goja/parser/lexer.go

package parser

import (
	"errors"
	"fmt"
	"strconv"
	"strings"
	"unicode"
	"unicode/utf16"
	"unicode/utf8"

	"golang.org/x/text/unicode/rangetable"

	"github.com/dop251/goja/file"
	"github.com/dop251/goja/token"
	"github.com/dop251/goja/unistring"
)

var (
	unicodeRangeIdNeg      = rangetable.Merge(unicode.Pattern_Syntax, unicode.Pattern_White_Space)
	unicodeRangeIdStartPos = rangetable.Merge(unicode.Letter, unicode.Nl, unicode.Other_ID_Start)
	unicodeRangeIdContPos  = rangetable.Merge(unicodeRangeIdStartPos, unicode.Mn, unicode.Mc, unicode.Nd, unicode.Pc, unicode.Other_ID_Continue)
)

func isDecimalDigit(chr rune) bool {
	return '0' <= chr && chr <= '9'
}

func IsIdentifier(s string) bool {
	if s == "" {
		return false
	}
	r, size := utf8.DecodeRuneInString(s)
	if !isIdentifierStart(r) {
		return false
	}
	for _, r := range s[size:] {
		if !isIdentifierPart(r) {
			return false
		}
	}
	return true
}

func digitValue(chr rune) int {
	switch {
	case '0' <= chr && chr <= '9':
		return int(chr - '0')
	case 'a' <= chr && chr <= 'f':
		return int(chr - 'a' + 10)
	case 'A' <= chr && chr <= 'F':
		return int(chr - 'A' + 10)
	}
	return 16 // Larger than any legal digit value
}

func isDigit(chr rune, base int) bool {
	return digitValue(chr) < base
}

func isIdStartUnicode(r rune) bool {
	return unicode.Is(unicodeRangeIdStartPos, r) && !unicode.Is(unicodeRangeIdNeg, r)
}

func isIdPartUnicode(r rune) bool {
	return unicode.Is(unicodeRangeIdContPos, r) && !unicode.Is(unicodeRangeIdNeg, r) || r == '\u200C' || r == '\u200D'
}

func isIdentifierStart(chr rune) bool {
	return chr == '$' || chr == '_' || chr == '\\' ||
		'a' <= chr && chr <= 'z' || 'A' <= chr && chr <= 'Z' ||
		chr >= utf8.RuneSelf && isIdStartUnicode(chr)
}

func isIdentifierPart(chr rune) bool {
	return chr == '$' || chr == '_' || chr == '\\' ||
		'a' <= chr && chr <= 'z' || 'A' <= chr && chr <= 'Z' ||
		'0' <= chr && chr <= '9' ||
		chr >= utf8.RuneSelf && isIdPartUnicode(chr)
}

func (self *_parser) scanIdentifier() (string, unistring.String, bool, string) {
	offset := self.chrOffset
	hasEscape := false
	isUnicode := false
	length := 0
	for isIdentifierPart(self.chr) {
		r := self.chr
		length++
		if r == '\\' {
			hasEscape = true
			distance := self.chrOffset - offset
			self.read()
			if self.chr != 'u' {
				return "", "", false, fmt.Sprintf("Invalid identifier escape character: %c (%s)", self.chr, string(self.chr))
			}
			var value rune
			if self._peek() == '{' {
				self.read()
				value = -1
				for value <= utf8.MaxRune {
					self.read()
					if self.chr == '}' {
						break
					}
					decimal, ok := hex2decimal(byte(self.chr))
					if !ok {
						return "", "", false, "Invalid Unicode escape sequence"
					}
					if value == -1 {
						value = decimal
					} else {
						value = value<<4 | decimal
					}
				}
				if value == -1 {
					return "", "", false, "Invalid Unicode escape sequence"
				}
			} else {
				for j := 0; j < 4; j++ {
					self.read()
					decimal, ok := hex2decimal(byte(self.chr))
					if !ok {
						return "", "", false, fmt.Sprintf("Invalid identifier escape character: %c (%s)", self.chr, string(self.chr))
					}
					value = value<<4 | decimal
				}
			}
			if value == '\\' {
				return "", "", false, fmt.Sprintf("Invalid identifier escape value: %c (%s)", value, string(value))
			} else if distance == 0 {
				if !isIdentifierStart(value) {
					return "", "", false, fmt.Sprintf("Invalid identifier escape value: %c (%s)", value, string(value))
				}
			} else if distance > 0 {
				if !isIdentifierPart(value) {
					return "", "", false, fmt.Sprintf("Invalid identifier escape value: %c (%s)", value, string(value))
				}
			}
			r = value
		}
		if r >= utf8.RuneSelf {
			isUnicode = true
			if r > 0xFFFF {
				length++
			}
		}
		self.read()
	}

