/** The global unescape method, as per ECMA-262 15.1.2.5. */
private Object js_unescape(Object[] args) {
String s = ScriptRuntime.toString(args, 0);
int firstEscapePos = s.indexOf('%');
if (firstEscapePos >= 0) {
int L = s.length();
char[] buf = s.toCharArray();
int destination = firstEscapePos;
for (int k = firstEscapePos; k != L; ) {
char c = buf[k];
++k;
if (c == '%' && k != L) {
int end, start;
if (buf[k] == 'u') {
start = k + 1;
end = k + 5;
} else {
start = k;
end = k + 2;
}
if (end <= L) {
int x = 0;
for (int i = start; i != end; ++i) {
x = Kit.xDigitToInt(buf[i], x);
}
if (x >= 0) {
c = (char) x;
k = end;
}
}
}
buf[destination] = c;
++destination;
}
s = new String(buf, 0, destination);
}
return s;
}
final int getToken() throws IOException {
int c;
retry:
for (; ; ) {
// Eat whitespace, possibly sensitive to newlines.
for (; ; ) {
c = getChar();
if (c == EOF_CHAR) {
tokenBeg = cursor - 1;
tokenEnd = cursor;
return Token.EOF;
} else if (c == '\n') {
dirtyLine = false;
tokenBeg = cursor - 1;
tokenEnd = cursor;
return Token.EOL;
} else if (!isJSSpace(c)) {
if (c != '-') {
dirtyLine = true;
}
break;
}
}
// Assume the token will be 1 char - fixed up below.
tokenBeg = cursor - 1;
tokenEnd = cursor;
if (c == '@') return Token.XMLATTR;
// identifier/keyword/instanceof?
// watch out for starting with a <backslash>
boolean identifierStart;
boolean isUnicodeEscapeStart = false;
if (c == '\\') {
c = getChar();
if (c == 'u') {
identifierStart = true;
isUnicodeEscapeStart = true;
stringBufferTop = 0;
} else {
identifierStart = false;
ungetChar(c);
c = '\\';
}
} else {
identifierStart = Character.isJavaIdentifierStart((char) c);
if (identifierStart) {
stringBufferTop = 0;
addToString(c);
}
}
if (identifierStart) {
boolean containsEscape = isUnicodeEscapeStart;
for (; ; ) {
if (isUnicodeEscapeStart) {
// strictly speaking we should probably push-back
// all the bad characters if the <backslash>uXXXX
// sequence is malformed. But since there isn't a
// correct context(is there?) for a bad Unicode
// escape sequence in an identifier, we can report
// an error here.
int escapeVal = 0;
for (int i = 0; i != 4; ++i) {
c = getChar();
escapeVal = Kit.xDigitToInt(c, escapeVal);
// Next check takes care about c < 0 and bad escape
if (escapeVal < 0) {
break;
}
}
if (escapeVal < 0) {
parser.addError("msg.invalid.escape");
return Token.ERROR;
}
addToString(escapeVal);
isUnicodeEscapeStart = false;
} else {
c = getChar();
if (c == '\\') {
c = getChar();
if (c == 'u') {
isUnicodeEscapeStart = true;
containsEscape = true;
} else {
parser.addError("msg.illegal.character");
return Token.ERROR;
}
} else {
if (c == EOF_CHAR
|| c == BYTE_ORDER_MARK
|| !Character.isJavaIdentifierPart((char) c)) {
break;
}
addToString(c);
}
}
}
ungetChar(c);
String str = getStringFromBuffer();
if (!containsEscape) {
// OPT we shouldn't have to make a string (object!) to
// check if it's a keyword.
// Return the corresponding token if it's a keyword
int result = stringToKeyword(str);
if (result != Token.EOF) {
if ((result == Token.LET || result == Token.YIELD)
&& parser.compilerEnv.getLanguageVersion() < Context.VERSION_1_7) {
// LET and YIELD are tokens only in 1.7 and later
string = result == Token.LET ? "let" : "yield";
result = Token.NAME;
}
if (result != Token.RESERVED) {
return result;
} else if (!parser.compilerEnv.isReservedKeywordAsIdentifier()) {
return result;
}
}
}
this.string = (String) allStrings.intern(str);
return Token.NAME;
}
// is it a number?
if (isDigit(c) || (c == '.' && isDigit(peekChar()))) {
isOctal = false;
stringBufferTop = 0;
int base = 10;
if (c == '0') {
c = getChar();
if (c == 'x' || c == 'X') {
base = 16;
c = getChar();
} else if (isDigit(c)) {
base = 8;
isOctal = true;
} else {
addToString('0');
}
}
if (base == 16) {
while (0 <= Kit.xDigitToInt(c, 0)) {
addToString(c);
c = getChar();
}
} else {
while ('0' <= c && c <= '9') {
/*
* We permit 08 and 09 as decimal numbers, which
* makes our behavior a superset of the ECMA
* numeric grammar. We might not always be so
* permissive, so we warn about it.
