-TREE-UI-CONFIG: Git's bug: git-diff-tree loads config with /* no "diff" UI options */ (since 20 years ago).
+	// https://github.com/git/git/blame/5d2e7709234afea1b6ddb25cd4f60d3d5fb3c200/builtin/diff-tree.c#L127
+	// Although document and manual say that "git-diff-tree" supports "diff.orderfile" option, but it is not actually supported.
+	// So we need to apply the diff.orderfile explicitly in our code.
+	GlobalConfig.DiffOrderFile, _ = configGet(ctx, "diff.orderfile")
+	return nil
+}
+
+func configGet(ctx context.Context, key string) (string, error) {
+	stdout, _, err := gitcmd.NewCommand("config", "--global", "--get").AddDynamicArguments(key).RunStdString(ctx)
+	if err != nil && !gitcmd.IsErrorExitCode(err, 1) {
+		return "", fmt.Errorf("failed to get git config %s, err: %w", key, err)
+	}
+	return strings.TrimRight(stdout, "\r\n"), nil
+}
+
+func configSet(ctx context.Context, key, value string) error {
+	stdout, _, err := gitcmd.NewCommand("config", "--global", "--get").
+		AddDynamicArguments(key).
+		RunStdString(ctx)
+	if err != nil && !gitcmd.IsErrorExitCode(err, 1) {
+		return fmt.Errorf("failed to get git config %s, err: %w", key, err)
+	}
+
+	currValue := strings.TrimSpace(stdout)
+	if currValue == value {
+		return nil
+	}
+
+	if _, _, err = gitcmd.NewCommand("config", "--global").
+		AddDynamicArguments(key, value).
+		RunStdString(ctx); err != nil {
+		return fmt.Errorf("failed to set git global config %s, err: %w", key, err)
+	}
+
+	return nil
+}
+
+func configSetNonExist(ctx context.Context, key, value string) error {
+	_, _, err := gitcmd.NewCommand("config", "--global", "--get").AddDynamicArguments(key).RunStdString(ctx)
+	if err == nil {
+		// already exist
+		return nil
+	}
+	if gitcmd.IsErrorExitCode(err, 1) {
+		// not exist, set new config
+		_, _, err = gitcmd.NewCommand("config", "--global").AddDynamicArguments(key, value).RunStdString(ctx)
+		if err != nil {
+			return fmt.Errorf("failed to set git global config %s, err: %w", key, err)
+		}
+		return nil
+	}
+
+	return fmt.Errorf("failed to get git config %s, err: %w", key, err)
+}
+
+func configAddNonExist(ctx context.Context, key, value string) error {
+	_, _, err := gitcmd.NewCommand("config", "--global", "--get").AddDynamicArguments(key, regexp.QuoteMeta(value)).RunStdString(ctx)
+	if err == nil {
+		// already exist
+		return nil
+	}
+	if gitcmd.IsErrorExitCode(err, 1) {
+		// not exist, add new config
+		_, _, err = gitcmd.NewCommand("config", "--global", "--add").AddDynamicArguments(key, value).RunStdString(ctx)
+		if err != nil {
+			return fmt.Errorf("failed to add git global config %s, err: %w", key, err)
+		}
+		return nil
+	}
+	return fmt.Errorf("failed to get git config %s, err: %w", key, err)
+}
+
+func configUnsetAll(ctx context.Context, key, value string) error {
+	_, _, err := gitcmd.NewCommand("config", "--global", "--get").AddDynamicArguments(key).RunStdString(ctx)
+	if err == nil {
+		// exist, need to remove
+		_, _, err = gitcmd.NewCommand("config", "--global", "--unset-all").AddDynamicArguments(key, regexp.QuoteMeta(value)).RunStdString(ctx)
+		if err != nil {
+			return fmt.Errorf("failed to unset git global config %s, err: %w", key, err)
+		}
+		return nil
+	}
+	if gitcmd.IsErrorExitCode(err, 1) {
+		// not exist
+		return nil
+	}
+	return fmt.Errorf("failed to get git config %s, err: %w", key, err)
+}

modules/git/config_managed.go

@@ -0,0 +1,30 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+
+	"gitea.dev/modules/git/gitcmd"
+)
+
+// ManagedConfigAdd add a git configuration key to a specific value for the given repository.
+func ManagedConfigAdd(ctx context.Context, repo RepositoryFacade, key, value string) error {
+	return LockConfigAndDo(ctx, repo, func(ctx context.Context) error {
+		_, _, err := gitcmd.NewCommand("config", "--add").
+			AddDynamicArguments(key, value).WithRepo(repo).RunStdString(ctx)
+		return err
+	})
+}
+
+// ManagedConfigSet updates a git configuration key to a specific value for the given repository.
+// If the key does not exist, it will be created.
+// If the key exists, it will be updated to the new value.
+func ManagedConfigSet(ctx context.Context, repo RepositoryFacade, key, value string) error {
+	return LockConfigAndDo(ctx, repo, func(ctx context.Context) error {
+		_, _, err := gitcmd.NewCommand("config").
+			AddDynamicArguments(key, value).WithRepo(repo).RunStdString(ctx)
+		return err
+	})
+}

modules/git/config_submodule.go

@@ -0,0 +1,75 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"fmt"
+	"io"
+	"strings"
+)
+
+// SubModule is a reference on git repository
+type SubModule struct {
+	Path   string
+	URL    string
+	Branch string // this field is newly added but not really used
+}
+
+// configParseSubModules this is not a complete parse for gitmodules file, it only
+// parses the url and path of submodules. At the moment it only parses well-formed gitmodules files.
+// In the future, there should be a complete implementation of https://git-scm.com/docs/git-config#_syntax
+func configParseSubModules(r io.Reader) (*ObjectCache[*SubModule], error) {
+	var subModule *SubModule
+	subModules := newObjectCache[*SubModule]()
+	scanner := bufio.NewScanner(r)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+
+		// Skip empty lines and comments
+		if line == "" || strings.HasPrefix(line, "#") || strings.HasPrefix(line, ";") {
+			continue
+		}
+
+		// Section header [section]
+		if strings.HasPrefix(line, "[") && strings.HasSuffix(line, "]") {
+			if subModule != nil {
+				subModules.Set(subModule.Path, subModule)
+			}
+			if strings.HasPrefix(line, "[submodule") {
+				subModule = &SubModule{}
+			} else {
+				subModule = nil
+			}
+			continue
+		}
+
+		if subModule == nil {
+			continue
+		}
+
+		parts := strings.SplitN(line, "=", 2)
+		if len(parts) != 2 {
+			continue
+		}
+		key := strings.TrimSpace(parts[0])
+		value := strings.TrimSpace(parts[1])
+		switch key {
+		case "path":
+			subModule.Path = value
+		case "url":
+			subModule.URL = value
+		case "branch":
+			subModule.Branch = value
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		return nil, fmt.Errorf("error reading file: %w", err)
+	}
+	if subModule != nil {
+		subModules.Set(subModule.Path, subModule)
+	}
+	return subModules, nil
+}

modules/git/config_submodule_test.go

@@ -0,0 +1,49 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestConfigSubmodule(t *testing.T) {
+	input := `
+[core]
+path = test
+
+[submodule "submodule1"]
+	path = path1
+	url =	https://gitea.io/foo/foo
+	#branch = b1
+
+[other1]
+branch = master
+
+[submodule "submodule2"]
+	path = path2
+	url =	https://gitea.io/bar/bar
+	branch = b2
+
+[other2]
+branch = main
+
+[submodule "submodule3"]
+	path = path3
+	url =	https://gitea.io/xxx/xxx
+`
+
+	subModules, err := configParseSubModules(strings.NewReader(input))
+	assert.NoError(t, err)
+	assert.Len(t, subModules.cache, 3)
+
+	sm1, _ := subModules.Get("path1")
+	assert.Equal(t, &SubModule{Path: "path1", URL: "https://gitea.io/foo/foo", Branch: ""}, sm1)
+	sm2, _ := subModules.Get("path2")
+	assert.Equal(t, &SubModule{Path: "path2", URL: "https://gitea.io/bar/bar", Branch: "b2"}, sm2)
+	sm3, _ := subModules.Get("path3")
+	assert.Equal(t, &SubModule{Path: "path3", URL: "https://gitea.io/xxx/xxx", Branch: ""}, sm3)
+}

modules/git/config_test.go

@@ -0,0 +1,73 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"os"
+	"strings"
+	"testing"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func gitConfigContains(sub string) bool {
+	if b, err := os.ReadFile(gitcmd.HomeDir() + "/.gitconfig"); err == nil {
+		return strings.Contains(string(b), sub)
+	}
+	return false
+}
+
+func TestGitConfig(t *testing.T) {
+	ctx := t.Context()
+	assert.False(t, gitConfigContains("key-a"))
+
+	assert.NoError(t, configSetNonExist(ctx, "test.key-a", "val-a"))
+	assert.True(t, gitConfigContains("key-a = val-a"))
+
+	assert.NoError(t, configSetNonExist(ctx, "test.key-a", "val-a-changed"))
+	assert.False(t, gitConfigContains("key-a = val-a-changed"))
+
+	assert.NoError(t, configSet(ctx, "test.key-a", "val-a-changed"))
+	assert.True(t, gitConfigContains("key-a = val-a-changed"))
+
+	assert.NoError(t, configAddNonExist(ctx, "test.key-b", "val-b"))
+	assert.True(t, gitConfigContains("key-b = val-b"))
+
+	assert.NoError(t, configAddNonExist(ctx, "test.key-b", "val-2b"))
+	assert.True(t, gitConfigContains("key-b = val-b"))
+	assert.True(t, gitConfigContains("key-b = val-2b"))
+
+	assert.NoError(t, configUnsetAll(ctx, "test.key-b", "val-b"))
+	assert.False(t, gitConfigContains("key-b = val-b"))
+	assert.True(t, gitConfigContains("key-b = val-2b"))
+
+	assert.NoError(t, configUnsetAll(ctx, "test.key-b", "val-2b"))
+	assert.False(t, gitConfigContains("key-b = val-2b"))
+
+	assert.NoError(t, configSet(ctx, "test.key-x", "*"))
+	assert.True(t, gitConfigContains("key-x = *"))
+	assert.NoError(t, configSetNonExist(ctx, "test.key-x", "*"))
+	assert.NoError(t, configUnsetAll(ctx, "test.key-x", "*"))
+	assert.False(t, gitConfigContains("key-x = *"))
+}
+
+func TestSyncGitConfig(t *testing.T) {
+	defer test.MockVariableValue(&setting.GitConfig)()
+
+	assert.Empty(t, setting.GitConfig.Options)
+	assert.NoError(t, syncGitConfig(t.Context()))
+	assert.True(t, gitConfigContains("commitGraph = true")) // builtin default config
+
+	setting.GitConfig.Options["sync-test.cfg-key-a"] = "CfgValA"
+	setting.GitConfig.Options["core.commitgraph"] = "false"
+	assert.NoError(t, syncGitConfig(t.Context()))
+	assert.True(t, gitConfigContains("[sync-test]"))
+	assert.True(t, gitConfigContains("cfg-key-a = CfgValA"))
+	assert.False(t, gitConfigContains("commitGraph"))        // builtin default config can be overridden
+	assert.True(t, gitConfigContains("commitgraph = false")) // git config key is case-insensitive
+}

modules/git/diff.go

@@ -0,0 +1,348 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"context"
+	"fmt"
+	"io"
+	"regexp"
+	"strconv"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/util"
+)
+
+// RawDiffType output format: diff or patch
+type RawDiffType string
+
+const (
+	RawDiffNormal RawDiffType = "diff"
+	RawDiffPatch  RawDiffType = "patch"
+)
+
+// GetRawDiff dumps diff results of repository in given commit ID to io.Writer.
+func GetRawDiff(ctx context.Context, repo *Repository, commitID string, diffType RawDiffType, writer io.Writer) (retErr error) {
+	cmd, err := getRepoRawDiffForFileCmd(ctx, repo, "", commitID, diffType, "")
+	if err != nil {
+		return fmt.Errorf("getRepoRawDiffForFileCmd: %w", err)
+	}
+	return cmd.WithStdoutCopy(writer).RunWithStderr(ctx)
+}
+
+// GetFileDiffCutAroundLine cuts the old or new part of the diff of a file around a specific line number
+func GetFileDiffCutAroundLine(
+	ctx context.Context, repo *Repository, startCommit, endCommit, treePath string,
+	line int64, old bool, numbersOfLine int,
+) (ret string, retErr error) {
+	cmd, err := getRepoRawDiffForFileCmd(ctx, repo, startCommit, endCommit, RawDiffNormal, treePath)
+	if err != nil {
+		return "", fmt.Errorf("getRepoRawDiffForFileCmd: %w", err)
+	}
+	stdoutReader, stdoutClose := cmd.MakeStdoutPipe()
+	defer stdoutClose()
+	cmd.WithPipelineFunc(func(ctx gitcmd.Context) error {
+		ret, err = CutDiffAroundLine(stdoutReader, line, old, numbersOfLine)
+		return err
+	})
+	return ret, cmd.RunWithStderr(ctx)
+}
+
+// getRepoRawDiffForFile returns an io.Reader for the diff results of file in given commit ID
+// and a "finish" function to wait for the git command and clean up resources after reading is done.
+func getRepoRawDiffForFileCmd(ctx context.Context, repo *Repository, startCommit, endCommit string, diffType RawDiffType, file string) (*gitcmd.Command, error) {
+	commit, err := repo.GetCommit(ctx, endCommit)
+	if err != nil {
+		return nil, err
+	}
+	var files []string
+	if len(file) > 0 {
+		files = append(files, file)
+	}
+
+	cmd := gitcmd.NewCommand().WithRepo(repo)
+	switch diffType {
+	case RawDiffNormal:
+		if len(startCommit) != 0 {
+			cmd.AddArguments("diff").
+				AddOptionFormat("--find-renames=%s", setting.Git.DiffRenameSimilarityThreshold).
+				AddDynamicArguments(startCommit, endCommit).AddDashesAndList(files...)
+		} else if commit.ParentCount() == 0 {
+			cmd.AddArguments("show").AddDynamicArguments(endCommit).AddDashesAndList(files...)
+		} else {
+			c, err := commit.Parent(ctx, repo, 0)
+			if err != nil {
+				return nil, err
+			}
+			cmd.AddArguments("diff").
+				AddOptionFormat("--find-renames=%s", setting.Git.DiffRenameSimilarityThreshold).
+				AddDynamicArguments(c.ID.String(), endCommit).AddDashesAndList(files...)
+		}
+	case RawDiffPatch:
+		if len(startCommit) != 0 {
+			query := fmt.Sprintf("%s...%s", endCommit, startCommit)
+			cmd.AddArguments("format-patch", "--no-signature", "--stdout", "--root").AddDynamicArguments(query).AddDashesAndList(files...)
+		} else if commit.ParentCount() == 0 {
+			cmd.AddArguments("format-patch", "--no-signature", "--stdout", "--root").AddDynamicArguments(endCommit).AddDashesAndList(files...)
+		} else {
+			c, err := commit.Parent(ctx, repo, 0)
+			if err != nil {
+				return nil, err
+			}
+			query := fmt.Sprintf("%s...%s", endCommit, c.ID.String())
+			cmd.AddArguments("format-patch", "--no-signature", "--stdout").AddDynamicArguments(query).AddDashesAndList(files...)
+		}
+	default:
+		return nil, util.NewInvalidArgumentErrorf("invalid diff type: %s", diffType)
+	}
+	return cmd, nil
+}
+
+// ParseDiffHunkString parse the diff hunk content and return
+func ParseDiffHunkString(diffHunk string) (leftLine, leftHunk, rightLine, rightHunk int) {
+	ss := strings.Split(diffHunk, "@@")
+	ranges := strings.Split(ss[1][1:], " ")
+	leftRange := strings.Split(ranges[0], ",")
+	leftLine, _ = strconv.Atoi(leftRange[0][1:])
+	if len(leftRange) > 1 {
+		leftHunk, _ = strconv.Atoi(leftRange[1])
+	}
+	if len(ranges) > 1 {
+		rightRange := strings.Split(ranges[1], ",")
+		rightLine, _ = strconv.Atoi(rightRange[0])
+		if len(rightRange) > 1 {
+			rightHunk, _ = strconv.Atoi(rightRange[1])
+		}
+	} else {
+		log.Debug("Parse line number failed: %v", diffHunk)
+		rightLine = leftLine
+		rightHunk = leftHunk
+	}
+	if rightLine == 0 {
+		// FIXME: GIT-DIFF-CUT-BUG search this tag to see details
+		// this is only a hacky patch, the rightLine&rightHunk might still be incorrect in some cases.
+		rightLine++
+	}
+	return leftLine, leftHunk, rightLine, rightHunk
+}
+
+// Example: @@ -1,8 +1,9 @@ => [..., 1, 8, 1, 9]
+var hunkRegex = regexp.MustCompile(`^@@ -(?P<beginOld>[0-9]+)(,(?P<endOld>[0-9]+))? \+(?P<beginNew>[0-9]+)(,(?P<endNew>[0-9]+))? @@`)
+
+const cmdDiffHead = "diff --git "
+
+func isHeader(lof string, inHunk bool) bool {
+	return strings.HasPrefix(lof, cmdDiffHead) || (!inHunk && (strings.HasPrefix(lof, "---") || strings.HasPrefix(lof, "+++")))
+}
+
+func NewGitDiffScanner(r io.Reader) *bufio.Scanner {
+	// TODO: GIT-DIFF-PARSE-LONG-LINE: ideally it shouldn't use bufio.Scanner which has a limit.
+	// It will cause errors if a line is very long.
+	scanner := bufio.NewScanner(r)
+	scanner.Buffer(nil, max(512*1024, int(setting.UI.MaxDisplayFileSize/16)))
+	return scanner
+}
+
+// CutDiffAroundLine cuts a diff of a file in way that only the given line + numberOfLine above it will be shown
+// it also recalculates hunks and adds the appropriate headers to the new diff.
+// Warning: Only one-file diffs are allowed.
+func CutDiffAroundLine(originalDiff io.Reader, line int64, old bool, numbersOfLine int) (string, error) {
+	if line == 0 || numbersOfLine == 0 {
+		// no line or num of lines => no diff
+		return "", nil
+	}
+
+	scanner := NewGitDiffScanner(originalDiff)
+	hunk := make([]string, 0)
+
+	// begin is the start of the hunk containing searched line
+	// end is the end of the hunk ...
+	// currentLine is the line number on the side of the searched line (differentiated by old)
+	// otherLine is the line number on the opposite side of the searched line (differentiated by old)
+	var begin, end, currentLine, otherLine int64
+	var headerLines int
+
+	inHunk := false
+
+	for scanner.Scan() {
+		lof := scanner.Text()
+		// Add header to enable parsing
+
+		if isHeader(lof, inHunk) {
+			if strings.HasPrefix(lof, cmdDiffHead) {
+				inHunk = false
+			}
+			hunk = append(hunk, lof)
+			headerLines++
+		}
+		if currentLine > line {
+			break
+		}
+		// Detect "hunk" with contains commented lof
+		if strings.HasPrefix(lof, "@@") {
+			inHunk = true
+			// Already got our hunk. End of hunk detected!
+			if len(hunk) > headerLines {
+				break
+			}
+			// A map with named groups of our regex to recognize them later more easily
+			submatches := hunkRegex.FindStringSubmatch(lof)
+			groups := make(map[string]string)
+			for i, name := range hunkRegex.SubexpNames() {
+				if i != 0 && name != "" {
+					groups[name] = submatches[i]
+				}
+			}
+			if old {
+				begin, _ = strconv.ParseInt(groups["beginOld"], 10, 64)
+				end, _ = strconv.ParseInt(groups["endOld"], 10, 64)
+				// init otherLine with begin of opposite side
+				otherLine, _ = strconv.ParseInt(groups["beginNew"], 10, 64)
+			} else {
+				begin, _ = strconv.ParseInt(groups["beginNew"], 10, 64)
+				if groups["endNew"] != "" {
+					end, _ = strconv.ParseInt(groups["endNew"], 10, 64)
+				} else {
+					end = 0
+				}
+				// init otherLine with begin of opposite side
+				otherLine, _ = strconv.ParseInt(groups["beginOld"], 10, 64)
+			}
+			end += begin // end is for real only the number of lines in hunk
+			// lof is between begin and end
+			if begin <= line && end >= line {
+				hunk = append(hunk, lof)
+				currentLine = begin
+				continue
+			}
+		} else if len(hunk) > headerLines {
+			hunk = append(hunk, lof)
+			// Count lines in context
+			switch lof[0] {
+			case '+':
+				if !old {
+					currentLine++
+				} else {
+					otherLine++
+				}
+			case '-':
+				if old {
+					currentLine++
+				} else {
+					otherLine++
+				}
+			case '\\':
+				// FIXME: handle `\ No newline at end of file`
+			default:
+				currentLine++
+				otherLine++
+			}
+		}
+	}
+	if err := scanner.Err(); err != nil {
+		return "", fmt.Errorf("CutDiffAroundLine: scan: %w", err)
+	}
+
+	// No hunk found
+	if currentLine == 0 {
+		return "", nil
+	}
+	// headerLines + hunkLine (1) = totalNonCodeLines
+	if len(hunk)-headerLines-1 <= numbersOfLine {
+		// No need to cut the hunk => return existing hunk
+		return strings.Join(hunk, "\n"), nil
+	}
+	var oldBegin, oldNumOfLines, newBegin, newNumOfLines int64
+	if old {
+		oldBegin = currentLine
+		newBegin = otherLine
+	} else {
+		oldBegin = otherLine
+		newBegin = currentLine
+	}
+	// headers + hunk header
+	newHunk := make([]string, headerLines)
+	// transfer existing headers
+	copy(newHunk, hunk[:headerLines])
+	// transfer last n lines
+	newHunk = append(newHunk, hunk[len(hunk)-numbersOfLine-1:]...)
+	// calculate newBegin, ... by counting lines
+	for i := len(hunk) - 1; i >= len(hunk)-numbersOfLine; i-- {
+		switch hunk[i][0] {
+		case '+':
+			newBegin--
+			newNumOfLines++
+		case '-':
+			oldBegin--
+			oldNumOfLines++
+		default:
+			oldBegin--
+			newBegin--
+			newNumOfLines++
+			oldNumOfLines++
+		}
+	}
+
+	// "git diff" outputs "@@ -1 +1,3 @@" for "OLD" => "A\nB\nC"
+	// FIXME: GIT-DIFF-CUT-BUG But there is a bug in CutDiffAroundLine, then the "Patch" stored in the comment model becomes "@@ -1,1 +0,4 @@"
+	// It may generate incorrect results for difference cases, for example: delete 2 line add 1 line, delete 2 line add 2 line etc, need to double check.
+	// For example: "L1\nL2" => "A\nB", then the patch shows "L2" as line 1 on the left (deleted part)
+
+	// construct the new hunk header
+	newHunk[headerLines] = fmt.Sprintf("@@ -%d,%d +%d,%d @@",
+		oldBegin, oldNumOfLines, newBegin, newNumOfLines)
+	return strings.Join(newHunk, "\n"), nil
+}
+
+// GetAffectedFiles returns the affected files between two commits
+func GetAffectedFiles(ctx context.Context, repo *Repository, branchName, oldCommitID, newCommitID string, env []string) ([]string, error) {
+	if oldCommitID == emptySha1ObjectID.String() || oldCommitID == emptySha256ObjectID.String() {
+		startCommitID, err := repo.GetCommitBranchStart(ctx, env, branchName, newCommitID)
+		if err != nil {
+			return nil, err
+		}
+		if startCommitID == "" {
+			return nil, fmt.Errorf("cannot find the start commit of %s", newCommitID)
+		}
+		oldCommitID = startCommitID
+	}
+
+	affectedFiles := make([]string, 0, 32)
+
+	// Run `git diff --name-only` to get the names of the changed files
+	cmd := gitcmd.NewCommand("diff", "--name-only").AddDynamicArguments(oldCommitID, newCommitID)
+	stdoutReader, stdoutReaderClose := cmd.MakeStdoutPipe()
+	defer stdoutReaderClose()
+	err := cmd.WithEnv(env).WithRepo(repo).
+		WithPipelineFunc(func(ctx gitcmd.Context) error {
+			// Now scan the output from the command
+			scanner := bufio.NewScanner(stdoutReader)
+			for scanner.Scan() {
+				path := strings.TrimSpace(scanner.Text())
+				if len(path) == 0 {
+					continue
+				}
+				affectedFiles = append(affectedFiles, path)
+			}
+			return scanner.Err()
+		}).
+		Run(ctx)
+	if err != nil {
+		log.Error("Unable to get affected files for commits from %s to %s in %s: %v", oldCommitID, newCommitID, repo.LogString(), err)
+	}
+
+	return affectedFiles, err
+}
+
+// GetReverseRawDiff dumps the reverse diff results of repository in given commit ID to io.Writer.
+func GetReverseRawDiff(ctx context.Context, repo RepositoryFacade, commitID string, writer io.Writer) error {
+	return gitcmd.NewCommand("show", "--pretty=format:revert %H%n", "-R").
+		AddDynamicArguments(commitID).
+		WithStdoutCopy(writer).
+		WithRepo(repo).RunWithStderr(ctx)
+}

