jsregexp.c

java script test programing source code

                --operatorSP;
                switch (operatorStack[operatorSP].op) {
                case REOP_ASSERT:
                case REOP_ASSERT_NOT:
                case REOP_LPAREN:
                    operand = NewRENode(state, operatorStack[operatorSP].op);
                    if (!operand)
                        goto out;
                    operand->u.parenIndex =
                        operatorStack[operatorSP].parenIndex;
                    JS_ASSERT(operandSP);
                    operand->kid = operandStack[operandSP - 1];
                    operandStack[operandSP - 1] = operand;
                    if (state->treeDepth == TREE_DEPTH_MAX) {
                        js_ReportCompileErrorNumber(state->context,
                                                    state->tokenStream,
                                                    JSREPORT_TS |
                                                    JSREPORT_ERROR,
                                                    JSMSG_REGEXP_TOO_COMPLEX);
                        goto out;
                    }
                    ++state->treeDepth;
                    /* FALL THROUGH */

                case REOP_LPARENNON:
                    state->result = operandStack[operandSP - 1];
                    if (!ParseQuantifier(state))
                        goto out;
                    operandStack[operandSP - 1] = state->result;
                    goto restartOperator;
                default:
                    if (!ProcessOp(state, &operatorStack[operatorSP],
                                   operandStack, operandSP))
                        goto out;
                    --operandSP;
                    break;
                }
            }
            break;

        case '{':
        {
            const jschar *errp = state->cp;

            if (ParseMinMaxQuantifier(state, JS_TRUE) < 0) {
                /*
                 * This didn't even scan correctly as a quantifier, so we should
                 * treat it as flat.
                 */
                op = REOP_CONCAT;
                goto pushOperator;
            }

            state->cp = errp;
            /* FALL THROUGH */
        }

        case '+':
        case '*':
        case '?':
            js_ReportCompileErrorNumberUC(state->context, state->tokenStream,
                                          JSREPORT_TS | JSREPORT_ERROR,
                                          JSMSG_BAD_QUANTIFIER, state->cp);
            result = JS_FALSE;
            goto out;

        default:
            /* Anything else is the start of the next term. */
            op = REOP_CONCAT;
pushOperator:
            if (operatorSP == operatorStackSize) {
                operatorStackSize += operatorStackSize;
                operatorStack = (REOpData *)
                    JS_realloc(state->context, operatorStack,
                               sizeof(REOpData) * operatorStackSize);
                if (!operatorStack)
                    goto out;
            }
            operatorStack[operatorSP].op = op;
            operatorStack[operatorSP].errPos = state->cp;
            operatorStack[operatorSP++].parenIndex = parenIndex;
            break;
        }
    }
out:
    if (operatorStack)
        JS_free(state->context, operatorStack);
    if (operandStack)
        JS_free(state->context, operandStack);
    return result;
}

/*
 * Hack two bits in CompilerState.flags, for use within FindParenCount to flag
 * its being on the stack, and to propagate errors to its callers.
 */
#define JSREG_FIND_PAREN_COUNT  0x8000
#define JSREG_FIND_PAREN_ERROR  0x4000

/*
 * Magic return value from FindParenCount and GetDecimalValue, to indicate
 * overflow beyond GetDecimalValue's max parameter, or a computed maximum if
 * its findMax parameter is non-null.
 */
#define OVERFLOW_VALUE          ((uintN)-1)

static uintN
FindParenCount(CompilerState *state)
{
    CompilerState temp;
    int i;

    if (state->flags & JSREG_FIND_PAREN_COUNT)
        return OVERFLOW_VALUE;

    /*
     * Copy state into temp, flag it so we never report an invalid backref,
     * and reset its members to parse the entire regexp.  This is obviously
     * suboptimal, but GetDecimalValue calls us only if a backref appears to
     * refer to a forward parenthetical, which is rare.
     */
    temp = *state;
    temp.flags |= JSREG_FIND_PAREN_COUNT;
    temp.cp = temp.cpbegin;
    temp.parenCount = 0;
    temp.classCount = 0;
    temp.progLength = 0;
    temp.treeDepth = 0;
    temp.classBitmapsMem = 0;
    for (i = 0; i < CLASS_CACHE_SIZE; i++)
        temp.classCache[i].start = NULL;

    if (!ParseRegExp(&temp)) {
        state->flags |= JSREG_FIND_PAREN_ERROR;
        return OVERFLOW_VALUE;
    }
    return temp.parenCount;
}

/*
 * Extract and return a decimal value at state->cp.  The initial character c
 * has already been read.  Return OVERFLOW_VALUE if the result exceeds max.
 * Callers who pass a non-null findMax should test JSREG_FIND_PAREN_ERROR in
 * state->flags to discover whether an error occurred under findMax.
 */
static uintN
GetDecimalValue(jschar c, uintN max, uintN (*findMax)(CompilerState *state),
                CompilerState *state)
{
    uintN value = JS7_UNDEC(c);
    JSBool overflow = (value > max && (!findMax || value > findMax(state)));

    /* The following restriction allows simpler overflow checks. */
    JS_ASSERT(max <= ((uintN)-1 - 9) / 10);
    while (state->cp < state->cpend) {
        c = *state->cp;
        if (!JS7_ISDEC(c))
            break;
        value = 10 * value + JS7_UNDEC(c);
        if (!overflow && value > max && (!findMax || value > findMax(state)))
            overflow = JS_TRUE;
        ++state->cp;
    }
    return overflow ? OVERFLOW_VALUE : value;
}

