diff: about unified diff encoder

modules/diferenco/diferenco_test.go

@@ -1,11 +1,14 @@
 package diferenco
 
 import (
+	"context"
 	"fmt"
 	"os"
 	"path/filepath"
 	"runtime"
 	"testing"
+
+	"github.com/antgroup/hugescm/modules/diferenco/color"
 )
 
 func TestDiff(t *testing.T) {
@@ -23,45 +26,53 @@ func TestDiff(t *testing.T) {
 		return
 	}
 	textB := string(bytesB)
-	sink := &Sink{
-		Index: make(map[string]int),
+	u, err := DoUnified(context.Background(), &Options{
+		From: &File{
+			Path: "a.txt",
+		},
+		To: nil,
+		A:  textA,
+		B:  textB,
+	})
+	if err != nil {
+		return
 	}
-	a := sink.ParseLines(textA)
-	b := sink.ParseLines(textB)
-	changes := OnpDiff(a, b)
-	i := 0
-	for _, c := range changes {
-		for ; i < c.P1; i++ {
-			fmt.Fprintf(os.Stderr, "  %s", sink.Lines[a[i]])
-		}
-		for j := c.P1; j < c.P1+c.Del; j++ {
-			fmt.Fprintf(os.Stderr, "\x1b[31m- %s\x1b[0m", sink.Lines[a[j]])
-		}
-		for j := c.P2; j < c.P2+c.Ins; j++ {
-			fmt.Fprintf(os.Stderr, "\x1b[32m+ %s\x1b[0m", sink.Lines[b[j]])
-		}
-		i += c.Del
+	fmt.Fprintf(os.Stderr, "%s\n", u)
+}
+
+func TestDiff2(t *testing.T) {
+	_, filename, _, _ := runtime.Caller(0)
+	dir := filepath.Dir(filename)
+	bytesA, err := os.ReadFile(filepath.Join(dir, "testdata/a.txt"))
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "read a error: %v\n", err)
+		return
 	}
-	for ; i < len(a); i++ {
-		fmt.Fprintf(os.Stderr, "  %s", sink.Lines[a[i]])
+	textA := string(bytesA)
+	bytesB, err := os.ReadFile(filepath.Join(dir, "testdata/b.txt"))
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "read b error: %v\n", err)
+		return
 	}
-	fmt.Fprintf(os.Stderr, "\n\npatience\n\n")
-
-	diffs := PatienceDiff(a, b)
-	for _, d := range diffs {
-		switch d.T {
-		case Delete:
-			for _, i := range d.E {
-				fmt.Fprintf(os.Stderr, "\x1b[31m-%s\x1b[0m", sink.Lines[i])
-			}
-		case Insert:
-			for _, i := range d.E {
-				fmt.Fprintf(os.Stderr, "\x1b[32m+%s\x1b[0m", sink.Lines[i])
-			}
-		default:
-			for _, i := range d.E {
-				fmt.Fprintf(os.Stderr, " %s", sink.Lines[i])
-			}
-		}
+	textB := string(bytesB)
+	u, err := DoUnified(context.Background(), &Options{
+		From: &File{
+			Path: "a.txt",
+			Hash: "4789568",
+			Mode: 0o10644,
+		},
+		To: &File{
+			Path: "b.txt",
+			Hash: "6547898",
+			Mode: 0o10644,
+		},
+		A: textA,
+		B: textB,
+	})
+	if err != nil {
+		return
 	}
+	e := NewUnifiedEncoder(os.Stderr)
+	e.SetColor(color.NewColorConfig())
+	_ = e.Encode([]*Unified{u})
 }

modules/diferenco/histogram.go

@@ -1,6 +1,8 @@
 // Refer to https://github.com/pascalkuthe/imara-diff reimplemented in Golang.
 package diferenco
 
+import "context"
+
 // https://stackoverflow.com/questions/32365271/whats-the-difference-between-git-diff-patience-and-git-diff-histogram/32367597#32367597
 // https://arxiv.org/abs/1902.02467
 
