Reduce memory footprint of Item struct

src/util/chars.go

@@ -3,63 +3,81 @@ package util
 import (
 	"unicode"
 	"unicode/utf8"
+	"unsafe"
 )
 
 type Chars struct {
-	runes []rune
-	bytes []byte
+	slice           []byte // or []rune
+	inBytes         bool
+	trimLengthKnown bool
+	trimLength      uint16
+
+	// XXX Piggybacking item index here is a horrible idea. But I'm trying to
+	// minimize the memory footprint by not wasting padded spaces.
+	Index int32
 }
 
 // ToChars converts byte array into rune array
-func ToChars(bytea []byte) Chars {
+func ToChars(bytes []byte) Chars {
 	var runes []rune
-	ascii := true
-	numBytes := len(bytea)
+	inBytes := true
+	numBytes := len(bytes)
 	for i := 0; i < numBytes; {
-		if bytea[i] < utf8.RuneSelf {
-			if !ascii {
-				runes = append(runes, rune(bytea[i]))
+		if bytes[i] < utf8.RuneSelf {
+			if !inBytes {
+				runes = append(runes, rune(bytes[i]))
 			}
 			i++
 		} else {
-			if ascii {
-				ascii = false
+			if inBytes {
+				inBytes = false
 				runes = make([]rune, i, numBytes)
 				for j := 0; j < i; j++ {
-					runes[j] = rune(bytea[j])
+					runes[j] = rune(bytes[j])
 				}
 			}
-			r, sz := utf8.DecodeRune(bytea[i:])
+			r, sz := utf8.DecodeRune(bytes[i:])
 			i += sz
 			runes = append(runes, r)
 		}
 	}
-	if ascii {
-		return Chars{bytes: bytea}
+	if inBytes {
+		return Chars{slice: bytes, inBytes: inBytes}
 	}
-	return Chars{runes: runes}
+	return RunesToChars(runes)
 }
 
 func RunesToChars(runes []rune) Chars {
-	return Chars{runes: runes}
+	return Chars{slice: *(*[]byte)(unsafe.Pointer(&runes)), inBytes: false}
+}
+
+func (chars *Chars) optionalRunes() []rune {
+	if chars.inBytes {
+		return nil
+	}
+	return *(*[]rune)(unsafe.Pointer(&chars.slice))
 }
 
 func (chars *Chars) Get(i int) rune {
-	if chars.runes != nil {
-		return chars.runes[i]
+	if runes := chars.optionalRunes(); runes != nil {
+		return runes[i]
 	}
-	return rune(chars.bytes[i])
+	return rune(chars.slice[i])
 }
 
 func (chars *Chars) Length() int {
-	if chars.runes != nil {
-		return len(chars.runes)
+	if runes := chars.optionalRunes(); runes != nil {
+		return len(runes)
 	}
-	return len(chars.bytes)
+	return len(chars.slice)
 }
 
 // TrimLength returns the length after trimming leading and trailing whitespaces
-func (chars *Chars) TrimLength() int {
+func (chars *Chars) TrimLength() uint16 {
+	if chars.trimLengthKnown {
+		return chars.trimLength
+	}
+	chars.trimLengthKnown = true
 	var i int
 	len := chars.Length()
 	for i = len - 1; i >= 0; i-- {
@@ -80,7 +98,8 @@ func (chars *Chars) TrimLength() int {
 			break
 		}
 	}
-	return i - j + 1
+	chars.trimLength = AsUint16(i - j + 1)
+	return chars.trimLength
 }
 
 func (chars *Chars) TrailingWhitespaces() int {
@@ -96,28 +115,40 @@ func (chars *Chars) TrailingWhitespaces() int {
 }
 
 func (chars *Chars) ToString() string {
-	if chars.runes != nil {
-		return string(chars.runes)
+	if runes := chars.optionalRunes(); runes != nil {
+		return string(runes)
 	}
-	return string(chars.bytes)
+	return string(chars.slice)
 }
 
 func (chars *Chars) ToRunes() []rune {
-	if chars.runes != nil {
-		return chars.runes
+	if runes := chars.optionalRunes(); runes != nil {
+		return runes
 	}
-	runes := make([]rune, len(chars.bytes))
-	for idx, b := range chars.bytes {
+	bytes := chars.slice
+	runes := make([]rune, len(bytes))
+	for idx, b := range bytes {
 		runes[idx] = rune(b)
 	}
 	return runes
 }
 
-func (chars *Chars) Slice(b int, e int) Chars {
-	if chars.runes != nil {
-		return Chars{runes: chars.runes[b:e]}
+func (chars *Chars) CopyRunes(dest []rune) {
+	if runes := chars.optionalRunes(); runes != nil {
+		copy(dest, runes)
+		return
 	}
-	return Chars{bytes: chars.bytes[b:e]}
+	for idx, b := range chars.slice {
+		dest[idx] = rune(b)
+	}
+	return
+}
+
+func (chars *Chars) Slice(b int, e int) Chars {
+	if runes := chars.optionalRunes(); runes != nil {
+		return RunesToChars(runes[b:e])
+	}
+	return Chars{slice: chars.slice[b:e], inBytes: true}
 }
 
 func (chars *Chars) Split(delimiter string) []Chars {