Move blame to gitrepo (#36161)

modules/git/blame.go

@@ -1,218 +0,0 @@
-// Copyright 2019 The Gitea Authors. All rights reserved.
-// SPDX-License-Identifier: MIT
-
-package git
-
-import (
-	"bufio"
-	"bytes"
-	"context"
-	"io"
-	"os"
-
-	"code.gitea.io/gitea/modules/git/gitcmd"
-	"code.gitea.io/gitea/modules/log"
-	"code.gitea.io/gitea/modules/setting"
-)
-
-// 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 {
-	output         io.WriteCloser
-	reader         io.ReadCloser
-	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.reader.Close()
-	_ = r.output.Close()
-	for _, cleanup := range r.cleanupFuncs {
-		if cleanup != nil {
-			cleanup()
-		}
-	}
-	return err
-}
-
-// CreateBlameReader creates reader for given repository, commit and file
-func CreateBlameReader(ctx context.Context, objectFormat ObjectFormat, repoPath string, commit *Commit, file string, bypassBlameIgnore bool) (rd *BlameReader, err error) {
-	var ignoreRevsFileName string
-	var ignoreRevsFileCleanup func()
-	defer func() {
-		if err != nil && ignoreRevsFileCleanup != nil {
-			ignoreRevsFileCleanup()
-		}
-	}()
-
-	cmd := gitcmd.NewCommand("blame", "--porcelain")
-
-	if DefaultFeatures().CheckVersionAtLeast("2.23") && !bypassBlameIgnore {
-		ignoreRevsFileName, ignoreRevsFileCleanup, err = tryCreateBlameIgnoreRevsFile(commit)
-		if err != nil && !IsErrNotExist(err) {
-			return nil, err
-		}
-		if ignoreRevsFileName != "" {
-			// 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)
-
-	done := make(chan error, 1)
-	reader, stdout, err := os.Pipe()
-	if err != nil {
-		return nil, err
-	}
-	go func() {
-		stderr := bytes.Buffer{}
-		// 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"
-		err := cmd.WithDir(repoPath).
-			WithUseContextTimeout(true).
-			WithStdout(stdout).
-			WithStderr(&stderr).
-			Run(ctx)
-		done <- err
-		_ = stdout.Close()
-		if err != nil {
-			log.Error("Error running git blame (dir: %v): %v, stderr: %v", repoPath, err, stderr.String())
-		}
-	}()
-
-	bufferedReader := bufio.NewReader(reader)
-	return &BlameReader{
-		output:         stdout,
-		reader:         reader,
-		bufferedReader: bufferedReader,
-		done:           done,
-		ignoreRevsFile: ignoreRevsFileName,
-		objectFormat:   objectFormat,
-		cleanupFuncs:   []func(){ignoreRevsFileCleanup},
-	}, nil
-}
-
-func tryCreateBlameIgnoreRevsFile(commit *Commit) (string, func(), error) {
-	entry, err := commit.GetTreeEntryByPath(".git-blame-ignore-revs")
-	if err != nil {
-		return "", nil, err
-	}
-
-	r, err := entry.Blob().DataAsync()
-	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/gitrepo/blame.go

@@ -4,9 +4,16 @@
 package gitrepo
 
 import (
+	"bufio"
+	"bytes"
 	"context"
+	"io"
+	"os"
 
+	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/git/gitcmd"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/setting"
 )
 