	literal := self.str[offset:self.chrOffset]
	var parsed unistring.String
	if hasEscape || isUnicode {
		var err string
		// TODO strict
		parsed, err = parseStringLiteral(literal, length, isUnicode, false)
		if err != "" {
			return "", "", false, err
		}
	} else {
		parsed = unistring.String(literal)
	}

	return literal, parsed, hasEscape, ""
}

// 7.2
func isLineWhiteSpace(chr rune) bool {
	switch chr {
	case '\u0009', '\u000b', '\u000c', '\u0020', '\u00a0', '\ufeff':
		return true
	case '\u000a', '\u000d', '\u2028', '\u2029':
		return false
	case '\u0085':
		return false
	}
	return unicode.IsSpace(chr)
}

// 7.3
func isLineTerminator(chr rune) bool {
	switch chr {
	case '\u000a', '\u000d', '\u2028', '\u2029':
		return true
	}
	return false
}

func isId(tkn token.Token) bool {
	switch tkn {
	case token.KEYWORD,
		token.BOOLEAN,
		token.NULL,
		token.THIS,
		token.IF,
		token.IN,
		token.OF,
		token.DO,

		token.VAR,
		token.LET,
		token.FOR,
		token.NEW,
		token.TRY,

		token.ELSE,
		token.CASE,
		token.VOID,
		token.WITH,

		token.CONST,
		token.WHILE,
		token.BREAK,
		token.CATCH,
		token.THROW,

		token.RETURN,
		token.TYPEOF,
		token.DELETE,
		token.SWITCH,

		token.DEFAULT,
		token.FINALLY,

		token.FUNCTION,
		token.CONTINUE,
		token.DEBUGGER,

		token.INSTANCEOF:

		return true
	}
	return false
}

type parserState struct {
	tok                                token.Token
	literal                            string
	parsedLiteral                      unistring.String
	implicitSemicolon, insertSemicolon bool
	chr                                rune
	chrOffset, offset                  int
	errorCount                         int
}

func (self *_parser) mark(state *parserState) *parserState {
	if state == nil {
		state = &parserState{}
	}
	state.tok, state.literal, state.parsedLiteral, state.implicitSemicolon, state.insertSemicolon, state.chr, state.chrOffset, state.offset =
		self.token, self.literal, self.parsedLiteral, self.implicitSemicolon, self.insertSemicolon, self.chr, self.chrOffset, self.offset

	state.errorCount = len(self.errors)
	return state
}

func (self *_parser) restore(state *parserState) {
	self.token, self.literal, self.parsedLiteral, self.implicitSemicolon, self.insertSemicolon, self.chr, self.chrOffset, self.offset =
		state.tok, state.literal, state.parsedLiteral, state.implicitSemicolon, state.insertSemicolon, state.chr, state.chrOffset, state.offset
	self.errors = self.errors[:state.errorCount]
}

