Merge pull request #90 from vmarkovtsev/patch-1

Fix DiffLevenshtein counting single runes as multiple edits

Author: sergi

diffmatchpatch/diff.go

@@ -1236,9 +1236,9 @@ func (dmp *DiffMatchPatch) DiffLevenshtein(diffs []Diff) int {
 	for _, aDiff := range diffs {
 		switch aDiff.Type {
 		case DiffInsert:
-			insertions += len(aDiff.Text)
+			insertions += utf8.RuneCountInString(aDiff.Text)
 		case DiffDelete:
-			deletions += len(aDiff.Text)
+			deletions += utf8.RuneCountInString(aDiff.Text)
 		case DiffEqual:
 			// A deletion and an insertion is one substitution.
 			levenshtein += max(insertions, deletions)

diffmatchpatch/diff_test.go

@@ -1153,9 +1153,9 @@ func TestDiffLevenshtein(t *testing.T) {
 	dmp := New()
 
 	for i, tc := range []TestCase{
-		{"Levenshtein with trailing equality", []Diff{{DiffDelete, "abc"}, {DiffInsert, "1234"}, {DiffEqual, "xyz"}}, 4},
-		{"Levenshtein with leading equality", []Diff{{DiffEqual, "xyz"}, {DiffDelete, "abc"}, {DiffInsert, "1234"}}, 4},
-		{"Levenshtein with middle equality", []Diff{{DiffDelete, "abc"}, {DiffEqual, "xyz"}, {DiffInsert, "1234"}}, 7},
+		{"Levenshtein with trailing equality", []Diff{{DiffDelete, "абв"}, {DiffInsert, "1234"}, {DiffEqual, "эюя"}}, 4},
+		{"Levenshtein with leading equality", []Diff{{DiffEqual, "эюя"}, {DiffDelete, "абв"}, {DiffInsert, "1234"}}, 4},
+		{"Levenshtein with middle equality", []Diff{{DiffDelete, "абв"}, {DiffEqual, "эюя"}, {DiffInsert, "1234"}}, 7},
 	} {
 		actual := dmp.DiffLevenshtein(tc.Diffs)
 		assert.Equal(t, tc.Expected, actual, fmt.Sprintf("Test case #%d, %s", i, tc.Name))