*/
if (base == 8 && c >= '8') {
parser.addWarning("msg.bad.octal.literal", c == '8' ? "8" : "9");
base = 10;
}
addToString(c);
c = getChar();
}
}
boolean isInteger = true;
if (base == 10 && (c == '.' || c == 'e' || c == 'E')) {
isInteger = false;
if (c == '.') {
do {
addToString(c);
c = getChar();
} while (isDigit(c));
}
if (c == 'e' || c == 'E') {
addToString(c);
c = getChar();
if (c == '+' || c == '-') {
addToString(c);
c = getChar();
}
if (!isDigit(c)) {
parser.addError("msg.missing.exponent");
return Token.ERROR;
}
do {
addToString(c);
c = getChar();
} while (isDigit(c));
}
}
ungetChar(c);
String numString = getStringFromBuffer();
this.string = numString;
double dval;
if (base == 10 && !isInteger) {
try {
// Use Java conversion to number from string...
dval = Double.valueOf(numString).doubleValue();
} catch (NumberFormatException ex) {
parser.addError("msg.caught.nfe");
return Token.ERROR;
}
} else {
dval = ScriptRuntime.stringToNumber(numString, 0, base);
}
this.number = dval;
return Token.NUMBER;
}
// is it a string?
if (c == '"' || c == '\'') {
// We attempt to accumulate a string the fast way, by
// building it directly out of the reader. But if there
// are any escaped characters in the string, we revert to
// building it out of a StringBuffer.
quoteChar = c;
stringBufferTop = 0;
c = getChar();
strLoop:
while (c != quoteChar) {
if (c == '\n' || c == EOF_CHAR) {
ungetChar(c);
tokenEnd = cursor;
parser.addError("msg.unterminated.string.lit");
return Token.ERROR;
}
if (c == '\\') {
// We've hit an escaped character
int escapeVal;
c = getChar();
switch (c) {
case 'b':
c = '\b';
break;
case 'f':
c = '\f';
break;
case 'n':
c = '\n';
break;
case 'r':
c = '\r';
break;
case 't':
c = '\t';
break;
// \v a late addition to the ECMA spec,
// it is not in Java, so use 0xb
case 'v':
c = 0xb;
break;
case 'u':
// Get 4 hex digits; if the u escape is not
// followed by 4 hex digits, use 'u' + the
// literal character sequence that follows.
int escapeStart = stringBufferTop;
addToString('u');
escapeVal = 0;
for (int i = 0; i != 4; ++i) {
c = getChar();
escapeVal = Kit.xDigitToInt(c, escapeVal);
if (escapeVal < 0) {
continue strLoop;
}
addToString(c);
}
// prepare for replace of stored 'u' sequence
// by escape value
stringBufferTop = escapeStart;
c = escapeVal;
break;
case 'x':
// Get 2 hex digits, defaulting to 'x'+literal
// sequence, as above.