 func LineBlame(ctx context.Context, repo Repository, revision, file string, line uint) (string, error) {
@@ -16,3 +23,204 @@ func LineBlame(ctx context.Context, repo Repository, revision, file string, line
 			AddOptionValues("-p", revision).
 			AddDashesAndList(file))
 }
+
+// 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 {
+	output         io.WriteCloser
+	reader         io.ReadCloser
+	bufferedReader *bufio.Reader
+	done           chan error
+	lastSha        *string
+	ignoreRevsFile string
+	objectFormat   git.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.reader.Close()
+	_ = r.output.Close()
+	for _, cleanup := range r.cleanupFuncs {
+		if cleanup != nil {
+			cleanup()
+		}
+	}
+	return err
+}
+
+// CreateBlameReader creates reader for given repository, commit and file
+func CreateBlameReader(ctx context.Context, objectFormat git.ObjectFormat, repo Repository, commit *git.Commit, file string, bypassBlameIgnore bool) (rd *BlameReader, err error) {
+	var ignoreRevsFileName string
+	var ignoreRevsFileCleanup func()
+	defer func() {
+		if err != nil && ignoreRevsFileCleanup != nil {
+			ignoreRevsFileCleanup()
+		}
+	}()
+
+	cmd := gitcmd.NewCommand("blame", "--porcelain")
+
+	if git.DefaultFeatures().CheckVersionAtLeast("2.23") && !bypassBlameIgnore {
+		ignoreRevsFileName, ignoreRevsFileCleanup, err = tryCreateBlameIgnoreRevsFile(commit)
+		if err != nil && !git.IsErrNotExist(err) {
+			return nil, err
+		}
+		if ignoreRevsFileName != "" {
+			// 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)
+
+	done := make(chan error, 1)
+	reader, stdout, err := os.Pipe()
+	if err != nil {
+		return nil, err
+	}
+	go func() {
+		stderr := bytes.Buffer{}
+		// 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"
+		err := RunCmd(ctx, repo, cmd.WithUseContextTimeout(true).
+			WithStdout(stdout).
+			WithStderr(&stderr),
+		)
+		done <- err
+		_ = stdout.Close()
+		if err != nil {
+			log.Error("Error running git blame (dir: %v): %v, stderr: %v", repoPath, err, stderr.String())
+		}
+	}()
+
+	bufferedReader := bufio.NewReader(reader)
+	return &BlameReader{
+		output:         stdout,
+		reader:         reader,
+		bufferedReader: bufferedReader,
+		done:           done,
+		ignoreRevsFile: ignoreRevsFileName,
+		objectFormat:   objectFormat,
+		cleanupFuncs:   []func(){ignoreRevsFileCleanup},
+	}, nil
+}
+
+func tryCreateBlameIgnoreRevsFile(commit *git.Commit) (string, func(), error) {
+	entry, err := commit.GetTreeEntryByPath(".git-blame-ignore-revs")
+	if err != nil {
+		return "", nil, err
+	}
+
+	r, err := entry.Blob().DataAsync()
+	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/gitrepo/blame_sha256_test.go

@@ -1,12 +1,13 @@
 // Copyright 2024 The Gitea Authors. All rights reserved.
 // SPDX-License-Identifier: MIT
 
-package git
+package gitrepo
 
 import (
 	"context"
 	"testing"
 
+	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/setting"
 
 	"github.com/stretchr/testify/assert"
@@ -17,13 +18,14 @@ func TestReadingBlameOutputSha256(t *testing.T) {
 	ctx, cancel := context.WithCancel(t.Context())
 	defer cancel()
 
-	if isGogit {
+	if git.DefaultFeatures().UsingGogit {
 		t.Skip("Skipping test since gogit does not support sha256")
 		return
 	}
 
 	t.Run("Without .git-blame-ignore-revs", func(t *testing.T) {
-		repo, err := OpenRepository(ctx, "./tests/repos/repo5_pulls_sha256")
+		storage := &mockRepository{path: "repo5_pulls_sha256"}
+		repo, err := OpenRepository(ctx, storage)
 		assert.NoError(t, err)
 		defer repo.Close()
 
