/*
* Convert any jsval v to an integer jsval if ToString(v)
* contains a base-10 integer that fits into 31 bits.
* Otherwise return v.
*/
static jsval
try_convert_to_jsint(JSContext *cx, jsval idval)
{
const jschar *cp;
JSString *jsstr;
jsstr = JS_ValueToString(cx, idval);
if (!jsstr)
return idval;
cp = JS_GetStringChars(jsstr);
if (JS7_ISDEC(*cp)) {
jsuint index = JS7_UNDEC(*cp++);
jsuint oldIndex = 0;
jsuint c = 0;
if (index != 0) {
while (JS7_ISDEC(*cp)) {
oldIndex = index;
c = JS7_UNDEC(*cp);
index = 10*index + c;
cp++;
}
}
if (*cp == 0 &&
(oldIndex < (JSVAL_INT_MAX / 10) ||
(oldIndex == (JSVAL_INT_MAX / 10) && c < (JSVAL_INT_MAX % 10)))) {
return INT_TO_JSVAL(index);
}
}
return idval;
}
/*
* Determine if the id represents an array index.
*
* An id is an array index according to ECMA by (15.4):
*
* "Array objects give special treatment to a certain class of property names.
* A property name P (in the form of a string value) is an array index if and
* only if ToString(ToUint32(P)) is equal to P and ToUint32(P) is not equal
* to 2^32-1."
*
* In our implementation, it would be sufficient to check for JSVAL_IS_INT(id)
* except that by using signed 32-bit integers we miss the top half of the
* valid range. This function checks the string representation itself; note
* that calling a standard conversion routine might allow strings such as
* "08" or "4.0" as array indices, which they are not.
*/
static JSBool
IdIsIndex(jsid id, jsuint *indexp)
{
JSString *str;
jschar *cp;
if (JSVAL_IS_INT(id)) {
jsint i;
i = JSVAL_TO_INT(id);
if (i < 0)
return JS_FALSE;
*indexp = (jsuint)i;
return JS_TRUE;
}
/* It must be a string. */
str = JSVAL_TO_STRING(id);
cp = JSSTRING_CHARS(str);
if (JS7_ISDEC(*cp) && JSSTRING_LENGTH(str) < sizeof(MAXSTR)) {
jsuint index = JS7_UNDEC(*cp++);
jsuint oldIndex = 0;
jsuint c = 0;
if (index != 0) {
while (JS7_ISDEC(*cp)) {
oldIndex = index;
c = JS7_UNDEC(*cp);
index = 10*index + c;
cp++;
}
}
/* Make sure all characters were consumed and that it couldn't
* have overflowed.
*/
if (*cp == 0 &&
(oldIndex < (MAXINDEX / 10) ||
(oldIndex == (MAXINDEX / 10) && c < (MAXINDEX % 10))))
{
*indexp = index;
return JS_TRUE;
}
}
return JS_FALSE;
}
JSTokenType
js_GetToken(JSContext *cx, JSTokenStream *ts)
{
JSTokenType tt;
JSToken *tp;
int32 c;
JSAtom *atom;
JSBool hadUnicodeEscape;
#define INIT_TOKENBUF(tb) ((tb)->ptr = (tb)->base)
#define FINISH_TOKENBUF(tb) if (!AddToTokenBuf(cx, tb, 0)) RETURN(TOK_ERROR)
#define TOKEN_LENGTH(tb) ((tb)->ptr - (tb)->base - 1)
#define RETURN(tt) { if (tt == TOK_ERROR) ts->flags |= TSF_ERROR; \
tp->pos.end.index = ts->linepos + \
(ts->linebuf.ptr - ts->linebuf.base) - \
ts->ungetpos; \
return (tp->type = tt); }
/* If there was a fatal error, keep returning TOK_ERROR. */
if (ts->flags & TSF_ERROR)
return TOK_ERROR;
/* Check for a pushed-back token resulting from mismatching lookahead. */
while (ts->lookahead != 0) {
ts->lookahead--;
ts->cursor = (ts->cursor + 1) & NTOKENS_MASK;
tt = CURRENT_TOKEN(ts).type;
if (tt != TOK_EOL || (ts->flags & TSF_NEWLINES))
return tt;
}
retry:
do {
c = GetChar(ts);
if (c == '\n') {
ts->flags &= ~TSF_DIRTYLINE;
if (ts->flags & TSF_NEWLINES)
break;
}
} while (JS_ISSPACE(c));
ts->cursor = (ts->cursor + 1) & NTOKENS_MASK;
