diff: reimplement diff to reduce resource consumption

Makefile

@@ -5,7 +5,7 @@ SOURCE_DIR    := $(abspath $(dir $(lastword ${MAKEFILE_LIST})))
 BUILD_DIR     := ${SOURCE_DIR}/_build
 BUILD_TIME    := $(shell date +'%Y-%m-%dT%H:%M:%S%z')
 BUILD_COMMIT  := $(shell git rev-parse --short HEAD 2>/dev/null || echo 'none')
-BUILD_VERSION := $(shell cat VERSION || echo '0.14.0')
+BUILD_VERSION := $(shell cat VERSION || echo '0.15.0')
 GO_PACKAGES   := $(shell go list ./... | grep -v '^${PKG}/mock/' | grep -v '^${PKG}/proto/')
 GO_LDFLAGS    := -ldflags '-X ${PKG}/pkg/version.version=${BUILD_VERSION} -X ${PKG}/pkg/version.buildTime=${BUILD_TIME} -X ${PKG}/pkg/version.buildCommit=${BUILD_COMMIT}'
 

VERSION

@@ -1 +1 @@
-0.14.0
+0.15.0

bali.toml

@@ -3,7 +3,7 @@ name = "zeta"
 summary = "HugeSCM - A next generation cloud-based version control system"
 description = "HugeSCM - A next generation cloud-based version control system"
 package-name = "alipay-linkc-zeta"
-version = "0.14.0"
+version = "0.15.0"
 license = "MIT"
 prefix = "/usr/local"
 packager = "江二"

cmd/zeta-mc/crate.toml

@@ -1,7 +1,7 @@
 name = "zeta-mc"
 description = "zeta-mc - Migrate Git repository to zeta"
 destination = "bin"
-version = "0.14.0"
+version = "0.15.0"
 goflags = [
     "-ldflags",
     "-X github.com/antgroup/hugescm/pkg/version.version=$BUILD_VERSION -X github.com/antgroup/hugescm/pkg/version.buildTime=$BUILD_TIME -X github.com/antgroup/hugescm/pkg/version.buildCommit=$BUILD_COMMIT",

cmd/zeta/crate.toml

@@ -1,7 +1,7 @@
 name = "zeta"
 description = "HugeSCM - A next generation cloud-based version control system"
 destination = "bin"
-version = "0.14.0"
+version = "0.15.0"
 goflags = [
     "-ldflags",
     "-X github.com/antgroup/hugescm/pkg/version.version=$BUILD_VERSION -X github.com/antgroup/hugescm/pkg/version.buildTime=$BUILD_TIME -X github.com/antgroup/hugescm/pkg/version.buildCommit=$BUILD_COMMIT",

modules/diferenco/diferenco.go

@@ -131,8 +131,8 @@ type StringDiff struct {
 	Text string
 }
 
-type Stats struct {
-	Del, Ins, Hunks int
+type FileStat struct {
+	Addition, Deletion, Hunks int
 }
 
 type Options struct {
@@ -167,7 +167,7 @@ func diffInternal(ctx context.Context, L1, L2 []int, a Algorithm) ([]Change, err
 	}
 }
 
-func DoStats(ctx context.Context, opts *Options) (*Stats, error) {
+func Stat(ctx context.Context, opts *Options) (*FileStat, error) {
 	sink := &Sink{
 		Index: make(map[string]int),
 	}
@@ -183,12 +183,12 @@ func DoStats(ctx context.Context, opts *Options) (*Stats, error) {
 	if err != nil {
 		return nil, err
 	}
-	stats := &Stats{
+	stats := &FileStat{
 		Hunks: len(changes),
 	}
 	for _, ch := range changes {
-		stats.Del += ch.Del
-		stats.Ins += ch.Ins
+		stats.Addition += ch.Ins
+		stats.Deletion += ch.Del
 	}
 	return stats, nil
 }