/*
 * Calculate the total size of the bitmap required for a class expression.
 */
static JSBool
CalculateBitmapSize(CompilerState *state, RENode *target, const jschar *src,
                    const jschar *end)
{
    uintN max = 0;
    JSBool inRange = JS_FALSE;
    jschar c, rangeStart = 0;
    uintN n, digit, nDigits, i;

    target->u.ucclass.bmsize = 0;
    target->u.ucclass.sense = JS_TRUE;

    if (src == end)
        return JS_TRUE;

    if (*src == '^') {
        ++src;
        target->u.ucclass.sense = JS_FALSE;
    }

    while (src != end) {
        uintN localMax = 0;
        switch (*src) {
        case '\\':
            ++src;
            c = *src++;
            switch (c) {
            case 'b':
                localMax = 0x8;
                break;
            case 'f':
                localMax = 0xC;
                break;
            case 'n':
                localMax = 0xA;
                break;
            case 'r':
                localMax = 0xD;
                break;
            case 't':
                localMax = 0x9;
                break;
            case 'v':
                localMax = 0xB;
                break;
            case 'c':
                if (src < end && RE_IS_LETTER(*src)) {
                    localMax = (jschar) (*src++ & 0x1F);
                } else {
                    --src;
                    localMax = '\\';
                }
                break;
            case 'x':
                nDigits = 2;
                goto lexHex;
            case 'u':
                nDigits = 4;
lexHex:
                n = 0;
                for (i = 0; (i < nDigits) && (src < end); i++) {
                    c = *src++;
                    if (!isASCIIHexDigit(c, &digit)) {
                        /*
                         * Back off to accepting the original
                         *'\' as a literal.
                         */
                        src -= i + 1;
                        n = '\\';
                        break;
                    }
                    n = (n << 4) | digit;
                }
                localMax = n;
                break;
            case 'd':
                if (inRange) {
                    JS_ReportErrorNumber(state->context,
                                         js_GetErrorMessage, NULL,
                                         JSMSG_BAD_CLASS_RANGE);
                    return JS_FALSE;
                }
                localMax = '9';
                break;
            case 'D':
            case 's':
            case 'S':
            case 'w':
            case 'W':
                if (inRange) {
                    JS_ReportErrorNumber(state->context,
                                         js_GetErrorMessage, NULL,
                                         JSMSG_BAD_CLASS_RANGE);
                    return JS_FALSE;
                }
                target->u.ucclass.bmsize = 65535;
                return JS_TRUE;
            case '0':
            case '1':
            case '2':
            case '3':
            case '4':
            case '5':
            case '6':
            case '7':
                /*
                 *  This is a non-ECMA extension - decimal escapes (in this
                 *  case, octal!) are supposed to be an error inside class
                 *  ranges, but supported here for backwards compatibility.
                 *
                 */
                n = JS7_UNDEC(c);
                c = *src;
                if ('0' <= c && c <= '7') {
                    src++;
                    n = 8 * n + JS7_UNDEC(c);
                    c = *src;
                    if ('0' <= c && c <= '7') {
                        src++;
                        i = 8 * n + JS7_UNDEC(c);
                        if (i <= 0377)
                            n = i;
                        else
                            src--;
                    }
                }
                localMax = n;
                break;

            default:
                localMax = c;
                break;
            }
            break;
        default:
            localMax = *src++;
            break;
        }
        if (state->flags & JSREG_FOLD) {
            c = JS_MAX(upcase((jschar) localMax), downcase((jschar) localMax));
            if (c > localMax)
                localMax = c;
        }
        if (inRange) {
            if (rangeStart > localMax) {
                JS_ReportErrorNumber(state->context,
                                     js_GetErrorMessage, NULL,
                                     JSMSG_BAD_CLASS_RANGE);
                return JS_FALSE;
            }
            inRange = JS_FALSE;
        } else {
            if (src < end - 1) {
                if (*src == '-') {
                    ++src;
                    inRange = JS_TRUE;
                    rangeStart = (jschar)localMax;
                    continue;
                }
            }
        }
        if (localMax > max)
            max = localMax;
    }
    target->u.ucclass.bmsize = max;
    return JS_TRUE;
}

/*
 *  item:       assertion               An item is either an assertion or
 *              quantatom               a quantified atom.
 *
 *  assertion:  '^'                     Assertions match beginning of string
 *                                      (or line if the class static property
 *                                      RegExp.multiline is true).
 *              '$'                     End of string (or line if the class
 *                                      static property RegExp.multiline is
 *                                      true).
 *              '\b'                    Word boundary (between \w and \W).
 *              '\B'                    Word non-boundary.
 *
 *  quantatom:  atom                    An unquantified atom.
 *              quantatom '{' n ',' m '}'
 *                                      Atom must occur between n and m times.
 *              quantatom '{' n ',' '}' Atom must occur at least n times.
 *              quantatom '{' n '}'     Atom must occur exactly n times.
 *              quantatom '*'           Zero or more times (same as {0,}).
 *              quantatom '+'           One or more times (same as {1,}).
 *              quantatom '?'           Zero or one time (same as {0,1}).
 *