func (self *_parser) peek() token.Token {
	implicitSemicolon, insertSemicolon, chr, chrOffset, offset := self.implicitSemicolon, self.insertSemicolon, self.chr, self.chrOffset, self.offset
	tok, _, _, _ := self.scan()
	self.implicitSemicolon, self.insertSemicolon, self.chr, self.chrOffset, self.offset = implicitSemicolon, insertSemicolon, chr, chrOffset, offset
	return tok
}

func (self *_parser) scan() (tkn token.Token, literal string, parsedLiteral unistring.String, idx file.Idx) {

	self.implicitSemicolon = false

	for {
		self.skipWhiteSpace()

		idx = self.idxOf(self.chrOffset)
		insertSemicolon := false

		switch chr := self.chr; {
		case isIdentifierStart(chr):
			var err string
			var hasEscape bool
			literal, parsedLiteral, hasEscape, err = self.scanIdentifier()
			if err != "" {
				tkn = token.ILLEGAL
				break
			}
			if len(parsedLiteral) > 1 {
				// Keywords are longer than 1 character, avoid lookup otherwise
				var strict bool
				tkn, strict = token.IsKeyword(string(parsedLiteral))
				if hasEscape {
					self.insertSemicolon = true
					if tkn != 0 && tkn != token.LET || parsedLiteral == "true" || parsedLiteral == "false" || parsedLiteral == "null" {
						tkn = token.KEYWORD
					} else {
						tkn = token.IDENTIFIER
					}
					return
				}
				switch tkn {

				case 0: // Not a keyword
					if parsedLiteral == "true" || parsedLiteral == "false" {
						self.insertSemicolon = true
						tkn = token.BOOLEAN
						return
					} else if parsedLiteral == "null" {
						self.insertSemicolon = true
						tkn = token.NULL
						return
					}

				case token.KEYWORD:
					if strict {
						// TODO If strict and in strict mode, then this is not a break
						break
					}
					return

				case
					token.THIS,
					token.BREAK,
					token.THROW, // A newline after a throw is not allowed, but we need to detect it
					token.RETURN,
					token.CONTINUE,
					token.DEBUGGER:
					self.insertSemicolon = true
					return

