Support `.git-blame-ignore-revs` file (#26395)

Closes #26329

This PR adds the ability to ignore revisions specified in the
`.git-blame-ignore-revs` file in the root of the repository.


![grafik](https://github.com/go-gitea/gitea/assets/1666336/9e91be0c-6e9c-431c-bbe9-5f80154251c8)

The banner is displayed in this case. I intentionally did not add a UI
way to bypass the ignore file (same behaviour as Github) but you can add
`?bypass-blame-ignore=true` to the url manually.

---------

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

docs/content/usage/blame.en-us.md

@@ -0,0 +1,38 @@
+---
+date: "2023-08-14T00:00:00+00:00"
+title: "Blame File View"
+slug: "blame"
+sidebar_position: 13
+toc: false
+draft: false
+aliases:
+  - /en-us/blame
+menu:
+  sidebar:
+    parent: "usage"
+    name: "Blame"
+    sidebar_position: 13
+    identifier: "blame"
+---
+
+# Blame File View
+
+Gitea supports viewing the line-by-line revision history for a file also known as blame view.
+You can also use [`git blame`](https://git-scm.com/docs/git-blame) on the command line to view the revision history of lines within a file.
+
+1. Navigate to and open the file whose line history you want to view.
+1. Click the `Blame` button in the file header bar.
+1. The new view shows the line-by-line revision history for a file with author and commit information on the left side.
+1. To navigate to an older commit, click the ![versions](/octicon-versions.svg) icon.
+
+## Ignore commits in the blame view
+
+All revisions specified in the `.git-blame-ignore-revs` file are hidden from the blame view.
+This is especially useful to hide reformatting changes and keep the benefits of `git blame`.
+Lines that were changed or added by an ignored commit will be blamed on the previous commit that changed that line or nearby lines.
+The `.git-blame-ignore-revs` file must be located in the root directory of the repository.
+For more information like the file format, see [the `git blame --ignore-revs-file` documentation](https://git-scm.com/docs/git-blame#Documentation/git-blame.txt---ignore-revs-fileltfilegt).
+
+### Bypassing `.git-blame-ignore-revs` in the blame view
+
+If the blame view for a file shows a message about ignored revisions, you can see the normal blame view by appending the url parameter `?bypass-blame-ignore=true`.

docs/static/octicon-versions.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 16 16" width="16" height="16"><path d="M7.75 14A1.75 1.75 0 0 1 6 12.25v-8.5C6 2.784 6.784 2 7.75 2h6.5c.966 0 1.75.784 1.75 1.75v8.5A1.75 1.75 0 0 1 14.25 14Zm-.25-1.75c0 .138.112.25.25.25h6.5a.25.25 0 0 0 .25-.25v-8.5a.25.25 0 0 0-.25-.25h-6.5a.25.25 0 0 0-.25.25ZM4.9 3.508a.75.75 0 0 1-.274 1.025.249.249 0 0 0-.126.217v6.5c0 .09.048.173.126.217a.75.75 0 0 1-.752 1.298A1.75 1.75 0 0 1 3 11.25v-6.5c0-.649.353-1.214.874-1.516a.75.75 0 0 1 1.025.274ZM1.625 5.533h.001a.249.249 0 0 0-.126.217v4.5c0 .09.048.173.126.217a.75.75 0 0 1-.752 1.298A1.748 1.748 0 0 1 0 10.25v-4.5a1.748 1.748 0 0 1 .873-1.516.75.75 0 1 1 .752 1.299Z"></path></svg>

modules/git/blame.go

@@ -13,6 +13,7 @@ import (
 	"regexp"
 
 	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/util"
 )
 
 // BlamePart represents block of blame - continuous lines with one sha
@@ -23,12 +24,16 @@ type BlamePart struct {
 
 // BlameReader returns part of file blame one by one
 type BlameReader struct {
-	cmd            *Command
 	output         io.WriteCloser
 	reader         io.ReadCloser
 	bufferedReader *bufio.Reader
 	done           chan error
 	lastSha        *string
+	ignoreRevsFile *string
+}
+
+func (r *BlameReader) UsesIgnoreRevs() bool {
+	return r.ignoreRevsFile != nil
 }
 
 var shaLineRegex = regexp.MustCompile("^([a-z0-9]{40})")
@@ -101,28 +106,44 @@ func (r *BlameReader) Close() error {
 	r.bufferedReader = nil
 	_ = r.reader.Close()
 	_ = r.output.Close()
+	if r.ignoreRevsFile != nil {
+		_ = util.Remove(*r.ignoreRevsFile)
+	}
 	return err
 }
 
