Move blame to gitrepo (#36161)

This commit is contained in:
Lunny Xiao
2025-12-16 16:14:14 -08:00
committed by GitHub
parent 84b74d7c3e
commit 4c67aac23b
5 changed files with 235 additions and 239 deletions

View File

@@ -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
}

View File

@@ -1,153 +0,0 @@
// Copyright 2024 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package git
import (
"context"
"testing"
"code.gitea.io/gitea/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 isGogit {
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")
assert.NoError(t, err)
defer repo.Close()
commit, err := repo.GetCommit("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, "./tests/repos/repo5_pulls_sha256", 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_sha256")
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()
assert.NoError(t, err)
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)
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)
}
})
}

View File

@@ -1,149 +0,0 @@
// Copyright 2019 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package git
import (
"context"
"testing"
"code.gitea.io/gitea/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) {
repo, err := OpenRepository(ctx, "./tests/repos/repo5_pulls")
assert.NoError(t, err)
defer repo.Close()
commit, err := repo.GetCommit("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, "./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{
{
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()
assert.NoError(t, err)
for _, c := range cases {
commit, err := repo.GetCommit(c.CommitID)
assert.NoError(t, err)
blameReader, err := CreateBlameReader(ctx, objectFormat, "./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)
}
})
}