				default:
					return

				}
			}
			self.insertSemicolon = true
			tkn = token.IDENTIFIER
			return
		case '0' <= chr && chr <= '9':
			self.insertSemicolon = true
			tkn, literal = self.scanNumericLiteral(false)
			return
		default:
			self.read()
			switch chr {
			case -1:
				if self.insertSemicolon {
					self.insertSemicolon = false
					self.implicitSemicolon = true
				}
				tkn = token.EOF
			case '\r', '\n', '\u2028', '\u2029':
				self.insertSemicolon = false
				self.implicitSemicolon = true
				continue
			case ':':
				tkn = token.COLON
			case '.':
				if digitValue(self.chr) < 10 {
					insertSemicolon = true
					tkn, literal = self.scanNumericLiteral(true)
				} else {
					if self.chr == '.' {
						self.read()
						if self.chr == '.' {
							self.read()
							tkn = token.ELLIPSIS
						} else {
							tkn = token.ILLEGAL
						}
					} else {
						tkn = token.PERIOD
					}
				}
			case ',':
				tkn = token.COMMA
			case ';':
				tkn = token.SEMICOLON
			case '(':
				tkn = token.LEFT_PARENTHESIS
			case ')':
				tkn = token.RIGHT_PARENTHESIS
				insertSemicolon = true
			case '[':
				tkn = token.LEFT_BRACKET
			case ']':
				tkn = token.RIGHT_BRACKET
				insertSemicolon = true
			case '{':
				tkn = token.LEFT_BRACE
			case '}':
				tkn = token.RIGHT_BRACE
				insertSemicolon = true
			case '+':
				tkn = self.switch3(token.PLUS, token.ADD_ASSIGN, '+', token.INCREMENT)
				if tkn == token.INCREMENT {
					insertSemicolon = true
				}
			case '-':
				tkn = self.switch3(token.MINUS, token.SUBTRACT_ASSIGN, '-', token.DECREMENT)
				if tkn == token.DECREMENT {
					insertSemicolon = true
				}
			case '*':
				tkn = self.switch2(token.MULTIPLY, token.MULTIPLY_ASSIGN)
			case '/':
				if self.chr == '/' {
					self.skipSingleLineComment()
					continue
				} else if self.chr == '*' {
					self.skipMultiLineComment()
					continue
				} else {
					// Could be division, could be RegExp literal
					tkn = self.switch2(token.SLASH, token.QUOTIENT_ASSIGN)
					insertSemicolon = true
				}
			case '%':
				tkn = self.switch2(token.REMAINDER, token.REMAINDER_ASSIGN)
			case '^':
				tkn = self.switch2(token.EXCLUSIVE_OR, token.EXCLUSIVE_OR_ASSIGN)
			case '<':
				tkn = self.switch4(token.LESS, token.LESS_OR_EQUAL, '<', token.SHIFT_LEFT, token.SHIFT_LEFT_ASSIGN)
			case '>':
				tkn = self.switch6(token.GREATER, token.GREATER_OR_EQUAL, '>', token.SHIFT_RIGHT, token.SHIFT_RIGHT_ASSIGN, '>', token.UNSIGNED_SHIFT_RIGHT, token.UNSIGNED_SHIFT_RIGHT_ASSIGN)
			case '=':
				if self.chr == '>' {
					self.read()
					if self.implicitSemicolon {
						tkn = token.ILLEGAL
					} else {
						tkn = token.ARROW
					}
				} else {
					tkn = self.switch2(token.ASSIGN, token.EQUAL)
					if tkn == token.EQUAL && self.chr == '=' {
						self.read()
						tkn = token.STRICT_EQUAL
					}
				}
			case '!':
				tkn = self.switch2(token.NOT, token.NOT_EQUAL)
				if tkn == token.NOT_EQUAL && self.chr == '=' {
					self.read()
					tkn = token.STRICT_NOT_EQUAL
				}
			case '&':
				tkn = self.switch3(token.AND, token.AND_ASSIGN, '&', token.LOGICAL_AND)
			case '|':
				tkn = self.switch3(token.OR, token.OR_ASSIGN, '|', token.LOGICAL_OR)
			case '~':
				tkn = token.BITWISE_NOT
			case '?':
				tkn = token.QUESTION_MARK
			case '"', '\'':
				insertSemicolon = true
				tkn = token.STRING
				var err string
				literal, parsedLiteral, err = self.scanString(self.chrOffset-1, true)
				if err != "" {
					tkn = token.ILLEGAL
				}
			case '`':
				tkn = token.BACKTICK
			default:
				self.errorUnexpected(idx, chr)
				tkn = token.ILLEGAL
			}
		}
		self.insertSemicolon = insertSemicolon
		return
	}
}

func (self *_parser) switch2(tkn0, tkn1 token.Token) token.Token {
	if self.chr == '=' {
		self.read()
		return tkn1
	}
	return tkn0
}

func (self *_parser) switch3(tkn0, tkn1 token.Token, chr2 rune, tkn2 token.Token) token.Token {
	if self.chr == '=' {
		self.read()
		return tkn1
	}
	if self.chr == chr2 {
		self.read()
		return tkn2
	}
	return tkn0
}

func (self *_parser) switch4(tkn0, tkn1 token.Token, chr2 rune, tkn2, tkn3 token.Token) token.Token {
	if self.chr == '=' {
		self.read()
		return tkn1
	}
	if self.chr == chr2 {
		self.read()
		if self.chr == '=' {
			self.read()
			return tkn3
		}
		return tkn2
	}
	return tkn0
}

