xmlschemastypes.c

libxml,在UNIX/LINUX下非常重要的一个库,为XML相关应用提供方便.目前上载的是最新版本,若要取得最新版本,请参考里面的readme.

#define SECS_PER_HOUR           (60 * SECS_PER_MIN)
#define SECS_PER_DAY            (24 * SECS_PER_HOUR)

static const long dayInYearByMonth[12] =
	{ 0, 31, 59, 90, 120, 151, 181, 212, 243, 273, 304, 334 };
static const long dayInLeapYearByMonth[12] =
	{ 0, 31, 60, 91, 121, 152, 182, 213, 244, 274, 305, 335 };

#define DAY_IN_YEAR(day, month, year)				\
        ((IS_LEAP(year) ?					\
                dayInLeapYearByMonth[month - 1] :		\
                dayInYearByMonth[month - 1]) + day)

#ifdef DEBUG
#define DEBUG_DATE(dt)                                                  \
    xmlGenericError(xmlGenericErrorContext,                             \
        "type=%o %04ld-%02u-%02uT%02u:%02u:%03f",                       \
        dt->type,dt->value.date.year,dt->value.date.mon,                \
        dt->value.date.day,dt->value.date.hour,dt->value.date.min,      \
        dt->value.date.sec);                                            \
    if (dt->value.date.tz_flag)                                         \
        if (dt->value.date.tzo != 0)                                    \
            xmlGenericError(xmlGenericErrorContext,                     \
                "%+05d\n",dt->value.date.tzo);                          \
        else                                                            \
            xmlGenericError(xmlGenericErrorContext, "Z\n");             \
    else                                                                \
        xmlGenericError(xmlGenericErrorContext,"\n")
#else
#define DEBUG_DATE(dt)
#endif

/**
 * _xmlSchemaParseGYear:
 * @dt:  pointer to a date structure
 * @str: pointer to the string to analyze
 *
 * Parses a xs:gYear without time zone and fills in the appropriate
 * field of the @dt structure. @str is updated to point just after the
 * xs:gYear. It is supposed that @dt->year is big enough to contain
 * the year.
 *
 * Returns 0 or the error code
 */
static int
_xmlSchemaParseGYear (xmlSchemaValDatePtr dt, const xmlChar **str) {
    const xmlChar *cur = *str, *firstChar;
    int isneg = 0, digcnt = 0;

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

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

    firstChar = cur;

    while ((*cur >= '0') && (*cur <= '9')) {
	dt->year = dt->year * 10 + (*cur - '0');
	cur++;
	digcnt++;
    }

    /* year must be at least 4 digits (CCYY); over 4
     * digits cannot have a leading zero. */
    if ((digcnt < 4) || ((digcnt > 4) && (*firstChar == '0')))
	return 1;

    if (isneg)
	dt->year = - dt->year;

    if (!VALID_YEAR(dt->year))
	return 2;

    *str = cur;
    return 0;
}

/**
 * PARSE_2_DIGITS:
 * @num:  the integer to fill in
 * @cur:  an #xmlChar *
 * @invalid: an integer
 *
 * Parses a 2-digits integer and updates @num with the value. @cur is
 * updated to point just after the integer.
 * In case of error, @invalid is set to %TRUE, values of @num and
 * @cur are undefined.
 */
#define PARSE_2_DIGITS(num, cur, invalid)			\
	if ((cur[0] < '0') || (cur[0] > '9') ||			\
	    (cur[1] < '0') || (cur[1] > '9'))			\
	    invalid = 1;					\
	else							\
	    num = (cur[0] - '0') * 10 + (cur[1] - '0');		\
	cur += 2;

/**
 * PARSE_FLOAT:
 * @num:  the double to fill in
 * @cur:  an #xmlChar *
 * @invalid: an integer
 *
 * Parses a float and updates @num with the value. @cur is
 * updated to point just after the float. The float must have a
 * 2-digits integer part and may or may not have a decimal part.
 * In case of error, @invalid is set to %TRUE, values of @num and
 * @cur are undefined.
 */
#define PARSE_FLOAT(num, cur, invalid)				\
	PARSE_2_DIGITS(num, cur, invalid);			\
	if (!invalid && (*cur == '.')) {			\
	    double mult = 1;				        \
	    cur++;						\
	    if ((*cur < '0') || (*cur > '9'))			\
		invalid = 1;					\
	    while ((*cur >= '0') && (*cur <= '9')) {		\
		mult /= 10;					\
		num += (*cur - '0') * mult;			\
		cur++;						\
	    }							\
	}