modules/git/diff_test.go

@@ -0,0 +1,184 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+const exampleDiff = `diff --git a/README.md b/README.md
+--- a/README.md
++++ b/README.md
+@@ -1,3 +1,6 @@
+ # gitea-github-migrator
++
++ Build Status
+- Latest Release
+ Docker Pulls
++ cut off
++ cut off`
+
+const breakingDiff = `diff --git a/aaa.sql b/aaa.sql
+index d8e4c92..19dc8ad 100644
+--- a/aaa.sql
++++ b/aaa.sql
+@@ -1,9 +1,10 @@
+ --some comment
+--- some comment 5
++--some coment 2
++-- some comment 3
+ create or replace procedure test(p1 varchar2)
+ is
+ begin
+---new comment
+ dbms_output.put_line(p1);
++--some other comment
+ end;
+ /
+`
+
+var issue17875Diff = `diff --git a/Geschäftsordnung.md b/Geschäftsordnung.md
+index d46c152..a7d2d55 100644
+--- a/Geschäftsordnung.md
++++ b/Geschäftsordnung.md
+@@ -1,5 +1,5 @@
+ ---
+-date: "23.01.2021"
++date: "30.11.2021"
+ ...
+ ` + `
+ # Geschäftsordnung
+@@ -16,4 +16,22 @@ Diese Geschäftsordnung regelt alle Prozesse des Vereins, solange diese nicht du
+ ` + `
+ ## § 3 Datenschutzverantwortlichkeit
+ ` + `
+-1. Der Verein bestellt eine datenschutzverantwortliche Person mit den Aufgaben nach Artikel 39 DSGVO.
+\ No newline at end of file
++1. Der Verein bestellt eine datenschutzverantwortliche Person mit den Aufgaben nach Artikel 39 DSGVO.
++
++## §4 Umgang mit der SARS-Cov-2-Pandemie
++
++1. Der Vorstand hat die Befugnis, in Rücksprache mit den Vereinsmitgliedern, verschiedene Hygienemaßnahmen für Präsenzveranstaltungen zu beschließen.
++
++2. Die Einführung, Änderung und Abschaffung dieser Maßnahmen sind nur zum Zweck der Eindämmung der SARS-Cov-2-Pandemie zulässig.
++
++3. Die Einführung, Änderung und Abschaffung von Maßnahmen nach Abs. 2 bedarf einer wissenschaftlichen Grundlage.
++
++4. Die Maßnahmen nach Abs. 2 setzen sich aus den folgenden Bausteinen inklusive einer ihrer Ausprägungen zusammen.
++
++	1. Maskenpflicht: Keine; Maskenpflicht, außer am Platz, oder wo Abstände nicht eingehalten werden können; Maskenpflicht, wenn Abstände nicht eingehalten werden können;  Maskenpflicht
++
++	2. Geimpft-, Genesen- oder Testnachweis: Kein Nachweis notwendig; Nachweis, dass Person geimpft, genesen oder tagesaktuell getestet ist (3G); Nachweis, dass Person geimpft oder genesen ist (2G); Nachweis, dass Person geimpft bzw. genesen und tagesaktuell getestet ist (2G+)
++
++	3. Online-Veranstaltung: Keine, parallele Online-Veranstaltung, ausschließlich Online-Veranstaltung
++
++5. Bei Präsenzveranstungen gelten außerdem die Hygienevorschriften des Veranstaltungsorts. Bei Regelkollision greift die restriktivere Regel.
+\ No newline at end of file`
+
+func TestCutDiffAroundLineIssue17875(t *testing.T) {
+	result, err := CutDiffAroundLine(strings.NewReader(issue17875Diff), 23, false, 3)
+	assert.NoError(t, err)
+	expected := `diff --git a/Geschäftsordnung.md b/Geschäftsordnung.md
+--- a/Geschäftsordnung.md
++++ b/Geschäftsordnung.md
+@@ -20,0 +21,3 @@
++## §4 Umgang mit der SARS-Cov-2-Pandemie
++
++1. Der Vorstand hat die Befugnis, in Rücksprache mit den Vereinsmitgliedern, verschiedene Hygienemaßnahmen für Präsenzveranstaltungen zu beschließen.`
+	assert.Equal(t, expected, result)
+}
+
+func TestCutDiffAroundLine(t *testing.T) {
+	result, err := CutDiffAroundLine(strings.NewReader(exampleDiff), 4, false, 3)
+	assert.NoError(t, err)
+	resultByLine := strings.Split(result, "\n")
+	assert.Len(t, resultByLine, 7)
+	// Check if headers got transferred
+	assert.Equal(t, "diff --git a/README.md b/README.md", resultByLine[0])
+	assert.Equal(t, "--- a/README.md", resultByLine[1])
+	assert.Equal(t, "+++ b/README.md", resultByLine[2])
+	// Check if hunk header is calculated correctly
+	assert.Equal(t, "@@ -2,2 +3,2 @@", resultByLine[3])
+	// Check if line got transferred
+	assert.Equal(t, "+ Build Status", resultByLine[4])
+
+	// Must be same result as before since old line 3 == new line 5
+	newResult, err := CutDiffAroundLine(strings.NewReader(exampleDiff), 3, true, 3)
+	assert.NoError(t, err)
+	assert.Equal(t, result, newResult, "Must be same result as before since old line 3 == new line 5")
+
+	newResult, err = CutDiffAroundLine(strings.NewReader(exampleDiff), 6, false, 300)
+	assert.NoError(t, err)
+	assert.Equal(t, exampleDiff, newResult)
+
+	emptyResult, err := CutDiffAroundLine(strings.NewReader(exampleDiff), 6, false, 0)
+	assert.NoError(t, err)
+	assert.Empty(t, emptyResult)
+
+	// Line is out of scope
+	emptyResult, err = CutDiffAroundLine(strings.NewReader(exampleDiff), 434, false, 0)
+	assert.NoError(t, err)
+	assert.Empty(t, emptyResult)
+
+	// Handle minus diffs properly
+	minusDiff, err := CutDiffAroundLine(strings.NewReader(breakingDiff), 2, false, 4)
+	assert.NoError(t, err)
+
+	expected := `diff --git a/aaa.sql b/aaa.sql
+--- a/aaa.sql
++++ b/aaa.sql
+@@ -1,9 +1,10 @@
+ --some comment
+--- some comment 5
++--some coment 2`
+	assert.Equal(t, expected, minusDiff)
+
+	// Handle minus diffs properly
+	minusDiff, err = CutDiffAroundLine(strings.NewReader(breakingDiff), 3, false, 4)
+	assert.NoError(t, err)
+
+	expected = `diff --git a/aaa.sql b/aaa.sql
+--- a/aaa.sql
++++ b/aaa.sql
+@@ -1,9 +1,10 @@
+ --some comment
+--- some comment 5
++--some coment 2
++-- some comment 3`
+
+	assert.Equal(t, expected, minusDiff)
+}
+
+func BenchmarkCutDiffAroundLine(b *testing.B) {
+	for b.Loop() {
+		CutDiffAroundLine(strings.NewReader(exampleDiff), 3, true, 3)
+	}
+}
+
+func ExampleCutDiffAroundLine() {
+	const diff = `diff --git a/README.md b/README.md
+--- a/README.md
++++ b/README.md
+@@ -1,3 +1,6 @@
+ # gitea-github-migrator
++
++ Build Status
+- Latest Release
+ Docker Pulls
++ cut off
++ cut off`
+	result, _ := CutDiffAroundLine(strings.NewReader(diff), 4, false, 3)
+	println(result)
+}
+
+func TestParseDiffHunkString(t *testing.T) {
+	leftLine, leftHunk, rightLine, rightHunk := ParseDiffHunkString("@@ -19,3 +19,5 @@ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER")
+	assert.Equal(t, 19, leftLine)
+	assert.Equal(t, 3, leftHunk)
+	assert.Equal(t, 19, rightLine)
+	assert.Equal(t, 5, rightHunk)
+}

modules/git/error.go

@@ -0,0 +1,126 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"fmt"
+	"strings"
+
+	"gitea.dev/modules/util"
+)
+
+// ErrNotExist commit not exist error
+type ErrNotExist struct {
+	ID      string
+	RelPath string
+}
+
+// IsErrNotExist if some error is ErrNotExist
+func IsErrNotExist(err error) bool {
+	_, ok := err.(ErrNotExist)
+	return ok
+}
+
+func (err ErrNotExist) Error() string {
+	return fmt.Sprintf("object does not exist [id: %s, rel_path: %s]", err.ID, err.RelPath)
+}
+
+func (err ErrNotExist) Unwrap() error {
+	return util.ErrNotExist
+}
+
+// ErrBranchNotExist represents a "BranchNotExist" kind of error.
+type ErrBranchNotExist struct {
+	Name string
+}
+
+// IsErrBranchNotExist checks if an error is a ErrBranchNotExist.
+func IsErrBranchNotExist(err error) bool {
+	_, ok := err.(ErrBranchNotExist)
+	return ok
+}
+
+func (err ErrBranchNotExist) Error() string {
+	return fmt.Sprintf("branch does not exist [name: %s]", err.Name)
+}
+
+func (err ErrBranchNotExist) Unwrap() error {
+	return util.ErrNotExist
+}
+
+// ErrPushOutOfDate represents an error if merging fails due to the base branch being updated
+type ErrPushOutOfDate struct {
+	StdOut string
+	StdErr string
+	Err    error
+}
+
+// IsErrPushOutOfDate checks if an error is a ErrPushOutOfDate.
+func IsErrPushOutOfDate(err error) bool {
+	_, ok := err.(*ErrPushOutOfDate)
+	return ok
+}
+
+func (err *ErrPushOutOfDate) Error() string {
+	return fmt.Sprintf("PushOutOfDate Error: %v: %s\n%s", err.Err, err.StdErr, err.StdOut)
+}
+
+// Unwrap unwraps the underlying error
+func (err *ErrPushOutOfDate) Unwrap() error {
+	return fmt.Errorf("%w - %s", err.Err, err.StdErr)
+}
+
+// ErrPushRejected represents an error if merging fails due to rejection from a hook
+type ErrPushRejected struct {
+	Message string
+	StdOut  string
+	StdErr  string
+	Err     error
+}
+
+// IsErrPushRejected checks if an error is a ErrPushRejected.
+func IsErrPushRejected(err error) bool {
+	_, ok := err.(*ErrPushRejected)
+	return ok
+}
+
+func (err *ErrPushRejected) Error() string {
+	return fmt.Sprintf("PushRejected Error: %v: %s\n%s", err.Err, err.StdErr, err.StdOut)
+}
+
+// Unwrap unwraps the underlying error
+func (err *ErrPushRejected) Unwrap() error {
+	return fmt.Errorf("%w - %s", err.Err, err.StdErr)
+}
+
+// GenerateMessage generates the remote message from the stderr
+func (err *ErrPushRejected) GenerateMessage() {
+	// The stderr is like this:
+	//
+	// > remote: error: push is rejected .....
+	// > To /work/gitea/tests/integration/gitea-integration-sqlite/gitea-repositories/user2/repo1.git
+	// >  ! [remote rejected] 44e67c77559211d21b630b902cdcc6ab9d4a4f51 -> develop (pre-receive hook declined)
+	// > error: failed to push some refs to '/work/gitea/tests/integration/gitea-integration-sqlite/gitea-repositories/user2/repo1.git'
+	//
+	// The local message contains sensitive information, so we only need the remote message
+	const prefixRemote = "remote: "
+	const prefixError = "error: "
+	pos := strings.Index(err.StdErr, prefixRemote)
+	if pos < 0 {
+		err.Message = "push is rejected"
+		return
+	}
+
+	messageBuilder := &strings.Builder{}
+	lines := strings.SplitSeq(err.StdErr, "\n")
+	for line := range lines {
+		line, ok := strings.CutPrefix(line, prefixRemote)
+		if !ok {
+			continue
+		}
+		line = strings.TrimPrefix(line, prefixError)
+		messageBuilder.WriteString(strings.TrimSpace(line) + "\n")
+	}
+	err.Message = strings.TrimSpace(messageBuilder.String())
+}

modules/git/fastimport.go

@@ -0,0 +1,67 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"os"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/util"
+)
+
+type FastImportInit struct {
+	Bare         bool
+	ObjectFormat string
+}
+
+type FastImportFile struct {
+	Mode    EntryMode
+	Path    string
+	Content string
+}
+
+type FastImportCommit struct {
+	Ref     string
+	Message string
+	Files   []FastImportFile
+}
+
+// ForceFastImportWithInit is for mainly for testing purpose
+func ForceFastImportWithInit(ctx context.Context, repoLocalPath string, commits []FastImportCommit, initOpts ...FastImportInit) (RepositoryFacade, error) {
+	repo := gitrepo.RepositoryUnmanaged(repoLocalPath)
+	initOpt := util.OptionalArg(initOpts, FastImportInit{Bare: true})
+	dirEntries, err := os.ReadDir(repoLocalPath)
+	if os.IsNotExist(err) || (err == nil && len(dirEntries) == 0) {
+		_ = os.MkdirAll(repoLocalPath, 0o755)
+		err := InitRepositoryLocal(ctx, repoLocalPath, initOpt.Bare, util.IfZero(initOpt.ObjectFormat, "sha1"))
+		if err != nil {
+			return nil, err
+		}
+	}
+	return repo, ForceFastImport(ctx, repo, commits)
+}
+
+// ForceFastImport is for mainly for testing purpose
+func ForceFastImport(ctx context.Context, repo RepositoryFacade, commits []FastImportCommit) error {
+	var buf bytes.Buffer
+	for i, c := range commits {
+		msg := util.IfZero(c.Message, fmt.Sprintf("commit %d", i+1))
+		_, _ = fmt.Fprintf(&buf, "reset %s\n", c.Ref)
+		_, _ = fmt.Fprintf(&buf, "commit %s\nmark :%d\ncommitter Gitea <gitea@example.com> 1500000000 +0000\n", c.Ref, i+1)
+		_, _ = fmt.Fprintf(&buf, "data %d\n%s\n", len(msg), msg)
+		for _, f := range c.Files {
+			mode := util.IfZero(f.Mode, EntryModeBlob)
+			_, _ = fmt.Fprintf(&buf, "M %s inline %s\ndata %d\n%s\n", mode.String(), f.Path, len(f.Content), f.Content)
+		}
+	}
+	buf.WriteString("done\n")
+	_, _, err := gitcmd.NewCommand("fast-import").AddArguments("--force", "--done").
+		WithRepo(repo).WithStdinBytes(buf.Bytes()).
+		RunStdString(ctx)
+	return err
+}

modules/git/fetch.go

@@ -0,0 +1,29 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/git/gitrepo"
+)
+
+// FetchRemoteCommit fetches a specific commit and its related objects from a remote
+// repository into the managed repository.
+//
+// If no reference (branch, tag, or other ref) points to the fetched commit, it will
+// be treated as unreachable and cleaned up by `git gc` after the default prune
+// expiration period (2 weeks). Ref: https://www.kernel.org/pub/software/scm/git/docs/git-gc.html
+//
+// This behavior is sufficient for temporary operations, such as determining the
+// merge base between commits.
+func FetchRemoteCommit(ctx context.Context, repo, remoteRepo RepositoryFacade, commitID string) error {
+	return LockWriteAndDo(ctx, repo, func(ctx context.Context) error {
+		return gitcmd.NewCommand("fetch", "--no-tags").
+			AddDynamicArguments(gitrepo.RepoLocalPath(remoteRepo)).
+			AddDynamicArguments(commitID).
+			WithRepo(repo).Run(ctx)
+	})
+}

modules/git/foreachref/format.go

@@ -0,0 +1,83 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package foreachref
+
+import (
+	"encoding/hex"
+	"fmt"
+	"io"
+	"strings"
+)
+
+var (
+	nullChar     = []byte("\x00")
+	dualNullChar = []byte("\x00\x00")
+)
+
+// Format supports specifying and parsing an output format for 'git
+// for-each-ref'. See git-for-each-ref(1) for available fields.
+type Format struct {
+	// fieldNames hold %(fieldname)s to be passed to the '--format' flag of
+	// for-each-ref. See git-for-each-ref(1) for available fields.
+	fieldNames []string
+
+	// fieldDelim is the character sequence that is used to separate fields
+	// for each reference. fieldDelim and refDelim should be selected to not
+	// interfere with each other and to not be present in field values.
+	fieldDelim []byte
+	// fieldDelimStr is a string representation of fieldDelim. Used to save
+	// us from repetitive reallocation whenever we need the delimiter as a
+	// string.
+	fieldDelimStr string
+	// refDelim is the character sequence used to separate reference from
+	// each other in the output. fieldDelim and refDelim should be selected
+	// to not interfere with each other and to not be present in field
+	// values.
+	refDelim []byte
+}
+
+// NewFormat creates a forEachRefFormat using the specified fieldNames. See
+// git-for-each-ref(1) for available fields.
+func NewFormat(fieldNames ...string) Format {
+	return Format{
+		fieldNames:    fieldNames,
+		fieldDelim:    nullChar,
+		fieldDelimStr: string(nullChar),
+		refDelim:      dualNullChar,
+	}
+}
+
+// Flag returns a for-each-ref --format flag value that captures the fieldNames.
+func (f Format) Flag() string {
+	var formatFlag strings.Builder
+	for i, field := range f.fieldNames {
+		// field key and field value
+		fmt.Fprintf(&formatFlag, "%s %%(%s)", field, field)
+
+		if i < len(f.fieldNames)-1 {
+			// note: escape delimiters to allow control characters as
+			// delimiters. For example, '%00' for null character or '%0a'
+			// for newline.
+			formatFlag.WriteString(f.hexEscaped(f.fieldDelim))
+		}
+	}
+	formatFlag.WriteString(f.hexEscaped(f.refDelim))
+	return formatFlag.String()
+}
+
+// Parser returns a Parser capable of parsing 'git for-each-ref' output produced
+// with this Format.
+func (f Format) Parser(r io.Reader) *Parser {
+	return NewParser(r, f)
+}
+
+// hexEscaped produces hex-escaped characters from a string. For example, "\n\0"
+// would turn into "%0a%00".
+func (f Format) hexEscaped(delim []byte) string {
+	var escaped strings.Builder
+	for i := range delim {
+		escaped.WriteString("%" + hex.EncodeToString([]byte{delim[i]}))
+	}
+	return escaped.String()
+}

modules/git/foreachref/format_test.go

@@ -0,0 +1,66 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package foreachref_test
+
+import (
+	"testing"
+
+	"gitea.dev/modules/git/foreachref"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestFormat_Flag(t *testing.T) {
+	tests := []struct {
+		name string
+
+		givenFormat foreachref.Format
+
+		wantFlag string
+	}{
+		{
+			name: "references are delimited by dual null chars",
+
+			// no reference fields requested
+			givenFormat: foreachref.NewFormat(),
+
+			// only a reference delimiter field in --format
+			wantFlag: "%00%00",
+		},
+
+		{
+			name: "a field is a space-separated key-value pair",
+
+			givenFormat: foreachref.NewFormat("refname:short"),
+
+			// only a reference delimiter field
+			wantFlag: "refname:short %(refname:short)%00%00",
+		},
+
+		{
+			name: "fields are separated by a null char field-delimiter",
+
+			givenFormat: foreachref.NewFormat("refname:short", "author"),
+
+			wantFlag: "refname:short %(refname:short)%00author %(author)%00%00",
+		},
+
+		{
+			name: "multiple fields",
+
+			givenFormat: foreachref.NewFormat("refname:lstrip=2", "objecttype", "objectname"),
+
+			wantFlag: "refname:lstrip=2 %(refname:lstrip=2)%00objecttype %(objecttype)%00objectname %(objectname)%00%00",
+		},
+	}
+
+	for _, test := range tests {
+		tc := test // don't close over loop variable
+		t.Run(tc.name, func(t *testing.T) {
+			gotFlag := tc.givenFormat.Flag()
+
+			require.Equal(t, tc.wantFlag, gotFlag, "unexpected for-each-ref --format string. wanted: '%s', got: '%s'", tc.wantFlag, gotFlag)
+		})
+	}
+}

modules/git/foreachref/parser.go

@@ -0,0 +1,135 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package foreachref
+
+import (
+	"bufio"
+	"bytes"
+	"fmt"
+	"io"
+	"strings"
+)
+
+// Parser parses 'git for-each-ref' output according to a given output Format.
+type Parser struct {
+	//  tokenizes 'git for-each-ref' output into "reference paragraphs".
+	scanner *bufio.Scanner
+
+	// format represents the '--format' string that describes the expected
+	// 'git for-each-ref' output structure.
+	format Format
+
+	// err holds the last encountered error during parsing.
+	err error
+}
+
+// NewParser creates a 'git for-each-ref' output parser that will parse all
+// references in the provided Reader. The references in the output are assumed
+// to follow the specified Format.
+func NewParser(r io.Reader, format Format) *Parser {
+	scanner := bufio.NewScanner(r)
+
+	// default Scanner.MaxScanTokenSize = 64 kiB may be too small for some references,
+	// so allow the buffer to be large enough in case the ref has long content (e.g.: a tag with long message)
+	// as long as it doesn't exceed some reasonable limit (4 MiB here, or MAX_DISPLAY_FILE_SIZE=8MiB), it is OK
+	// there are still some choices: 1. add a config option for the limit; 2. don't use scanner and write our own parser to fully handle large contents
+	scanner.Buffer(nil, 4*1024*1024)
+
+	// in addition to the reference delimiter we specified in the --format,
+	// `git for-each-ref` will always add a newline after every reference.
+	refDelim := make([]byte, 0, len(format.refDelim)+1)
+	refDelim = append(refDelim, format.refDelim...)
+	refDelim = append(refDelim, '\n')
+
+	// Split input into delimiter-separated "reference blocks".
+	scanner.Split(
+		func(data []byte, atEOF bool) (advance int, token []byte, err error) {
+			// Scan until delimiter, marking end of reference.
+			delimIdx := bytes.Index(data, refDelim)
+			if delimIdx >= 0 {
+				token := data[:delimIdx]
+				advance := delimIdx + len(refDelim)
+				return advance, token, nil
+			}
+			// If we're at EOF, we have a final, non-terminated reference. Return it.
+			if atEOF {
+				return len(data), data, nil
+			}
+			// Not yet a full field. Request more data.
+			return 0, nil, nil
+		})
+
+	return &Parser{
+		scanner: scanner,
+		format:  format,
+		err:     nil,
+	}
+}
+
+// Next returns the next reference as a collection of key-value pairs. nil
+// denotes EOF but is also returned on errors. The Err method should always be
+// consulted after Next returning nil.
+//
+// It could, for example return something like:
+//
+//	{ "objecttype": "tag", "refname:short": "v1.16.4", "object": "f460b7543ed500e49c133c2cd85c8c55ee9dbe27" }
+func (p *Parser) Next() map[string]string {
+	if !p.scanner.Scan() {
+		if err := p.scanner.Err(); err != nil {
+			p.err = err
+		}
+		return nil
+	}
+	fields, err := p.parseRef(p.scanner.Text())
+	if err != nil {
+		p.err = err
+		return nil
+	}
+	return fields
+}
+
+// Err returns the latest encountered parsing error.
+func (p *Parser) Err() error {
+	return p.err
+}
+
+// parseRef parses out all key-value pairs from a single reference block, such as
+//
+//	"objecttype tag\0refname:short v1.16.4\0object f460b7543ed500e49c133c2cd85c8c55ee9dbe27"
+func (p *Parser) parseRef(refBlock string) (map[string]string, error) {
+	if refBlock == "" {
+		// must be at EOF
+		return nil, nil //nolint:nilnil // return nil to signal EOF
+	}
+
+	fieldValues := make(map[string]string, len(p.format.fieldNames))
+
+	fields := strings.Split(refBlock, p.format.fieldDelimStr)
+	if len(fields) != len(p.format.fieldNames) {
+		return nil, fmt.Errorf("unexpected number of reference fields: wanted %d, was %d",
+			len(fields), len(p.format.fieldNames))
+	}
+	for i, field := range fields {
+		var fieldKey string
+		var fieldVal string
+		before, after, ok := strings.Cut(field, " ")
+		if ok {
+			fieldKey = before
+			fieldVal = after
+		} else {
+			// could be the case if the requested field had no value
+			fieldKey = field
+		}
+
+		// enforce the format order of fields
+		if p.format.fieldNames[i] != fieldKey {
+			return nil, fmt.Errorf("unexpected field name at position %d: wanted: '%s', was: '%s'",
+				i, p.format.fieldNames[i], fieldKey)
+		}
+
+		fieldValues[fieldKey] = fieldVal
+	}
+
+	return fieldValues, nil
+}