func (self *_parser) switch6(tkn0, tkn1 token.Token, chr2 rune, tkn2, tkn3 token.Token, chr3 rune, tkn4, tkn5 token.Token) token.Token {
	if self.chr == '=' {
		self.read()
		return tkn1
	}
	if self.chr == chr2 {
		self.read()
		if self.chr == '=' {
			self.read()
			return tkn3
		}
		if self.chr == chr3 {
			self.read()
			if self.chr == '=' {
				self.read()
				return tkn5
			}
			return tkn4
		}
		return tkn2
	}
	return tkn0
}

func (self *_parser) _peek() rune {
	if self.offset < self.length {
		return rune(self.str[self.offset])
	}
	return -1
}

func (self *_parser) read() {
	if self.offset < self.length {
		self.chrOffset = self.offset
		chr, width := rune(self.str[self.offset]), 1
		if chr >= utf8.RuneSelf { // !ASCII
			chr, width = utf8.DecodeRuneInString(self.str[self.offset:])
			if chr == utf8.RuneError && width == 1 {
				self.error(self.chrOffset, "Invalid UTF-8 character")
			}
		}
		self.offset += width
		self.chr = chr
	} else {
		self.chrOffset = self.length
		self.chr = -1 // EOF
	}
}

func (self *_parser) skipSingleLineComment() {
	for self.chr != -1 {
		self.read()
		if isLineTerminator(self.chr) {
			return
		}
	}
}

func (self *_parser) skipMultiLineComment() {
	self.read()
	for self.chr >= 0 {
		chr := self.chr
		self.read()
		if chr == '*' && self.chr == '/' {
			self.read()
			return
		}
	}

	self.errorUnexpected(0, self.chr)
}

func (self *_parser) skipWhiteSpace() {
	for {
		switch self.chr {
		case ' ', '\t', '\f', '\v', '\u00a0', '\ufeff':
			self.read()
			continue
		case '\r':
			if self._peek() == '\n' {
				self.read()
			}
			fallthrough
		case '\u2028', '\u2029', '\n':
			if self.insertSemicolon {
				return
			}
			self.read()
			continue
		}
		if self.chr >= utf8.RuneSelf {
			if unicode.IsSpace(self.chr) {
				self.read()
				continue
			}
		}
		break
	}
}

func (self *_parser) skipLineWhiteSpace() {
	for isLineWhiteSpace(self.chr) {
		self.read()
	}
}

func (self *_parser) scanMantissa(base int) {
	for digitValue(self.chr) < base {
		self.read()
	}
}

func (self *_parser) scanEscape(quote rune) (int, bool) {

	var length, base uint32
	chr := self.chr
	switch chr {
	case '0', '1', '2', '3', '4', '5', '6', '7':
		//    Octal:
		length, base = 3, 8
	case 'a', 'b', 'f', 'n', 'r', 't', 'v', '\\', '"', '\'':
		self.read()
		return 1, false
	case '\r':
		self.read()
		if self.chr == '\n' {
			self.read()
			return 2, false
		}
		return 1, false
	case '\n':
		self.read()
		return 1, false
	case '\u2028', '\u2029':
		self.read()
		return 1, true
	case 'x':
		self.read()
		length, base = 2, 16
	case 'u':
		self.read()
		if self.chr == '{' {
			self.read()
			length, base = 0, 16
		} else {
			length, base = 4, 16
		}
	default:
		self.read() // Always make progress
	}

	if base > 0 {
		var value uint32
		if length > 0 {
			for ; length > 0 && self.chr != quote && self.chr >= 0; length-- {
				digit := uint32(digitValue(self.chr))
				if digit >= base {
					break
				}
				value = value*base + digit
				self.read()
			}
		} else {
			for self.chr != quote && self.chr >= 0 && value < utf8.MaxRune {
				if self.chr == '}' {
					self.read()
					break
				}
				digit := uint32(digitValue(self.chr))
				if digit >= base {
					break
				}
				value = value*base + digit
				self.read()
			}
		}
		chr = rune(value)
	}
	if chr >= utf8.RuneSelf {
		if chr > 0xFFFF {
			return 2, true
		}
		return 1, true
	}
	return 1, false
}

