Speed improvements

tokenize/handlers_builtin.go

@@ -77,6 +77,7 @@ var A = struct {
 	AnyByte         Handler
 	AnyRune         Handler
 	ValidRune       Handler
+	ValidRunes      Handler
 	InvalidRune     Handler
 	Str             func(string) Handler
 	StrNoCase       func(string) Handler
@@ -166,6 +167,7 @@ var A = struct {
 	AnyByte:         MatchAnyByte(),
 	AnyRune:         MatchAnyRune(),
 	ValidRune:       MatchValidRune(),
+	ValidRunes:      MatchValidRunes(),
 	InvalidRune:     MatchInvalidRune(),
 	Str:             MatchStr,
 	StrNoCase:       MatchStrNoCase,
@@ -990,36 +992,21 @@ func MatchNotFollowedBy(lookAhead Handler, handler Handler) Handler {
 }
 
 func MatchInOptionalBlanks(handler Handler) Handler {
+	blanks := MatchBlanks()
 	return func(tokenAPI *API) bool {
-		skipBlanks(tokenAPI)
+		tokenAPI.Output.Suspend()
+		blanks(tokenAPI)
+		tokenAPI.Output.Resume()
 		if !handler(tokenAPI) {
 			return false
 		}
-		skipBlanks(tokenAPI)
+		tokenAPI.Output.Suspend()
+		blanks(tokenAPI)
+		tokenAPI.Output.Resume()
 		return true
 	}
 }
 
-func skipBlanks(tokenAPI *API) {
-	for {
-		bs, err := tokenAPI.Input.Byte.PeekMulti(0, 128)
-		for i, b := range bs {
-			if b != ' ' && b != '\t' {
-				if i > 0 {
-					tokenAPI.Input.Byte.MoveCursorMulti(bs[:i]...)
-				}
-				return
-			}
-		}
-		if err != nil {
-			if len(bs) > 0 {
-				tokenAPI.Input.Byte.MoveCursorMulti(bs...)
-			}
-			return
-		}
-	}
-}
-
 // MakeInputFlusher creates a Handler that will flush the input buffer when the
 // provided handler matches.
 //
@@ -1182,6 +1169,40 @@ func MatchValidRune() Handler {
 	}
 }
 
+func MatchValidRunes() Handler {
+	return func(tokenAPI *API) bool {
+		rs := make([]rune, 256, 256)
+		ok := false
+		for {
+			bs, err := tokenAPI.Input.Byte.PeekMulti(0, 256)
+			end := 0
+			offset := 0
+			for offset <= 256-utf8.UTFMax {
+				r, w := utf8.DecodeRune(bs[offset:])
+				if r == utf8.RuneError {
+					if end > 0 {
+						tokenAPI.Input.Rune.AcceptMulti(rs[:end]...)
+					}
+					return ok
+				}
+				ok = true
+				offset += w
+				rs[end] = r
+				end++
+			}
+			if end > 0 {
+				tokenAPI.Input.Rune.AcceptMulti(rs[:end]...)
+			}
+			if err != nil {
+				if err == io.EOF {
+					return ok
+				}
+				return false
+			}
+		}
+	}
+}
+
 // MatchInvalidRune creates a Handler function that checks if an invalid
 // UTF8 rune can be read from the input.
 func MatchInvalidRune() Handler {