modules/git/foreachref/parser_test.go

@@ -0,0 +1,227 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package foreachref_test
+
+import (
+	"errors"
+	"fmt"
+	"io"
+	"strings"
+	"testing"
+
+	"gitea.dev/modules/git/foreachref"
+	"gitea.dev/modules/json"
+
+	"github.com/stretchr/testify/require"
+)
+
+type refSlice = []map[string]string
+
+func TestParser(t *testing.T) {
+	tests := []struct {
+		name string
+
+		givenFormat foreachref.Format
+		givenInput  io.Reader
+
+		wantRefs    refSlice
+		wantErr     bool
+		expectedErr error
+	}{
+		// this would, for example, be the result when running `git
+		// for-each-ref refs/tags` on a repo without tags.
+		{
+			name: "no references on empty input",
+
+			givenFormat: foreachref.NewFormat("refname:short"),
+			givenInput:  strings.NewReader(``),
+
+			wantRefs: []map[string]string{},
+		},
+
+		// note: `git for-each-ref` will add a newline between every
+		// reference (in addition to the ref-delimiter we've chosen)
+		{
+			name: "single field requested, single reference in output",
+
+			givenFormat: foreachref.NewFormat("refname:short"),
+			givenInput:  strings.NewReader("refname:short v0.0.1\x00\x00" + "\n"),
+
+			wantRefs: []map[string]string{
+				{"refname:short": "v0.0.1"},
+			},
+		},
+		{
+			name: "single field requested, multiple references in output",
+
+			givenFormat: foreachref.NewFormat("refname:short"),
+			givenInput: strings.NewReader(
+				"refname:short v0.0.1\x00\x00" + "\n" +
+					"refname:short v0.0.2\x00\x00" + "\n" +
+					"refname:short v0.0.3\x00\x00" + "\n"),
+
+			wantRefs: []map[string]string{
+				{"refname:short": "v0.0.1"},
+				{"refname:short": "v0.0.2"},
+				{"refname:short": "v0.0.3"},
+			},
+		},
+
+		{
+			name: "multiple fields requested for each reference",
+
+			givenFormat: foreachref.NewFormat("refname:short", "objecttype", "objectname"),
+			givenInput: strings.NewReader(
+
+				"refname:short v0.0.1\x00objecttype commit\x00objectname 7b2c5ac9fc04fc5efafb60700713d4fa609b777b\x00\x00" + "\n" +
+					"refname:short v0.0.2\x00objecttype commit\x00objectname a1f051bc3eba734da4772d60e2d677f47cf93ef4\x00\x00" + "\n" +
+					"refname:short v0.0.3\x00objecttype commit\x00objectname ef82de70bb3f60c65fb8eebacbb2d122ef517385\x00\x00" + "\n",
+			),
+
+			wantRefs: []map[string]string{
+				{
+					"refname:short": "v0.0.1",
+					"objecttype":    "commit",
+					"objectname":    "7b2c5ac9fc04fc5efafb60700713d4fa609b777b",
+				},
+				{
+					"refname:short": "v0.0.2",
+					"objecttype":    "commit",
+					"objectname":    "a1f051bc3eba734da4772d60e2d677f47cf93ef4",
+				},
+				{
+					"refname:short": "v0.0.3",
+					"objecttype":    "commit",
+					"objectname":    "ef82de70bb3f60c65fb8eebacbb2d122ef517385",
+				},
+			},
+		},
+
+		{
+			name: "must handle multi-line fields such as 'content'",
+
+			givenFormat: foreachref.NewFormat("refname:short", "contents", "author"),
+			givenInput: strings.NewReader(
+				"refname:short v0.0.1\x00contents Create new buffer if not present yet (#549)\n\nFixes a nil dereference when ProcessFoo is used\nwith multiple commands.\x00author Foo Bar <foo@bar.com> 1507832733 +0200\x00\x00" + "\n" +
+					"refname:short v0.0.2\x00contents Update CI config (#651)\n\n\x00author John Doe <john.doe@foo.com> 1521643174 +0000\x00\x00" + "\n" +
+					"refname:short v0.0.3\x00contents Fixed code sample for bash completion (#687)\n\n\x00author Foo Baz <foo@baz.com> 1524836750 +0200\x00\x00" + "\n",
+			),
+
+			wantRefs: []map[string]string{
+				{
+					"refname:short": "v0.0.1",
+					"contents":      "Create new buffer if not present yet (#549)\n\nFixes a nil dereference when ProcessFoo is used\nwith multiple commands.",
+					"author":        "Foo Bar <foo@bar.com> 1507832733 +0200",
+				},
+				{
+					"refname:short": "v0.0.2",
+					"contents":      "Update CI config (#651)\n\n",
+					"author":        "John Doe <john.doe@foo.com> 1521643174 +0000",
+				},
+				{
+					"refname:short": "v0.0.3",
+					"contents":      "Fixed code sample for bash completion (#687)\n\n",
+					"author":        "Foo Baz <foo@baz.com> 1524836750 +0200",
+				},
+			},
+		},
+
+		{
+			name: "must handle fields without values",
+
+			givenFormat: foreachref.NewFormat("refname:short", "object", "objecttype"),
+			givenInput: strings.NewReader(
+				"refname:short v0.0.1\x00object \x00objecttype commit\x00\x00" + "\n" +
+					"refname:short v0.0.2\x00object \x00objecttype commit\x00\x00" + "\n" +
+					"refname:short v0.0.3\x00object \x00objecttype commit\x00\x00" + "\n",
+			),
+
+			wantRefs: []map[string]string{
+				{
+					"refname:short": "v0.0.1",
+					"object":        "",
+					"objecttype":    "commit",
+				},
+				{
+					"refname:short": "v0.0.2",
+					"object":        "",
+					"objecttype":    "commit",
+				},
+				{
+					"refname:short": "v0.0.3",
+					"object":        "",
+					"objecttype":    "commit",
+				},
+			},
+		},
+
+		{
+			name: "must fail when the number of fields in the input doesn't match expected format",
+
+			givenFormat: foreachref.NewFormat("refname:short", "objecttype", "objectname"),
+			givenInput: strings.NewReader(
+				"refname:short v0.0.1\x00objecttype commit\x00\x00" + "\n" +
+					"refname:short v0.0.2\x00objecttype commit\x00\x00" + "\n" +
+					"refname:short v0.0.3\x00objecttype commit\x00\x00" + "\n",
+			),
+
+			wantErr:     true,
+			expectedErr: errors.New("unexpected number of reference fields: wanted 2, was 3"),
+		},
+
+		{
+			name: "must fail input fields don't match expected format",
+
+			givenFormat: foreachref.NewFormat("refname:short", "objectname"),
+			givenInput: strings.NewReader(
+				"refname:short v0.0.1\x00objecttype commit\x00\x00" + "\n" +
+					"refname:short v0.0.2\x00objecttype commit\x00\x00" + "\n" +
+					"refname:short v0.0.3\x00objecttype commit\x00\x00" + "\n",
+			),
+
+			wantErr:     true,
+			expectedErr: errors.New("unexpected field name at position 1: wanted: 'objectname', was: 'objecttype'"),
+		},
+	}
+
+	for _, test := range tests {
+		tc := test // don't close over loop variable
+		t.Run(tc.name, func(t *testing.T) {
+			parser := tc.givenFormat.Parser(tc.givenInput)
+
+			//
+			// parse references from input
+			//
+			gotRefs := make([]map[string]string, 0)
+			for {
+				ref := parser.Next()
+				if ref == nil {
+					break
+				}
+				gotRefs = append(gotRefs, ref)
+			}
+			err := parser.Err()
+
+			//
+			// verify expectations
+			//
+			if tc.wantErr {
+				require.Error(t, err)
+				require.EqualError(t, err, tc.expectedErr.Error())
+			} else {
+				require.NoError(t, err, "for-each-ref parser unexpectedly failed with: %v", err)
+				require.Equal(t, tc.wantRefs, gotRefs, "for-each-ref parser produced unexpected reference set. wanted: %v, got: %v", pretty(tc.wantRefs), pretty(gotRefs))
+			}
+		})
+	}
+}
+
+func pretty(v any) string {
+	data, err := json.MarshalIndent(v, "", "  ")
+	if err != nil {
+		// shouldn't happen
+		panic(fmt.Sprintf("json-marshalling failed: %v", err))
+	}
+	return string(data)
+}

modules/git/fsck.go

@@ -0,0 +1,16 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"time"
+
+	"gitea.dev/modules/git/gitcmd"
+)
+
+// Fsck verifies the connectivity and validity of the objects in the database
+func Fsck(ctx context.Context, repo RepositoryFacade, timeout time.Duration, args gitcmd.TrustedCmdArgs) error {
+	return gitcmd.NewCommand("fsck").AddArguments(args...).WithTimeout(timeout).WithRepo(repo).Run(ctx)
+}

modules/git/git.go

@@ -0,0 +1,215 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"os"
+	"path/filepath"
+	"runtime"
+	"strings"
+
+	"gitea.dev/modules/cache"
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/globallock"
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/tempdir"
+	"gitea.dev/modules/testlogger"
+
+	"github.com/hashicorp/go-version"
+)
+
+const RequiredVersion = "2.25.0" // the minimum Git version required
+
+type Features struct {
+	gitVersion *version.Version
+
+	UsingGogit                 bool
+	SupportProcReceive         bool           // >= 2.29
+	SupportHashSha256          bool           // >= 2.42, SHA-256 repositories no longer an ‘experimental curiosity’
+	SupportedObjectFormats     []ObjectFormat // sha1, sha256
+	SupportCheckAttrOnBare     bool           // >= 2.40
+	SupportCatFileBatchCommand bool           // >= 2.36, support `git cat-file --batch-command`
+	SupportGitMergeTree        bool           // >= 2.40 // we also need "--merge-base"
+}
+
+type GlobalConfigStruct struct {
+	DiffOrderFile string
+}
+
+var (
+	defaultFeatures *Features
+	GlobalConfig    *GlobalConfigStruct
+)
+
+func (f *Features) CheckVersionAtLeast(atLeast string) bool {
+	return f.gitVersion.Compare(version.Must(version.NewVersion(atLeast))) >= 0
+}
+
+// VersionInfo returns git version information
+func (f *Features) VersionInfo() string {
+	return f.gitVersion.Original()
+}
+
+func DefaultFeatures() *Features {
+	if defaultFeatures == nil {
+		if !setting.IsProd || setting.IsInTesting {
+			log.Warn("git.DefaultFeatures is called before git.InitXxx, initializing with default values")
+		}
+		if err := InitSimple(); err != nil {
+			log.Fatal("git.InitSimple failed: %v", err)
+		}
+	}
+	return defaultFeatures
+}
+
+func loadGitVersionFeatures() (*Features, error) {
+	stdout, _, runErr := gitcmd.NewCommand("version").RunStdString(context.Background())
+	if runErr != nil {
+		return nil, runErr
+	}
+
+	ver, err := parseGitVersionLine(strings.TrimSpace(stdout))
+	if err != nil {
+		return nil, err
+	}
+
+	features := &Features{gitVersion: ver, UsingGogit: isGogit}
+	features.SupportProcReceive = features.CheckVersionAtLeast("2.29")
+	features.SupportHashSha256 = features.CheckVersionAtLeast("2.42") && !isGogit
+	features.SupportedObjectFormats = []ObjectFormat{Sha1ObjectFormat}
+	if features.SupportHashSha256 {
+		features.SupportedObjectFormats = append(features.SupportedObjectFormats, Sha256ObjectFormat)
+	}
+	features.SupportCheckAttrOnBare = features.CheckVersionAtLeast("2.40")
+	features.SupportCatFileBatchCommand = features.CheckVersionAtLeast("2.36")
+	features.SupportGitMergeTree = features.CheckVersionAtLeast("2.40") // we also need "--merge-base"
+	return features, nil
+}
+
+func parseGitVersionLine(s string) (*version.Version, error) {
+	fields := strings.Fields(s)
+	if len(fields) < 3 {
+		return nil, fmt.Errorf("invalid git version: %q", s)
+	}
+
+	// version output is like: "git version {versionString}"
+	// versionString can be:
+	// * "2.5.3"
+	// * "2.29.3.windows.1"
+	// * "2.28.0.618.gf4bc123cb7": https://github.com/go-gitea/gitea/issues/12731
+	versionString := fields[2]
+	versionFields := strings.Split(versionString, ".")
+	if len(versionFields) > 3 {
+		versionFields = versionFields[:3]
+	}
+	return version.NewVersion(strings.Join(versionFields, "."))
+}
+
+func checkGitVersionCompatibility(gitVer *version.Version) error {
+	badVersions := []struct {
+		Version *version.Version
+		Reason  string
+	}{
+		{version.Must(version.NewVersion("2.43.1")), "regression bug of GIT_FLUSH"},
+	}
+	for _, bad := range badVersions {
+		if gitVer.Equal(bad.Version) {
+			return errors.New(bad.Reason)
+		}
+	}
+	return nil
+}
+
+func ensureGitVersion() error {
+	if !DefaultFeatures().CheckVersionAtLeast(RequiredVersion) {
+		moreHint := "get git: https://git-scm.com/downloads"
+		if runtime.GOOS == "linux" {
+			// there are a lot of CentOS/RHEL users using old git, so we add a special hint for them
+			if _, err := os.Stat("/etc/redhat-release"); err == nil {
+				// ius.io is the recommended official(git-scm.com) method to install git
+				moreHint = "get git: https://git-scm.com/downloads/linux and https://ius.io"
+			}
+		}
+		return fmt.Errorf("installed git version %q is not supported, Gitea requires git version >= %q, %s", DefaultFeatures().gitVersion.Original(), RequiredVersion, moreHint)
+	}
+
+	if err := checkGitVersionCompatibility(DefaultFeatures().gitVersion); err != nil {
+		return fmt.Errorf("installed git version %s has a known compatibility issue with Gitea: %w, please upgrade (or downgrade) git", DefaultFeatures().gitVersion.String(), err)
+	}
+	return nil
+}
+
+// InitSimple initializes git module with a very simple step, no config changes, no global command arguments.
+// This method doesn't change anything to filesystem. At the moment, it is only used by some Gitea sub-commands.
+func InitSimple() error {
+	if setting.Git.HomePath == "" {
+		return errors.New("unable to init Git's HomeDir, incorrect initialization of the setting and git modules")
+	}
+
+	if defaultFeatures != nil && (!setting.IsProd || setting.IsInTesting) {
+		log.Warn("git module has been initialized already, duplicate init may work but it's better to fix it")
+	}
+
+	if err := gitcmd.SetExecutablePath(setting.Git.Path); err != nil {
+		return err
+	}
+
+	var err error
+	defaultFeatures, err = loadGitVersionFeatures()
+	if err != nil {
+		return err
+	}
+	if err = ensureGitVersion(); err != nil {
+		return err
+	}
+
+	// when git works with gnupg (commit signing), there should be a stable home for gnupg commands
+	if _, ok := os.LookupEnv("GNUPGHOME"); !ok {
+		_ = os.Setenv("GNUPGHOME", filepath.Join(gitcmd.HomeDir(), ".gnupg"))
+	}
+	return nil
+}
+
+// InitFull initializes git module with version check and change global variables, sync gitconfig.
+// It should only be called once at the beginning of the program initialization (TestMain/GlobalInitInstalled) as this code makes unsynchronized changes to variables.
+func InitFull() (err error) {
+	if err = InitSimple(); err != nil {
+		return err
+	}
+	return syncGitConfig(context.Background())
+}
+
+// RunGitTests helps to init the git module and run tests.
+// FIXME: GIT-PACKAGE-DEPENDENCY: the dependency is not right, setting.Git.HomePath is initialized in this package but used in gitcmd package
+func RunGitTests(m interface{ Run() int }) {
+	os.Exit(runGitTests(m))
+}
+
+func runGitTests(m interface{ Run() int }) int {
+	_ = cache.Init()
+	gitHomePath, cleanup, err := tempdir.OsTempDir("gitea-test").MkdirTempRandom("git-home")
+	if err != nil {
+		return testlogger.MainErrorf("unable to create temp dir: %v", err)
+	}
+	defer cleanup()
+
+	setting.Git.HomePath = gitHomePath
+	if err = InitFull(); err != nil {
+		return testlogger.MainErrorf("failed to call Init: %v", err)
+	}
+	return m.Run()
+}
+
+func LockConfigAndDo(ctx context.Context, repo RepositoryFacade, fn func(ctx context.Context) error) error {
+	return globallock.LockAndDo(ctx, "repo-config:"+repo.GitRepoManagedID(), fn)
+}
+
+func LockWriteAndDo(ctx context.Context, repo RepositoryFacade, fn func(ctx context.Context) error) error {
+	return globallock.LockAndDo(ctx, "repo-write:"+repo.GitRepoManagedID(), fn)
+}

modules/git/git_test.go

@@ -0,0 +1,37 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"testing"
+
+	"github.com/hashicorp/go-version"
+	"github.com/stretchr/testify/assert"
+)
+
+func TestParseGitVersion(t *testing.T) {
+	v, err := parseGitVersionLine("git version 2.29.3")
+	assert.NoError(t, err)
+	assert.Equal(t, "2.29.3", v.String())
+
+	v, err = parseGitVersionLine("git version 2.29.3.windows.1")
+	assert.NoError(t, err)
+	assert.Equal(t, "2.29.3", v.String())
+
+	v, err = parseGitVersionLine("git version 2.28.0.618.gf4bc123cb7")
+	assert.NoError(t, err)
+	assert.Equal(t, "2.28.0", v.String())
+
+	_, err = parseGitVersionLine("git version")
+	assert.Error(t, err)
+
+	_, err = parseGitVersionLine("git version windows")
+	assert.Error(t, err)
+}
+
+func TestCheckGitVersionCompatibility(t *testing.T) {
+	assert.NoError(t, checkGitVersionCompatibility(version.Must(version.NewVersion("2.43.0"))))
+	assert.ErrorContains(t, checkGitVersionCompatibility(version.Must(version.NewVersion("2.43.1"))), "regression bug of GIT_FLUSH")
+	assert.NoError(t, checkGitVersionCompatibility(version.Must(version.NewVersion("2.43.2"))))
+}