c = getChar();
escapeVal = Kit.xDigitToInt(c, 0);
if (escapeVal < 0) {
addToString('x');
continue strLoop;
} else {
int c1 = c;
c = getChar();
escapeVal = Kit.xDigitToInt(c, escapeVal);
if (escapeVal < 0) {
addToString('x');
addToString(c1);
continue strLoop;
} else {
// got 2 hex digits
c = escapeVal;
}
}
break;
case '\n':
// Remove line terminator after escape to follow
// SpiderMonkey and C/C++
c = getChar();
continue strLoop;
default:
if ('0' <= c && c < '8') {
int val = c - '0';
c = getChar();
if ('0' <= c && c < '8') {
val = 8 * val + c - '0';
c = getChar();
if ('0' <= c && c < '8' && val <= 037) {
// c is 3rd char of octal sequence only
// if the resulting val <= 0377
val = 8 * val + c - '0';
c = getChar();
}
}
ungetChar(c);
c = val;
}
}
}
addToString(c);
c = getChar();
}
String str = getStringFromBuffer();
this.string = (String) allStrings.intern(str);
return Token.STRING;
}
switch (c) {
case ';':
return Token.SEMI;
case '[':
return Token.LB;
case ']':
return Token.RB;
case '{':
return Token.LC;
case '}':
return Token.RC;
case '(':
return Token.LP;
case ')':
return Token.RP;
case ',':
return Token.COMMA;
case '?':
return Token.HOOK;
case ':':
if (matchChar(':')) {
return Token.COLONCOLON;
} else {
return Token.COLON;
}
case '.':
if (matchChar('.')) {
return Token.DOTDOT;
} else if (matchChar('(')) {
return Token.DOTQUERY;
} else {
return Token.DOT;
}
case '|':
if (matchChar('|')) {
return Token.OR;
} else if (matchChar('=')) {
return Token.ASSIGN_BITOR;
} else {
return Token.BITOR;
}
case '^':
if (matchChar('=')) {
return Token.ASSIGN_BITXOR;
} else {
return Token.BITXOR;
}
case '&':
if (matchChar('&')) {
return Token.AND;
} else if (matchChar('=')) {
return Token.ASSIGN_BITAND;
} else {
return Token.BITAND;
}
case '=':
if (matchChar('=')) {
if (matchChar('=')) {
return Token.SHEQ;
} else {
return Token.EQ;
}
} else {
return Token.ASSIGN;
}
case '!':
if (matchChar('=')) {
if (matchChar('=')) {
return Token.SHNE;
} else {
return Token.NE;
}
} else {
return Token.NOT;
}
case '<':
/* NB:treat HTML begin-comment as comment-till-eol */
if (matchChar('!')) {
if (matchChar('-')) {
if (matchChar('-')) {
tokenBeg = cursor - 4;
skipLine();
commentType = Token.CommentType.HTML;
return Token.COMMENT;
}
ungetCharIgnoreLineEnd('-');
}
ungetCharIgnoreLineEnd('!');
}
if (matchChar('<')) {
if (matchChar('=')) {
return Token.ASSIGN_LSH;
} else {
return Token.LSH;
}
} else {
if (matchChar('=')) {
return Token.LE;
} else {
return Token.LT;
}
}
case '>':
if (matchChar('>')) {
if (matchChar('>')) {
if (matchChar('=')) {
return Token.ASSIGN_URSH;
} else {
return Token.URSH;
}
} else {
if (matchChar('=')) {
return Token.ASSIGN_RSH;
} else {
return Token.RSH;
}
}
} else {
if (matchChar('=')) {
return Token.GE;
} else {
return Token.GT;
}
}
case '*':
if (matchChar('=')) {
return Token.ASSIGN_MUL;
} else {
return Token.MUL;
}
case '/':
markCommentStart();
// is it a // comment?
if (matchChar('/')) {
tokenBeg = cursor - 2;
skipLine();
commentType = Token.CommentType.LINE;
return Token.COMMENT;
}
// is it a /* or /** comment?
if (matchChar('*')) {
boolean lookForSlash = false;
tokenBeg = cursor - 2;
if (matchChar('*')) {
lookForSlash = true;
commentType = Token.CommentType.JSDOC;
} else {
commentType = Token.CommentType.BLOCK_COMMENT;
}
for (; ; ) {
c = getChar();
if (c == EOF_CHAR) {
tokenEnd = cursor - 1;
parser.addError("msg.unterminated.comment");
return Token.COMMENT;
} else if (c == '*') {
lookForSlash = true;
} else if (c == '/') {
if (lookForSlash) {
tokenEnd = cursor;
return Token.COMMENT;
}
} else {
lookForSlash = false;
tokenEnd = cursor;
}
}
}
if (matchChar('=')) {
return Token.ASSIGN_DIV;
} else {
return Token.DIV;
}
case '%':
if (matchChar('=')) {
return Token.ASSIGN_MOD;
} else {
return Token.MOD;
}
case '~':
return Token.BITNOT;
case '+':
if (matchChar('=')) {
return Token.ASSIGN_ADD;
} else if (matchChar('+')) {
return Token.INC;
} else {
return Token.ADD;
}
case '-':
if (matchChar('=')) {
c = Token.ASSIGN_SUB;
} else if (matchChar('-')) {
if (!dirtyLine) {
// treat HTML end-comment after possible whitespace
// after line start as comment-until-eol
if (matchChar('>')) {
markCommentStart("--");
skipLine();
commentType = Token.CommentType.HTML;
return Token.COMMENT;
}
}
c = Token.DEC;
} else {
c = Token.SUB;
}
dirtyLine = true;
return c;
default:
parser.addError("msg.illegal.character");
return Token.ERROR;
}
}
}