xmlschemastypes.c.svn-base

这是一个用于解析xml文件的类库。使用这个类库

 * Converts a base64 encoded character to its base 64 value.
 *
 * Returns 0-63 (value), 64 (pad), or -1 (not recognized)
 */
static int
_xmlSchemaBase64Decode (const xmlChar ch) {
    if (('A' <= ch) && (ch <= 'Z')) return ch - 'A';
    if (('a' <= ch) && (ch <= 'z')) return ch - 'a' + 26;
    if (('0' <= ch) && (ch <= '9')) return ch - '0' + 52;
    if ('+' == ch) return 62;
    if ('/' == ch) return 63;
    if ('=' == ch) return 64;
    return -1;
}

/****************************************************************
 *								*
 *	XML Schema Dates/Times Datatypes Handling		*
 *								*
 ****************************************************************/

/**
 * PARSE_DIGITS:
 * @num:  the integer to fill in
 * @cur:  an #xmlChar *
 * @num_type: an integer flag
 *
 * Parses a digits integer and updates @num with the value. @cur is
 * updated to point just after the integer.
 * In case of error, @num_type is set to -1, values of @num and
 * @cur are undefined.
 */
#define PARSE_DIGITS(num, cur, num_type)	                \
	if ((*cur < '0') || (*cur > '9'))			\
	    num_type = -1;					\
        else                                                    \
	    while ((*cur >= '0') && (*cur <= '9')) {		\
	        num = num * 10 + (*cur - '0');		        \
	        cur++;                                          \
            }

/**
 * PARSE_NUM:
 * @num:  the double to fill in
 * @cur:  an #xmlChar *
 * @num_type: an integer flag
 *
 * Parses a float or integer and updates @num with the value. @cur is
 * updated to point just after the number. If the number is a float,
 * then it must have an integer part and a decimal part; @num_type will
 * be set to 1. If there is no decimal part, @num_type is set to zero.
 * In case of error, @num_type is set to -1, values of @num and
 * @cur are undefined.
 */
#define PARSE_NUM(num, cur, num_type)				\
        num = 0;                                                \
	PARSE_DIGITS(num, cur, num_type);	                \
	if (!num_type && (*cur == '.')) {			\
	    double mult = 1;				        \
	    cur++;						\
	    if ((*cur < '0') || (*cur > '9'))			\
		num_type = -1;					\
            else                                                \
                num_type = 1;                                   \
	    while ((*cur >= '0') && (*cur <= '9')) {		\
		mult /= 10;					\
		num += (*cur - '0') * mult;			\
		cur++;						\
	    }							\
	}

/**
 * xmlSchemaValidateDates:
 * @type: the expected type or XML_SCHEMAS_UNKNOWN
 * @dateTime:  string to analyze
 * @val:  the return computed value
 *
 * Check that @dateTime conforms to the lexical space of one of the date types.
 * if true a value is computed and returned in @val.
 *
 * Returns 0 if this validates, a positive error code number otherwise
 *         and -1 in case of internal or API error.
 */
static int
xmlSchemaValidateDates (xmlSchemaValType type,
	                const xmlChar *dateTime, xmlSchemaValPtr *val) {
    xmlSchemaValPtr dt;
    int ret;
    const xmlChar *cur = dateTime;

#define RETURN_TYPE_IF_VALID(t)					\
    if (IS_TZO_CHAR(*cur)) {					\
	ret = _xmlSchemaParseTimeZone(&(dt->value.date), &cur);	\
	if (ret == 0) {						\
	    if (*cur != 0)					\
		goto error;					\
	    dt->type = t;					\
	    goto done;						\
	}							\
    }

    if (dateTime == NULL)
	return -1;

    if ((*cur != '-') && (*cur < '0') && (*cur > '9'))
	return 1;

    dt = xmlSchemaNewValue(XML_SCHEMAS_UNKNOWN);
    if (dt == NULL)
	return -1;

    if ((cur[0] == '-') && (cur[1] == '-')) {
	/*
	 * It's an incomplete date (xs:gMonthDay, xs:gMonth or
	 * xs:gDay)
	 */
	cur += 2;

	/* is it an xs:gDay? */
	if (*cur == '-') {
	    if (type == XML_SCHEMAS_GMONTH)
		goto error;
	  ++cur;
	    ret = _xmlSchemaParseGDay(&(dt->value.date), &cur);
	    if (ret != 0)
		goto error;