modules/git/gitcmd/command.go

@@ -0,0 +1,582 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2016 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"bytes"
+	"context"
+	"errors"
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/git/internal" //nolint:depguard // only this file can use the internal type CmdArg, other files and packages should use AddXxx functions
+	"gitea.dev/modules/gtprof"
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/process"
+	"gitea.dev/modules/util"
+)
+
+// TrustedCmdArgs returns the trusted arguments for git command.
+// It's mainly for passing user-provided and trusted arguments to git command
+// In most cases, it shouldn't be used. Use AddXxx function instead
+type TrustedCmdArgs []internal.CmdArg
+
+// Command represents a command with its subcommands or arguments.
+type Command struct {
+	callerInfo string
+	prog       string
+	args       []string
+	preErrors  []error
+	configArgs []string
+
+	// Dir is the working dir for the git command, however:
+	// FIXME: GIT-DIR-ARGUMENT: this could be incorrect in many cases, for example:
+	// * /some/path/.git
+	// * /some/path/.git/gitea-data/data/repositories/user/repo.git
+	// If "user/repo.git" is invalid/broken, then running git command in it will use "/some/path/.git", and produce unexpected results
+	// The correct approach is to use `--git-dir" global argument or "GIT_DIR=..." environment variable.
+	// Actually, when working with a bare repo, the current directory should not be the git dir,
+	// otherwise some git commands might overwrite git dir internal files by a repo file.
+	gitDir string
+
+	cmd *process.Cmd
+
+	cmdCtx       context.Context
+	cmdCancel    process.CancelCauseFunc
+	cmdFinished  process.FinishedFunc
+	cmdStartTime time.Time
+
+	pipelineFunc func(Context) error
+
+	parentPipeFiles   []*os.File
+	parentPipeReaders []*os.File
+	childrenPipeFiles []*os.File
+
+	cmdEnv     []string
+	cmdTimeout time.Duration
+
+	// only os.Pipe and in-memory buffers can work with Stdin safely, see https://github.com/golang/go/issues/77227 if the command would exit unexpectedly
+	cmdStdin  io.Reader
+	cmdStdout io.Writer
+	cmdStderr io.Writer
+
+	cmdManagedStderr *bytes.Buffer
+}
+
+func logArgSanitize(arg string) string {
+	if filepath.IsAbs(arg) {
+		base := filepath.Base(arg)
+		dir := filepath.Dir(arg)
+		return ".../" + filepath.Join(filepath.Base(dir), base)
+	}
+	return util.SanitizeCredentialURLs(arg)
+}
+
+func (c *Command) LogString() string {
+	// WARNING: this function is for debugging purposes only. It's much better than old code (which only joins args with space),
+	// It's impossible to make a simple and 100% correct implementation of argument quoting for different platforms here.
+	debugQuote := func(s string) string {
+		if strings.ContainsAny(s, " `'\"\t\r\n") {
+			return fmt.Sprintf("%q", s)
+		}
+		return s
+	}
+	a := make([]string, 0, len(c.args)+1)
+	a = append(a, debugQuote(c.prog))
+	for i := 0; i < len(c.args); i++ {
+		a = append(a, debugQuote(logArgSanitize(c.args[i])))
+	}
+	return strings.Join(a, " ")
+}
+
+func (c *Command) ProcessState() string {
+	if c.cmd == nil {
+		return ""
+	}
+	return c.cmd.ProcessState.String()
+}
+
+// NewCommand creates and returns a new Git Command based on given command and arguments.
+// Each argument should be safe to be trusted. User-provided arguments should be passed to AddDynamicArguments instead.
+func NewCommand(args ...internal.CmdArg) *Command {
+	cargs := make([]string, 0, len(args))
+	for _, arg := range args {
+		cargs = append(cargs, string(arg))
+	}
+	return &Command{
+		prog: GitExecutable,
+		args: cargs,
+	}
+}
+
+func (c *Command) handlePreErrorBrokenCommand(arg string) {
+	c.preErrors = append(c.preErrors, util.ErrorWrap(ErrBrokenCommand, `broken git command argument %q`, arg))
+}
+
+// isSafeArgumentValue checks if the argument is safe to be used as a value (not an option)
+func isSafeArgumentValue(s string) bool {
+	return s == "" || s[0] != '-'
+}
+
+// isValidArgumentOption checks if the argument is a valid option (starting with '-').
+// It doesn't check whether the option is supported or not
+func isValidArgumentOption(s string) bool {
+	return s != "" && s[0] == '-'
+}
+
+// AddArguments adds new git arguments (option/value) to the command. It only accepts string literals, or trusted CmdArg.
+// Type CmdArg is in the internal package, so it can not be used outside of this package directly,
+// it makes sure that user-provided arguments won't cause RCE risks.
+// User-provided arguments should be passed by other AddXxx functions
+func (c *Command) AddArguments(args ...internal.CmdArg) *Command {
+	for _, arg := range args {
+		c.args = append(c.args, string(arg))
+	}
+	return c
+}
+
+// AddOptionValues adds a new option with a list of non-option values
+// For example: AddOptionValues("--opt", val) means 2 arguments: {"--opt", val}.
+// The values are treated as dynamic argument values. It equals to: AddArguments("--opt") then AddDynamicArguments(val).
+func (c *Command) AddOptionValues(opt internal.CmdArg, args ...string) *Command {
+	if !isValidArgumentOption(string(opt)) {
+		c.handlePreErrorBrokenCommand(string(opt))
+		return c
+	}
+	c.args = append(c.args, string(opt))
+	c.AddDynamicArguments(args...)
+	return c
+}
+
+// AddOptionFormat adds a new option with a format string and arguments
+// For example: AddOptionFormat("--opt=%s %s", val1, val2) means 1 argument: {"--opt=val1 val2"}.
+func (c *Command) AddOptionFormat(opt string, args ...any) *Command {
+	if !isValidArgumentOption(opt) {
+		c.handlePreErrorBrokenCommand(opt)
+		return c
+	}
+	// a quick check to make sure the format string matches the number of arguments, to find low-level mistakes ASAP
+	if strings.Count(strings.ReplaceAll(opt, "%%", ""), "%") != len(args) {
+		c.handlePreErrorBrokenCommand(opt)
+		return c
+	}
+	s := fmt.Sprintf(opt, args...)
+	c.args = append(c.args, s)
+	return c
+}
+
+// AddDynamicArguments adds new dynamic argument values to the command.
+// The arguments may come from user input and can not be trusted, so no leading '-' is allowed to avoid passing options.
+// TODO: in the future, this function can be renamed to AddArgumentValues
+func (c *Command) AddDynamicArguments(args ...string) *Command {
+	for _, arg := range args {
+		if !isSafeArgumentValue(arg) {
+			c.handlePreErrorBrokenCommand(arg)
+		}
+	}
+	if len(c.preErrors) != 0 {
+		return c
+	}
+	c.args = append(c.args, args...)
+	return c
+}
+
+// AddDashesAndList adds the "--" and then add the list as arguments, it's usually for adding file list
+// At the moment, this function can be only called once, maybe in future it can be refactored to support multiple calls (if necessary)
+func (c *Command) AddDashesAndList(list ...string) *Command {
+	c.args = append(c.args, "--")
+	// Some old code also checks `arg != ""`, IMO it's not necessary.
+	// If the check is needed, the list should be prepared before the call to this function
+	c.args = append(c.args, list...)
+	return c
+}
+
+func (c *Command) AddConfig(key, value string) *Command {
+	kv := key + "=" + value
+	if !isSafeArgumentValue(kv) {
+		c.handlePreErrorBrokenCommand(kv)
+	} else {
+		c.configArgs = append(c.configArgs, "-c", kv)
+	}
+	return c
+}
+
+// ToTrustedCmdArgs converts a list of strings (trusted as argument) to TrustedCmdArgs
+// In most cases, it shouldn't be used. Use NewCommand().AddXxx() function instead
+func ToTrustedCmdArgs(args []string) TrustedCmdArgs {
+	ret := make(TrustedCmdArgs, len(args))
+	for i, arg := range args {
+		ret[i] = internal.CmdArg(arg)
+	}
+	return ret
+}
+
+func commonBaseEnvs() []string {
+	envs := []string{
+		// Make Gitea use internal git config only, to prevent conflicts with user's git config
+		// It's better to use GIT_CONFIG_GLOBAL, but it requires git >= 2.32, so we still use HOME at the moment.
+		"HOME=" + HomeDir(),
+		// Avoid using system git config, it would cause problems (eg: use macOS osxkeychain to show a modal dialog, auto installing lfs hooks)
+		// This might be a breaking change in 1.24, because some users said that they have put some configs like "receive.certNonceSeed" in "/etc/gitconfig"
+		// For these users, they need to migrate the necessary configs to Gitea's git config file manually.
+		"GIT_CONFIG_NOSYSTEM=1",
+		// Ignore replace references (https://git-scm.com/docs/git-replace)
+		"GIT_NO_REPLACE_OBJECTS=1",
+	}
+
+	// some environment variables should be passed to git command
+	passThroughEnvKeys := []string{
+		"GNUPGHOME", // git may call gnupg to do commit signing
+	}
+	for _, key := range passThroughEnvKeys {
+		if val, ok := os.LookupEnv(key); ok {
+			envs = append(envs, key+"="+val)
+		}
+	}
+	return envs
+}
+
+// CommonGitCmdEnvs returns the common environment variables for a "git" command.
+func CommonGitCmdEnvs() []string {
+	return append(commonBaseEnvs(), []string{
+		"LC_ALL=C",              // ensure git output is in English, error messages are parsed in English
+		"GIT_TERMINAL_PROMPT=0", // avoid prompting for credentials interactively, supported since git v2.3
+	}...)
+}
+
+// CommonCmdServEnvs is like CommonGitCmdEnvs, but it only returns minimal required environment variables for the "gitea serv" command
+func CommonCmdServEnvs() []string {
+	return commonBaseEnvs()
+}
+
+var ErrBrokenCommand = errors.New("git command is broken")
+
+func (c *Command) WithDir(dir string) *Command {
+	c.gitDir = dir
+	return c
+}
+
+func (c *Command) WithRepo(repo gitrepo.RepositoryFacade) *Command {
+	c.gitDir = gitrepo.RepoLocalPath(repo)
+	return c
+}
+
+func (c *Command) WithEnv(env []string) *Command {
+	c.cmdEnv = env
+	return c
+}
+
+func (c *Command) WithTimeout(timeout time.Duration) *Command {
+	c.cmdTimeout = timeout
+	return c
+}
+
+func (c *Command) makeStdoutStderr(w *io.Writer) (PipeReader, func()) {
+	pr, pw, err := os.Pipe()
+	if err != nil {
+		c.preErrors = append(c.preErrors, err)
+		return &pipeNull{err}, func() {}
+	}
+	c.childrenPipeFiles = append(c.childrenPipeFiles, pw)
+	c.parentPipeFiles = append(c.parentPipeFiles, pr)
+	c.parentPipeReaders = append(c.parentPipeReaders, pr)
+	*w /* stdout, stderr */ = pw
+	return &pipeReader{f: pr}, func() { pr.Close() }
+}
+
+// MakeStdinPipe creates a writer for the command's stdin.
+// The returned closer function must be called by the caller to close the pipe.
+func (c *Command) MakeStdinPipe() (writer PipeWriter, closer func()) {
+	pr, pw, err := os.Pipe()
+	if err != nil {
+		c.preErrors = append(c.preErrors, err)
+		return &pipeNull{err}, func() {}
+	}
+	c.childrenPipeFiles = append(c.childrenPipeFiles, pr)
+	c.parentPipeFiles = append(c.parentPipeFiles, pw)
+	c.cmdStdin = pr
+	return &pipeWriter{pw}, func() { pw.Close() }
+}
+
+// MakeStdoutPipe creates a reader for the command's stdout.
+// The returned closer function must be called by the caller to close the pipe.
+// After the pipe reader is closed, the unread data will be discarded.
+//
+// If the process (git command) still tries to write after the pipe is closed, the Wait error will be "signal: broken pipe".
+// WithPipelineFunc + Run won't return "broken pipe" error in this case if the callback returns no error.
+// But if you are calling Start / Wait family functions, you should either drain the pipe before close it, or handle the Wait error correctly.
+func (c *Command) MakeStdoutPipe() (reader PipeReader, closer func()) {
+	return c.makeStdoutStderr(&c.cmdStdout)
+}
+
+// MakeStderrPipe is like MakeStdoutPipe, but for stderr.
+func (c *Command) MakeStderrPipe() (reader PipeReader, closer func()) {
+	return c.makeStdoutStderr(&c.cmdStderr)
+}
+
+func (c *Command) MakeStdinStdoutPipe() (stdin PipeWriter, stdout PipeReader, closer func()) {
+	stdin, stdinClose := c.MakeStdinPipe()
+	stdout, stdoutClose := c.MakeStdoutPipe()
+	return stdin, stdout, func() {
+		stdinClose()
+		stdoutClose()
+	}
+}
+
+func (c *Command) WithStdinBytes(stdin []byte) *Command {
+	c.cmdStdin = bytes.NewReader(stdin)
+	return c
+}
+
+func (c *Command) WithStdoutBuffer(w PipeBufferWriter) *Command {
+	c.cmdStdout = w
+	return c
+}
+
+// WithStdinCopy and WithStdoutCopy are general functions that accept any io.Reader / io.Writer.
+// In this case, Golang exec.Cmd will start new internal goroutines to do io.Copy between pipes and provided Reader/Writer.
+// If the reader or writer is blocked and never returns, then the io.Copy won't finish, then exec.Cmd.Wait won't return, which may cause deadlocks.
+// A typical deadlock example is:
+// * `r,w:=io.Pipe(); cmd.Stdin=r; defer w.Close(); cmd.Run()`: the Run() will never return because stdin reader is blocked forever and w.Close() will never be called.
+// If the reader/writer won't block forever (for example: read from a file or buffer), then these functions are safe to use.
+func (c *Command) WithStdinCopy(w io.Reader) *Command {
+	c.cmdStdin = w
+	return c
+}
+
+func (c *Command) WithStdoutCopy(w io.Writer) *Command {
+	c.cmdStdout = w
+	return c
+}
+
+// WithPipelineFunc sets the pipeline function for the command.
+// The pipeline function will be called in the Run / Wait function after the command is started successfully.
+// The function can read/write from/to the command's stdio pipes (if any).
+// The pipeline function can cancel (kill) the command by calling ctx.CancelPipeline before the command finishes.
+// The returned error of Run / Wait can be joined errors from the pipeline function, context cause, and command exit error.
+// Caller can get the pipeline function's error (if any) by UnwrapPipelineError.
+func (c *Command) WithPipelineFunc(f func(ctx Context) error) *Command {
+	c.pipelineFunc = f
+	return c
+}
+
+// WithParentCallerInfo can be used to set the caller info (usually function name) of the parent function of the caller.
+// For most cases, "Run" family functions can get its caller info automatically
+// But if you need to call "Run" family functions in a wrapper function: "FeatureFunc -> GeneralWrapperFunc -> RunXxx",
+// then you can to call this function in GeneralWrapperFunc to set the caller info of FeatureFunc.
+// The caller info can only be set once.
+func (c *Command) WithParentCallerInfo(optInfo ...string) *Command {
+	if c.callerInfo != "" {
+		return c
+	}
+	if len(optInfo) > 0 {
+		c.callerInfo = optInfo[0]
+		return c
+	}
+	skip := 1 /*parent "wrap/run" functions*/ + 1 /*this function*/
+	c.callerInfo = util.CallerFuncName(skip)
+	return c
+}
+
+func (c *Command) Start(ctx context.Context) (retErr error) {
+	if c.cmd != nil {
+		// this is a programming error, it will cause serious deadlock problems, so it must be fixed.
+		panic("git command has already been started")
+	}
+
+	defer func() {
+		c.closePipeFiles(c.childrenPipeFiles)
+		if retErr != nil {
+			// release the pipes to avoid resource leak since the command failed to start
+			c.closePipeFiles(c.parentPipeFiles)
+			// if error occurs, we must also finish the task, otherwise, cmdFinished will be called in "Wait" function
+			if c.cmdFinished != nil {
+				c.cmdFinished()
+			}
+		}
+	}()
+
+	if len(c.preErrors) != 0 {
+		// In most cases, such error shouldn't happen. If it happens, log it as error level with more details
+		err := errors.Join(c.preErrors...)
+		log.Error("git command: %s, error: %s", c.LogString(), err)
+		return err
+	}
+
+	cmdLogString := c.LogString()
+	if c.callerInfo == "" {
+		c.WithParentCallerInfo()
+	}
+	// these logs are for debugging purposes only, so no guarantee of correctness or stability
+	desc := fmt.Sprintf("git.Run(by:%s, repo:%s): %s", c.callerInfo, logArgSanitize(c.gitDir), cmdLogString)
+	log.Debug("git.Command: %s", desc)
+
+	_, span := gtprof.GetTracer().Start(ctx, gtprof.TraceSpanGitRun)
+	defer span.End()
+	span.SetAttributeString(gtprof.TraceAttrFuncCaller, c.callerInfo)
+	span.SetAttributeString(gtprof.TraceAttrGitCommand, cmdLogString)
+
+	if c.cmdTimeout <= 0 {
+		c.cmdCtx, c.cmdCancel, c.cmdFinished = process.GetManager().AddContext(ctx, desc)
+	} else {
+		c.cmdCtx, c.cmdCancel, c.cmdFinished = process.GetManager().AddContextTimeout(ctx, c.cmdTimeout, desc)
+	}
+
+	c.cmdStartTime = time.Now()
+
+	c.cmd = process.CommandContext(c.cmdCtx, c.prog, append(c.configArgs, c.args...)...)
+	if c.cmdEnv == nil {
+		c.cmd.Env = os.Environ()
+	} else {
+		c.cmd.Env = c.cmdEnv
+	}
+
+	c.cmd.Env = append(c.cmd.Env, CommonGitCmdEnvs()...)
+	c.cmd.Dir = c.gitDir
+	c.cmd.Stdout = c.cmdStdout
+	c.cmd.Stdin = c.cmdStdin
+	c.cmd.Stderr = c.cmdStderr
+	c.cmd.WithOnCancelGracefully(func() error {
+		// Need to close the pipes to notify all sub processes to exit.
+		// Especially on Windows: there is no process group, and we didn't implement process job object (like process group).
+		c.closePipeFiles(c.parentPipeFiles)
+		return nil
+	})
+	return c.cmd.Start()
+}
+
+func (c *Command) closePipeFiles(files []*os.File) {
+	for _, f := range files {
+		_ = f.Close()
+	}
+}
+
+func (c *Command) discardPipeReaders(files []*os.File) {
+	for _, f := range files {
+		_, _ = io.Copy(io.Discard, f)
+	}
+}
+
+func (c *Command) Wait() error {
+	defer func() {
+		// The reader in another goroutine might be still reading the stdout, so we shouldn't close the pipes here
+		// MakeStdoutPipe returns a closer function to force callers to close the pipe correctly
+		// Here we only need to mark the command as finished
+		c.cmdFinished()
+	}()
+
+	if c.pipelineFunc != nil {
+		errPipeline := c.pipelineFunc(&cmdContext{Context: c.cmdCtx, cmd: c})
+
+		if context.Cause(c.cmdCtx) == nil {
+			// if the context is not canceled explicitly, we need to discard the unread data,
+			// and wait for the command to exit normally, and then get its exit code
+			c.discardPipeReaders(c.parentPipeReaders)
+		} // else: canceled command will be killed, and the exit code is caused by kill
+
+		// after the pipeline function returns, we can safely close the pipes, then wait for the command to exit
+		c.closePipeFiles(c.parentPipeFiles)
+		errWait := c.cmd.Wait()
+		errCause := context.Cause(c.cmdCtx) // in case the cause is set during Wait(), get the final cancel cause
+
+		if unwrapped, ok := UnwrapPipelineError(errCause); ok {
+			if unwrapped != errPipeline {
+				panic("unwrapped context pipeline error should be the same one returned by pipeline function")
+			}
+			if unwrapped == nil {
+				// the pipeline function declares that there is no error, and it cancels (kills) the command ahead,
+				// so we should ignore the errors from "wait" and "cause"
+				errWait, errCause = nil, nil
+			}
+		}
+
+		// some legacy code still need to access the error returned by pipeline function by "==" but not "errors.Is"
+		// so we need to make sure the original error is able to be unwrapped by UnwrapPipelineError
+		return errors.Join(wrapPipelineError(errPipeline), errCause, errWait)
+	}
+
+	// there might be other goroutines using the context or pipes, so we just wait for the command to finish
+	errWait := c.cmd.Wait()
+	elapsed := time.Since(c.cmdStartTime)
+	if elapsed > time.Second {
+		log.Debug("slow git.Command.Run: %s (%s)", c, elapsed) // TODO: no need to log this for long-running commands
+	}
+
+	// Here the logic is different from "PipelineFunc" case,
+	// because PipelineFunc can return error if it fails, it knows whether it succeeds or fails.
+	// But in normal case, the caller just runs the git command, the command's exit code is the source of truth.
+	// If the caller need to know whether the command error is caused by cancellation, it should check the "err" by itself.
+	errCause := context.Cause(c.cmdCtx)
+	return errors.Join(errCause, errWait)
+}
+
+func (c *Command) StartWithStderr(ctx context.Context) RunStdError {
+	if c.cmdStderr != nil {
+		panic("caller-provided stderr receiver doesn't work with managed stderr buffer")
+	}
+	c.cmdManagedStderr = &bytes.Buffer{}
+	c.cmdStderr = c.cmdManagedStderr
+	err := c.WithParentCallerInfo().Start(ctx)
+	if err != nil {
+		return &runStdError{err: err}
+	}
+	return nil
+}
+
+func (c *Command) WaitWithStderr() RunStdError {
+	if c.cmdManagedStderr == nil {
+		panic("managed stderr buffer is not initialized")
+	}
+	errWait := c.Wait()
+	if errWait == nil {
+		// if no exec error but only stderr output, the stderr output is still saved in "c.cmdManagedStderr" and can be read later
+		return nil
+	}
+	return NewRunStdError(errWait, util.UnsafeBytesToString(c.cmdManagedStderr.Bytes()))
+}
+
+func (c *Command) RunWithStderr(ctx context.Context) RunStdError {
+	if err := c.WithParentCallerInfo().StartWithStderr(ctx); err != nil {
+		return &runStdError{err: err}
+	}
+	return c.WaitWithStderr()
+}
+
+func (c *Command) Run(ctx context.Context) (err error) {
+	if err = c.WithParentCallerInfo().Start(ctx); err != nil {
+		return err
+	}
+	return c.Wait()
+}
+
+// RunStdString runs the command and returns stdout/stderr as string. and store stderr to returned error (err combined with stderr).
+func (c *Command) RunStdString(ctx context.Context) (stdout, stderr string, runErr RunStdError) {
+	stdoutBytes, stderrBytes, runErr := c.WithParentCallerInfo().runStdBytes(ctx)
+	return util.UnsafeBytesToString(stdoutBytes), util.UnsafeBytesToString(stderrBytes), runErr
+}
+
+// RunStdBytes runs the command and returns stdout/stderr as bytes. and store stderr to returned error (err combined with stderr).
+func (c *Command) RunStdBytes(ctx context.Context) (stdout, stderr []byte, runErr RunStdError) {
+	return c.WithParentCallerInfo().runStdBytes(ctx)
+}
+
+func (c *Command) runStdBytes(ctx context.Context) ([]byte, []byte, RunStdError) {
+	if c.cmdStdout != nil || c.cmdStderr != nil {
+		// it must panic here, otherwise there would be bugs if developers set other Stdin/Stderr by mistake, and it would be very difficult to debug
+		panic("stdout and stderr field must be nil when using RunStdBytes")
+	}
+	stdoutBuf := &bytes.Buffer{}
+	err := c.WithStdoutBuffer(stdoutBuf).RunWithStderr(ctx)
+	return stdoutBuf.Bytes(), c.cmdManagedStderr.Bytes(), err
+}
+
+func (c *Command) DebugKill() {
+	_ = c.cmd.Process.Kill()
+}

modules/git/gitcmd/command_test.go

@@ -0,0 +1,142 @@
+// Copyright 2022 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"strings"
+	"testing"
+	"time"
+
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/tempdir"
+	"gitea.dev/modules/testlogger"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func testMain(m *testing.M) int {
+	// FIXME: GIT-PACKAGE-DEPENDENCY: the dependency is not right.
+	// "setting.Git.HomePath" is initialized in "git" package but really used in "gitcmd" package
+	gitHomePath, cleanup, err := tempdir.OsTempDir("gitea-test").MkdirTempRandom("git-home")
+	if err != nil {
+		return testlogger.MainErrorf("failed to create temp dir: %v", err)
+	}
+	defer cleanup()
+
+	setting.Git.HomePath = gitHomePath
+	return m.Run()
+}
+
+func TestMain(m *testing.M) {
+	os.Exit(testMain(m))
+}
+
+func TestRunWithContextStd(t *testing.T) {
+	{
+		cmd := NewCommand("--version")
+		stdout, stderr, err := cmd.RunStdString(t.Context())
+		assert.NoError(t, err)
+		assert.Empty(t, stderr)
+		assert.Contains(t, stdout, "git version")
+	}
+
+	{
+		cmd := NewCommand("ls-tree", "no-such")
+		stdout, stderr, err := cmd.RunStdString(t.Context())
+		if assert.Error(t, err) {
+			assert.Equal(t, stderr, err.Stderr())
+			stderrLower := strings.ToLower(stderr) // see: IsStdErrorNotValidObjectName
+			assert.Equal(t, "fatal: not a valid object name no-such\n", stderrLower)
+			// FIXME: GIT-CMD-STDERR: it is a bad design, the stderr should not be put in the error message
+			errLower := strings.ToLower(err.Error())
+			assert.Equal(t, "exit status 128 - fatal: not a valid object name no-such", errLower)
+			assert.Empty(t, stdout)
+		}
+	}
+
+	{
+		cmd := NewCommand("ls-tree", "no-such")
+		stdout, stderr, err := cmd.RunStdBytes(t.Context())
+		if assert.Error(t, err) {
+			assert.Equal(t, string(stderr), err.Stderr())
+			stderrLower := strings.ToLower(err.Stderr()) // see: IsStdErrorNotValidObjectName
+			assert.Equal(t, "fatal: not a valid object name no-such\n", stderrLower)
+			// FIXME: GIT-CMD-STDERR: it is a bad design, the stderr should not be put in the error message
+			errLower := strings.ToLower(err.Error())
+			assert.Equal(t, "exit status 128 - fatal: not a valid object name no-such", errLower)
+			assert.Empty(t, stdout)
+		}
+	}
+
+	{
+		cmd := NewCommand()
+		cmd.AddDynamicArguments("-test")
+		assert.ErrorIs(t, cmd.Run(t.Context()), ErrBrokenCommand)
+
+		cmd = NewCommand()
+		cmd.AddDynamicArguments("--test")
+		assert.ErrorIs(t, cmd.Run(t.Context()), ErrBrokenCommand)
+	}
+
+	{
+		subCmd := "version"
+		cmd := NewCommand().AddDynamicArguments(subCmd) // for test purpose only, the sub-command should never be dynamic for production
+		stdout, stderr, err := cmd.RunStdString(t.Context())
+		assert.NoError(t, err)
+		assert.Empty(t, stderr)
+		assert.Contains(t, stdout, "git version")
+	}
+}
+
+func TestGitArgument(t *testing.T) {
+	assert.True(t, isValidArgumentOption("-x"))
+	assert.True(t, isValidArgumentOption("--xx"))
+	assert.False(t, isValidArgumentOption(""))
+	assert.False(t, isValidArgumentOption("x"))
+
+	assert.True(t, isSafeArgumentValue(""))
+	assert.True(t, isSafeArgumentValue("x"))
+	assert.False(t, isSafeArgumentValue("-x"))
+}
+
+func TestCommandString(t *testing.T) {
+	cmd := NewCommand("a", "-m msg", "it's a test", `say "hello"`)
+	assert.Equal(t, cmd.prog+` a "-m msg" "it's a test" "say \"hello\""`, cmd.LogString())
+
+	cmd = NewCommand("url: https://a:b@c/", "/root/dir-a/dir-b")
+	assert.Equal(t, cmd.prog+` "url: https://(masked)@c/" .../dir-a/dir-b`, cmd.LogString())
+
+	cmd = NewCommand("url: a:b@c/", "/root/dir-a/dir-b")
+	assert.Equal(t, cmd.prog+` "url: (masked)@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)
+}
+
+func TestRunWithContextTimeout(t *testing.T) {
+	t.Run("NoTimeout", func(t *testing.T) {
+		// 'git --version' does not block so it must be finished before the timeout triggered.
+		err := NewCommand("--version").Run(t.Context())
+		require.NoError(t, err)
+	})
+	t.Run("WithTimeout", func(t *testing.T) {
+		cmd := NewCommand("hash-object", "--stdin")
+		_, _, pipeClose := cmd.MakeStdinStdoutPipe()
+		defer pipeClose()
+		err := cmd.WithTimeout(1 * time.Millisecond).Run(t.Context())
+		require.ErrorIs(t, err, context.DeadlineExceeded)
+	})
+}