tp = &CURRENT_TOKEN(ts);
tp->ptr = ts->linebuf.ptr - 1;
tp->pos.begin.index = ts->linepos + (tp->ptr - ts->linebuf.base);
tp->pos.begin.lineno = tp->pos.end.lineno = ts->lineno;
if (c == EOF)
RETURN(TOK_EOF);
if (c != '-' && c != '\n')
ts->flags |= TSF_DIRTYLINE;
hadUnicodeEscape = JS_FALSE;
if (JS_ISIDENT_START(c) ||
(c == '\\' &&
(c = GetUnicodeEscape(ts),
hadUnicodeEscape = JS_ISIDENT_START(c)))) {
INIT_TOKENBUF(&ts->tokenbuf);
for (;;) {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
if (c == '\\') {
c = GetUnicodeEscape(ts);
if (!JS_ISIDENT(c))
break;
hadUnicodeEscape = JS_TRUE;
} else {
if (!JS_ISIDENT(c))
break;
}
}
UngetChar(ts, c);
FINISH_TOKENBUF(&ts->tokenbuf);
atom = js_AtomizeChars(cx,
ts->tokenbuf.base,
TOKEN_LENGTH(&ts->tokenbuf),
0);
if (!atom)
RETURN(TOK_ERROR);
if (!hadUnicodeEscape && ATOM_KEYWORD(atom)) {
struct keyword *kw = ATOM_KEYWORD(atom);
if (JSVERSION_IS_ECMA(cx->version) || kw->version <= cx->version) {
tp->t_op = (JSOp) kw->op;
RETURN(kw->tokentype);
}
}
tp->t_op = JSOP_NAME;
tp->t_atom = atom;
RETURN(TOK_NAME);
}
if (JS7_ISDEC(c) || (c == '.' && JS7_ISDEC(PeekChar(ts)))) {
jsint radix;
const jschar *endptr;
jsdouble dval;
radix = 10;
INIT_TOKENBUF(&ts->tokenbuf);
if (c == '0') {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
if (JS_TOLOWER(c) == 'x') {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
radix = 16;
} else if (JS7_ISDEC(c)) {
radix = 8;
}
}
while (JS7_ISHEX(c)) {
if (radix < 16) {
if (JS7_ISLET(c))
break;
/*
* We permit 08 and 09 as decimal numbers, which makes our
* behaviour a superset of the ECMA numeric grammar. We might
* not always be so permissive, so we warn about it.
*/
if (radix == 8 && c >= '8') {
if (!js_ReportCompileErrorNumber(cx, ts, NULL,
JSREPORT_WARNING,
JSMSG_BAD_OCTAL,
c == '8' ? "08" : "09")) {
RETURN(TOK_ERROR);
}
radix = 10;
}
}
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
}
if (radix == 10 && (c == '.' || JS_TOLOWER(c) == 'e')) {
if (c == '.') {
do {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
} while (JS7_ISDEC(c));
}
if (JS_TOLOWER(c) == 'e') {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
if (c == '+' || c == '-') {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
}
if (!JS7_ISDEC(c)) {
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_MISSING_EXPONENT);
RETURN(TOK_ERROR);
}
do {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
} while (JS7_ISDEC(c));
}
}
UngetChar(ts, c);
FINISH_TOKENBUF(&ts->tokenbuf);
if (radix == 10) {
if (!js_strtod(cx, ts->tokenbuf.base, &endptr, &dval)) {
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_OUT_OF_MEMORY);
RETURN(TOK_ERROR);
}
} else {
if (!js_strtointeger(cx, ts->tokenbuf.base, &endptr, radix, &dval)) {
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_OUT_OF_MEMORY);
RETURN(TOK_ERROR);
}
}
tp->t_dval = dval;
RETURN(TOK_NUMBER);
}
if (c == '"' || c == '\'') {
int32 val, qc = c;
INIT_TOKENBUF(&ts->tokenbuf);
while ((c = GetChar(ts)) != qc) {
if (c == '\n' || c == EOF) {
UngetChar(ts, c);
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_UNTERMINATED_STRING);
RETURN(TOK_ERROR);
}
if (c == '\\') {
switch (c = GetChar(ts)) {
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;
case 'v': c = '\v'; break;