/**
 * _xmlSchemaParseGMonth:
 * @dt:  pointer to a date structure
 * @str: pointer to the string to analyze
 *
 * Parses a xs:gMonth without time zone and fills in the appropriate
 * field of the @dt structure. @str is updated to point just after the
 * xs:gMonth.
 *
 * Returns 0 or the error code
 */
static int
_xmlSchemaParseGMonth (xmlSchemaValDatePtr dt, const xmlChar **str) {
    const xmlChar *cur = *str;
    int ret = 0;

    PARSE_2_DIGITS(dt->mon, cur, ret);
    if (ret != 0)
	return ret;

    if (!VALID_MONTH(dt->mon))
	return 2;

    *str = cur;
    return 0;
}

/**
 * _xmlSchemaParseGDay:
 * @dt:  pointer to a date structure
 * @str: pointer to the string to analyze
 *
 * Parses a xs:gDay without time zone and fills in the appropriate
 * field of the @dt structure. @str is updated to point just after the
 * xs:gDay.
 *
 * Returns 0 or the error code
 */
static int
_xmlSchemaParseGDay (xmlSchemaValDatePtr dt, const xmlChar **str) {
    const xmlChar *cur = *str;
    int ret = 0;

    PARSE_2_DIGITS(dt->day, cur, ret);
    if (ret != 0)
	return ret;

    if (!VALID_DAY(dt->day))
	return 2;

    *str = cur;
    return 0;
}

/**
 * _xmlSchemaParseTime:
 * @dt:  pointer to a date structure
 * @str: pointer to the string to analyze
 *
 * Parses a xs:time without time zone and fills in the appropriate
 * fields of the @dt structure. @str is updated to point just after the
 * xs:time.
 * In case of error, values of @dt fields are undefined.
 *
 * Returns 0 or the error code
 */
static int
_xmlSchemaParseTime (xmlSchemaValDatePtr dt, const xmlChar **str) {
    const xmlChar *cur = *str;
    unsigned int hour = 0; /* use temp var in case str is not xs:time */
    int ret = 0;

    PARSE_2_DIGITS(hour, cur, ret);
    if (ret != 0)
	return ret;

    if (*cur != ':')
	return 1;
    cur++;

    /* the ':' insures this string is xs:time */
    dt->hour = hour;

    PARSE_2_DIGITS(dt->min, cur, ret);
    if (ret != 0)
	return ret;

    if (*cur != ':')
	return 1;
    cur++;

    PARSE_FLOAT(dt->sec, cur, ret);
    if (ret != 0)
	return ret;

    if (!VALID_TIME(dt))
	return 2;

    *str = cur;
    return 0;
}

/**
 * _xmlSchemaParseTimeZone:
 * @dt:  pointer to a date structure
 * @str: pointer to the string to analyze
 *
 * Parses a time zone without time zone and fills in the appropriate
 * field of the @dt structure. @str is updated to point just after the
 * time zone.
 *
 * Returns 0 or the error code
 */
static int
_xmlSchemaParseTimeZone (xmlSchemaValDatePtr dt, const xmlChar **str) {
    const xmlChar *cur = *str;
    int ret = 0;

    if (str == NULL)
	return -1;

    switch (*cur) {
    case 0:
	dt->tz_flag = 0;
	dt->tzo = 0;
	break;

    case 'Z':
	dt->tz_flag = 1;
	dt->tzo = 0;
	cur++;
	break;

    case '+':
    case '-': {
	int isneg = 0, tmp = 0;
	isneg = (*cur == '-');

	cur++;

	PARSE_2_DIGITS(tmp, cur, ret);
	if (ret != 0)
	    return ret;
	if (!VALID_HOUR(tmp))
	    return 2;

	if (*cur != ':')
	    return 1;
	cur++;

	dt->tzo = tmp * 60;

	PARSE_2_DIGITS(tmp, cur, ret);
	if (ret != 0)
	    return ret;
	if (!VALID_MIN(tmp))
	    return 2;

	dt->tzo += tmp;
	if (isneg)
	    dt->tzo = - dt->tzo;

	if (!VALID_TZO(dt->tzo))
	    return 2;

	dt->tz_flag = 1;
	break;
      }
    default:
	return 1;
    }

    *str = cur;
    return 0;
}

/**
 * _xmlSchemaBase64Decode:
 * @ch: a character
 *
 * 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);