modules/git/gitcmd/context.go

@@ -0,0 +1,32 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"context"
+)
+
+type Context interface {
+	context.Context
+
+	// CancelPipeline is a helper function to cancel the command context (kill the command) with a specific error cause,
+	// it returns the same error for convenience to break the PipelineFunc easily
+	CancelPipeline(err error) error
+
+	// In the future, this interface will be extended to support stdio pipe readers/writers
+}
+
+type cmdContext struct {
+	context.Context
+	cmd *Command
+}
+
+func (c *cmdContext) CancelPipeline(err error) error {
+	// pipelineError is used to distinguish between:
+	// * context canceled by pipeline caller with/without error (normal cancellation)
+	// * context canceled by parent context (still context.Canceled error)
+	// * other causes
+	c.cmd.cmdCancel(pipelineError{err})
+	return err
+}

modules/git/gitcmd/env.go

@@ -0,0 +1,40 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"fmt"
+	"os/exec"
+
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/setting"
+)
+
+var GitExecutable = "git" // the command name of git, will be updated to an absolute path during initialization
+
+// SetExecutablePath changes the path of git executable and checks the file permission and version.
+func SetExecutablePath(path string) error {
+	// If path is empty, we use the default value of GitExecutable "git" to search for the location of git.
+	if path != "" {
+		GitExecutable = path
+	}
+	absPath, err := exec.LookPath(GitExecutable)
+	if err != nil {
+		return fmt.Errorf("git not found: %w", err)
+	}
+	GitExecutable = absPath
+	return nil
+}
+
+// HomeDir is the home dir for git to store the global config file used by Gitea internally
+func HomeDir() string {
+	if setting.Git.HomePath == "" {
+		// strict check, make sure the git module is initialized correctly.
+		// attention: when the git module is called in gitea sub-command (serv/hook), the log module might not obviously show messages to users/developers.
+		// for example: if there is gitea git hook code calling NewCommand before git.InitXxx, the integration test won't show the real failure reasons.
+		log.Fatal("Unable to init Git's HomeDir, incorrect initialization of the setting and git modules")
+		return ""
+	}
+	return setting.Git.HomePath
+}

modules/git/gitcmd/error.go

@@ -0,0 +1,166 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"os/exec"
+	"strings"
+
+	"gitea.dev/modules/regexplru"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/util"
+)
+
+type RunStdError interface {
+	error
+	Unwrap() error
+	Stderr() string
+}
+
+type runStdError struct {
+	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 {
+	// FIXME: GIT-CMD-STDERR: it is a bad design, the stderr should not be put in the error message
+	// But a lot of code only checks `strings.Contains(err.Error(), "git error")`
+	if r.errMsg == "" {
+		r.errMsg = fmt.Sprintf("%s - %s", r.err.Error(), strings.TrimSpace(r.stderr))
+	}
+	return r.errMsg
+}
+
+func (r *runStdError) Unwrap() error {
+	return r.err
+}
+
+func (r *runStdError) Stderr() string {
+	return r.stderr
+}
+
+func NewRunStdError(err error, stderr string) RunStdError {
+	return &runStdError{err: err, stderr: util.NormalizeStringEOL(stderr)}
+}
+
+func ErrorAsStderr(err error) (string, bool) {
+	if runErr, ok := errors.AsType[RunStdError](err); ok {
+		return runErr.Stderr(), true
+	}
+	return "", false
+}
+
+func IsErrorExitCode(err error, code int) bool {
+	if exitError, ok := errors.AsType[*exec.ExitError](err); ok {
+		return exitError.ExitCode() == code
+	}
+	return false
+}
+
+func IsErrorSignalKilled(err error) bool {
+	var exitError *exec.ExitError
+	return errors.As(err, &exitError) && exitError.String() == "signal: killed"
+}
+
+func IsErrorCanceledOrKilled(err error) bool {
+	// When "cancel()" a git command's context, the returned error of "Run()" could be one of them:
+	// - context.Canceled
+	// - *exec.ExitError: "signal: killed"
+	// TODO: in the future, we need to use unified error type from gitcmd.Run to check whether it is manually canceled
+	return errors.Is(err, context.Canceled) || IsErrorSignalKilled(err)
+}
+
+type StderrCheck interface {
+	internalOnly()
+}
+
+type (
+	StderrPrefix string
+	StderrRegexp string
+)
+
+func (StderrPrefix) internalOnly() {}
+func (StderrRegexp) internalOnly() {}
+
+const (
+	StderrNotValidObjectName StderrPrefix = "fatal: not a valid object name"
+	StderrNotTreeObject      StderrPrefix = "fatal: not a tree object"
+	StderrPathSpec           StderrPrefix = "fatal: pathspec"
+	StderrBadRevision        StderrPrefix = "fatal: bad revision"
+	StderrNoSuchPath         StderrPrefix = "fatal: no such path"
+
+	StderrNoSuchRemote1 StderrPrefix = "fatal: no such remote" // git < 2.30, exit status 128
+	StderrNoSuchRemote2 StderrPrefix = "error: no such remote" // git >= 2.30. exit status 2
+
+	StderrAuthenticationFailed StderrPrefix = "fatal: Authentication failed for"
+	StderrCouldNotReadUsername StderrPrefix = "fatal: could not read Username"
+
+	StderrUnknownRevisionOrPath StderrRegexp = "^fatal: .*: unknown revision or path not in the working tree"
+	StderrNoMergeBase           StderrRegexp = "^fatal: .*: no merge base"
+	StderrFileNoEnoughLines     StderrRegexp = `^fatal: file .* has only \d+ lines?`
+)
+
+func matchStderrCheck(stderr string, checkIntf StderrCheck) (match bool) {
+	switch check := any(checkIntf).(type) {
+	case StderrPrefix:
+		checkLen := len(check)
+		if len(stderr) >= checkLen {
+			// Git is lowercasing the "fatal: Not a valid object name" error message
+			// ref: https://lore.kernel.org/git/pull.2052.git.1771836302101.gitgitgadget@gmail.com
+			match = util.AsciiEqualFold(stderr[:checkLen], string(check))
+		}
+	case StderrRegexp:
+		re, err := regexplru.SystemCache().GetCompiled(string(check))
+		if err != nil {
+			setting.PanicInDevOrTesting("invalid stderr regexp %s", check)
+		} else {
+			match = re.MatchString(stderr)
+		}
+	default:
+		setting.PanicInDevOrTesting("invalid stderr type %T", checkIntf)
+	}
+	return match
+}
+
+func IsStderr(err error, checks ...StderrCheck) bool {
+	stderr, ok := ErrorAsStderr(err)
+	if !ok {
+		return false
+	}
+
+	for line := range strings.SplitSeq(stderr, "\n") { // git can emit multiple-line message in stderr
+		for _, checkIntf := range checks {
+			if matchStderrCheck(line, checkIntf) {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+type pipelineError struct {
+	error
+}
+
+func (e pipelineError) Unwrap() error {
+	return e.error
+}
+
+func wrapPipelineError(err error) error {
+	if err == nil {
+		return nil
+	}
+	return pipelineError{err}
+}
+
+func UnwrapPipelineError(err error) (error, bool) { //nolint:revive // this is for error unwrapping
+	if pe, ok := errors.AsType[pipelineError](err); ok {
+		return pe.error, true
+	}
+	return nil, false
+}

modules/git/gitcmd/error_test.go

@@ -0,0 +1,26 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestIsStderr(t *testing.T) {
+	cases := []struct {
+		check  StderrCheck
+		stderr string
+	}{
+		{StderrUnknownRevisionOrPath, "fatal: ambiguous argument 'origin': unknown revision or path not in the working tree...."},
+		{StderrNoMergeBase, "fatal: origin/main..HEAD: no merge base...."},
+		{StderrFileNoEnoughLines, "fatal: file foo/bar has only 1 line"},
+		{StderrAuthenticationFailed, "Cloning into 'repo'...\r\nremote: Invalid username or token.\r\nfatal: Authentication failed for 'https://host/repo.git/'\r\n"},
+	}
+	for _, tc := range cases {
+		assert.True(t, IsStderr(&runStdError{stderr: tc.stderr}, tc.check), "stderr: %s", tc.stderr)
+	}
+	assert.False(t, IsStderr(&runStdError{stderr: "remote: fatal: Authentication failed for 'https://host/repo.git/'\n"}, StderrAuthenticationFailed))
+}

modules/git/gitcmd/pipe.go

@@ -0,0 +1,88 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitcmd
+
+import (
+	"io"
+	"os"
+)
+
+type PipeBufferReader interface {
+	Read(p []byte) (n int, err error)
+	Bytes() []byte
+}
+
+type PipeBufferWriter interface {
+	Write(p []byte) (n int, err error)
+	Bytes() []byte
+}
+
+// PipeReader should be used in the same goroutine as command's Wait
+// When Reader in one goroutine, command's Wait in another goroutine, then the command exits, the pipe will be closed:
+// * If the Reader goroutine reads faster, it will read all remaining data and then get io.EOF
+//   - But this io.EOF doesn't mean the Reader has gotten complete data, the data might still be corrupted
+//
+// * If the Reader goroutine reads slower, it will get os.ErrClosed because the os.Pipe is closed ahead when the command exits
+//
+// When using 2 goroutines, no clear solution to distinguish these two cases or make Reader knows whether the data is complete
+// It should avoid using Reader in a different goroutine than the command if the Read error needs to be handled.
+type PipeReader interface {
+	io.ReadCloser
+	internalOnly()
+}
+
+type pipeReader struct {
+	f *os.File
+}
+
+func (r *pipeReader) internalOnly() {}
+
+func (r *pipeReader) Read(p []byte) (n int, err error) {
+	return r.f.Read(p)
+}
+
+func (r *pipeReader) Close() error {
+	return r.f.Close()
+}
+
+type PipeWriter interface {
+	io.WriteCloser
+	internalOnly()
+}
+
+type pipeWriter struct {
+	f *os.File
+}
+
+func (w *pipeWriter) internalOnly() {}
+
+func (w *pipeWriter) Close() error {
+	return w.f.Close()
+}
+
+func (w *pipeWriter) Write(p []byte) (n int, err error) {
+	return w.f.Write(p)
+}
+
+func (w *pipeWriter) DrainBeforeClose() error {
+	return nil
+}
+
+type pipeNull struct {
+	err error
+}
+
+func (p *pipeNull) internalOnly() {}
+
+func (p *pipeNull) Read([]byte) (n int, err error) {
+	return 0, p.err
+}
+
+func (p *pipeNull) Write([]byte) (n int, err error) {
+	return 0, p.err
+}
+
+func (p *pipeNull) Close() error {
+	return nil
+}

modules/git/gitrepo.go

@@ -0,0 +1,51 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"io"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/reqctx"
+	"gitea.dev/modules/util"
+)
+
+// contextKey is a value for use with context.WithValue.
+type contextKey struct {
+	key string
+}
+
+// RepositoryFromContextOrOpen attempts to get the repository from the context or just opens it
+// The caller must call Closer.Close()
+func RepositoryFromContextOrOpen(ctx context.Context, repo RepositoryFacade) (*Repository, io.Closer, error) {
+	reqCtx := reqctx.FromContext(ctx)
+	if reqCtx != nil {
+		gitRepo, err := RepositoryFromRequestContextOrOpen(reqCtx, repo)
+		return gitRepo, util.NopCloser{}, err
+	}
+	gitRepo, err := OpenRepository(ctx, repo)
+	return gitRepo, gitRepo, err
+}
+
+// RepositoryFromRequestContextOrOpen opens the repository at the given relative path in the provided request context.
+// Caller shouldn't close the git repo manually, the git repo will be automatically closed when the request context is done.
+func RepositoryFromRequestContextOrOpen(ctx reqctx.RequestContext, repo RepositoryFacade) (*Repository, error) {
+	ck := contextKey{key: repo.GitRepoLocation()}
+	if gitRepo, ok := ctx.Value(ck).(*Repository); ok {
+		return gitRepo, nil
+	}
+	gitRepo, err := OpenRepository(ctx, repo)
+	if err != nil {
+		return nil, err
+	}
+	ctx.AddCloser(gitRepo)
+	ctx.SetContextValue(ck, gitRepo)
+	return gitRepo, nil
+}
+
+func UpdateServerInfo(ctx context.Context, repo RepositoryFacade) error {
+	_, _, err := gitcmd.NewCommand("update-server-info").WithRepo(repo).RunStdBytes(ctx)
+	return err
+}

modules/git/gitrepo/dbrepo.go

@@ -0,0 +1,28 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitrepo
+
+import (
+	"strings"
+
+	"gitea.dev/modules/util"
+)
+
+func RepoCodeGitRepoRelativePath(ownerName, repoName string) string {
+	return util.PathJoinRelX(strings.ToLower(ownerName), strings.ToLower(repoName)+".git")
+}
+
+func RepoWikiGitRepoRelativePath(ownerName, repoName string) string {
+	return util.PathJoinRelX(strings.ToLower(ownerName), strings.ToLower(repoName)+".wiki.git")
+}
+
+// CodeRepoByName returns an unmanaged repository facade for the code repository of the given owner and repository name.
+// Usually it is used for migration fixes or repository adoption/creation/rename/transfer.
+func CodeRepoByName(ownerName, repoName string) RepositoryFacade {
+	return RepositoryUnmanaged(RepoCodeGitRepoRelativePath(ownerName, repoName))
+}
+
+func WikiRepoByName(ownerName, repoName string) RepositoryFacade {
+	return RepositoryUnmanaged(RepoWikiGitRepoRelativePath(ownerName, repoName))
+}

modules/git/gitrepo/gitrepo.go

@@ -0,0 +1,116 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package gitrepo
+
+import (
+	"io/fs"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync/atomic"
+
+	"gitea.dev/modules/setting"
+)
+
+type RepositoryFacade interface {
+	// GitRepoManagedID returns a "managed id", which should be a cache-key-friendly string.
+	// e.g.: ID with prefix&suffix or UUID
+	GitRepoManagedID() string
+
+	// GitRepoLocation returns the location for the git repository.
+	// * relative path: will be converted to an absolute path by setting.RepoRootPath
+	// * absolute path: will be used as-is
+	// * in the future: maybe URI for more flexible definitions
+	GitRepoLocation() string
+
+	LogString() string
+}
+
+// RepoLocalPath returns an absolute path for a RepositoryFacade.
+// TODO: most of the calls to this function should be replaced with a "Repo FS" in the future
+// to handle file accesses in the git repo (e.g.: read, write, list, remove).
+func RepoLocalPath(repo RepositoryFacade) string {
+	repoLoc := repo.GitRepoLocation()
+	if filepath.IsAbs(repoLoc) {
+		return repoLoc
+	}
+	if setting.RepoRootPath == "" {
+		panic("repo root path is not initialized")
+	}
+	// the repo root path and the repo loc should all have been cleaned, so we can safely join them together
+	return setting.RepoRootPath + string(filepath.Separator) + filepath.FromSlash(repoLoc)
+}
+
+func UserLocalPath(userName string) string {
+	if setting.RepoRootPath == "" {
+		panic("repo root path is not initialized")
+	}
+	return filepath.Join(setting.RepoRootPath, filepath.Clean(strings.ToLower(userName)))
+}
+
+func repoLogNameByLocation(loc string) string {
+	t := filepath.FromSlash(loc)
+	// hide the parent paths, then the name should be safe for end users
+	return ".../" + filepath.Base(filepath.Dir(t)) + "/" + filepath.Base(t)
+}
+
+type repositoryUnmanaged struct {
+	loc     string
+	logName atomic.Pointer[string]
+}
+
+func (r *repositoryUnmanaged) LogString() string {
+	s := r.logName.Load()
+	if s == nil {
+		s = new(repoLogNameByLocation(r.loc))
+		r.logName.Store(s)
+	}
+	return *s
+}
+
+func (r *repositoryUnmanaged) GitRepoManagedID() string {
+	panic("this repo is not managed by Gitea, can't be used in this managed context")
+}
+
+func (r *repositoryUnmanaged) GitRepoLocation() string {
+	return r.loc
+}
+
+// RepositoryUnmanaged returns a RepositoryFacade for a repository that might not be managed by Gitea.
+// If the path is not absolute, then it is relative to setting.RepoRootPath
+// This function is mainly for maintaining the owner's repo when the repo is not managed yet.
+// e.g.: init, clone, transfer, rename, adopt, etc., and temp repo creation and modification.
+func RepositoryUnmanaged(s string) RepositoryFacade {
+	return &repositoryUnmanaged{loc: filepath.Clean(s)}
+}
+
+type repositoryManaged struct {
+	id, loc string
+	logName atomic.Pointer[string]
+}
+
+func (r *repositoryManaged) LogString() string {
+	s := r.logName.Load()
+	if s == nil {
+		s = new(repoLogNameByLocation(r.loc))
+		r.logName.Store(s)
+	}
+	return *s
+}
+
+func (r *repositoryManaged) GitRepoManagedID() string {
+	return r.id
+}
+
+func (r *repositoryManaged) GitRepoLocation() string {
+	return r.loc
+}
+
+func RepositoryManaged(id, loc string) RepositoryFacade {
+	return &repositoryManaged{id: id, loc: filepath.Clean(loc)}
+}
+
+func RepoLocalFS(repo RepositoryFacade) fs.FS {
+	return os.DirFS(RepoLocalPath(repo))
+}

modules/git/gpg.go

@@ -0,0 +1,78 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2017 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"strings"
+	"sync/atomic"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/process"
+	"gitea.dev/modules/util"
+)
+
+type CommitSignSettings struct {
+	Sign  bool
+	Email string
+	Name  string
+
+	Format string // default to GPG
+	KeyID  string // GPG key id or SSH key file
+
+	cachedPublicKeyContent atomic.Pointer[string]
+}
+
+func (css *CommitSignSettings) PublicKeyContent(ctx context.Context) (string, error) {
+	cached := css.cachedPublicKeyContent.Load()
+	if cached != nil {
+		return *cached, nil
+	}
+
+	if css.Format == SigningKeyFormatSSH {
+		content, err := os.ReadFile(css.KeyID)
+		if err != nil {
+			return "", fmt.Errorf("unable to read SSH public key file: %s, %w", css.KeyID, err)
+		}
+		s := string(content)
+		css.cachedPublicKeyContent.Store(&s)
+		return s, nil
+	}
+
+	content, stderr, err := process.CommandContext(ctx, "gpg", "-a", "--export", css.KeyID).OutputString()
+	if err != nil {
+		return "", fmt.Errorf("unable to get default signing key: %s, %s, %w", css.KeyID, stderr, err)
+	}
+	css.cachedPublicKeyContent.Store(&content)
+	return content, nil
+}
+
+var GlobalCommitSignSettings = util.OnceValue[*CommitSignSettings]{
+	Func: func() *CommitSignSettings {
+		ctx := context.Background()
+		css := &CommitSignSettings{}
+
+		// all errors are ignored because the keys might not exist
+		// "--type=bool" resolves a valueless "commit.gpgsign" to true
+		value, _, _ := gitcmd.NewCommand("config", "--global", "--default", "false", "--type=bool", "--get", "commit.gpgsign").RunStdString(ctx)
+		css.Sign = strings.TrimSpace(value) == "true"
+
+		signingKey, _, _ := gitcmd.NewCommand("config", "--global", "--get", "user.signingkey").RunStdString(ctx)
+		css.KeyID = strings.TrimSpace(signingKey)
+		css.Sign = css.Sign && css.KeyID != ""
+
+		format, _, _ := gitcmd.NewCommand("config", "--global", "--default", SigningKeyFormatOpenPGP, "--get", "gpg.format").RunStdString(ctx)
+		css.Format = strings.TrimSpace(format)
+
+		defaultEmail, _, _ := gitcmd.NewCommand("config", "--global", "--get", "user.email").RunStdString(ctx)
+		css.Email = strings.TrimSpace(defaultEmail)
+
+		defaultName, _, _ := gitcmd.NewCommand("config", "--global", "--get", "user.name").RunStdString(ctx)
+		css.Name = strings.TrimSpace(defaultName)
+		return css
+	},
+}

modules/git/gpg_test.go

@@ -0,0 +1,33 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestGlobalCommitSignSettings(t *testing.T) {
+	defer test.MockVariableValue(&setting.Git.HomePath)()
+	defer GlobalCommitSignSettings.Reset()
+
+	signWithGitConfig := func(gitConfig string) bool {
+		setting.Git.HomePath = t.TempDir()
+		require.NoError(t, os.WriteFile(filepath.Join(setting.Git.HomePath, ".gitconfig"), []byte(gitConfig), 0o600))
+		GlobalCommitSignSettings.Reset()
+		return GlobalCommitSignSettings.Value().Sign
+	}
+
+	assert.False(t, signWithGitConfig("[user]\n\tsigningkey = KEY\n"), "must not sign when commit.gpgsign is unset")
+	assert.True(t, signWithGitConfig("[user]\n\tsigningkey = KEY\n[commit]\n\tgpgsign = true\n"))
+	assert.True(t, signWithGitConfig("[user]\n\tsigningkey = KEY\n[commit]\n\tgpgsign\n"), "git reads a valueless commit.gpgsign as true")
+	assert.False(t, signWithGitConfig("[user]\n\tsigningkey = KEY\n[commit]\n\tgpgsign = nonsense\n"), "must not sign when git cannot parse commit.gpgsign")
+}

modules/git/grep.go

@@ -0,0 +1,140 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"context"
+	"errors"
+	"fmt"
+	"slices"
+	"strconv"
+	"strings"
+	"time"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/util"
+)
+
+type GrepResult struct {
+	Filename    string
+	LineNumbers []int
+	LineCodes   []string
+}
+
+type GrepModeType string
+
+const (
+	GrepModeExact  GrepModeType = "exact"
+	GrepModeWords  GrepModeType = "words"
+	GrepModeRegexp GrepModeType = "regexp"
+)
+
+type GrepOptions struct {
+	RefName           string
+	MaxResultLimit    int
+	ContextLineNumber int
+	GrepMode          GrepModeType
+	MaxLineLength     int // the maximum length of a line to parse, exceeding chars will be truncated
+	PathspecList      []string
+}
+
+// grepSearchTimeout is the timeout for git grep search, it should be long enough to get results
+// but not too long to cause performance issues
+const grepSearchTimeout = 30 * time.Second
+
+func GrepSearch(ctx context.Context, repo *Repository, search string, opts GrepOptions) ([]*GrepResult, error) {
+	/*
+	 The output is like this ( "^@" means \x00):
+
+	 HEAD:.air.toml
+	 6^@bin = "gitea"
+
+	 HEAD:.changelog.yml
+	 2^@repo: go-gitea/gitea
+	*/
+	var results []*GrepResult
+	cmd := gitcmd.NewCommand("grep", "--null", "--break", "--heading", "--line-number", "--full-name")
+	cmd.AddOptionValues("--context", strconv.Itoa(opts.ContextLineNumber))
+	switch opts.GrepMode {
+	case GrepModeExact:
+		cmd.AddArguments("--fixed-strings")
+		cmd.AddOptionValues("-e", strings.TrimLeft(search, "-"))
+	case GrepModeRegexp:
+		cmd.AddArguments("--perl-regexp")
+		cmd.AddOptionValues("-e", strings.TrimLeft(search, "-"))
+	default: /* words */
+		words := strings.Fields(search)
+		cmd.AddArguments("--fixed-strings", "--ignore-case")
+		for i, word := range words {
+			cmd.AddOptionValues("-e", strings.TrimLeft(word, "-"))
+			if i < len(words)-1 {
+				cmd.AddOptionValues("--and")
+			}
+		}
+	}
+	cmd.AddDynamicArguments(util.IfZero(opts.RefName, "HEAD"))
+	cmd.AddDashesAndList(opts.PathspecList...)
+	opts.MaxResultLimit = util.IfZero(opts.MaxResultLimit, 50)
+
+	stdoutReader, stdoutReaderClose := cmd.MakeStdoutPipe()
+	defer stdoutReaderClose()
+	err := cmd.WithRepo(repo).
+		WithTimeout(grepSearchTimeout).
+		WithPipelineFunc(func(ctx gitcmd.Context) error {
+			isInBlock := false
+			rd := bufio.NewReaderSize(stdoutReader, util.IfZero(opts.MaxLineLength, 16*1024))
+			var res *GrepResult
+			for {
+				lineBytes, isPrefix, err := rd.ReadLine()
+				if isPrefix {
+					lineBytes = slices.Clone(lineBytes)
+					for isPrefix && err == nil {
+						_, isPrefix, err = rd.ReadLine()
+					}
+				}
+				if len(lineBytes) == 0 && err != nil {
+					break
+				}
+				line := string(lineBytes) // the memory of lineBytes is mutable
+				if !isInBlock {
+					if _ /* ref */, filename, ok := strings.Cut(line, ":"); ok {
+						isInBlock = true
+						res = &GrepResult{Filename: filename}
+						results = append(results, res)
+					}
+					continue
+				}
+				if line == "" {
+					if len(results) >= opts.MaxResultLimit {
+						return ctx.CancelPipeline(nil)
+					}
+					isInBlock = false
+					continue
+				}
+				if line == "--" {
+					continue
+				}
+				if lineNum, lineCode, ok := strings.Cut(line, "\x00"); ok {
+					lineNumInt, _ := strconv.Atoi(lineNum)
+					res.LineNumbers = append(res.LineNumbers, lineNumInt)
+					res.LineCodes = append(res.LineCodes, lineCode)
+				}
+			}
+			return nil
+		}).
+		RunWithStderr(ctx)
+	// git grep exits by cancel (killed), usually it is caused by the limit of results
+	if gitcmd.IsErrorExitCode(err, -1) && err.Stderr() == "" {
+		return results, nil
+	}
+	// git grep exits with 1 if no results are found
+	if gitcmd.IsErrorExitCode(err, 1) && err.Stderr() == "" {
+		return nil, nil
+	}
+	if err != nil && !errors.Is(err, context.Canceled) {
+		return nil, fmt.Errorf("unable to run git grep: %w", err)
+	}
+	return results, nil
+}