func (self *_parser) scanString(offset int, parse bool) (literal string, parsed unistring.String, err string) {
	// " ' /
	quote := rune(self.str[offset])
	length := 0
	isUnicode := false
	for self.chr != quote {
		chr := self.chr
		if chr == '\n' || chr == '\r' || chr == '\u2028' || chr == '\u2029' || chr < 0 {
			goto newline
		}
		self.read()
		if chr == '\\' {
			if self.chr == '\n' || self.chr == '\r' || self.chr == '\u2028' || self.chr == '\u2029' || self.chr < 0 {
				if quote == '/' {
					goto newline
				}
				self.scanNewline()
			} else {
				l, u := self.scanEscape(quote)
				length += l
				if u {
					isUnicode = true
				}
			}
			continue
		} else if chr == '[' && quote == '/' {
			// Allow a slash (/) in a bracket character class ([...])
			// TODO Fix this, this is hacky...
			quote = -1
		} else if chr == ']' && quote == -1 {
			quote = '/'
		}
		if chr >= utf8.RuneSelf {
			isUnicode = true
			if chr > 0xFFFF {
				length++
			}
		}
		length++
	}

	// " ' /
	self.read()
	literal = self.str[offset:self.chrOffset]
	if parse {
		// TODO strict
		parsed, err = parseStringLiteral(literal[1:len(literal)-1], length, isUnicode, false)
	}
	return

newline:
	self.scanNewline()
	errStr := "String not terminated"
	if quote == '/' {
		errStr = "Invalid regular expression: missing /"
		self.error(self.idxOf(offset), errStr)
	}
	return "", "", errStr
}

func (self *_parser) scanNewline() {
	if self.chr == '\r' {
		self.read()
		if self.chr != '\n' {
			return
		}
	}
	self.read()
}

func (self *_parser) parseTemplateCharacters() (literal string, parsed unistring.String, finished bool, parseErr, err string) {
	offset := self.chrOffset
	var end int
	length := 0
	isUnicode := false
	hasCR := false
	for {
		chr := self.chr
		if chr < 0 {
			goto unterminated
		}
		self.read()
		if chr == '`' {
			finished = true
			end = self.chrOffset - 1
			break
		}
		if chr == '\\' {
			if self.chr == '\n' || self.chr == '\r' || self.chr == '\u2028' || self.chr == '\u2029' || self.chr < 0 {
				if self.chr == '\r' {
					hasCR = true
				}
				self.scanNewline()
			} else {
				if self.chr == '8' || self.chr == '9' {
					if parseErr == "" {
						parseErr = "\\8 and \\9 are not allowed in template strings."
					}
				}
				l, u := self.scanEscape('`')
				length += l
				if u {
					isUnicode = true
				}
			}
			continue
		}
		if chr == '$' && self.chr == '{' {
			self.read()
			end = self.chrOffset - 2
			break
		}
		if chr >= utf8.RuneSelf {
			isUnicode = true
			if chr > 0xFFFF {
				length++
			}
		} else if chr == '\r' {
			hasCR = true
			if self.chr == '\n' {
				length--
			}
		}
		length++
	}
	literal = self.str[offset:end]
	if hasCR {
		literal = normaliseCRLF(literal)
	}
	if parseErr == "" {
		parsed, parseErr = parseStringLiteral(literal, length, isUnicode, true)
	}
	self.insertSemicolon = true
	return
unterminated:
	err = err_UnexpectedEndOfInput
	finished = true
	return
}

func normaliseCRLF(s string) string {
	var buf strings.Builder
	buf.Grow(len(s))
	for i := 0; i < len(s); i++ {
		if s[i] == '\r' {
			buf.WriteByte('\n')
			if i < len(s)-1 && s[i+1] == '\n' {
				i++
			}
		} else {
			buf.WriteByte(s[i])
		}
	}
	return buf.String()
}