modules/diferenco/diferenco_test.go

@@ -32,7 +32,7 @@ func TestDiff(t *testing.T) {
 		now := time.Now()
 		u, err := DoUnified(context.Background(), &Options{
 			From: &File{
-				Path: "a.txt",
+				Name: "a.txt",
 			},
 			To: nil,
 			S1: textA,
@@ -64,12 +64,12 @@ func TestPatchFD(t *testing.T) {
 	textB := string(bytesB)
 	u, err := DoUnified(context.Background(), &Options{
 		From: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "4789568",
 			Mode: 0o10644,
 		},
 		To: &File{
-			Path: "b.txt",
+			Name: "b.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -101,12 +101,12 @@ func TestPatch(t *testing.T) {
 	textB := string(bytesB)
 	u, err := DoUnified(context.Background(), &Options{
 		From: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "4789568",
 			Mode: 0o10644,
 		},
 		To: &File{
-			Path: "b.txt",
+			Name: "b.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -133,7 +133,7 @@ func TestPatchNew(t *testing.T) {
 	u, err := DoUnified(context.Background(), &Options{
 		From: nil,
 		To: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -159,7 +159,7 @@ func TestPatchDelete(t *testing.T) {
 	textA := string(bytesA)
 	u, err := DoUnified(context.Background(), &Options{
 		From: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -189,7 +189,7 @@ foo bar
 31df1778815171897c907daf454c4419cfaa46f9`
 	u, err := DoUnified(context.Background(), &Options{
 		From: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -222,12 +222,12 @@ func TestPatchScss(t *testing.T) {
 	textB := string(bytesB)
 	u, err := DoUnified(context.Background(), &Options{
 		From: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "4789568",
 			Mode: 0o10644,
 		},
 		To: &File{
-			Path: "b.txt",
+			Name: "b.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -259,12 +259,12 @@ func TestPatchCss(t *testing.T) {
 	textB := string(bytesB)
 	u, err := DoUnified(context.Background(), &Options{
 		From: &File{
-			Path: "a.txt",
+			Name: "a.txt",
 			Hash: "4789568",
 			Mode: 0o10644,
 		},
 		To: &File{
-			Path: "b.txt",
+			Name: "b.txt",
 			Hash: "6547898",
 			Mode: 0o10644,
 		},
@@ -278,3 +278,27 @@ func TestPatchCss(t *testing.T) {
 	e.SetColor(color.NewColorConfig())
 	_ = e.Encode([]*Unified{u})
 }
+
+func TestShowPatch(t *testing.T) {
+	patch := []*Unified{
+		{
+			From: &File{
+				Name: "docs/a.png",
+				Hash: "1ab12893fc666524ed79caae503e12c20a748e2f92db7730c8be09d981970f96",
+				Mode: 33188,
+			},
+			IsBinary: true,
+		},
+		{
+			To: &File{
+				Name: "images/windows7.iso",
+				Hash: "adba50d9794b9ef3f7ec8cbc680f7f1fa3fbf9df0ac8d1f9b9ccab6d941bc11b",
+				Mode: 33188,
+			},
+			IsFragments: true,
+		},
+	}
+	e := NewUnifiedEncoder(os.Stderr)
+	e.SetColor(color.NewColorConfig())
+	_ = e.Encode(patch)
+}

modules/diferenco/histogram_test.go

@@ -32,8 +32,8 @@ func TestHistogram(t *testing.T) {
 	a := sink.SplitLines(textA)
 	b := sink.SplitLines(textB)
 	changes, _ := HistogramDiff(context.Background(), a, b)
-	u := sink.ToUnified(&File{Path: "a.txt"}, &File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
+	u := sink.ToUnified(&File{Name: "a.txt"}, &File{Name: "b.txt"}, changes, a, b, DefaultContextLines)
 	e := NewUnifiedEncoder(os.Stderr)
 	e.SetColor(color.NewColorConfig())
-	e.Encode([]*Unified{u})
+	_ = e.Encode([]*Unified{u})
 }

modules/diferenco/minimal_test.go

@@ -32,8 +32,8 @@ func TestMinimalDiff(t *testing.T) {
 	a := sink.SplitLines(textA)
 	b := sink.SplitLines(textB)
 	changes, _ := MinimalDiff(context.Background(), a, b)
-	u := sink.ToUnified(&File{Path: "a.txt"}, &File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
+	u := sink.ToUnified(&File{Name: "a.txt"}, &File{Name: "b.txt"}, changes, a, b, DefaultContextLines)
 	e := NewUnifiedEncoder(os.Stderr)
 	e.SetColor(color.NewColorConfig())
-	e.Encode([]*Unified{u})
+	_ = e.Encode([]*Unified{u})
 }

modules/diferenco/myers_test.go

@@ -70,7 +70,7 @@ func TestMyersDiff2(t *testing.T) {
 	a := sink.SplitLines(textA)
 	b := sink.SplitLines(textB)
 	changes, _ := MyersDiff(context.Background(), a, b)
-	u := sink.ToUnified(&File{Path: "a.txt"}, &File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
+	u := sink.ToUnified(&File{Name: "a.txt"}, &File{Name: "b.txt"}, changes, a, b, DefaultContextLines)
 	fmt.Fprintf(os.Stderr, "diff:\n%s\n", u.String())
 }
 
