Cache runes in chars mode

Author: triallax

interp/functions.go

@@ -450,22 +450,6 @@ func (p *interp) sprintf(format string, args []value) (string, error) {
 	return fmt.Sprintf(format, converted...), nil
 }
 
-func substrChars(s string, pos int) string {
-	// Count characters till we get to pos.
-	chars := 1
-	start := 0
-	for start = range s {
-		chars++
-		if chars > pos {
-			break
-		}
-	}
-	if pos >= chars {
-		start = len(s)
-	}
-	return s[start:]
-}
-
 func substrLengthChars(s string, pos, length int) string {
 	// Count characters till we get to pos.
 	chars := 1

interp/value.go

@@ -22,7 +22,8 @@ const (
 type value struct {
 	typ valueType // Type of value
 	s   string    // String value (for typeStr and typeNumStr)
-	n   float64   // Numeric value (for typeNum)
+	r   *[]rune
+	n   float64 // Numeric value (for typeNum)
 }
 
 // Create a new null value
@@ -37,12 +38,16 @@ func num(n float64) value {
 
 // Create a new string value
 func str(s string) value {
-	return value{typ: typeStr, s: s}
+	return value{typ: typeStr, s: s, r: new([]rune)}
+}
+
+func strFromRunes(runes []rune) value {
+	return value{typ: typeStr, s: string(runes), r: &runes}
 }
 
 // Create a new value to represent a "numeric string" from an input field
 func numStr(s string) value {
-	return value{typ: typeNumStr, s: s}
+	return value{typ: typeNumStr, s: s, r: new([]rune)}
 }
 
 // Create a numeric value from a Go bool
@@ -53,6 +58,18 @@ func boolean(b bool) value {
 	return num(0)
 }
 
+func (v value) runes(floatFormat string) []rune {
+	switch v.typ {
+	case typeStr, typeNumStr:
+		if *v.r == nil {
+			*v.r = []rune(v.s)
+		}
+		return *v.r
+	default: // typeNum, typeNull
+		return []rune(v.str(floatFormat))
+	}
+}
+
 // String returns a string representation of v for debugging.
 func (v value) String() string {
 	switch v.typ {

interp/vm.go

@@ -1001,12 +1001,12 @@ func (p *interp) callBuiltin(builtinOp compiler.BuiltinOp) error {
 		p.push(num(float64(length)))
 
 	case compiler.BuiltinLengthArg:
-		s := p.toString(p.peekTop())
+		s := p.peekTop()
 		var length int
 		if p.chars {
-			length = utf8.RuneCountInString(s)
+			length = len(s.runes(p.convertFormat))
 		} else {
-			length = len(s)
+			length = len(p.toString(s))
 		}
 		p.replaceTop(num(float64(length)))
 
@@ -1066,9 +1066,18 @@ func (p *interp) callBuiltin(builtinOp compiler.BuiltinOp) error {
 		sValue, posValue := p.peekPop()
 		pos := int(posValue.num())
 		s := p.toString(sValue)
-		var substr string
+		var substr value
 		if p.chars {
-			substr = substrChars(s, pos)
+			runes := sValue.runes(p.convertFormat)
+			if pos > len(runes) {
+				pos = len(runes) + 1
+			}
+			if pos < 1 {
+				pos = 1
+			}
+			length := len(runes) - pos + 1
+			runes = runes[pos-1 : pos-1+length]
+			substr = strFromRunes(runes)
 		} else {
 			if pos > len(s) {
 				pos = len(s) + 1
@@ -1077,19 +1086,34 @@ func (p *interp) callBuiltin(builtinOp compiler.BuiltinOp) error {
 				pos = 1
 			}
 			length := len(s) - pos + 1
-			substr = s[pos-1 : pos-1+length]
+			substr = str(s[pos-1 : pos-1+length])
 		}
-		p.replaceTop(str(substr))
+		p.replaceTop(substr)
 
 	case compiler.BuiltinSubstrLength:
 		posValue, lengthValue := p.popTwo()
 		length := int(lengthValue.num())
 		pos := int(posValue.num())
-		s := p.toString(p.peekTop())
-		var substr string
+		sValue := p.peekTop()
+		var substr value
 		if p.chars {
-			substr = substrLengthChars(s, pos, length)
+			runes := sValue.runes(p.convertFormat)
+			if pos > len(runes) {
+				pos = len(runes) + 1
+			}
+			if pos < 1 {
+				pos = 1
+			}
+			maxLength := len(runes) - pos + 1
+			if length < 0 {
+				length = 0
+			}
+			if length > maxLength {
+				length = maxLength
+			}
+			substr = strFromRunes(runes[pos-1 : pos-1+length])
 		} else {
+			s := p.toString(sValue)
 			if pos > len(s) {
 				pos = len(s) + 1
 			}
@@ -1103,9 +1127,9 @@ func (p *interp) callBuiltin(builtinOp compiler.BuiltinOp) error {
 			if length > maxLength {
 				length = maxLength
 			}
-			substr = s[pos-1 : pos-1+length]
+			substr = str(s[pos-1 : pos-1+length])
 		}
-		p.replaceTop(str(substr))
+		p.replaceTop(substr)
 
 	case compiler.BuiltinSystem:
 		if p.noExec {