modules/git/grep_test.go

@@ -0,0 +1,86 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/test"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestGrepSearch(t *testing.T) {
+	defer test.MockVariableValue(&setting.RepoRootPath, t.TempDir())()
+	repo, err := OpenRepositoryLocal(t.Context(), filepath.Join(testReposDir, "language_stats_repo"))
+	assert.NoError(t, err)
+	defer repo.Close()
+
+	res, err := GrepSearch(t.Context(), repo, "void", GrepOptions{})
+	assert.NoError(t, err)
+	assert.Equal(t, []*GrepResult{
+		{
+			Filename:    "java-hello/main.java",
+			LineNumbers: []int{3},
+			LineCodes:   []string{" public static void main(String[] args)"},
+		},
+		{
+			Filename:    "main.vendor.java",
+			LineNumbers: []int{3},
+			LineCodes:   []string{" public static void main(String[] args)"},
+		},
+	}, res)
+
+	res, err = GrepSearch(t.Context(), repo, "void", GrepOptions{PathspecList: []string{":(glob)java-hello/*"}})
+	assert.NoError(t, err)
+	assert.Equal(t, []*GrepResult{
+		{
+			Filename:    "java-hello/main.java",
+			LineNumbers: []int{3},
+			LineCodes:   []string{" public static void main(String[] args)"},
+		},
+	}, res)
+
+	res, err = GrepSearch(t.Context(), repo, "void", GrepOptions{PathspecList: []string{":(glob,exclude)java-hello/*"}})
+	assert.NoError(t, err)
+	assert.Equal(t, []*GrepResult{
+		{
+			Filename:    "main.vendor.java",
+			LineNumbers: []int{3},
+			LineCodes:   []string{" public static void main(String[] args)"},
+		},
+	}, res)
+
+	res, err = GrepSearch(t.Context(), repo, "void", GrepOptions{MaxResultLimit: 1})
+	assert.NoError(t, err)
+	assert.Equal(t, []*GrepResult{
+		{
+			Filename:    "java-hello/main.java",
+			LineNumbers: []int{3},
+			LineCodes:   []string{" public static void main(String[] args)"},
+		},
+	}, res)
+
+	res, err = GrepSearch(t.Context(), repo, "void", GrepOptions{MaxResultLimit: 1, MaxLineLength: 39})
+	assert.NoError(t, err)
+	assert.Equal(t, []*GrepResult{
+		{
+			Filename:    "java-hello/main.java",
+			LineNumbers: []int{3},
+			LineCodes:   []string{" public static void main(String[] arg"},
+		},
+	}, res)
+
+	res, err = GrepSearch(t.Context(), repo, "no-such-content", GrepOptions{})
+	assert.NoError(t, err)
+	assert.Empty(t, res)
+
+	nonExistingRepo := &Repository{RepositoryBase: RepositoryBase{repoFacade: gitrepo.RepositoryUnmanaged("no-such-git-repo")}}
+	res, err = GrepSearch(t.Context(), nonExistingRepo, "no-such-content", GrepOptions{})
+	assert.Error(t, err)
+	assert.Empty(t, res)
+}

modules/git/hook.go

@@ -0,0 +1,113 @@
+// Copyright 2015 The Gogs Authors. All rights reserved.
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"errors"
+	"os"
+	"path/filepath"
+	"slices"
+	"strings"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/util"
+)
+
+// hookNames is a list of Git server hooks' name that are supported.
+var hookNames = []string{
+	"pre-receive",
+	"update",
+	"post-receive",
+}
+
+// ErrNotValidHook error when a git hook is not valid
+var ErrNotValidHook = errors.New("not a valid Git hook")
+
+// IsValidHookName returns true if given name is a valid Git hook.
+func IsValidHookName(name string) bool {
+	return slices.Contains(hookNames, name)
+}
+
+// Hook represents a Git hook.
+type Hook struct {
+	name     string
+	IsActive bool   // Indicates whether repository has this hook.
+	Content  string // Content of hook if it's active.
+	Sample   string // Sample content from Git.
+	path     string // Hook file path.
+}
+
+// GetHook returns a Git hook by given name and repository.
+func GetHook(repo RepositoryFacade, name string) (*Hook, error) {
+	repoPath := gitrepo.RepoLocalPath(repo)
+	if !IsValidHookName(name) {
+		return nil, ErrNotValidHook
+	}
+	h := &Hook{
+		name: name,
+		path: filepath.Join(repoPath, filepath.Join("hooks", name+".d", name)),
+	}
+	if data, err := os.ReadFile(h.path); err == nil {
+		h.IsActive = true
+		h.Content = string(data)
+		return h, nil
+	} else if !os.IsNotExist(err) {
+		return nil, err
+	}
+
+	samplePath := filepath.Join(repoPath, "hooks", name+".sample")
+	if data, err := os.ReadFile(samplePath); err == nil {
+		h.Sample = string(data)
+	}
+	return h, nil
+}
+
+// Name return the name of the hook
+func (h *Hook) Name() string {
+	return h.name
+}
+
+// Update updates hook settings.
+func (h *Hook) Update() error {
+	if len(strings.TrimSpace(h.Content)) == 0 {
+		// empty content means to remove the file
+		err := util.RemoveWithRetry(h.path)
+		if err != nil && !os.IsNotExist(err) {
+			return err
+		}
+		h.IsActive = false
+		return nil
+	}
+	d := filepath.Dir(h.path)
+	if err := os.MkdirAll(d, os.ModePerm); err != nil {
+		return err
+	}
+
+	err := os.WriteFile(h.path, []byte(strings.ReplaceAll(h.Content, "\r", "")), os.ModePerm)
+	if err != nil {
+		return err
+	}
+	h.IsActive = true
+	return nil
+}
+
+// ListHooks returns a list of Git hooks of given repository.
+func ListHooks(repo RepositoryFacade) (_ []*Hook, err error) {
+	exist, err := util.IsDir(filepath.Join(gitrepo.RepoLocalPath(repo), "hooks"))
+	if err != nil {
+		return nil, err
+	} else if !exist {
+		return nil, errors.New("hooks path does not exist")
+	}
+
+	hooks := make([]*Hook, len(hookNames))
+	for i, name := range hookNames {
+		hooks[i], err = GetHook(repo, name)
+		if err != nil {
+			return nil, err
+		}
+	}
+	return hooks, nil
+}

modules/git/hooks.go

@@ -0,0 +1,245 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"os"
+	"path/filepath"
+	"runtime"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/setting"
+	"gitea.dev/modules/util"
+)
+
+func getHookTemplates() (hookNames, hookTpls, giteaHookTpls []string) {
+	hookNames = []string{"pre-receive", "update", "post-receive"}
+	hookTpls = []string{
+		// for pre-receive
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+data=$(cat)
+exitcodes=""
+hookname=$(basename $0)
+GIT_DIR=${GIT_DIR:-$(dirname $0)/..}
+
+for hook in ${GIT_DIR}/hooks/${hookname}.d/*; do
+  test -x "${hook}" && test -f "${hook}" || continue
+  echo "${data}" | "${hook}"
+  exitcodes="${exitcodes} $?"
+done
+
+for i in ${exitcodes}; do
+  [ ${i} -eq 0 ] || exit ${i}
+done
+`, setting.ScriptType),
+
+		// for update
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+exitcodes=""
+hookname=$(basename $0)
+GIT_DIR=${GIT_DIR:-$(dirname $0/..)}
+
+for hook in ${GIT_DIR}/hooks/${hookname}.d/*; do
+  test -x "${hook}" && test -f "${hook}" || continue
+  "${hook}" $1 $2 $3
+  exitcodes="${exitcodes} $?"
+done
+
+for i in ${exitcodes}; do
+  [ ${i} -eq 0 ] || exit ${i}
+done
+`, setting.ScriptType),
+
+		// for post-receive
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+data=$(cat)
+exitcodes=""
+hookname=$(basename $0)
+GIT_DIR=${GIT_DIR:-$(dirname $0)/..}
+
+for hook in ${GIT_DIR}/hooks/${hookname}.d/*; do
+  test -x "${hook}" && test -f "${hook}" || continue
+  echo "${data}" | "${hook}"
+  exitcodes="${exitcodes} $?"
+done
+
+for i in ${exitcodes}; do
+  [ ${i} -eq 0 ] || exit ${i}
+done
+`, setting.ScriptType),
+	}
+
+	giteaHookTpls = []string{
+		// for pre-receive
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+%s hook --config=%s pre-receive
+`, setting.ScriptType, util.ShellEscape(setting.AppPath), util.ShellEscape(setting.CustomConf)),
+
+		// for update
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+%s hook --config=%s update $1 $2 $3
+`, setting.ScriptType, util.ShellEscape(setting.AppPath), util.ShellEscape(setting.CustomConf)),
+
+		// for post-receive
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+%s hook --config=%s post-receive
+`, setting.ScriptType, util.ShellEscape(setting.AppPath), util.ShellEscape(setting.CustomConf)),
+	}
+
+	// although only new git (>=2.29) supports proc-receive, it's still good to create its hook, in case the user upgrades git
+	hookNames = append(hookNames, "proc-receive")
+	hookTpls = append(hookTpls,
+		fmt.Sprintf(`#!/usr/bin/env %s
+# AUTO GENERATED BY GITEA, DO NOT MODIFY
+%s hook --config=%s proc-receive
+`, setting.ScriptType, util.ShellEscape(setting.AppPath), util.ShellEscape(setting.CustomConf)))
+	giteaHookTpls = append(giteaHookTpls, "")
+
+	return hookNames, hookTpls, giteaHookTpls
+}
+
+// CreateDelegateHooks creates all the hooks scripts for the repo
+func CreateDelegateHooks(_ context.Context, repo RepositoryFacade) (err error) {
+	return createDelegateHooks(filepath.Join(gitrepo.RepoLocalPath(repo), "hooks"))
+}
+
+func createDelegateHooks(hookDir string) (err error) {
+	hookNames, hookTpls, giteaHookTpls := getHookTemplates()
+
+	for i, hookName := range hookNames {
+		oldHookPath := filepath.Join(hookDir, hookName)
+		newHookPath := filepath.Join(hookDir, hookName+".d", "gitea")
+
+		hookDDir := filepath.Join(hookDir, hookName+".d")
+		if err := os.MkdirAll(hookDDir, 0o755); err != nil {
+			return fmt.Errorf("create hooks dir '%s': %w", hookDDir, err)
+		}
+		if err := os.Chmod(hookDDir, 0o755); err != nil {
+			return fmt.Errorf("chmod hooks dir '%s': %w", hookDDir, err)
+		}
+
+		// WARNING: This will override all old server-side hooks
+		if err = util.RemoveWithRetry(oldHookPath); err != nil && !os.IsNotExist(err) {
+			return fmt.Errorf("unable to pre-remove old hook file '%s' prior to rewriting: %w ", oldHookPath, err)
+		}
+		if err = os.WriteFile(oldHookPath, []byte(hookTpls[i]), 0o755); err != nil {
+			return fmt.Errorf("write old hook file '%s': %w", oldHookPath, err)
+		}
+
+		if err = ensureExecutable(oldHookPath); err != nil {
+			return fmt.Errorf("Unable to set %s executable. Error %w", oldHookPath, err)
+		}
+
+		if err = util.RemoveWithRetry(newHookPath); err != nil && !os.IsNotExist(err) {
+			return fmt.Errorf("unable to pre-remove new hook file '%s' prior to rewriting: %w", newHookPath, err)
+		}
+		if err = os.WriteFile(newHookPath, []byte(giteaHookTpls[i]), 0o755); err != nil {
+			return fmt.Errorf("write new hook file '%s': %w", newHookPath, err)
+		}
+
+		if err = ensureExecutable(newHookPath); err != nil {
+			return fmt.Errorf("Unable to set %s executable. Error %w", oldHookPath, err)
+		}
+	}
+
+	return nil
+}
+
+func checkExecutable(filename string) bool {
+	// windows has no concept of a executable bit
+	if runtime.GOOS == "windows" {
+		return true
+	}
+	fileInfo, err := os.Stat(filename)
+	if err != nil {
+		return false
+	}
+	return (fileInfo.Mode() & 0o100) > 0
+}
+
+func ensureExecutable(filename string) error {
+	fileInfo, err := os.Stat(filename)
+	if err != nil {
+		return err
+	}
+	if (fileInfo.Mode() & 0o100) > 0 {
+		return nil
+	}
+	mode := fileInfo.Mode() | 0o100
+	return os.Chmod(filename, mode)
+}
+
+// CheckDelegateHooks checks the hooks scripts for the repo
+func CheckDelegateHooks(_ context.Context, repo RepositoryFacade) ([]string, error) {
+	return checkDelegateHooks(filepath.Join(gitrepo.RepoLocalPath(repo), "hooks"))
+}
+
+func checkDelegateHooks(hookDir string) ([]string, error) {
+	hookNames, hookTpls, giteaHookTpls := getHookTemplates()
+
+	results := make([]string, 0, 10)
+
+	for i, hookName := range hookNames {
+		oldHookPath := filepath.Join(hookDir, hookName)
+		newHookPath := filepath.Join(hookDir, hookName+".d", "gitea")
+
+		cont := false
+		isExist, err := util.IsExist(oldHookPath)
+		if err != nil {
+			results = append(results, fmt.Sprintf("unable to check if %s exists. Error: %v", oldHookPath, err))
+		}
+		if err == nil && !isExist {
+			results = append(results, fmt.Sprintf("old hook file %s does not exist", oldHookPath))
+			cont = true
+		}
+		isExist, err = util.IsExist(oldHookPath + ".d")
+		if err != nil {
+			results = append(results, fmt.Sprintf("unable to check if %s exists. Error: %v", oldHookPath+".d", err))
+		}
+		if err == nil && !isExist {
+			results = append(results, fmt.Sprintf("hooks directory %s does not exist", oldHookPath+".d"))
+			cont = true
+		}
+		isExist, err = util.IsExist(newHookPath)
+		if err != nil {
+			results = append(results, fmt.Sprintf("unable to check if %s exists. Error: %v", newHookPath, err))
+		}
+		if err == nil && !isExist {
+			results = append(results, fmt.Sprintf("new hook file %s does not exist", newHookPath))
+			cont = true
+		}
+		if cont {
+			continue
+		}
+		contents, err := os.ReadFile(oldHookPath)
+		if err != nil {
+			return results, err
+		}
+		if string(contents) != hookTpls[i] {
+			results = append(results, fmt.Sprintf("old hook file %s is out of date", oldHookPath))
+		}
+		if !checkExecutable(oldHookPath) {
+			results = append(results, fmt.Sprintf("old hook file %s is not executable", oldHookPath))
+		}
+		contents, err = os.ReadFile(newHookPath)
+		if err != nil {
+			return results, err
+		}
+		if string(contents) != giteaHookTpls[i] {
+			results = append(results, fmt.Sprintf("new hook file %s is out of date", newHookPath))
+		}
+		if !checkExecutable(newHookPath) {
+			results = append(results, fmt.Sprintf("new hook file %s is not executable", newHookPath))
+		}
+	}
+	return results, nil
+}

modules/git/hooks_test.go

@@ -0,0 +1,34 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/stretchr/testify/require"
+)
+
+func TestCreateDelegateHooksPermissions(t *testing.T) {
+	hookDir := t.TempDir()
+	existingHookDir := filepath.Join(hookDir, "post-receive.d")
+	require.NoError(t, os.MkdirAll(existingHookDir, 0o777))
+	require.NoError(t, os.Chmod(existingHookDir, 0o777))
+
+	require.NoError(t, createDelegateHooks(hookDir))
+
+	hookNames, _, _ := getHookTemplates()
+	for _, hookName := range hookNames {
+		for _, path := range []string{
+			filepath.Join(hookDir, hookName),
+			filepath.Join(hookDir, hookName+".d"),
+			filepath.Join(hookDir, hookName+".d", "gitea"),
+		} {
+			info, err := os.Stat(path)
+			require.NoError(t, err)
+			require.Equal(t, os.FileMode(0o755), info.Mode().Perm(), path)
+		}
+	}
+}

modules/git/internal/cmdarg.go

@@ -0,0 +1,9 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package internal
+
+// CmdArg represents a command argument for git command, and it will be used for the git command directly without any further processing.
+// In most cases, you should use the "AddXxx" functions to add arguments, but not use this type directly.
+// Casting a risky (user-provided) string to CmdArg would cause security issues if it's injected with a "--xxx" argument.
+type CmdArg string

modules/git/key.go

@@ -0,0 +1,55 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+
+	"gitea.dev/modules/setting"
+)
+
+// Based on https://git-scm.com/docs/git-config#Documentation/git-config.txt-gpgformat
+const (
+	SigningKeyFormatOpenPGP = "openpgp" // for GPG keys, the expected default of git cli
+	SigningKeyFormatSSH     = "ssh"
+)
+
+// SigningKey represents an instance key info which will be used to sign git commits.
+// FIXME: need to refactor it to a new name, this name conflicts with the variable names for "asymkey.GPGKey" in many places.
+type SigningKey struct {
+	KeyID  string
+	Format string
+}
+
+func (s *SigningKey) String() string {
+	// Do not expose KeyID
+	// In case the key is a file path and the struct is rendered in a template, then the server path will be exposed.
+	setting.PanicInDevOrTesting("don't call SigningKey.String() - it exposes the KeyID which might be a local file path")
+	return "SigningKey:" + s.Format
+}
+
+// GetSigningKey returns the KeyID and git Signature for the repo
+func GetSigningKey(ctx context.Context) (*SigningKey, *Signature) {
+	if setting.Repository.Signing.SigningKey == "none" {
+		return nil, nil
+	}
+
+	if setting.Repository.Signing.SigningKey == "default" || setting.Repository.Signing.SigningKey == "" {
+		commitSignSettings := GlobalCommitSignSettings.Value()
+		if !commitSignSettings.Sign {
+			return nil, nil
+		}
+		sigKey := &SigningKey{KeyID: commitSignSettings.KeyID, Format: commitSignSettings.Format}
+		sig := &Signature{Name: commitSignSettings.Name, Email: commitSignSettings.Email}
+		return sigKey, sig
+	}
+
+	if setting.Repository.Signing.SigningKey == "" {
+		return nil, nil
+	}
+
+	sigKey := &SigningKey{KeyID: setting.Repository.Signing.SigningKey, Format: setting.Repository.Signing.SigningFormat}
+	sig := &Signature{Name: setting.Repository.Signing.SigningName, Email: setting.Repository.Signing.SigningEmail}
+	return sigKey, sig
+}

modules/git/languagestats/language_stats.go

@@ -0,0 +1,65 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package languagestats
+
+import (
+	"context"
+	"strings"
+	"unicode"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/git/attribute"
+)
+
+const (
+	fileSizeLimit int64 = 16 * 1024   // 16 KiB
+	bigFileSize   int64 = 1024 * 1024 // 1 MiB
+)
+
+// mergeLanguageStats mergers language names with different cases. The name with most upper case letters is used.
+func mergeLanguageStats(stats map[string]int64) map[string]int64 {
+	names := map[string]struct {
+		uniqueName string
+		upperCount int
+	}{}
+
+	countUpper := func(s string) (count int) {
+		for _, r := range s {
+			if unicode.IsUpper(r) {
+				count++
+			}
+		}
+		return count
+	}
+
+	for name := range stats {
+		cnt := countUpper(name)
+		lower := strings.ToLower(name)
+		if cnt >= names[lower].upperCount {
+			names[lower] = struct {
+				uniqueName string
+				upperCount int
+			}{uniqueName: name, upperCount: cnt}
+		}
+	}
+
+	res := make(map[string]int64, len(names))
+	for name, num := range stats {
+		res[names[strings.ToLower(name)].uniqueName] += num
+	}
+	return res
+}
+
+// GetFileLanguage tries to get the (linguist) language of the file content
+func GetFileLanguage(ctx context.Context, gitRepo *git.Repository, treeish, treePath string) (string, error) {
+	attributesMap, err := attribute.CheckAttributes(ctx, gitRepo, treeish, attribute.CheckAttributeOpts{
+		Attributes: []string{attribute.LinguistLanguage, attribute.GitlabLanguage},
+		Filenames:  []string{treePath},
+	})
+	if err != nil {
+		return "", err
+	}
+
+	return attributesMap[treePath].GetLanguage().Value(), nil
+}