func hex2decimal(chr byte) (value rune, ok bool) {
	{
		chr := rune(chr)
		switch {
		case '0' <= chr && chr <= '9':
			return chr - '0', true
		case 'a' <= chr && chr <= 'f':
			return chr - 'a' + 10, true
		case 'A' <= chr && chr <= 'F':
			return chr - 'A' + 10, true
		}
		return
	}
}

func parseNumberLiteral(literal string) (value interface{}, err error) {
	// TODO Is Uint okay? What about -MAX_UINT
	value, err = strconv.ParseInt(literal, 0, 64)
	if err == nil {
		return
	}

	parseIntErr := err // Save this first error, just in case

	value, err = strconv.ParseFloat(literal, 64)
	if err == nil {
		return
	} else if err.(*strconv.NumError).Err == strconv.ErrRange {
		// Infinity, etc.
		return value, nil
	}

	err = parseIntErr

	if err.(*strconv.NumError).Err == strconv.ErrRange {
		if len(literal) > 2 && literal[0] == '0' && (literal[1] == 'X' || literal[1] == 'x') {
			// Could just be a very large number (e.g. 0x8000000000000000)
			var value float64
			literal = literal[2:]
			for _, chr := range literal {
				digit := digitValue(chr)
				if digit >= 16 {
					goto error
				}
				value = value*16 + float64(digit)
			}
			return value, nil
		}
	}

error:
	return nil, errors.New("Illegal numeric literal")
}

func parseStringLiteral(literal string, length int, unicode, strict bool) (unistring.String, string) {
	var sb strings.Builder
	var chars []uint16
	if unicode {
		chars = make([]uint16, 1, length+1)
		chars[0] = unistring.BOM
	} else {
		sb.Grow(length)
	}
	str := literal
	for len(str) > 0 {
		switch chr := str[0]; {
		// We do not explicitly handle the case of the quote
		// value, which can be: " ' /
		// This assumes we're already passed a partially well-formed literal
		case chr >= utf8.RuneSelf:
			chr, size := utf8.DecodeRuneInString(str)
			if chr <= 0xFFFF {
				chars = append(chars, uint16(chr))
			} else {
				first, second := utf16.EncodeRune(chr)
				chars = append(chars, uint16(first), uint16(second))
			}
			str = str[size:]
			continue
		case chr != '\\':
			if unicode {
				chars = append(chars, uint16(chr))
			} else {
				sb.WriteByte(chr)
			}
			str = str[1:]
			continue
		}