@@ -47,7 +49,7 @@ func TestReadingBlameOutputSha256(t *testing.T) {
 		}
 
 		for _, bypass := range []bool{false, true} {
-			blameReader, err := CreateBlameReader(ctx, Sha256ObjectFormat, "./tests/repos/repo5_pulls_sha256", commit, "README.md", bypass)
+			blameReader, err := CreateBlameReader(ctx, git.Sha256ObjectFormat, storage, commit, "README.md", bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()
@@ -68,7 +70,8 @@ func TestReadingBlameOutputSha256(t *testing.T) {
 	})
 
 	t.Run("With .git-blame-ignore-revs", func(t *testing.T) {
-		repo, err := OpenRepository(ctx, "./tests/repos/repo6_blame_sha256")
+		storage := &mockRepository{path: "repo6_blame_sha256"}
+		repo, err := OpenRepository(ctx, storage)
 		assert.NoError(t, err)
 		defer repo.Close()
 
@@ -131,7 +134,7 @@ func TestReadingBlameOutputSha256(t *testing.T) {
 		for _, c := range cases {
 			commit, err := repo.GetCommit(c.CommitID)
 			assert.NoError(t, err)
-			blameReader, err := CreateBlameReader(ctx, objectFormat, "./tests/repos/repo6_blame_sha256", commit, "blame.txt", c.Bypass)
+			blameReader, err := CreateBlameReader(ctx, objectFormat, storage, commit, "blame.txt", c.Bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()

modules/gitrepo/blame_test.go

@@ -1,12 +1,13 @@
 // Copyright 2019 The Gitea Authors. All rights reserved.
 // SPDX-License-Identifier: MIT
 
-package git
+package gitrepo
 
 import (
 	"context"
 	"testing"
 
+	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/setting"
 
 	"github.com/stretchr/testify/assert"
@@ -18,10 +19,10 @@ func TestReadingBlameOutput(t *testing.T) {
 	defer cancel()
 
 	t.Run("Without .git-blame-ignore-revs", func(t *testing.T) {
-		repo, err := OpenRepository(ctx, "./tests/repos/repo5_pulls")
+		storage := &mockRepository{path: "repo5_pulls"}
+		repo, err := OpenRepository(ctx, storage)
 		assert.NoError(t, err)
 		defer repo.Close()
-
 		commit, err := repo.GetCommit("f32b0a9dfd09a60f616f29158f772cedd89942d2")
 		assert.NoError(t, err)
 
@@ -42,7 +43,7 @@ func TestReadingBlameOutput(t *testing.T) {
 		}
 
 		for _, bypass := range []bool{false, true} {
-			blameReader, err := CreateBlameReader(ctx, Sha1ObjectFormat, "./tests/repos/repo5_pulls", commit, "README.md", bypass)
+			blameReader, err := CreateBlameReader(ctx, git.Sha1ObjectFormat, storage, commit, "README.md", bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()
@@ -63,7 +64,8 @@ func TestReadingBlameOutput(t *testing.T) {
 	})
 
 	t.Run("With .git-blame-ignore-revs", func(t *testing.T) {
-		repo, err := OpenRepository(ctx, "./tests/repos/repo6_blame")
+		storage := &mockRepository{path: "repo6_blame"}
+		repo, err := OpenRepository(ctx, storage)
 		assert.NoError(t, err)
 		defer repo.Close()
 
@@ -127,7 +129,7 @@ func TestReadingBlameOutput(t *testing.T) {
 			commit, err := repo.GetCommit(c.CommitID)
 			assert.NoError(t, err)
 
-			blameReader, err := CreateBlameReader(ctx, objectFormat, "./tests/repos/repo6_blame", commit, "blame.txt", c.Bypass)
+			blameReader, err := CreateBlameReader(ctx, objectFormat, storage, commit, "blame.txt", c.Bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()

routers/web/repo/blame.go

@@ -17,6 +17,7 @@ import (
 	"code.gitea.io/gitea/modules/charset"
 	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/git/languagestats"
+	"code.gitea.io/gitea/modules/gitrepo"
 	"code.gitea.io/gitea/modules/highlight"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/setting"
@@ -99,7 +100,7 @@ func RefBlame(ctx *context.Context) {
 }
 
 type blameResult struct {
-	Parts                []*git.BlamePart
+	Parts                []*gitrepo.BlamePart
 	UsesIgnoreRevs       bool
 	FaultyIgnoreRevsFile bool
 }
@@ -107,7 +108,7 @@ type blameResult struct {
 func performBlame(ctx *context.Context, repo *repo_model.Repository, commit *git.Commit, file string, bypassBlameIgnore bool) (*blameResult, error) {
 	objectFormat := ctx.Repo.GetObjectFormat()
 