modules/git/languagestats/language_stats_get.go

@@ -0,0 +1,209 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package languagestats
+
+import (
+	"bytes"
+	"context"
+	"io"
+
+	"gitea.dev/modules/analyze"
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/git/attribute"
+	"gitea.dev/modules/log"
+	"gitea.dev/modules/optional"
+
+	"github.com/go-enry/go-enry/v2"
+)
+
+// GetLanguageStats calculates language stats for git repository at specified commit
+func GetLanguageStats(ctx context.Context, repo *git.Repository, commitID string) (map[string]int64, error) {
+	// We will feed the commit IDs in order into cat-file --batch, followed by blobs as necessary.
+	// so let's create a batch stdin and stdout
+	batch, cancel, err := repo.CatFileBatch()
+	if err != nil {
+		return nil, err
+	}
+	defer cancel()
+
+	commitInfo, batchReader, err := batch.QueryContent(commitID)
+	if err != nil {
+		return nil, err
+	}
+	if commitInfo.Type != "commit" {
+		log.Debug("Unable to get commit for: %s. Err: %v", commitID, err)
+		return nil, git.ErrNotExist{ID: commitID}
+	}
+
+	sha, err := git.NewIDFromString(commitInfo.ID)
+	if err != nil {
+		log.Debug("Unable to get commit for: %s. Err: %v", commitID, err)
+		return nil, git.ErrNotExist{ID: commitID}
+	}
+
+	commit, err := git.CommitFromReader(sha, io.LimitReader(batchReader, commitInfo.Size))
+	if err != nil {
+		log.Debug("Unable to get commit for: %s. Err: %v", commitID, err)
+		return nil, err
+	}
+	if _, err = batchReader.Discard(1); err != nil {
+		return nil, err
+	}
+
+	entries, err := commit.Tree().ListEntriesRecursiveWithSize(ctx, repo)
+	if err != nil {
+		return nil, err
+	}
+
+	checker, err := attribute.NewBatchChecker(ctx, repo, commitID, attribute.LinguistAttributes)
+	if err != nil {
+		return nil, err
+	}
+	defer checker.Close()
+
+	contentBuf := bytes.Buffer{}
+	var content []byte
+
+	// sizes contains the current calculated size of all files by language
+	sizes := make(map[string]int64)
+	// by default we will only count the sizes of programming languages or markup languages
+	// unless they are explicitly set using linguist-language
+	includedLanguage := map[string]bool{}
+	// or if there's only one language in the repository
+	firstExcludedLanguage := ""
+	firstExcludedLanguageSize := int64(0)
+
+	for _, f := range entries {
+		select {
+		case <-ctx.Done():
+			return sizes, ctx.Err()
+		default:
+		}
+
+		contentBuf.Reset()
+		content = contentBuf.Bytes()
+
+		if f.GetSize(ctx, repo) == 0 {
+			continue
+		}
+
+		isVendored := optional.None[bool]()
+		isDocumentation := optional.None[bool]()
+		isDetectable := optional.None[bool]()
+
+		attrs, err := checker.CheckPath(f.Name())
+		attrLinguistGenerated := optional.None[bool]()
+		if err == nil {
+			if isVendored = attrs.GetVendored(); isVendored.ValueOrDefault(false) {
+				continue
+			}
+
+			if attrLinguistGenerated = attrs.GetGenerated(); attrLinguistGenerated.ValueOrDefault(false) {
+				continue
+			}
+
+			if isDocumentation = attrs.GetDocumentation(); isDocumentation.ValueOrDefault(false) {
+				continue
+			}
+
+			if isDetectable = attrs.GetDetectable(); !isDetectable.ValueOrDefault(true) {
+				continue
+			}
+
+			if hasLanguage := attrs.GetLanguage(); hasLanguage.Value() != "" {
+				language := hasLanguage.Value()
+
+				if canonicalLanguage, ok := enry.GetLanguageByAlias(language); ok {
+					language = canonicalLanguage
+				}
+
+				// group languages, such as Pug -> HTML; SCSS -> CSS
+				group := enry.GetLanguageGroup(language)
+				if len(group) != 0 {
+					language = group
+				}
+
+				// this language will always be added to the size
+				sizes[language] += f.GetSize(ctx, repo)
+				continue
+			}
+		}
+
+		if (!isVendored.Has() && analyze.IsVendor(f.Name())) ||
+			enry.IsDotFile(f.Name()) ||
+			(!isDocumentation.Has() && enry.IsDocumentation(f.Name())) ||
+			(!isDetectable.Has() && enry.IsConfiguration(f.Name())) {
+			continue
+		}
+
+		// If content can not be read or file is too big just do detection by filename
+
+		if f.GetSize(ctx, repo) <= bigFileSize {
+			info, _, err := batch.QueryContent(f.ID.String())
+			if err != nil {
+				return nil, err
+			}
+
+			sizeToRead := info.Size
+			discard := int64(1)
+			if info.Size > fileSizeLimit {
+				sizeToRead = fileSizeLimit
+				discard = info.Size - fileSizeLimit + 1
+			}
+
+			_, err = contentBuf.ReadFrom(io.LimitReader(batchReader, sizeToRead))
+			if err != nil {
+				return nil, err
+			}
+			content = contentBuf.Bytes()
+			if err := git.DiscardFull(batchReader, discard); err != nil {
+				return nil, err
+			}
+		}
+
+		// if "generated" attribute is set, use it, otherwise use enry.IsGenerated to guess
+		var isGenerated bool
+		if attrLinguistGenerated.Has() {
+			isGenerated = attrLinguistGenerated.Value()
+		} else {
+			isGenerated = enry.IsGenerated(f.Name(), content)
+		}
+		if isGenerated {
+			continue
+		}
+
+		// FIXME: Why can't we split this and the IsGenerated tests to avoid reading the blob unless absolutely necessary?
+		// - eg. do the all the detection tests using filename first before reading content.
+		language := analyze.GetCodeLanguage(f.Name(), content)
+		if language == "" {
+			continue
+		}
+
+		// group languages, such as Pug -> HTML; SCSS -> CSS
+		group := enry.GetLanguageGroup(language)
+		if group != "" {
+			language = group
+		}
+
+		included, checked := includedLanguage[language]
+		if !checked {
+			langType := enry.GetLanguageType(language)
+			included = langType == enry.Programming || langType == enry.Markup
+			includedLanguage[language] = included
+		}
+		if included || isDetectable.ValueOrDefault(false) {
+			sizes[language] += f.GetSize(ctx, repo)
+		} else if len(sizes) == 0 && (firstExcludedLanguage == "" || firstExcludedLanguage == language) {
+			firstExcludedLanguage = language
+			firstExcludedLanguageSize += f.GetSize(ctx, repo)
+		}
+	}
+
+	// If there are no included languages add the first excluded language
+	if len(sizes) == 0 && firstExcludedLanguage != "" {
+		sizes[firstExcludedLanguage] = firstExcludedLanguageSize
+	}
+
+	return mergeLanguageStats(sizes), nil
+}

modules/git/languagestats/language_stats_test.go

@@ -0,0 +1,44 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package languagestats
+
+import (
+	"testing"
+
+	"gitea.dev/modules/git"
+	"gitea.dev/modules/setting"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestRepository_GetLanguageStats(t *testing.T) {
+	setting.AppDataPath = t.TempDir()
+	repoPath := "../tests/repos/language_stats_repo"
+	gitRepo, err := git.OpenRepositoryLocal(t.Context(), repoPath)
+	require.NoError(t, err)
+	defer gitRepo.Close()
+
+	stats, err := GetLanguageStats(t.Context(), gitRepo, "8fee858da5796dfb37704761701bb8e800ad9ef3")
+	require.NoError(t, err)
+
+	assert.Equal(t, map[string]int64{
+		"Python": 134,
+		"Java":   112,
+	}, stats)
+}
+
+func TestMergeLanguageStats(t *testing.T) {
+	assert.Equal(t, map[string]int64{
+		"PHP":    1,
+		"python": 10,
+		"JAVA":   700,
+	}, mergeLanguageStats(map[string]int64{
+		"PHP":    1,
+		"python": 10,
+		"Java":   100,
+		"java":   200,
+		"JAVA":   400,
+	}))
+}

modules/git/languagestats/main_test.go

@@ -0,0 +1,14 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package languagestats
+
+import (
+	"testing"
+
+	"gitea.dev/modules/git"
+)
+
+func TestMain(m *testing.M) {
+	git.RunGitTests(m)
+}

modules/git/last_commit_cache.go

@@ -0,0 +1,74 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+
+	"gitea.dev/modules/cache"
+	"gitea.dev/modules/log"
+)
+
+func getCacheKey(repo RepositoryFacade, commitID, entryPath string) string {
+	return cache.SafeCacheKey(fmt.Sprintf("git-last-commit:%s:%s", repo.GitRepoManagedID(), commitID), entryPath)
+}
+
+// LastCommitCache represents a cache to store last commit
+type LastCommitCache struct {
+	ttlFn       func() int64
+	repo        *Repository
+	commitCache map[string]*Commit
+	cache       cache.StringCache
+}
+
+// Put puts the last commit id with commit and entry path
+func (c *LastCommitCache) Put(ref, entryPath, commitID string) error {
+	log.Debug("LastCommitCache save: [%s:%s:%s]", ref, entryPath, commitID)
+	return c.cache.Put(getCacheKey(c.repo, ref, entryPath), commitID, c.ttlFn())
+}
+
+// Get gets the last commit information by commit id and entry path
+func (c *LastCommitCache) Get(ctx context.Context, ref, entryPath string) (*Commit, error) {
+	lastCommitID, ok := c.cache.Get(getCacheKey(c.repo, ref, entryPath))
+	if !ok || lastCommitID == "" {
+		return nil, nil //nolint:nilnil // return nil when cache miss
+	}
+
+	log.Debug("LastCommitCache hit level 1: [%s:%s:%s]", ref, entryPath, lastCommitID)
+	if lastCommit, ok := c.commitCache[lastCommitID]; ok {
+		log.Debug("LastCommitCache hit level 2: [%s:%s:%s]", ref, entryPath, lastCommitID)
+		return lastCommit, nil
+	}
+
+	lastCommit, err := c.repo.GetCommit(ctx, lastCommitID)
+	if err != nil {
+		return nil, err
+	}
+	if c.commitCache == nil {
+		c.commitCache = make(map[string]*Commit)
+	}
+	c.commitCache[lastCommitID] = lastCommit
+	return lastCommit, nil
+}
+
+// GetCommitByPath gets the last commit for the entry in the provided commit
+func (c *LastCommitCache) GetCommitByPath(ctx context.Context, entryCommitID ObjectID, entryPath string) (*Commit, error) {
+	entryCommitIDStr := entryCommitID.String()
+	lastCommit, err := c.Get(ctx, entryCommitIDStr, entryPath)
+	if err != nil || lastCommit != nil {
+		return lastCommit, err
+	}
+
+	lastCommit, err = c.repo.getCommitByPathWithID(ctx, entryCommitID, entryPath)
+	if err != nil {
+		return nil, err
+	}
+
+	if err := c.Put(entryCommitIDStr, entryPath, lastCommit.ID.String()); err != nil {
+		log.Error("Unable to cache %s as the last commit for %q in %s %s. Error %v", lastCommit.ID.String(), entryPath, entryCommitID, c.repo.LogString(), err)
+	}
+
+	return lastCommit, nil
+}

modules/git/last_commit_cache_gogit.go

@@ -0,0 +1,63 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build gogit
+
+package git
+
+import (
+	"context"
+
+	"github.com/go-git/go-git/v5/plumbing"
+	cgobject "github.com/go-git/go-git/v5/plumbing/object/commitgraph"
+)
+
+// CacheCommit will cache the commit from the gitRepository
+func (c *Commit) CacheCommit(ctx context.Context, gitRepo *Repository) error {
+	commitNodeIndex, closer := gitRepo.CommitNodeIndex()
+	defer closer()
+
+	index, err := commitNodeIndex.Get(plumbing.Hash(c.ID.RawValue()))
+	if err != nil {
+		return err
+	}
+
+	return c.recursiveCache(ctx, gitRepo, index, c.Tree(), "", 1)
+}
+
+func (c *Commit) recursiveCache(ctx context.Context, gitRepo *Repository, index cgobject.CommitNode, tree *Tree, treePath string, level int) error {
+	if level == 0 {
+		return nil
+	}
+
+	entries, err := tree.ListEntries(ctx, gitRepo)
+	if err != nil {
+		return err
+	}
+
+	entryPaths := make([]string, len(entries))
+	entryMap := make(map[string]*TreeEntry)
+	for i, entry := range entries {
+		entryPaths[i] = entry.Name()
+		entryMap[entry.Name()] = entry
+	}
+
+	commits, err := getLastCommitForPathsByCommitNode(ctx, gitRepo, index, treePath, entryPaths)
+	if err != nil {
+		return err
+	}
+
+	for entry := range commits {
+		if entryMap[entry].IsDir() {
+			subTree, err := tree.SubTree(ctx, gitRepo, entry)
+			if err != nil {
+				return err
+			}
+			if err := c.recursiveCache(ctx, gitRepo, index, subTree, entry, level-1); err != nil {
+				return err
+			}
+		}
+	}
+
+	return nil
+}

modules/git/last_commit_cache_nogogit.go

@@ -0,0 +1,50 @@
+// Copyright 2020 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build !gogit
+
+package git
+
+import (
+	"context"
+)
+
+// CacheCommit will cache the commit from the gitRepository
+func (c *Commit) CacheCommit(ctx context.Context, gitRepo *Repository) error {
+	return c.recursiveCache(ctx, gitRepo, c.Tree(), "", 1)
+}
+
+func (c *Commit) recursiveCache(ctx context.Context, gitRepo *Repository, tree *Tree, treePath string, level int) error {
+	if level == 0 {
+		return nil
+	}
+	entries, err := tree.ListEntries(ctx, gitRepo)
+	if err != nil {
+		return err
+	}
+
+	entryPaths := make([]string, len(entries))
+	for i, entry := range entries {
+		entryPaths[i] = entry.Name()
+	}
+
+	_, err = walkGitLog(ctx, gitRepo, c, treePath, entryPaths...)
+	if err != nil {
+		return err
+	}
+
+	for _, treeEntry := range entries {
+		// entryMap won't contain "" therefore skip this.
+		if treeEntry.IsDir() {
+			subTree, err := tree.SubTree(ctx, gitRepo, treeEntry.Name())
+			if err != nil {
+				return err
+			}
+			if err := c.recursiveCache(ctx, gitRepo, subTree, treeEntry.Name(), level-1); err != nil {
+				return err
+			}
+		}
+	}
+
+	return nil
+}

modules/git/localfs.go

@@ -0,0 +1,66 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"os"
+	"path/filepath"
+
+	"gitea.dev/modules/git/gitrepo"
+	"gitea.dev/modules/util"
+)
+
+// IsRepositoryExist returns true if the repository directory exists in the disk
+func IsRepositoryExist(ctx context.Context, repo RepositoryFacade) (bool, error) {
+	return util.IsExist(gitrepo.RepoLocalPath(repo))
+}
+
+// DeleteRepository deletes the repository directory from the disk, it will return
+// nil if the repository does not exist.
+func DeleteRepository(ctx context.Context, repo RepositoryFacade) error {
+	return util.RemoveAllWithRetry(gitrepo.RepoLocalPath(repo))
+}
+
+// RenameRepository renames a repository's name on disk
+func RenameRepository(ctx context.Context, repo, newRepo RepositoryFacade) error {
+	dstDir := gitrepo.RepoLocalPath(newRepo)
+	if err := os.MkdirAll(filepath.Dir(dstDir), os.ModePerm); err != nil {
+		return fmt.Errorf("Failed to create dir %s: %w", filepath.Dir(dstDir), err)
+	}
+
+	if err := util.RenameWithRetry(gitrepo.RepoLocalPath(repo), dstDir); err != nil {
+		return fmt.Errorf("rename repository directory: %w", err)
+	}
+	return nil
+}
+
+func InitRepository(ctx context.Context, repo RepositoryFacade, objectFormatName string) error {
+	return InitRepositoryLocal(ctx, gitrepo.RepoLocalPath(repo), true, objectFormatName)
+}
+
+func IsRepoFileExist(ctx context.Context, repo RepositoryFacade, relativeFilePath string) (bool, error) {
+	absoluteFilePath := filepath.Join(gitrepo.RepoLocalPath(repo), relativeFilePath)
+	return util.IsExist(absoluteFilePath)
+}
+
+func IsRepoDirExist(ctx context.Context, repo RepositoryFacade, relativeDirPath string) (bool, error) {
+	absoluteDirPath := filepath.Join(gitrepo.RepoLocalPath(repo), relativeDirPath)
+	return util.IsDir(absoluteDirPath)
+}
+
+func RemoveRepoFileOrDir(ctx context.Context, repo RepositoryFacade, relativeFileOrDirPath string) error {
+	absoluteFilePath := filepath.Join(gitrepo.RepoLocalPath(repo), relativeFileOrDirPath)
+	return util.RemoveWithRetry(absoluteFilePath)
+}
+
+func CreateRepoFile(ctx context.Context, repo RepositoryFacade, relativeFilePath string) (io.WriteCloser, error) {
+	absoluteFilePath := filepath.Join(gitrepo.RepoLocalPath(repo), relativeFilePath)
+	if err := os.MkdirAll(filepath.Dir(absoluteFilePath), os.ModePerm); err != nil {
+		return nil, err
+	}
+	return os.Create(absoluteFilePath)
+}

modules/git/log_name_status_nogogit.go

@@ -0,0 +1,390 @@
+// Copyright 2021 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+//go:build !gogit
+
+package git
+
+import (
+	"bufio"
+	"bytes"
+	"context"
+	"errors"
+	"io"
+	"path"
+	"sort"
+	"strings"
+
+	"gitea.dev/modules/container"
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/log"
+)
+
+// logNameStatusRepo opens git log --raw in the provided repo and returns a parser
+func logNameStatusRepo(ctx context.Context, repo RepositoryFacade, head, treepath string, paths ...string) *logNameStatusRepoParser {
+	cmd := gitcmd.NewCommand()
+	cmd.AddArguments("log", "--name-status", "-c", "--format=commit%x00%H %P%x00", "--parents", "--no-renames", "-t", "-z").AddDynamicArguments(head)
+
+	var files []string
+	if len(paths) < 70 {
+		if treepath != "" {
+			files = append(files, treepath)
+			for _, pth := range paths {
+				if pth != "" {
+					files = append(files, path.Join(treepath, pth))
+				}
+			}
+		} else {
+			for _, pth := range paths {
+				if pth != "" {
+					files = append(files, pth)
+				}
+			}
+		}
+	} else if treepath != "" {
+		files = append(files, treepath)
+	}
+	// Use the :(literal) pathspec magic to handle edge cases with files named like ":file.txt" or "*.jpg"
+	for i, file := range files {
+		files[i] = ":(literal)" + file
+	}
+	cmd.AddDashesAndList(files...)
+
+	stdoutReader, stdoutReaderClose := cmd.MakeStdoutPipe()
+	ctx, ctxCancel := context.WithCancel(ctx)
+	go func() {
+		err := cmd.WithRepo(repo).RunWithStderr(ctx)
+		if err != nil && !errors.Is(err, context.Canceled) && !errors.Is(err, context.DeadlineExceeded) {
+			log.Error("Unable to run git command %v: %v", cmd.LogString(), err)
+		}
+	}()
+
+	bufReader := bufio.NewReaderSize(stdoutReader, 32*1024)
+	return &logNameStatusRepoParser{
+		treepath: treepath,
+		paths:    paths,
+		rd:       bufReader,
+		close: func() {
+			ctxCancel()
+			stdoutReaderClose()
+		},
+	}
+}
+
+// logNameStatusRepoParser parses a git log raw output from LogRawRepo
+type logNameStatusRepoParser struct {
+	treepath string
+	paths    []string
+	next     []byte
+	buffull  bool
+	rd       *bufio.Reader
+	close    func()
+}
+
+// logNameStatusCommitData represents a commit artifact from git log raw
+type logNameStatusCommitData struct {
+	CommitID  string
+	ParentIDs []string
+	Paths     []bool
+}
+
+// walkNext returns the next LogStatusCommitData
+func (g *logNameStatusRepoParser) walkNext(treepath string, paths2ids map[string]int, changed []bool, maxpathlen int) (*logNameStatusCommitData, error) {
+	var err error
+	if len(g.next) == 0 {
+		g.buffull = false
+		g.next, err = g.rd.ReadSlice('\x00')
+		switch {
+		case errors.Is(err, bufio.ErrBufferFull):
+			g.buffull = true
+		case err != nil:
+			return nil, err
+		}
+	}
+
+	ret := logNameStatusCommitData{}
+	if bytes.Equal(g.next, []byte("commit\000")) {
+		g.next, err = g.rd.ReadSlice('\x00')
+		switch {
+		case errors.Is(err, bufio.ErrBufferFull):
+			g.buffull = true
+		case err != nil:
+			return nil, err
+		}
+	}
+
+	// Our "line" must look like: <commitid> SP (<parent> SP) * NUL
+	commitIDs := string(g.next)
+	if g.buffull {
+		more, err := g.rd.ReadString('\x00')
+		if err != nil {
+			return nil, err
+		}
+		commitIDs += more
+	}
+	commitIDs = commitIDs[:len(commitIDs)-1]
+	splitIDs := strings.Split(commitIDs, " ")
+	ret.CommitID = splitIDs[0]
+	if len(splitIDs) > 1 {
+		ret.ParentIDs = splitIDs[1:]
+	}
+
+	// now read the next "line"
+	g.buffull = false
+	g.next, err = g.rd.ReadSlice('\x00')
+	if err != nil {
+		if err == bufio.ErrBufferFull {
+			g.buffull = true
+		} else if err != io.EOF {
+			return nil, err
+		}
+	}
+
+	if err == io.EOF || !(g.next[0] == '\n' || g.next[0] == '\000') {
+		return &ret, nil
+	}
+
+	// Ok we have some changes.
+	// This line will look like: NL <fname> NUL
+	//
+	// Subsequent lines will not have the NL - so drop it here - g.bufffull must also be false at this point too.
+	if g.next[0] == '\n' {
+		g.next = g.next[1:]
+	} else {
+		g.buffull = false
+		g.next, err = g.rd.ReadSlice('\x00')
+		if err != nil {
+			if err == bufio.ErrBufferFull {
+				g.buffull = true
+			} else if err != io.EOF {
+				return nil, err
+			}
+		}
+		if len(g.next) == 0 {
+			return &ret, nil
+		}
+		if g.next[0] == '\x00' {
+			g.buffull = false
+			g.next, err = g.rd.ReadSlice('\x00')
+			if err != nil {
+				if err == bufio.ErrBufferFull {
+					g.buffull = true
+				} else if err != io.EOF {
+					return nil, err
+				}
+			}
+		}
+	}
+
+	fnameBuf := make([]byte, 4096)
+
+diffloop:
+	for {
+		if err == io.EOF || bytes.Equal(g.next, []byte("commit\000")) {
+			return &ret, nil
+		}
+		g.next, err = g.rd.ReadSlice('\x00')
+		if err != nil {
+			switch err {
+			case bufio.ErrBufferFull:
+				g.buffull = true
+			case io.EOF:
+				return &ret, nil
+			default:
+				return nil, err
+			}
+		}
+		copy(fnameBuf, g.next)
+		if len(fnameBuf) < len(g.next) {
+			fnameBuf = append(fnameBuf, g.next[len(fnameBuf):]...)
+		} else {
+			fnameBuf = fnameBuf[:len(g.next)]
+		}
+		if err != nil {
+			if err != bufio.ErrBufferFull {
+				return nil, err
+			}
+			more, err := g.rd.ReadBytes('\x00')
+			if err != nil {
+				return nil, err
+			}
+			fnameBuf = append(fnameBuf, more...)
+		}
+
+		// read the next line
+		g.buffull = false
+		g.next, err = g.rd.ReadSlice('\x00')
+		if err != nil {
+			if err == bufio.ErrBufferFull {
+				g.buffull = true
+			} else if err != io.EOF {
+				return nil, err
+			}
+		}
+
+		if treepath != "" {
+			if !bytes.HasPrefix(fnameBuf, []byte(treepath)) {
+				fnameBuf = fnameBuf[:cap(fnameBuf)]
+				continue diffloop
+			}
+		}
+		fnameBuf = fnameBuf[len(treepath) : len(fnameBuf)-1]
+		if len(fnameBuf) > maxpathlen {
+			fnameBuf = fnameBuf[:cap(fnameBuf)]
+			continue diffloop
+		}
+		if len(fnameBuf) > 0 {
+			if len(treepath) > 0 {
+				if fnameBuf[0] != '/' || bytes.IndexByte(fnameBuf[1:], '/') >= 0 {
+					fnameBuf = fnameBuf[:cap(fnameBuf)]
+					continue diffloop
+				}
+				fnameBuf = fnameBuf[1:]
+			} else if bytes.IndexByte(fnameBuf, '/') >= 0 {
+				fnameBuf = fnameBuf[:cap(fnameBuf)]
+				continue diffloop
+			}
+		}
+
+		idx, ok := paths2ids[string(fnameBuf)]
+		if !ok {
+			fnameBuf = fnameBuf[:cap(fnameBuf)]
+			continue diffloop
+		}
+		if ret.Paths == nil {
+			ret.Paths = changed
+		}
+		changed[idx] = true
+	}
+}
+
+var walkGitLogDebugBeforeNext func() // is used to simulate various edge git process cases
+
+// walkGitLog walks the git log --name-status for the head commit in the provided treepath and files
+func walkGitLog(ctx context.Context, repo *Repository, head *Commit, treepath string, paths ...string) (map[string]string, error) {
+	headRef := head.ID.String()
+	tree, err := head.SubTree(ctx, repo, treepath)
+	if err != nil {
+		return nil, err
+	}
+
+	entries, err := tree.ListEntries(ctx, repo)
+	if err != nil {
+		return nil, err
+	}
+
+	if len(paths) == 0 {
+		paths = make([]string, 0, len(entries)+1)
+		paths = append(paths, "")
+		for _, entry := range entries {
+			paths = append(paths, entry.Name())
+		}
+	} else {
+		sort.Strings(paths)
+		if paths[0] != "" {
+			paths = append([]string{""}, paths...)
+		}
+		// remove duplicates
+		for i := len(paths) - 1; i > 0; i-- {
+			if paths[i] == paths[i-1] {
+				paths = append(paths[:i-1], paths[i:]...)
+			}
+		}
+	}
+
+	path2idx := map[string]int{}
+	maxpathlen := len(treepath)
+
+	for i := range paths {
+		path2idx[paths[i]] = i
+		pthlen := len(paths[i]) + len(treepath) + 1
+		if pthlen > maxpathlen {
+			maxpathlen = pthlen
+		}
+	}
+
+	g := logNameStatusRepo(ctx, repo, head.ID.String(), treepath, paths...)
+	// don't use defer g.cancel() here as g may change its value - instead wrap in a func
+	defer func() { g.close() }()
+
+	results := make([]string, len(paths))
+	remaining := len(paths)
+	nextRestart := min((len(paths)*3)/4, 70)
+	lastEmptyParent := head.ID.String()
+	commitSinceLastEmptyParent := uint64(0)
+	commitSinceNextRestart := uint64(0)
+	parentRemaining := make(container.Set[string])
+
+	changed := make([]bool, len(paths))
+
+heaploop:
+	for {
+		if walkGitLogDebugBeforeNext != nil {
+			walkGitLogDebugBeforeNext()
+		}
+		current, err := g.walkNext(treepath, path2idx, changed, maxpathlen)
+		if ctx.Err() != nil {
+			break heaploop // context is either canceled or deadline exceeded - break the loop and return what we have so far
+		} else if errors.Is(err, io.EOF) {
+			break heaploop // reached to the end of log output
+		} else if err != nil {
+			return nil, err // other unknown errors
+		}
+		parentRemaining.Remove(current.CommitID)
+		for i, found := range current.Paths {
+			if !found {
+				continue
+			}
+			changed[i] = false
+			if results[i] == "" {
+				results[i] = current.CommitID
+				if err := repo.LastCommitCache.Put(headRef, path.Join(treepath, paths[i]), current.CommitID); err != nil {
+					return nil, err
+				}
+				delete(path2idx, paths[i])
+				remaining--
+				if results[0] == "" {
+					results[0] = current.CommitID
+					if err := repo.LastCommitCache.Put(headRef, treepath, current.CommitID); err != nil {
+						return nil, err
+					}
+					delete(path2idx, "")
+					remaining--
+				}
+			}
+		}
+
+		if remaining <= 0 {
+			break heaploop
+		}
+		commitSinceLastEmptyParent++
+		if len(parentRemaining) == 0 {
+			lastEmptyParent = current.CommitID
+			commitSinceLastEmptyParent = 0
+		}
+		if remaining <= nextRestart {
+			commitSinceNextRestart++
+			if 4*commitSinceNextRestart > 3*commitSinceLastEmptyParent {
+				remainingPaths := make([]string, 0, len(paths))
+				for i, pth := range paths {
+					if results[i] == "" {
+						remainingPaths = append(remainingPaths, pth)
+					}
+				}
+				g.close()
+				g = logNameStatusRepo(ctx, repo, lastEmptyParent, treepath, remainingPaths...)
+				parentRemaining = make(container.Set[string])
+				nextRestart = (remaining * 3) / 4
+				continue heaploop
+			}
+		}
+		parentRemaining.AddMultiple(current.ParentIDs...)
+	}
+
+	resultsMap := map[string]string{}
+	for i, pth := range paths {
+		resultsMap[pth] = results[i]
+	}
+
+	return resultsMap, nil
+}