		if len(str) <= 1 {
			panic("len(str) <= 1")
		}
		chr := str[1]
		var value rune
		if chr >= utf8.RuneSelf {
			str = str[1:]
			var size int
			value, size = utf8.DecodeRuneInString(str)
			str = str[size:] // \ + <character>
			if value == '\u2028' || value == '\u2029' {
				continue
			}
		} else {
			str = str[2:] // \<character>
			switch chr {
			case 'b':
				value = '\b'
			case 'f':
				value = '\f'
			case 'n':
				value = '\n'
			case 'r':
				value = '\r'
			case 't':
				value = '\t'
			case 'v':
				value = '\v'
			case 'x', 'u':
				size := 0
				switch chr {
				case 'x':
					size = 2
				case 'u':
					if str == "" || str[0] != '{' {
						size = 4
					}
				}
				if size > 0 {
					if len(str) < size {
						return "", fmt.Sprintf("invalid escape: \\%s: len(%q) != %d", string(chr), str, size)
					}
					for j := 0; j < size; j++ {
						decimal, ok := hex2decimal(str[j])
						if !ok {
							return "", fmt.Sprintf("invalid escape: \\%s: %q", string(chr), str[:size])
						}
						value = value<<4 | decimal
					}
				} else {
					str = str[1:]
					var val rune
					value = -1
					for ; size < len(str); size++ {
						if str[size] == '}' {
							if size == 0 {
								return "", fmt.Sprintf("invalid escape: \\%s", string(chr))
							}
							size++
							value = val
							break
						}
						decimal, ok := hex2decimal(str[size])
						if !ok {
							return "", fmt.Sprintf("invalid escape: \\%s: %q", string(chr), str[:size+1])
						}
						val = val<<4 | decimal
						if val > utf8.MaxRune {
							return "", fmt.Sprintf("undefined Unicode code-point: %q", str[:size+1])
						}
					}
					if value == -1 {
						return "", fmt.Sprintf("unterminated \\u{: %q", str)
					}
				}
				str = str[size:]
				if chr == 'x' {
					break
				}
				if value > utf8.MaxRune {
					panic("value > utf8.MaxRune")
				}
			case '0':
				if len(str) == 0 || '0' > str[0] || str[0] > '7' {
					value = 0
					break
				}
				fallthrough
			case '1', '2', '3', '4', '5', '6', '7':
				if strict {
					return "", "Octal escape sequences are not allowed in this context"
				}
				value = rune(chr) - '0'
				j := 0
				for ; j < 2; j++ {
					if len(str) < j+1 {
						break
					}
					chr := str[j]
					if '0' > chr || chr > '7' {
						break
					}
					decimal := rune(str[j]) - '0'
					value = (value << 3) | decimal
				}
				str = str[j:]
			case '\\':
				value = '\\'
			case '\'', '"':
				value = rune(chr)
			case '\r':
				if len(str) > 0 {
					if str[0] == '\n' {
						str = str[1:]
					}
				}
				fallthrough
			case '\n':
				continue
			default:
				value = rune(chr)
			}
		}
		if unicode {
			if value <= 0xFFFF {
				chars = append(chars, uint16(value))
			} else {
				first, second := utf16.EncodeRune(value)
				chars = append(chars, uint16(first), uint16(second))
			}
		} else {
			if value >= utf8.RuneSelf {
				return "", "Unexpected unicode character"
			}
			sb.WriteByte(byte(value))
		}
	}

	if unicode {
		if len(chars) != length+1 {
			panic(fmt.Errorf("unexpected unicode length while parsing '%s'", literal))
		}
		return unistring.FromUtf16(chars), ""
	}
	if sb.Len() != length {
		panic(fmt.Errorf("unexpected length while parsing '%s'", literal))
	}
	return unistring.String(sb.String()), ""
}

func (self *_parser) scanNumericLiteral(decimalPoint bool) (token.Token, string) {

	offset := self.chrOffset
	tkn := token.NUMBER

	if decimalPoint {
		offset--
		self.scanMantissa(10)
	} else {
		if self.chr == '0' {
			self.read()
			base := 0
			switch self.chr {
			case 'x', 'X':
				base = 16
			case 'o', 'O':
				base = 8
			case 'b', 'B':
				base = 2
			case '.', 'e', 'E':
				// no-op
			default:
				// legacy octal
				self.scanMantissa(8)
				goto end
			}
			if base > 0 {
				self.read()
				if !isDigit(self.chr, base) {
					return token.ILLEGAL, self.str[offset:self.chrOffset]
				}
				self.scanMantissa(base)
				goto end
			}
		} else {
			self.scanMantissa(10)
		}
		if self.chr == '.' {
			self.read()
			self.scanMantissa(10)
		}
	}

	if self.chr == 'e' || self.chr == 'E' {
		self.read()
		if self.chr == '-' || self.chr == '+' {
			self.read()
		}
		if isDecimalDigit(self.chr) {
			self.read()
			self.scanMantissa(10)
		} else {
			return token.ILLEGAL, self.str[offset:self.chrOffset]
		}
	}
end:
	if isIdentifierStart(self.chr) || isDecimalDigit(self.chr) {
		return token.ILLEGAL, self.str[offset:self.chrOffset]
	}

	return tkn, self.str[offset:self.chrOffset]
}