 // CreateBlameReader creates reader for given repository, commit and file
-func CreateBlameReader(ctx context.Context, repoPath, commitID, file string) (*BlameReader, error) {
-	cmd := NewCommandContextNoGlobals(ctx, "blame", "--porcelain").
-		AddDynamicArguments(commitID).
+func CreateBlameReader(ctx context.Context, repoPath string, commit *Commit, file string, bypassBlameIgnore bool) (*BlameReader, error) {
+	var ignoreRevsFile *string
+	if CheckGitVersionAtLeast("2.23") == nil && !bypassBlameIgnore {
+		ignoreRevsFile = tryCreateBlameIgnoreRevsFile(commit)
+	}
+
+	cmd := NewCommandContextNoGlobals(ctx, "blame", "--porcelain")
+	if ignoreRevsFile != nil {
+		// 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", *ignoreRevsFile)
+	}
+	cmd.AddDynamicArguments(commit.ID.String()).
 		AddDashesAndList(file).
 		SetDescription(fmt.Sprintf("GetBlame [repo_path: %s]", repoPath))
 	reader, stdout, err := os.Pipe()
 	if err != nil {
+		if ignoreRevsFile != nil {
+			_ = util.Remove(*ignoreRevsFile)
+		}
 		return nil, err
 	}
 
 	done := make(chan error, 1)
 
-	go func(cmd *Command, dir string, stdout io.WriteCloser, done chan error) {
+	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.Run(&RunOpts{
 			UseContextTimeout: true,
-			Dir:               dir,
+			Dir:               repoPath,
 			Stdout:            stdout,
 			Stderr:            &stderr,
 		})
@@ -131,15 +152,42 @@ func CreateBlameReader(ctx context.Context, repoPath, commitID, file string) (*B
 		if err != nil {
 			log.Error("Error running git blame (dir: %v): %v, stderr: %v", repoPath, err, stderr.String())
 		}
-	}(cmd, repoPath, stdout, done)
+	}()
 
 	bufferedReader := bufio.NewReader(reader)
 
 	return &BlameReader{
-		cmd:            cmd,
 		output:         stdout,
 		reader:         reader,
 		bufferedReader: bufferedReader,
 		done:           done,
+		ignoreRevsFile: ignoreRevsFile,
 	}, nil
 }
+
+func tryCreateBlameIgnoreRevsFile(commit *Commit) *string {
+	entry, err := commit.GetTreeEntryByPath(".git-blame-ignore-revs")
+	if err != nil {
+		return nil
+	}
+
+	r, err := entry.Blob().DataAsync()
+	if err != nil {
+		return nil
+	}
+	defer r.Close()
+
+	f, err := os.CreateTemp("", "gitea_git-blame-ignore-revs")
+	if err != nil {
+		return nil
+	}
+
+	_, err = io.Copy(f, r)
+	_ = f.Close()
+	if err != nil {
+		_ = util.Remove(f.Name())
+		return nil
+	}
+
+	return util.ToPointer(f.Name())
+}

modules/git/blame_test.go

@@ -14,27 +14,127 @@ func TestReadingBlameOutput(t *testing.T) {
 	ctx, cancel := context.WithCancel(context.Background())
 	defer cancel()
 