default:
if ('0' <= c && c < '8') {
val = JS7_UNDEC(c);
c = PeekChar(ts);
if ('0' <= c && c < '8') {
val = 8 * val + JS7_UNDEC(c);
GetChar(ts);
c = PeekChar(ts);
if ('0' <= c && c < '8') {
int32 save = val;
val = 8 * val + JS7_UNDEC(c);
if (val <= 0377)
GetChar(ts);
else
val = save;
}
}
c = (jschar)val;
} else if (c == 'u') {
jschar cp[4];
if (PeekChars(ts, 4, cp) &&
JS7_ISHEX(cp[0]) && JS7_ISHEX(cp[1]) &&
JS7_ISHEX(cp[2]) && JS7_ISHEX(cp[3])) {
c = (((((JS7_UNHEX(cp[0]) << 4)
+ JS7_UNHEX(cp[1])) << 4)
+ JS7_UNHEX(cp[2])) << 4)
+ JS7_UNHEX(cp[3]);
SkipChars(ts, 4);
}
} else if (c == 'x') {
jschar cp[2];
if (PeekChars(ts, 2, cp) &&
JS7_ISHEX(cp[0]) && JS7_ISHEX(cp[1])) {
c = (JS7_UNHEX(cp[0]) << 4) + JS7_UNHEX(cp[1]);
SkipChars(ts, 2);
}
} else if (c == '\n' && JSVERSION_IS_ECMA(cx->version)) {
/* ECMA follows C by removing escaped newlines. */
continue;
}
break;
}
}
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
}
FINISH_TOKENBUF(&ts->tokenbuf);
atom = js_AtomizeChars(cx,
ts->tokenbuf.base,
TOKEN_LENGTH(&ts->tokenbuf),
0);
if (!atom)
RETURN(TOK_ERROR);
tp->pos.end.lineno = ts->lineno;
tp->t_op = JSOP_STRING;
tp->t_atom = atom;
RETURN(TOK_STRING);
}
switch (c) {
case '\n':
c = TOK_EOL;
break;
case ';': c = TOK_SEMI; break;
case '.': c = TOK_DOT; break;
case '[': c = TOK_LB; break;
case ']': c = TOK_RB; break;
case '{': c = TOK_LC; break;
case '}': c = TOK_RC; break;
case '(': c = TOK_LP; break;
case ')': c = TOK_RP; break;
case ',': c = TOK_COMMA; break;
case '?': c = TOK_HOOK; break;
case ':':
/*
* Default so compiler can modify to JSOP_GETTER if 'p getter: v' in an
* object initializer, likewise for setter.
*/
tp->t_op = JSOP_NOP;
c = TOK_COLON;
break;
case '|':
if (MatchChar(ts, c)) {
c = TOK_OR;
} else if (MatchChar(ts, '=')) {
tp->t_op = JSOP_BITOR;
c = TOK_ASSIGN;
} else {
c = TOK_BITOR;
}
break;
case '^':
if (MatchChar(ts, '=')) {
tp->t_op = JSOP_BITXOR;
c = TOK_ASSIGN;
} else {
c = TOK_BITXOR;
}
break;
case '&':
if (MatchChar(ts, c)) {
c = TOK_AND;
} else if (MatchChar(ts, '=')) {
tp->t_op = JSOP_BITAND;
c = TOK_ASSIGN;
} else {
c = TOK_BITAND;
}
break;
case '=':
if (MatchChar(ts, c)) {
#if JS_HAS_TRIPLE_EQOPS
tp->t_op = MatchChar(ts, c) ? JSOP_NEW_EQ : (JSOp)cx->jsop_eq;
#else
tp->t_op = cx->jsop_eq;
#endif
c = TOK_EQOP;
} else {
tp->t_op = JSOP_NOP;
c = TOK_ASSIGN;
}
break;
case '!':
if (MatchChar(ts, '=')) {
#if JS_HAS_TRIPLE_EQOPS
tp->t_op = MatchChar(ts, '=') ? JSOP_NEW_NE : (JSOp)cx->jsop_ne;
#else
tp->t_op = cx->jsop_ne;
#endif
c = TOK_EQOP;
} else {
tp->t_op = JSOP_NOT;
c = TOK_UNARYOP;
}
break;
case '<':
/* NB: treat HTML begin-comment as comment-till-end-of-line */
if (MatchChar(ts, '!')) {
if (MatchChar(ts, '-')) {
if (MatchChar(ts, '-'))
goto skipline;
UngetChar(ts, '-');
}
UngetChar(ts, '!');
}
if (MatchChar(ts, c)) {
tp->t_op = JSOP_LSH;
c = MatchChar(ts, '=') ? TOK_ASSIGN : TOK_SHOP;
} else {
tp->t_op = MatchChar(ts, '=') ? JSOP_LE : JSOP_LT;
c = TOK_RELOP;
}
break;
case '>':
if (MatchChar(ts, c)) {
tp->t_op = MatchChar(ts, c) ? JSOP_URSH : JSOP_RSH;
c = MatchChar(ts, '=') ? TOK_ASSIGN : TOK_SHOP;
} else {
tp->t_op = MatchChar(ts, '=') ? JSOP_GE : JSOP_GT;
c = TOK_RELOP;
}
break;
case '*':
tp->t_op = JSOP_MUL;
c = MatchChar(ts, '=') ? TOK_ASSIGN : TOK_STAR;