@@ -141,29 +143,40 @@ type changesOut struct {
 	changes []Change
 }
 
-func (h *histogram[E]) run(beforce []E, beforePos int, after []E, afterPos int, o *changesOut) {
+func (h *histogram[E]) run(ctx context.Context, beforce []E, beforePos int, after []E, afterPos int, o *changesOut) error {
 	for {
+		select {
+		case <-ctx.Done():
+			return ctx.Err()
+		default:
+		}
 		if len(beforce) == 0 {
 			if len(after) != 0 {
 				o.changes = append(o.changes, Change{P1: beforePos, P2: afterPos, Ins: len(after)})
 			}
-			return
+			return nil
 		}
 		if len(after) == 0 {
 			o.changes = append(o.changes, Change{P1: beforePos, P2: afterPos, Del: len(beforce)})
-			return
+			return nil
 		}
 		h.populate(beforce)
 		lcs := findLcs(beforce, after, h)
 		if lcs == nil {
-			o.changes = append(o.changes, onpDiff(beforce, beforePos, after, afterPos)...)
-			return
+			changes, err := onpDiff(ctx, beforce, beforePos, after, afterPos)
+			if err != nil {
+				return err
+			}
+			o.changes = append(o.changes, changes...)
+			return nil
 		}
 		if lcs.length == 0 {
 			o.changes = append(o.changes, Change{P1: beforePos, P2: afterPos, Del: len(beforce), Ins: len(after)})
-			return
+			return nil
+		}
+		if err := h.run(ctx, beforce[:lcs.beforeStart], beforePos, after[:lcs.afterStart], afterPos, o); err != nil {
+			return err
 		}
-		h.run(beforce[:lcs.beforeStart], beforePos, after[:lcs.afterStart], afterPos, o)
 		e1 := lcs.beforeStart + lcs.length
 		beforce = beforce[e1:]
 		beforePos += e1
@@ -173,7 +186,7 @@ func (h *histogram[E]) run(beforce []E, beforePos int, after []E, afterPos int,
 	}
 }
 
-func HistogramDiff[E comparable](L1, L2 []E) []Change {
+func HistogramDiff[E comparable](ctx context.Context, L1, L2 []E) ([]Change, error) {
 	prefix := commonPrefixLength(L1, L2)
 	L1 = L1[prefix:]
 	L2 = L2[prefix:]
@@ -184,6 +197,8 @@ func HistogramDiff[E comparable](L1, L2 []E) []Change {
 		tokenOccurances: make(map[E][]int, len(L1)),
 	}
 	o := &changesOut{changes: make([]Change, 0, 100)}
-	h.run(L1, prefix, L2, prefix, o)
-	return o.changes
+	if err := h.run(ctx, L1, prefix, L2, prefix, o); err != nil {
+		return nil, err
+	}
+	return o.changes, nil
 }

modules/diferenco/histogram_test.go

@@ -1,6 +1,7 @@
 package diferenco
 
 import (
+	"context"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -28,7 +29,7 @@ func TestHistogram(t *testing.T) {
 	}
 	a := sink.ParseLines(textA)
 	b := sink.ParseLines(textB)
-	changes := HistogramDiff(a, b)
+	changes, _ := HistogramDiff(context.Background(), a, b)
 	i := 0
 	for _, c := range changes {
 		for ; i < c.P1; i++ {

modules/diferenco/myers.go

@@ -6,19 +6,22 @@
 
 package diferenco
 
-import "slices"
+import (
+	"context"
+	"slices"
+)
 
-func MyersDiff[E comparable](seq1, seq2 []E) []Change {
+func MyersDiff[E comparable](ctx context.Context, seq1, seq2 []E) ([]Change, error) {
 	// These are common special cases.
 	// The early return improves performance dramatically.
 	if len(seq1) == 0 && len(seq2) == 0 {
-		return []Change{}
+		return []Change{}, nil
 	}
 	if len(seq1) == 0 {
-		return []Change{{Ins: len(seq2)}}
+		return []Change{{Ins: len(seq2)}}, nil
 	}
 	if len(seq2) == 0 {
-		return []Change{{Del: len(seq1)}}
+		return []Change{{Del: len(seq1)}}, nil
 	}
 	seqX := seq1
 	seqY := seq2
@@ -48,6 +51,11 @@ func MyersDiff[E comparable](seq1, seq2 []E) []Change {
 	k := 0
 outer:
 	for {
+		select {
+		case <-ctx.Done():
+			return nil, ctx.Err()
+		default:
+		}
 		d++
 		// The paper has `for (k = -d; k <= d; k += 2)`, but we can ignore diagonals that cannot influence the result.
 		lowerBound := -min(d, len(seqY)+(d%2))
@@ -110,7 +118,7 @@ outer:
 		path = path.pre
 	}
 	slices.Reverse(changes)
-	return changes
+	return changes, nil
 }
 