-	blameReader, err := CreateBlameReader(ctx, "./tests/repos/repo5_pulls", "f32b0a9dfd09a60f616f29158f772cedd89942d2", "README.md")
-	assert.NoError(t, err)
-	defer blameReader.Close()
-
-	parts := []*BlamePart{
-		{
-			"72866af952e98d02a73003501836074b286a78f6",
-			[]string{
-				"# test_repo",
-				"Test repository for testing migration from github to gitea",
-			},
-		},
-		{
-			"f32b0a9dfd09a60f616f29158f772cedd89942d2",
-			[]string{"", "Do not make any changes to this repo it is used for unit testing"},
-		},
-	}
-
-	for _, part := range parts {
-		actualPart, err := blameReader.NextPart()
+	t.Run("Without .git-blame-ignore-revs", func(t *testing.T) {
+		repo, err := OpenRepository(ctx, "./tests/repos/repo5_pulls")
 		assert.NoError(t, err)
-		assert.Equal(t, part, actualPart)
-	}
+		defer repo.Close()
+
+		commit, err := repo.GetCommit("f32b0a9dfd09a60f616f29158f772cedd89942d2")
+		assert.NoError(t, err)
+
+		parts := []*BlamePart{
+			{
+				"72866af952e98d02a73003501836074b286a78f6",
+				[]string{
+					"# test_repo",
+					"Test repository for testing migration from github to gitea",
+				},
+			},
+			{
+				"f32b0a9dfd09a60f616f29158f772cedd89942d2",
+				[]string{"", "Do not make any changes to this repo it is used for unit testing"},
+			},
+		}
+
+		for _, bypass := range []bool{false, true} {
+			blameReader, err := CreateBlameReader(ctx, "./tests/repos/repo5_pulls", 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) {
+		repo, err := OpenRepository(ctx, "./tests/repos/repo6_blame")
+		assert.NoError(t, err)
+		defer repo.Close()
+
+		full := []*BlamePart{
+			{
+				"af7486bd54cfc39eea97207ca666aa69c9d6df93",
+				[]string{"line", "line"},
+			},
+			{
+				"45fb6cbc12f970b04eacd5cd4165edd11c8d7376",
+				[]string{"changed line"},
+			},
+			{
+				"af7486bd54cfc39eea97207ca666aa69c9d6df93",
+				[]string{"line", "line", ""},
+			},
+		}
+
+		cases := []struct {
+			CommitID       string
+			UsesIgnoreRevs bool
+			Bypass         bool
+			Parts          []*BlamePart
+		}{
+			{
+				CommitID:       "544d8f7a3b15927cddf2299b4b562d6ebd71b6a7",
+				UsesIgnoreRevs: true,
+				Bypass:         false,
+				Parts: []*BlamePart{
+					{
+						"af7486bd54cfc39eea97207ca666aa69c9d6df93",
+						[]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,
+			},
+		}
+
+		for _, c := range cases {
+			commit, err := repo.GetCommit(c.CommitID)
+			assert.NoError(t, err)
+
+			blameReader, err := CreateBlameReader(ctx, "./tests/repos/repo6_blame", 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/tests/repos/repo6_blame/HEAD

@@ -0,0 +1 @@
+ref: refs/heads/master

modules/git/tests/repos/repo6_blame/config

@@ -0,0 +1,4 @@
+[core]
+	repositoryformatversion = 0
+	filemode = true
+	bare = true

modules/git/tests/repos/repo6_blame/refs/heads/master

@@ -0,0 +1 @@
+544d8f7a3b15927cddf2299b4b562d6ebd71b6a7

options/locale/locale_en-US.ini

@@ -1007,6 +1007,8 @@ delete_preexisting = Delete pre-existing files
 delete_preexisting_content = Delete files in %s
 delete_preexisting_success = Deleted unadopted files in %s
 blame_prior = View blame prior to this change
+blame.ignore_revs = Ignoring revisions in <a href="%s">.git-blame-ignore-revs</a>. Click <a href="%s">here to bypass</a> and see the normal blame view.
+blame.ignore_revs.failed = Failed to ignore revisions in <a href="%s">.git-blame-ignore-revs</a>.
 author_search_tooltip = Shows a maximum of 30 users
 
 transfer.accept = Accept Transfer

routers/web/repo/blame.go

@@ -8,9 +8,9 @@ import (
 	gotemplate "html/template"
 	"net/http"
 	"net/url"
+	"strconv"
 	"strings"
 
-	repo_model "code.gitea.io/gitea/models/repo"
 	user_model "code.gitea.io/gitea/models/user"
 	"code.gitea.io/gitea/modules/charset"
 	"code.gitea.io/gitea/modules/context"
@@ -45,10 +45,6 @@ func RefBlame(ctx *context.Context) {
 		return
 	}
 
-	userName := ctx.Repo.Owner.Name
-	repoName := ctx.Repo.Repository.Name
-	commitID := ctx.Repo.CommitID
-
 	branchLink := ctx.Repo.RepoLink + "/src/" + ctx.Repo.BranchNameSubURL()
 	treeLink := branchLink
 	rawLink := ctx.Repo.RepoLink + "/raw/" + ctx.Repo.BranchNameSubURL()
@@ -101,26 +97,16 @@ func RefBlame(ctx *context.Context) {
 		return
 	}
 
-	blameReader, err := git.CreateBlameReader(ctx, repo_model.RepoPath(userName, repoName), commitID, fileName)
+	bypassBlameIgnore, _ := strconv.ParseBool(ctx.FormString("bypass-blame-ignore"))
+
+	result, err := performBlame(ctx, ctx.Repo.Repository.RepoPath(), ctx.Repo.Commit, fileName, bypassBlameIgnore)
 	if err != nil {
 		ctx.NotFound("CreateBlameReader", err)
 		return
 	}
-	defer blameReader.Close()
 