	    RETURN_TYPE_IF_VALID(XML_SCHEMAS_GDAY);

	    goto error;
	}

	/*
	 * it should be an xs:gMonthDay or xs:gMonth
	 */
	ret = _xmlSchemaParseGMonth(&(dt->value.date), &cur);
	if (ret != 0)
	    goto error;

        /*
         * a '-' char could indicate this type is xs:gMonthDay or
         * a negative time zone offset. Check for xs:gMonthDay first.
         * Also the first three char's of a negative tzo (-MM:SS) can
         * appear to be a valid day; so even if the day portion
         * of the xs:gMonthDay verifies, we must insure it was not
         * a tzo.
         */
        if (*cur == '-') {
            const xmlChar *rewnd = cur;
            cur++;

  	    ret = _xmlSchemaParseGDay(&(dt->value.date), &cur);
            if ((ret == 0) && ((*cur == 0) || (*cur != ':'))) {

                /*
                 * we can use the VALID_MDAY macro to validate the month
                 * and day because the leap year test will flag year zero
                 * as a leap year (even though zero is an invalid year).
                 */
                if (VALID_MDAY((&(dt->value.date)))) {

	            RETURN_TYPE_IF_VALID(XML_SCHEMAS_GMONTHDAY);

                    goto error;
                }
            }

            /*
             * not xs:gMonthDay so rewind and check if just xs:gMonth
             * with an optional time zone.
             */
            cur = rewnd;
        }

	RETURN_TYPE_IF_VALID(XML_SCHEMAS_GMONTH);

	goto error;
    }

    /*
     * It's a right-truncated date or an xs:time.
     * Try to parse an xs:time then fallback on right-truncated dates.
     */
    if ((*cur >= '0') && (*cur <= '9')) {
	ret = _xmlSchemaParseTime(&(dt->value.date), &cur);
	if (ret == 0) {
	    /* it's an xs:time */
	    RETURN_TYPE_IF_VALID(XML_SCHEMAS_TIME);
	}
    }

    /* fallback on date parsing */
    cur = dateTime;

    ret = _xmlSchemaParseGYear(&(dt->value.date), &cur);
    if (ret != 0)
	goto error;

    /* is it an xs:gYear? */
    RETURN_TYPE_IF_VALID(XML_SCHEMAS_GYEAR);

    if (*cur != '-')
	goto error;
    cur++;

    ret = _xmlSchemaParseGMonth(&(dt->value.date), &cur);
    if (ret != 0)
	goto error;

    /* is it an xs:gYearMonth? */
    RETURN_TYPE_IF_VALID(XML_SCHEMAS_GYEARMONTH);

    if (*cur != '-')
	goto error;
    cur++;

    ret = _xmlSchemaParseGDay(&(dt->value.date), &cur);
    if ((ret != 0) || !VALID_DATE((&(dt->value.date))))
	goto error;

    /* is it an xs:date? */
    RETURN_TYPE_IF_VALID(XML_SCHEMAS_DATE);

    if (*cur != 'T')
	goto error;
    cur++;

    /* it should be an xs:dateTime */
    ret = _xmlSchemaParseTime(&(dt->value.date), &cur);
    if (ret != 0)
	goto error;

    ret = _xmlSchemaParseTimeZone(&(dt->value.date), &cur);
    if ((ret != 0) || (*cur != 0) || !VALID_DATETIME((&(dt->value.date))))
	goto error;


    dt->type = XML_SCHEMAS_DATETIME;

done:
#if 1
    if ((type != XML_SCHEMAS_UNKNOWN) && (type != dt->type))
        goto error;
#else
    /*
     * insure the parsed type is equal to or less significant (right
     * truncated) than the desired type.
     */
    if ((type != XML_SCHEMAS_UNKNOWN) && (type != dt->type)) {