break;
case '/':
if (MatchChar(ts, '/')) {
skipline:
while ((c = GetChar(ts)) != EOF && c != '\n')
/* skip to end of line */;
UngetChar(ts, c);
goto retry;
}
if (MatchChar(ts, '*')) {
while ((c = GetChar(ts)) != EOF &&
!(c == '*' && MatchChar(ts, '/'))) {
/* Ignore all characters until comment close. */
}
if (c == EOF) {
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_UNTERMINATED_COMMENT);
RETURN(TOK_ERROR);
}
goto retry;
}
#if JS_HAS_REGEXPS
if (ts->flags & TSF_REGEXP) {
JSObject *obj;
uintN flags;
INIT_TOKENBUF(&ts->tokenbuf);
while ((c = GetChar(ts)) != '/') {
if (c == '\n' || c == EOF) {
UngetChar(ts, c);
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_UNTERMINATED_REGEXP);
RETURN(TOK_ERROR);
}
if (c == '\\') {
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
c = GetChar(ts);
}
if (!AddToTokenBuf(cx, &ts->tokenbuf, (jschar)c))
RETURN(TOK_ERROR);
}
FINISH_TOKENBUF(&ts->tokenbuf);
for (flags = 0; ; ) {
if (MatchChar(ts, 'g'))
flags |= JSREG_GLOB;
else if (MatchChar(ts, 'i'))
flags |= JSREG_FOLD;
else if (MatchChar(ts, 'm'))
flags |= JSREG_MULTILINE;
else
break;
}
c = PeekChar(ts);
if (JS7_ISLET(c)) {
tp->ptr = ts->linebuf.ptr - 1;
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_BAD_REGEXP_FLAG);
(void) GetChar(ts);
RETURN(TOK_ERROR);
}
obj = js_NewRegExpObject(cx, ts,
ts->tokenbuf.base,
TOKEN_LENGTH(&ts->tokenbuf),
flags);
if (!obj)
RETURN(TOK_ERROR);
atom = js_AtomizeObject(cx, obj, 0);
if (!atom)
RETURN(TOK_ERROR);
tp->t_op = JSOP_OBJECT;
tp->t_atom = atom;
RETURN(TOK_OBJECT);
}
#endif /* JS_HAS_REGEXPS */
tp->t_op = JSOP_DIV;
c = MatchChar(ts, '=') ? TOK_ASSIGN : TOK_DIVOP;
break;
case '%':
tp->t_op = JSOP_MOD;
c = MatchChar(ts, '=') ? TOK_ASSIGN : TOK_DIVOP;
break;
case '~':
tp->t_op = JSOP_BITNOT;
c = TOK_UNARYOP;
break;
case '+':
if (MatchChar(ts, '=')) {
tp->t_op = JSOP_ADD;
c = TOK_ASSIGN;
} else if (MatchChar(ts, c)) {
c = TOK_INC;
} else {
tp->t_op = JSOP_POS;
c = TOK_PLUS;
}
break;
case '-':
if (MatchChar(ts, '=')) {
tp->t_op = JSOP_SUB;
c = TOK_ASSIGN;
} else if (MatchChar(ts, c)) {
if (PeekChar(ts) == '>' && !(ts->flags & TSF_DIRTYLINE))
goto skipline;
c = TOK_DEC;
} else {
tp->t_op = JSOP_NEG;
c = TOK_MINUS;
}
ts->flags |= TSF_DIRTYLINE;
break;
#if JS_HAS_SHARP_VARS
case '#':
{
uint32 n;
c = GetChar(ts);
if (!JS7_ISDEC(c)) {
UngetChar(ts, c);
goto badchar;
}
n = (uint32)JS7_UNDEC(c);
for (;;) {
c = GetChar(ts);
if (!JS7_ISDEC(c))
break;
n = 10 * n + JS7_UNDEC(c);
if (n >= ATOM_INDEX_LIMIT) {
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_SHARPVAR_TOO_BIG);
RETURN(TOK_ERROR);
}
}
tp->t_dval = (jsdouble) n;
if (JS_HAS_STRICT_OPTION(cx) &&
(c == '=' || c == '#')) {
char buf[20];
JS_snprintf(buf, sizeof buf, "#%u%c", n, c);
if (!js_ReportCompileErrorNumber(cx, ts, NULL,
JSREPORT_WARNING |
JSREPORT_STRICT,
JSMSG_DEPRECATED_USAGE,
buf)) {
RETURN(TOK_ERROR);
}
}
if (c == '=')
RETURN(TOK_DEFSHARP);
if (c == '#')
RETURN(TOK_USESHARP);
goto badchar;
}
badchar:
#endif /* JS_HAS_SHARP_VARS */
default:
js_ReportCompileErrorNumber(cx, ts, NULL, JSREPORT_ERROR,
JSMSG_ILLEGAL_CHARACTER);
RETURN(TOK_ERROR);
}
JS_ASSERT(c < TOK_LIMIT);
RETURN((JSTokenType)c);
#undef INIT_TOKENBUF
#undef FINISH_TOKENBUF
#undef TOKEN_LENGTH
#undef RETURN
}