new diff lib: diferenco (WIP)

modules/diferenco/diferenco.go

@@ -0,0 +1,92 @@
+package diferenco
+
+import "slices"
+
+// https://github.com/Wilfred/difftastic/wiki/Line-Based-Diffs
+// https://neil.fraser.name/writing/diff/
+// https://prettydiff.com/2/guide/unrelated_diff.xhtml
+
+// Operation defines the operation of a diff item.
+type Operation int8
+
+const (
+	// Delete item represents a delete hunk.
+	Delete Operation = -1
+	// Insert item represents an insert hunk.
+	Insert Operation = 1
+	// Equal item represents an equal hunk.
+	Equal Operation = 0
+)
+
+// commonPrefixLength returns the length of the common prefix of two T slices.
+func commonPrefixLength[E comparable](a, b []E) int {
+	n := min(len(a), len(b))
+	i := 0
+	for i < n && a[i] == b[i] {
+		i++
+	}
+	return i
+}
+
+// commonSuffixLength returns the length of the common suffix of two rune slices.
+func commonSuffixLength[E comparable](a, b []E) int {
+	i1, i2 := len(a), len(b)
+	n := min(i1, i2)
+	i := 0
+	for i < n && a[i1-1-i] == b[i2-1-i] {
+		i++
+	}
+	return i
+}
+
+func slicesHasSuffix[E comparable](a, suffix []E) bool {
+	return len(a) >= len(suffix) && slices.Equal(a[len(a)-len(suffix):], suffix)
+}
+
+func slicesHasPrefix[E comparable](a, prefix []E) bool {
+	return len(a) >= len(prefix) && slices.Equal(a[:len(prefix)], prefix)
+}
+
+// slicesIndex is the equivalent of strings.Index for rune slices.
+func slicesIndex[E comparable](s1, s2 []E) int {
+	last := len(s1) - len(s2)
+	for i := 0; i <= last; i++ {
+		if slices.Equal(s1[i:i+len(s2)], s2) {
+			return i
+		}
+	}
+	return -1
+}
+
+// slicesIndexOf returns the index of pattern in target, starting at target[i].
+func slicesIndexOf[E comparable](target, pattern []E, i int) int {
+	if i > len(target)-1 {
+		return -1
+	}
+	if i <= 0 {
+		return slicesIndex(target, pattern)
+	}
+	ind := slicesIndex(target[i:], pattern)
+	if ind == -1 {
+		return -1
+	}
+	return ind + i
+}
+
+type Change struct {
+	P1  int // before: position in before
+	P2  int // after: position in after
+	Del int // number of elements that deleted from a
+	Ins int // number of elements that inserted into b
+}
+
+type Dfio[E comparable] struct {
+	T Operation
+	E []E
+}
+
+// StringDiff represents one diff operation
+type StringDiff struct {
+	Type Operation
+	Text string
+}

modules/diferenco/diferenco_test.go

@@ -0,0 +1,67 @@
+package diferenco
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"runtime"
+	"testing"
+)
+
+func TestDiff(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
+	}
+	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
+	}
+	textB := string(bytesB)
+	sink := &Sink{
+		Index: make(map[string]int),
+	}
+	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
+	}
+	for ; i < len(a); i++ {
+		fmt.Fprintf(os.Stderr, "  %s", sink.Lines[a[i]])
+	}
+	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])
+			}
+		}
+	}
+}