modules/git/main_test.go

@@ -0,0 +1,25 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/git/gitrepo"
+)
+
+const testReposDir = "tests/repos/"
+
+func mockRepository(repoPath string) RepositoryFacade {
+	if !filepath.IsAbs(repoPath) {
+		// resolve repository path relative to the unit test fixture directory
+		repoPath, _ = filepath.Abs(filepath.Join(testReposDir, repoPath))
+	}
+	return gitrepo.RepositoryUnmanaged(repoPath)
+}
+
+func TestMain(m *testing.M) {
+	RunGitTests(m)
+}

modules/git/merge.go

@@ -0,0 +1,26 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"fmt"
+	"strings"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/util"
+)
+
+// MergeBase checks and returns merge base of two commits.
+func MergeBase(ctx context.Context, repo RepositoryFacade, baseCommitID, headCommitID string) (string, error) {
+	mergeBase, stderr, err := gitcmd.NewCommand("merge-base").
+		AddDashesAndList(baseCommitID, headCommitID).WithRepo(repo).RunStdString(ctx)
+	if err != nil {
+		if gitcmd.IsErrorExitCode(err, 1) && strings.TrimSpace(stderr) == "" {
+			return "", util.NewNotExistErrorf("merge-base for %s and %s doesn't exist", baseCommitID, headCommitID)
+		}
+		return "", fmt.Errorf("get merge-base of %s and %s failed: %w", baseCommitID, headCommitID, err)
+	}
+	return strings.TrimSpace(mergeBase), nil
+}

modules/git/merge_tree.go

@@ -0,0 +1,59 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bufio"
+	"context"
+	"fmt"
+
+	"gitea.dev/modules/git/gitcmd"
+	"gitea.dev/modules/util"
+)
+
+const MaxConflictedDetectFiles = 10
+
+// MergeTree performs a merge between two commits (baseRef and headRef) with an optional merge base.
+// It returns the resulting tree hash, a list of conflicted files (if any), and an error if the operation fails.
+// If there are no conflicts, the list of conflicted files will be nil.
+func MergeTree(ctx context.Context, repo RepositoryFacade, baseRef, headRef, mergeBase string) (treeID string, isErrHasConflicts bool, conflictFiles []string, _ error) {
+	cmd := gitcmd.NewCommand("merge-tree", "--write-tree", "-z", "--name-only", "--no-messages").
+		AddOptionFormat("--merge-base=%s", mergeBase).
+		AddDynamicArguments(baseRef, headRef)
+
+	stdout, stdoutClose := cmd.MakeStdoutPipe()
+	defer stdoutClose()
+	cmd.WithPipelineFunc(func(ctx gitcmd.Context) error {
+		// https://git-scm.com/docs/git-merge-tree/2.38.0#OUTPUT
+		// For a conflicted merge, the output is:
+		// <OID of toplevel tree>NUL
+		// <Conflicted file name 1>NUL
+		// <Conflicted file name 2>NUL
+		// ...
+		scanner := bufio.NewScanner(stdout)
+		scanner.Split(util.BufioScannerSplit(0))
+		for scanner.Scan() {
+			line := scanner.Text()
+			if treeID == "" { // first line is tree ID
+				treeID = line
+				continue
+			}
+			conflictFiles = append(conflictFiles, line)
+			if len(conflictFiles) >= MaxConflictedDetectFiles {
+				break
+			}
+		}
+		return scanner.Err()
+	})
+
+	err := cmd.WithRepo(repo).RunWithStderr(ctx)
+	// For a successful, non-conflicted merge, the exit status is 0. When the merge has conflicts, the exit status is 1.
+	// A merge can have conflicts without having individual files conflict
+	// https://git-scm.com/docs/git-merge-tree/2.38.0#_mistakes_to_avoid
+	isErrHasConflicts = gitcmd.IsErrorExitCode(err, 1)
+	if err == nil || isErrHasConflicts {
+		return treeID, isErrHasConflicts, conflictFiles, nil
+	}
+	return "", false, nil, fmt.Errorf("run merge-tree failed: %w", err)
+}

modules/git/merge_tree_test.go

@@ -0,0 +1,82 @@
+// Copyright 2026 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/git/gitcmd"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func prepareRepoDirRenameConflict(t *testing.T) string {
+	repoDir := filepath.Join(t.TempDir(), "repo-dir-rename-conflict.git")
+	require.NoError(t, gitcmd.NewCommand("init", "--bare").AddDynamicArguments(repoDir).Run(t.Context()))
+	stdin := `blob
+mark :1
+data 2
+b
+
+blob
+mark :2
+data 2
+c
+
+reset refs/heads/master
+commit refs/heads/master
+mark :3
+author test <test@example.com> 1769202331 -0800
+committer test <test@example.com> 1769202331 -0800
+data 2
+O
+M 100644 :1 z/b
+M 100644 :2 z/c
+
+commit refs/heads/split
+mark :4
+author test <test@example.com> 1769202336 -0800
+committer test <test@example.com> 1769202336 -0800
+data 2
+A
+from :3
+M 100644 :2 w/c
+M 100644 :1 y/b
+D z/b
+D z/c
+
+blob
+mark :5
+data 2
+d
+
+commit refs/heads/add
+mark :6
+author test <test@example.com> 1769202342 -0800
+committer test <test@example.com> 1769202342 -0800
+data 2
+B
+from :3
+M 100644 :5 z/d
+`
+	require.NoError(t, gitcmd.NewCommand("fast-import").WithDir(repoDir).WithStdinBytes([]byte(stdin)).Run(t.Context()))
+	return repoDir
+}
+
+func TestMergeTreeDirectoryRenameConflictWithoutFiles(t *testing.T) {
+	repoDir := prepareRepoDirRenameConflict(t)
+	require.DirExists(t, repoDir)
+	repo := mockRepository(repoDir)
+
+	mergeBase, err := MergeBase(t.Context(), repo, "add", "split")
+	require.NoError(t, err)
+
+	treeID, conflicted, conflictedFiles, err := MergeTree(t.Context(), repo, "add", "split", mergeBase)
+	require.NoError(t, err)
+	assert.True(t, conflicted)
+	assert.Empty(t, conflictedFiles)
+	assert.Equal(t, "5e3dd4cfc5b11e278a35b2daa83b7274175e3ab1", treeID)
+}

modules/git/notes.go

@@ -0,0 +1,89 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"context"
+	"path"
+	"strings"
+
+	"gitea.dev/modules/setting"
+)
+
+// NotesRef is the git ref where Gitea will look for git-notes data.
+// The value ("refs/notes/commits") is the default ref used by git-notes.
+const NotesRef = "refs/notes/commits"
+
+// Note stores information about a note created using git-notes.
+type Note struct {
+	refCommit *Commit
+
+	BlobMessage CommitMessage // if the blob is too large, the message will be truncated
+	BlobSize    int64
+	TreePath    string
+}
+
+// GetNote retrieves the git-notes data for a given commit.
+func GetNote(ctx context.Context, repo *Repository, commitID string) (*Note, error) {
+	noteCommit, err := repo.GetCommit(ctx, NotesRef)
+	if err != nil {
+		return nil, err
+	}
+
+	// A note for a commit is stored in a blob in the notes commit tree, with the path being the commit ID.
+	// The path can be "FullCommitID" or a fanout path like "ab/cdef...." or "ab/cd/ef.....".
+	tree := noteCommit.Tree()
+	entryName := commitID
+	var entry *TreeEntry
+	var treePathBuf strings.Builder
+	for len(entryName) > 2 {
+		entry, err = tree.GetTreeEntryByPath(ctx, repo, entryName)
+		if err == nil {
+			treePathBuf.WriteString(entryName)
+			break
+		} else if IsErrNotExist(err) {
+			fanoutDir, fanoutName := entryName[0:2], entryName[2:]
+			tree, err = tree.SubTree(ctx, repo, fanoutDir)
+			if err != nil {
+				return nil, err
+			}
+			treePathBuf.WriteString(fanoutDir)
+			treePathBuf.WriteByte('/')
+			entryName = fanoutName
+		} else {
+			return nil, err
+		}
+	}
+	if entry == nil {
+		return nil, ErrNotExist{ID: commitID}
+	}
+
+	treePath := treePathBuf.String()
+	blob := entry.Blob(repo)
+	note := &Note{TreePath: treePath, refCommit: noteCommit}
+	note.BlobMessage.MessageRaw, err = blob.GetBlobContent(ctx, setting.UI.MaxDisplayFileSize)
+	if err != nil {
+		return nil, err
+	}
+	note.BlobSize = blob.Size(ctx) // it should be called after the get blob content, then the "size" is cached
+	return note, nil
+}
+
+func GetNoteWithLastCommit(ctx context.Context, repo *Repository, commitID string) (*Note, *Commit, error) {
+	note, err := GetNote(ctx, repo, commitID)
+	if err != nil {
+		return nil, nil, err
+	}
+	parentPath, entryName := path.Split(note.TreePath)
+	parentPath = strings.Trim(parentPath, "/")
+	lastCommits, err := GetLastCommitForPaths(ctx, repo, note.refCommit, parentPath, []string{entryName})
+	if err != nil {
+		return nil, nil, err
+	}
+	lastCommit := lastCommits[entryName]
+	if lastCommit == nil {
+		return nil, nil, ErrNotExist{ID: commitID}
+	}
+	return note, lastCommit, nil
+}

modules/git/notes_test.go

@@ -0,0 +1,46 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"path/filepath"
+	"testing"
+
+	"gitea.dev/modules/git/gitrepo"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestGetNote(t *testing.T) {
+	repo, err := OpenRepositoryLocal(t.Context(), filepath.Join(testReposDir, "repo1_bare"))
+	assert.NoError(t, err)
+	defer repo.Close()
+
+	note, err := GetNote(t.Context(), repo, "95bb4d39648ee7e325106df01a621c530863a653")
+	assert.NoError(t, err)
+	assert.Equal(t, "Note contents\n", note.BlobMessage.MessageUTF8())
+	assert.EqualValues(t, len(note.BlobMessage.MessageRaw), note.BlobSize)
+
+	_, err = GetNote(t.Context(), repo, "non_existent_sha")
+	assert.ErrorAs(t, err, &ErrNotExist{})
+}
+
+func TestGetNoteNestedWithCache(t *testing.T) {
+	repoPath, _ := filepath.Abs(filepath.Join(testReposDir, "repo3_notes"))
+	repo, err := OpenRepository(t.Context(), gitrepo.RepositoryManaged("repo3_notes", repoPath))
+	assert.NoError(t, err)
+	defer repo.Close()
+
+	note, lastCommit, err := GetNoteWithLastCommit(t.Context(), repo, "ba0a96fa63532d6c5087ecef070b0250ed72fa47")
+	assert.NoError(t, err)
+	assert.Equal(t, "Note 1", note.BlobMessage.MessageUTF8())
+	assert.Equal(t, "ba0a96fa63532d6c5087ecef070b0250ed72fa47", note.TreePath)
+	assert.Equal(t, "Filip Navara", lastCommit.Author.Name)
+
+	note, lastCommit, err = GetNoteWithLastCommit(t.Context(), repo, "3e668dbfac39cbc80a9ff9c61eb565d944453ba4")
+	assert.NoError(t, err)
+	assert.Equal(t, "Note 2", note.BlobMessage.MessageUTF8())
+	assert.Equal(t, "3e/66/8dbfac39cbc80a9ff9c61eb565d944453ba4", note.TreePath)
+	assert.Equal(t, "Filip Navara", lastCommit.Author.Name)
+}

modules/git/object_format.go

@@ -0,0 +1,165 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"crypto/sha1"
+	"crypto/sha256"
+	"regexp"
+	"strconv"
+)
+
+// sha1Pattern can be used to determine if a string is a valid sha
+var sha1Pattern = regexp.MustCompile(`^[0-9a-f]{4,40}$`)
+
+// sha256Pattern can be used to determine if a string is a valid sha
+var sha256Pattern = regexp.MustCompile(`^[0-9a-f]{4,64}$`)
+
+type ObjectFormat interface {
+	// Name returns the name of the object format
+	Name() string
+	// EmptyObjectID creates a new empty ObjectID from an object format hash name
+	EmptyObjectID() ObjectID
+	// EmptyTree is the hash of an empty tree
+	EmptyTree() ObjectID
+	// FullLength is the length of the hash's hex string
+	FullLength() int
+	// IsValid returns true if the input is a valid hash
+	IsValid(input string) bool
+	// MustID creates a new ObjectID from a byte slice
+	MustID(b []byte) ObjectID
+	// ComputeHash compute the hash for a given ObjectType and content
+	ComputeHash(t ObjectType, content []byte) ObjectID
+}
+
+type Sha1ObjectFormatImpl struct{}
+
+var (
+	emptySha1ObjectID = &Sha1Hash{}
+	// emptySha1Tree: 4b825dc642cb6eb9a060e54bf8d69288fbee4904
+	emptySha1Tree = &Sha1Hash{0x4b, 0x82, 0x5d, 0xc6, 0x42, 0xcb, 0x6e, 0xb9, 0xa0, 0x60, 0xe5, 0x4b, 0xf8, 0xd6, 0x92, 0x88, 0xfb, 0xee, 0x49, 0x04}
+)
+
+func (Sha1ObjectFormatImpl) Name() string { return "sha1" }
+func (Sha1ObjectFormatImpl) EmptyObjectID() ObjectID {
+	return emptySha1ObjectID
+}
+
+func (Sha1ObjectFormatImpl) EmptyTree() ObjectID {
+	return emptySha1Tree
+}
+func (Sha1ObjectFormatImpl) FullLength() int { return 40 }
+func (Sha1ObjectFormatImpl) IsValid(input string) bool {
+	return sha1Pattern.MatchString(input)
+}
+
+func (Sha1ObjectFormatImpl) MustID(b []byte) ObjectID {
+	var id Sha1Hash
+	copy(id[0:20], b)
+	return &id
+}
+
+// ComputeHash compute the hash for a given ObjectType and content
+func (h Sha1ObjectFormatImpl) ComputeHash(t ObjectType, content []byte) ObjectID {
+	hasher := sha1.New()
+	_, _ = hasher.Write(t.Bytes())
+	_, _ = hasher.Write([]byte(" "))
+	_, _ = hasher.Write([]byte(strconv.FormatInt(int64(len(content)), 10)))
+	_, _ = hasher.Write([]byte{0})
+	_, _ = hasher.Write(content)
+	return h.MustID(hasher.Sum(nil))
+}
+
+type Sha256ObjectFormatImpl struct{}
+
+var (
+	emptySha256ObjectID = &Sha256Hash{}
+	emptySha256Tree     = &Sha256Hash{
+		0x6e, 0xf1, 0x9b, 0x41, 0x22, 0x5c, 0x53, 0x69, 0xf1, 0xc1,
+		0x04, 0xd4, 0x5d, 0x8d, 0x85, 0xef, 0xa9, 0xb0, 0x57, 0xb5,
+		0x3b, 0x14, 0xb4, 0xb9, 0xb9, 0x39, 0xdd, 0x74, 0xde, 0xcc,
+		0x53, 0x21,
+	}
+)
+
+func (Sha256ObjectFormatImpl) Name() string { return "sha256" }
+func (Sha256ObjectFormatImpl) EmptyObjectID() ObjectID {
+	return emptySha256ObjectID
+}
+
+func (Sha256ObjectFormatImpl) EmptyTree() ObjectID {
+	return emptySha256Tree
+}
+func (Sha256ObjectFormatImpl) FullLength() int { return 64 }
+func (Sha256ObjectFormatImpl) IsValid(input string) bool {
+	return sha256Pattern.MatchString(input)
+}
+
+func (Sha256ObjectFormatImpl) MustID(b []byte) ObjectID {
+	var id Sha256Hash
+	copy(id[0:32], b)
+	return &id
+}
+
+// ComputeHash compute the hash for a given ObjectType and content
+func (h Sha256ObjectFormatImpl) ComputeHash(t ObjectType, content []byte) ObjectID {
+	hasher := sha256.New()
+	_, _ = hasher.Write(t.Bytes())
+	_, _ = hasher.Write([]byte(" "))
+	_, _ = hasher.Write([]byte(strconv.FormatInt(int64(len(content)), 10)))
+	_, _ = hasher.Write([]byte{0})
+	_, _ = hasher.Write(content)
+	return h.MustID(hasher.Sum(nil))
+}
+
+type invalidObjectFormatImpl struct{}
+
+var emptyInvalidObjectID = &Sha1Hash{}
+
+func (h invalidObjectFormatImpl) Name() string {
+	return "invalid-object-format"
+}
+
+func (h invalidObjectFormatImpl) EmptyObjectID() ObjectID {
+	return emptyInvalidObjectID
+}
+
+func (h invalidObjectFormatImpl) EmptyTree() ObjectID {
+	return emptyInvalidObjectID
+}
+
+func (h invalidObjectFormatImpl) FullLength() int {
+	return len(emptyInvalidObjectID) * 2
+}
+
+func (h invalidObjectFormatImpl) IsValid(input string) bool {
+	return false
+}
+
+func (h invalidObjectFormatImpl) MustID(b []byte) ObjectID {
+	return emptyInvalidObjectID
+}
+
+func (h invalidObjectFormatImpl) ComputeHash(t ObjectType, content []byte) ObjectID {
+	return emptyInvalidObjectID
+}
+
+var (
+	Sha1ObjectFormat    ObjectFormat = Sha1ObjectFormatImpl{}
+	Sha256ObjectFormat  ObjectFormat = Sha256ObjectFormatImpl{}
+	invalidObjectFormat ObjectFormat = invalidObjectFormatImpl{}
+)
+
+func ObjectFormatFromName(name string) ObjectFormat {
+	for _, objectFormat := range DefaultFeatures().SupportedObjectFormats {
+		if name == objectFormat.Name() {
+			return objectFormat
+		}
+	}
+	return invalidObjectFormat
+}
+
+func IsValidObjectFormat(name string) bool {
+	return ObjectFormatFromName(name) != invalidObjectFormat
+}

modules/git/object_id.go

@@ -0,0 +1,106 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"bytes"
+	"encoding/hex"
+	"fmt"
+)
+
+type ObjectID interface {
+	String() string
+	RefName() RefName
+	IsZero() bool
+	RawValue() []byte
+	Type() ObjectFormat
+}
+
+type Sha1Hash [20]byte
+
+var _ ObjectID = (*Sha1Hash)(nil)
+
+func (h *Sha1Hash) String() string {
+	return hex.EncodeToString(h[:])
+}
+
+func (h *Sha1Hash) RefName() RefName {
+	return RefName(h.String())
+}
+
+func (h *Sha1Hash) IsZero() bool {
+	empty := Sha1Hash{}
+	return bytes.Equal(empty[:], h[:])
+}
+func (h *Sha1Hash) RawValue() []byte { return h[:] }
+func (*Sha1Hash) Type() ObjectFormat { return Sha1ObjectFormat }
+
+func MustIDFromString(hexHash string) ObjectID {
+	id, err := NewIDFromString(hexHash)
+	if err != nil {
+		panic(err)
+	}
+	return id
+}
+
+type Sha256Hash [32]byte
+
+var _ ObjectID = (*Sha256Hash)(nil)
+
+func (h *Sha256Hash) String() string {
+	return hex.EncodeToString(h[:])
+}
+
+func (h *Sha256Hash) RefName() RefName {
+	return RefName(h.String())
+}
+
+func (h *Sha256Hash) IsZero() bool {
+	empty := Sha256Hash{}
+	return bytes.Equal(empty[:], h[:])
+}
+func (h *Sha256Hash) RawValue() []byte { return h[:] }
+func (*Sha256Hash) Type() ObjectFormat { return Sha256ObjectFormat }
+
+func NewIDFromString(hexHash string) (ObjectID, error) {
+	var theObjectFormat ObjectFormat
+	for _, objectFormat := range DefaultFeatures().SupportedObjectFormats {
+		if len(hexHash) == objectFormat.FullLength() {
+			theObjectFormat = objectFormat
+			break
+		}
+	}
+
+	if theObjectFormat == nil {
+		return nil, fmt.Errorf("length %d has no matched object format: %s", len(hexHash), hexHash)
+	}
+
+	b, err := hex.DecodeString(hexHash)
+	if err != nil {
+		return nil, err
+	}
+
+	if len(b) != theObjectFormat.FullLength()/2 {
+		return theObjectFormat.EmptyObjectID(), fmt.Errorf("length must be %d: %v", theObjectFormat.FullLength(), b)
+	}
+	return theObjectFormat.MustID(b), nil
+}
+
+func IsEmptyCommitID(commitID string) bool {
+	if commitID == "" {
+		return true
+	}
+
+	id, err := NewIDFromString(commitID)
+	if err != nil {
+		return false
+	}
+
+	return id.IsZero()
+}
+
+// ComputeBlobHash compute the hash for a given blob content
+func ComputeBlobHash(hashType ObjectFormat, content []byte) ObjectID {
+	return hashType.ComputeHash(ObjectBlob, content)
+}

modules/git/object_id_gogit.go

@@ -0,0 +1,30 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+//go:build gogit
+
+package git
+
+import (
+	"github.com/go-git/go-git/v5/plumbing"
+	"github.com/go-git/go-git/v5/plumbing/hash"
+)
+
+func ParseGogitHash(h plumbing.Hash) ObjectID {
+	switch hash.Size {
+	case 20:
+		return Sha1ObjectFormat.MustID(h[:])
+	case 32:
+		return Sha256ObjectFormat.MustID(h[:])
+	}
+
+	return nil
+}
+
+func ParseGogitHashArray(objectIDs []plumbing.Hash) []ObjectID {
+	ret := make([]ObjectID, len(objectIDs))
+	for i, h := range objectIDs {
+		ret[i] = ParseGogitHash(h)
+	}
+
+	return ret
+}