 type SnakePath struct {

modules/diferenco/myers_test.go

@@ -1,6 +1,7 @@
 package diferenco
 
 import (
+	"context"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -28,7 +29,7 @@ func TestMyersDiff(t *testing.T) {
 	}
 	a := sink.ParseLines(textA)
 	b := sink.ParseLines(textB)
-	changes := MyersDiff(a, b)
+	changes, _ := MyersDiff(context.Background(), a, b)
 	i := 0
 	for _, c := range changes {
 		for ; i < c.P1; i++ {
@@ -68,8 +69,8 @@ func TestMyersDiff2(t *testing.T) {
 	}
 	a := sink.ParseLines(textA)
 	b := sink.ParseLines(textB)
-	changes := MyersDiff(a, b)
-	u := sink.ToUnified(File{Path: "a.txt"}, File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
+	changes, _ := MyersDiff(context.Background(), a, b)
+	u := sink.ToUnified(&File{Path: "a.txt"}, &File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
 	fmt.Fprintf(os.Stderr, "diff:\n%s\n", u.String())
 }
 
@@ -89,7 +90,7 @@ func TestMyersDiff3(t *testing.T) {
 	}
 	a := sink.ParseLines(textA)
 	b := sink.ParseLines(textB)
-	changes := MyersDiff(a, b)
-	u := sink.ToUnified(File{Path: "a.txt"}, File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
+	changes, _ := MyersDiff(context.Background(), a, b)
+	u := sink.ToUnified(&File{Path: "a.txt"}, &File{Path: "b.txt"}, changes, a, b, DefaultContextLines)
 	fmt.Fprintf(os.Stderr, "diff:\n%s\n", u.String())
 }

modules/diferenco/onp.go

@@ -9,7 +9,9 @@
 // "An O(NP) Sequence Comparison Algorithm" August 1989.
 package diferenco
 
-func onpDiff[E comparable](L1 []E, P1 int, L2 []E, P2 int) []Change {
+import "context"
+
+func onpDiff[E comparable](ctx context.Context, L1 []E, P1 int, L2 []E, P2 int) ([]Change, error) {
 	m, n := len(L1), len(L2)
 	c := &onpCtx[E]{L1: L1, L2: L2, P1: P1, P2: P2}
 	if n >= m {
@@ -21,7 +23,7 @@ func onpDiff[E comparable](L1 []E, P1 int, L2 []E, P2 int) []Change {
 		c.xchg = true
 	}
 	c.Δ = c.N - c.M
-	return c.compare()
+	return c.compare(ctx)
 }
 
 type onpCtx[E comparable] struct {
@@ -33,7 +35,12 @@ type onpCtx[E comparable] struct {
 	xchg   bool
 }
 
-func (c *onpCtx[E]) compare() []Change {
+func (c *onpCtx[E]) compare(ctx context.Context) ([]Change, error) {
+	select {
+	case <-ctx.Done():
+		return nil, ctx.Err()
+	default:
+	}
 	c.fp = make([]point, (c.M+1)+(c.N+1)+1)
 	for i := range c.fp {
 		c.fp[i].y = -1
@@ -71,7 +78,7 @@ func (c *onpCtx[E]) compare() []Change {
 			changes = append(changes, Change{y + c.P1, x + c.P2, c.N - y, c.M - x})
 		}
 	}
-	return changes
+	return changes, nil
 }
 
 func (c *onpCtx[E]) snake(k int) {
@@ -140,13 +147,13 @@ type onpLcs struct {
 
 // OnpDiff returns the differences between data.
 // It makes O(NP) (the worst case) calls to data.Equal.
-func OnpDiff[E comparable](L1, L2 []E) []Change {
+func OnpDiff[E comparable](ctx context.Context, L1, L2 []E) ([]Change, error) {
 	//return myersDiff(L1, 0, L2, 0)
 	prefix := commonPrefixLength(L1, L2)
 	L1 = L1[prefix:]
 	L2 = L2[prefix:]
 	suffix := commonSuffixLength(L1, L2)
 	L1 = L1[:len(L1)-suffix]
 	L2 = L2[:len(L2)-suffix]
-	return onpDiff(L1, prefix, L2, prefix)
+	return onpDiff(ctx, L1, prefix, L2, prefix)
 }

modules/diferenco/onp_test.go

@@ -1,6 +1,7 @@
 package diferenco
 
 import (
+	"context"
 	"fmt"
 	"os"
 	"path/filepath"
@@ -28,7 +29,7 @@ func TestONP(t *testing.T) {
 	}
 	a := sink.ParseLines(textA)
 	b := sink.ParseLines(textB)
-	changes := OnpDiff(a, b)
+	changes, _ := OnpDiff(context.Background(), a, b)
 	i := 0
 	for _, c := range changes {
 		for ; i < c.P1; i++ {