-	blameReader, err := git.CreateBlameReader(ctx, objectFormat, repo.RepoPath(), commit, file, bypassBlameIgnore)
+	blameReader, err := gitrepo.CreateBlameReader(ctx, objectFormat, repo, commit, file, bypassBlameIgnore)
 	if err != nil {
 		return nil, err
 	}
@@ -123,7 +124,7 @@ func performBlame(ctx *context.Context, repo *repo_model.Repository, commit *git
 		if len(r.Parts) == 0 && r.UsesIgnoreRevs {
 			// try again without ignored revs
 
-			blameReader, err = git.CreateBlameReader(ctx, objectFormat, repo.RepoPath(), commit, file, true)
+			blameReader, err = gitrepo.CreateBlameReader(ctx, objectFormat, repo, commit, file, true)
 			if err != nil {
 				return nil, err
 			}
@@ -143,12 +144,12 @@ func performBlame(ctx *context.Context, repo *repo_model.Repository, commit *git
 	return r, nil
 }
 
-func fillBlameResult(br *git.BlameReader, r *blameResult) error {
+func fillBlameResult(br *gitrepo.BlameReader, r *blameResult) error {
 	r.UsesIgnoreRevs = br.UsesIgnoreRevs()
 
-	previousHelper := make(map[string]*git.BlamePart)
+	previousHelper := make(map[string]*gitrepo.BlamePart)
 
-	r.Parts = make([]*git.BlamePart, 0, 5)
+	r.Parts = make([]*gitrepo.BlamePart, 0, 5)
 	for {
 		blamePart, err := br.NextPart()
 		if err != nil {
@@ -173,7 +174,7 @@ func fillBlameResult(br *git.BlameReader, r *blameResult) error {
 	return nil
 }
 
-func processBlameParts(ctx *context.Context, blameParts []*git.BlamePart) map[string]*user_model.UserCommit {
+func processBlameParts(ctx *context.Context, blameParts []*gitrepo.BlamePart) map[string]*user_model.UserCommit {
 	// store commit data by SHA to look up avatar info etc
 	commitNames := make(map[string]*user_model.UserCommit)
 	// and as blameParts can reference the same commits multiple
@@ -220,7 +221,7 @@ func processBlameParts(ctx *context.Context, blameParts []*git.BlamePart) map[st
 	return commitNames
 }
 
-func renderBlameFillFirstBlameRow(repoLink string, avatarUtils *templates.AvatarUtils, part *git.BlamePart, commit *user_model.UserCommit, br *blameRow) {
+func renderBlameFillFirstBlameRow(repoLink string, avatarUtils *templates.AvatarUtils, part *gitrepo.BlamePart, commit *user_model.UserCommit, br *blameRow) {
 	if commit.User != nil {
 		br.Avatar = avatarUtils.Avatar(commit.User, 18)
 	} else {
@@ -234,7 +235,7 @@ func renderBlameFillFirstBlameRow(repoLink string, avatarUtils *templates.Avatar
 	br.CommitSince = templates.TimeSince(commit.Author.When)
 }
 
-func renderBlame(ctx *context.Context, blameParts []*git.BlamePart, commitNames map[string]*user_model.UserCommit) {
+func renderBlame(ctx *context.Context, blameParts []*gitrepo.BlamePart, commitNames map[string]*user_model.UserCommit) {
 	language, err := languagestats.GetFileLanguage(ctx, ctx.Repo.GitRepo, ctx.Repo.CommitID, ctx.Repo.TreePath)
 	if err != nil {
 		log.Error("Unable to get file language for %-v:%s. Error: %v", ctx.Repo.Repository, ctx.Repo.TreePath, err)