@@ -91,6 +91,6 @@ func TestMyersDiff3(t *testing.T) {
 	a := sink.SplitLines(textA)
 	b := sink.SplitLines(textB)
 	changes, _ := MyersDiff(context.Background(), a, b)
-	u := sink.ToUnified(&File{Path: "a.txt"}, &File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
+	u := sink.ToUnified(&File{Name: "a.txt"}, &File{Name: "b.txt"}, changes, a, b, DefaultContextLines)
 	fmt.Fprintf(os.Stderr, "diff:\n%s\n", u.String())
 }

modules/diferenco/unified.go

@@ -14,25 +14,35 @@ import (
 const DefaultContextLines = 3
 
 type File struct {
-	Path string
-	Hash string
-	Mode uint32
+	Name string `json:"name"`
+	Hash string `json:"hash"`
+	Mode uint32 `json:"mode"`
 }
 
 // unified represents a set of edits as a unified diff.
 type Unified struct {
 	// From is the name of the original file.
-	From *File
+	From *File `json:"from,omitempty"`
 	// To is the name of the modified file.
-	To *File
+	To *File `json:"to,omitempty"`
 	// IsBinary returns true if this patch is representing a binary file.
-	IsBinary bool
+	IsBinary bool `json:"binary"`
 	// Fragments returns true if this patch is representing a fragments file.
-	IsFragments bool
+	IsFragments bool `json:"fragments"`
 	// Message prefix, eg: warning: something
-	Message string
+	Message string `json:"message"`
 	// Hunks is the set of edit Hunks needed to transform the file content.
-	Hunks []*Hunk
+	Hunks []*Hunk `json:"hunks,omitempty"`
+}
+
+func (u Unified) Stat() FileStat {
+	s := FileStat{Hunks: len(u.Hunks)}
+	for _, h := range u.Hunks {
+		ins, del := h.Stat()
+		s.Addition += ins
+		s.Deletion += del
+	}
+	return s
 }
 
 // String converts a unified diff to the standard textual form for that diff.
@@ -43,12 +53,12 @@ func (u Unified) String() string {
 	}
 	b := new(strings.Builder)
 	if u.From != nil {
-		fmt.Fprintf(b, "--- %s\n", u.From.Path)
+		fmt.Fprintf(b, "--- %s\n", u.From.Name)
 	} else {
 		fmt.Fprintf(b, "--- /dev/null\n")
 	}
 	if u.To != nil {
-		fmt.Fprintf(b, "+++ %s\n", u.To.Path)
+		fmt.Fprintf(b, "+++ %s\n", u.To.Name)
 	} else {
 		fmt.Fprintf(b, "+++ /dev/null\n")
 	}
@@ -104,16 +114,29 @@ func (u Unified) String() string {
 // Hunk represents a contiguous set of line edits to apply.
 type Hunk struct {
 	// The line in the original source where the hunk starts.
-	FromLine int
+	FromLine int `json:"from_line"`
 	// The line in the original source where the hunk finishes.
-	ToLine int
+	ToLine int `json:"to_line"`
 	// The set of line based edits to apply.
-	Lines []Line
+	Lines []Line `json:"lines,omitempty"`
+}
+
+func (h Hunk) Stat() (int, int) {
+	var ins, del int
+	for _, l := range h.Lines {
+		switch l.Kind {
+		case Delete:
+			del++
+		case Insert:
+			ins++
+		}
+	}
+	return ins, del
 }
 
 type Line struct {
-	Kind    Operation
-	Content string
+	Kind    Operation `json:"kind"`
+	Content string    `json:"content"`
 }
 
 func DoUnified(ctx context.Context, opts *Options) (*Unified, error) {