-	blameParts := make([]git.BlamePart, 0)
-
-	for {
-		blamePart, err := blameReader.NextPart()
-		if err != nil {
-			ctx.NotFound("NextPart", err)
-			return
-		}
-		if blamePart == nil {
-			break
-		}
-		blameParts = append(blameParts, *blamePart)
-	}
+	ctx.Data["UsesIgnoreRevs"] = result.UsesIgnoreRevs
+	ctx.Data["FaultyIgnoreRevsFile"] = result.FaultyIgnoreRevsFile
 
 	// Get Topics of this repo
 	renderRepoTopics(ctx)
@@ -128,16 +114,77 @@ func RefBlame(ctx *context.Context) {
 		return
 	}
 
-	commitNames, previousCommits := processBlameParts(ctx, blameParts)
+	commitNames, previousCommits := processBlameParts(ctx, result.Parts)
 	if ctx.Written() {
 		return
 	}
 
-	renderBlame(ctx, blameParts, commitNames, previousCommits)
+	renderBlame(ctx, result.Parts, commitNames, previousCommits)
 
 	ctx.HTML(http.StatusOK, tplRepoHome)
 }
 
+type blameResult struct {
+	Parts                []git.BlamePart
+	UsesIgnoreRevs       bool
+	FaultyIgnoreRevsFile bool
+}
+
+func performBlame(ctx *context.Context, repoPath string, commit *git.Commit, file string, bypassBlameIgnore bool) (*blameResult, error) {
+	blameReader, err := git.CreateBlameReader(ctx, repoPath, commit, file, bypassBlameIgnore)
+	if err != nil {
+		return nil, err
+	}
+
+	r := &blameResult{}
+	if err := fillBlameResult(blameReader, r); err != nil {
+		_ = blameReader.Close()
+		return nil, err
+	}
+
+	err = blameReader.Close()
+	if err != nil {
+		if len(r.Parts) == 0 && r.UsesIgnoreRevs {
+			// try again without ignored revs
+
+			blameReader, err = git.CreateBlameReader(ctx, repoPath, commit, file, true)
+			if err != nil {
+				return nil, err
+			}
+
+			r := &blameResult{
+				FaultyIgnoreRevsFile: true,
+			}
+			if err := fillBlameResult(blameReader, r); err != nil {
+				_ = blameReader.Close()
+				return nil, err
+			}
+
+			return r, blameReader.Close()
+		}
+		return nil, err
+	}
+	return r, nil
+}
+
+func fillBlameResult(br *git.BlameReader, r *blameResult) error {
+	r.UsesIgnoreRevs = br.UsesIgnoreRevs()
+
+	r.Parts = make([]git.BlamePart, 0, 5)
+	for {
+		blamePart, err := br.NextPart()
+		if err != nil {
+			return fmt.Errorf("BlameReader.NextPart failed: %w", err)
+		}
+		if blamePart == nil {
+			break
+		}
+		r.Parts = append(r.Parts, *blamePart)
+	}
+
+	return nil
+}
+
 func processBlameParts(ctx *context.Context, blameParts []git.BlamePart) (map[string]*user_model.UserCommit, map[string]string) {
 	// store commit data by SHA to look up avatar info etc
 	commitNames := make(map[string]*user_model.UserCommit)

templates/repo/blame.tmpl

@@ -1,3 +1,15 @@
+{{if or .UsesIgnoreRevs .FaultyIgnoreRevsFile}}
+	{{$revsFileLink := URLJoin .RepoLink "src" .BranchNameSubURL "/.git-blame-ignore-revs"}}
+	{{if .UsesIgnoreRevs}}
+		<div class="ui info message">
+			<p>{{.locale.Tr "repo.blame.ignore_revs" $revsFileLink (print $revsFileLink "?bypass-blame-ignore=true") | Str2html}}</p>
+		</div>
+	{{else}}
+		<div class="ui error message">
+			<p>{{.locale.Tr "repo.blame.ignore_revs.failed" $revsFileLink | Str2html}}</p>
+		</div>
+	{{end}}
+{{end}}
 <div class="{{TabSizeClass .Editorconfig .FileName}} non-diff-file-content">
 	<h4 class="file-header ui top attached header gt-df gt-ac gt-sb gt-fw">
 		<div class="file-header-left gt-df gt-ac gt-py-3 gt-pr-4">