yatwiki/diff/diff.go

// Paul's Simple Diff Algorithm v 0.1
// (C) Paul Butler 2007 <http://www.paulbutler.org/>
// May be used and distributed under the zlib/libpng license.
// https://github.com/paulgb/simplediff/
package diff

import (
	"html"
	"io"
	"strings"
)

var Separator string = "\n"

type Operation int8

const (
	OP_INSERTED  Operation = 1
	OP_UNCHANGED           = 2
	OP_REMOVED             = 3
)

type Instruction struct {
	Op      Operation
	Content []string
}

func (this *Instruction) RenderHTML() string {
	sc := strings.Join(this.Content, Separator)
	if len(sc) == 0 {
		return ""
	}

	switch this.Op {
	case OP_INSERTED:
		return `<ins>` + sc + `</ins>`
	case OP_REMOVED:
		return `<del>` + sc + `</del>`
	default: //OP_UNCHANGED:
		return sc
	}
}

type twoints struct {
	O, N int
}

func Diff(o, n []string) []Instruction {

	if len(o) == 0 && len(n) == 0 {
		return []Instruction{}
	}
	if len(o) == 0 {
		return []Instruction{{OP_INSERTED, n}}
	}
	if len(n) == 0 {
		return []Instruction{{OP_REMOVED, o}}
	}

	maxl := 0
	omax := 0
	nmax := 0
	matrix := make(map[twoints]int, len(o)+1)

	for oindex, ovalue := range o {

		// Find keys in new, where value matchines ovalue
		nkeys := make([]int, 0)
		for nindex, nvalue := range n {
			if nvalue == ovalue {
				nkeys = append(nkeys, nindex)
			}
		}

		// Build diff matrix
		for _, nindex := range nkeys {
			matrix[twoints{oindex, nindex}] = matrix[twoints{oindex - 1, nindex - 1}] + 1

			if matrix[twoints{oindex, nindex}] > maxl {
				maxl = matrix[twoints{oindex, nindex}]
				omax = oindex + 1 - maxl
				nmax = nindex + 1 - maxl
			}
		}
	}

	if maxl == 0 {
		return []Instruction{
			{OP_REMOVED, o},
			{OP_INSERTED, n},
		}
	}

	ret := Diff(o[:omax], n[:nmax])
	ret = append(ret, Instruction{OP_UNCHANGED, n[nmax : nmax+maxl]})
	ret = append(ret, Diff(o[omax+maxl:], n[nmax+maxl:])...)

	return ret
}

// HtmlDiff splits o and n by newlines, and writes an HTML-safe version to the dest writer.
func HtmlDiff(o, n string, dest io.Writer) {
	ops := Diff(strings.Split(html.EscapeString(o), Separator), strings.Split(html.EscapeString(n), Separator))

	for _, op := range ops {
		dest.Write([]byte(op.RenderHTML()))
	}
}