allow precomputation of suffix array

diff.go

@@ -191,7 +191,7 @@ func Diff(old, new io.Reader, patch io.Writer) error {
 		return err
 	}
 
-	pbuf, err := diffBytes(obuf, nbuf)
+	pbuf, err := diffBytes(nil, obuf, nbuf)
 	if err != nil {
 		return err
 	}
@@ -200,18 +200,56 @@ func Diff(old, new io.Reader, patch io.Writer) error {
 	return err
 }
 
-func diffBytes(obuf, nbuf []byte) ([]byte, error) {
+// BufSufData contains a buffer and its corresponding suffix array.
+type BufSufData struct {
+	Buf []byte
+	Suf []int
+}
+
+// ComputeSuf computes a BufSufData from reader using qsufsort algorithm.
+func ComputeSuf(reader io.Reader) (*BufSufData, error) {
+	buf, err := ioutil.ReadAll(reader)
+	if err != nil {
+		return nil, err
+	}
+	return &BufSufData{
+		Buf: buf,
+		Suf: qsufsort(buf),
+	}, nil
+}
+
+// DiffWithSuf does the same operations as Diff with the exception that
+// it optionally takes obufsuf (*BufSufData of old io.Reader) to avoid
+// recreating the suffix array.
+func DiffWithSuf(obufsuf *BufSufData, new io.Reader, patch io.Writer) error {
+	nbuf, err := ioutil.ReadAll(new)
+	if err != nil {
+		return err
+	}
+
+	pbuf, err := diffBytes(obufsuf.Suf, obufsuf.Buf, nbuf)
+	if err != nil {
+		return err
+	}
+
+	_, err = patch.Write(pbuf)
+	return err
+}
+
+func diffBytes(I []int, obuf, nbuf []byte) ([]byte, error) {
 	var patch seekBuffer
-	err := diff(obuf, nbuf, &patch)
+	err := diff(I, obuf, nbuf, &patch)
 	if err != nil {
 		return nil, err
 	}
 	return patch.buf, nil
 }
 
-func diff(obuf, nbuf []byte, patch io.WriteSeeker) error {
+func diff(I []int, obuf, nbuf []byte, patch io.WriteSeeker) error {
 	var lenf int
-	I := qsufsort(obuf)
+	if I == nil {
+		I = qsufsort(obuf)
+	}
 	db := make([]byte, len(nbuf))
 	eb := make([]byte, len(nbuf))
 	var dblen, eblen int

diff_test.go

@@ -2,27 +2,30 @@ package binarydist
 
 import (
 	"bytes"
+	"io"
 	"io/ioutil"
 	"os"
 	"os/exec"
 	"testing"
 )
 
-var diffT = []struct {
-	old *os.File
-	new *os.File
-}{
-	{
-		old: mustWriteRandFile("test.old", 1e3, 1),
-		new: mustWriteRandFile("test.new", 1e3, 2),
-	},
-	{
-		old: mustOpen("testdata/sample.old"),
-		new: mustOpen("testdata/sample.new"),
-	},
-}
+func testFunc(t *testing.T, fDiff func(old, new io.Reader, patch io.Writer) error) {
+	t.Helper()
+
+	diffT := []struct {
+		old *os.File
+		new *os.File
+	}{
+		{
+			old: mustWriteRandFile("test.old", 1e3, 1),
+			new: mustWriteRandFile("test.new", 1e3, 2),
+		},
+		{
+			old: mustOpen("testdata/sample.old"),
+			new: mustOpen("testdata/sample.new"),
+		},
+	}
 
-func TestDiff(t *testing.T) {
 	for _, s := range diffT {
 		got, err := ioutil.TempFile("/tmp", "bspatch.")
 		if err != nil {
@@ -43,7 +46,7 @@ func TestDiff(t *testing.T) {
 			panic(err)
 		}
 
-		err = Diff(s.old, s.new, got)
+		err = fDiff(s.old, s.new, got)
 		if err != nil {
 			t.Fatal("err", err)
 		}
@@ -65,3 +68,19 @@ func TestDiff(t *testing.T) {
 		}
 	}
 }
+
+func TestDiff(t *testing.T) {
+	testFunc(t, Diff)
+}
+
+func TestDiffWithSuf(t *testing.T) {
+	fDiff := func(old, new io.Reader, patch io.Writer) error {
+		oldSufStruct, err := ComputeSuf(old)
+		if err != nil {
+			return err
+		}
+
+		return DiffWithSuf(oldSufStruct, new, patch)
+	}
+	testFunc(t, fDiff)
+}