        /* time only matches time */
        if ((type == XML_SCHEMAS_TIME) && (dt->type == XML_SCHEMAS_TIME))
            goto error;

        if ((type == XML_SCHEMAS_DATETIME) &&
            ((dt->type != XML_SCHEMAS_DATE) ||
             (dt->type != XML_SCHEMAS_GYEARMONTH) ||
             (dt->type != XML_SCHEMAS_GYEAR)))
            goto error;

        if ((type == XML_SCHEMAS_DATE) &&
            ((dt->type != XML_SCHEMAS_GYEAR) ||
             (dt->type != XML_SCHEMAS_GYEARMONTH)))
            goto error;

        if ((type == XML_SCHEMAS_GYEARMONTH) && (dt->type != XML_SCHEMAS_GYEAR))
            goto error;

        if ((type == XML_SCHEMAS_GMONTHDAY) && (dt->type != XML_SCHEMAS_GMONTH))
            goto error;
    }
#endif

    if (val != NULL)
        *val = dt;
    else
	xmlSchemaFreeValue(dt);

    return 0;

error:
    if (dt != NULL)
	xmlSchemaFreeValue(dt);
    return 1;
}

/**
 * xmlSchemaValidateDuration:
 * @type: the predefined type
 * @duration:  string to analyze
 * @val:  the return computed value
 *
 * Check that @duration conforms to the lexical space of the duration type.
 * if true a value is computed and returned in @val.
 *
 * Returns 0 if this validates, a positive error code number otherwise
 *         and -1 in case of internal or API error.
 */
static int
xmlSchemaValidateDuration (xmlSchemaTypePtr type ATTRIBUTE_UNUSED,
	                   const xmlChar *duration, xmlSchemaValPtr *val) {
    const xmlChar  *cur = duration;
    xmlSchemaValPtr dur;
    int isneg = 0;
    unsigned int seq = 0;
    double         num;
    int            num_type = 0;  /* -1 = invalid, 0 = int, 1 = floating */
    const xmlChar  desig[]  = {'Y', 'M', 'D', 'H', 'M', 'S'};
    const double   multi[]  = { 0.0, 0.0, 86400.0, 3600.0, 60.0, 1.0, 0.0};

    if (duration == NULL)
	return -1;

    if (*cur == '-') {
        isneg = 1;
        cur++;
    }

    /* duration must start with 'P' (after sign) */
    if (*cur++ != 'P')
	return 1;

    if (*cur == 0)
	return 1;

    dur = xmlSchemaNewValue(XML_SCHEMAS_DURATION);
    if (dur == NULL)
	return -1;

    while (*cur != 0) {

        /* input string should be empty or invalid date/time item */
        if (seq >= sizeof(desig))
            goto error;

        /* T designator must be present for time items */
        if (*cur == 'T') {
            if (seq <= 3) {
                seq = 3;
                cur++;
            } else
                return 1;
        } else if (seq == 3)
            goto error;

        /* parse the number portion of the item */
        PARSE_NUM(num, cur, num_type);

        if ((num_type == -1) || (*cur == 0))
            goto error;

        /* update duration based on item type */
        while (seq < sizeof(desig)) {
            if (*cur == desig[seq]) {

                /* verify numeric type; only seconds can be float */
                if ((num_type != 0) && (seq < (sizeof(desig)-1)))
                    goto error;

                switch (seq) {
                    case 0:
                        dur->value.dur.mon = (long)num * 12;
                        break;
                    case 1:
                        dur->value.dur.mon += (long)num;
                        break;
                    default:
                        /* convert to seconds using multiplier */
                        dur->value.dur.sec += num * multi[seq];
                        seq++;
                        break;
                }

                break;          /* exit loop */
            }
            /* no date designators found? */
            if (++seq == 3)
                goto error;
        }
        cur++;
    }

    if (isneg) {
        dur->value.dur.mon = -dur->value.dur.mon;
        dur->value.dur.day = -dur->value.dur.day;
        dur->value.dur.sec = -dur->value.dur.sec;
    }

    if (val != NULL)
        *val = dur;
    else
	xmlSchemaFreeValue(dur);

    return 0;

error:
    if (dur != NULL)
	xmlSchemaFreeValue(dur);
    return 1;
}

/**
 * xmlSchemaStrip:
 * @value: a value
 *
 * Removes the leading and ending spaces of a string
 *
 * Returns the new string or NULL if no change was required.
 */
static xmlChar *
xmlSchemaStrip(const xmlChar *value) {
    const xmlChar *start = value, *end, *f;

    if (value == NULL) return(NULL);
    while ((*start != 0) && (IS_BLANK_CH(*start))) start++;
    end = start;
    while (*end != 0) end++;
    f = end;
    end--;