ure.c

OpenLdap是LDAP的开源项目

     * Return the number of characters consumed.
     */
    return sp - pp;
}

/*
 * Collect a hex number with 1 to 4 digits and return the number
 * of characters used.
 */
static unsigned long
_ure_hex(ucs2_t *np, unsigned long limit, ucs4_t *n)
{
    ucs2_t i;
    ucs2_t *sp, *ep;
    ucs4_t nn;

    sp = np;
    ep = sp + limit;

    for (nn = 0, i = 0; i < 4 && sp < ep; i++, sp++) {
        if (*sp >= '0' && *sp <= '9')
          nn = (nn << 4) + (*sp - '0');
        else if (*sp >= 'A' && *sp <= 'F')
          nn = (nn << 4) + ((*sp - 'A') + 10);
        else if (*sp >= 'a' && *sp <= 'f')
          nn = (nn << 4) + ((*sp - 'a') + 10);
        else
          /*
           * Encountered something that is not a hex digit.
           */
          break;
    }

    /*
     * Assign the character code collected and return the number of
     * characters used.
     */
    *n = nn;

    return sp - np;
}

/*
 * Insert a range into a character class, removing duplicates and ordering
 * them in increasing range-start order.
 */
static void
_ure_add_range(_ure_ccl_t *ccl, _ure_range_t *r, _ure_buffer_t *b)
{
    ucs2_t i;
    ucs4_t tmp;
    _ure_range_t *rp;

    /*
     * If the `casefold' flag is set, then make sure both endpoints of the
     * range are converted to lower case.
     */
    if (b->flags & _URE_DFA_CASEFOLD) {
        r->min_code = _ure_tolower(r->min_code);
        r->max_code = _ure_tolower(r->max_code);
    }

    /*
     * Swap the range endpoints if they are not in increasing order.
     */
    if (r->min_code > r->max_code) {
        tmp = r->min_code;
        r->min_code = r->max_code;
        r->max_code = tmp;
    }

    for (i = 0, rp = ccl->ranges;
         i < ccl->ranges_used && r->min_code < rp->min_code; i++, rp++) ;

    /*
     * Check for a duplicate.
     */
    if (i < ccl->ranges_used &&
        r->min_code == rp->min_code && r->max_code == rp->max_code)
      return;

    if (ccl->ranges_used == ccl->ranges_size) {
        if (ccl->ranges_size == 0)
          ccl->ranges = (_ure_range_t *) malloc(sizeof(_ure_range_t) << 3);
        else
          ccl->ranges = (_ure_range_t *)
              realloc((char *) ccl->ranges,
                      sizeof(_ure_range_t) * (ccl->ranges_size + 8));
        ccl->ranges_size += 8;
    }

    rp = ccl->ranges + ccl->ranges_used;

    if (i < ccl->ranges_used)
      _ure_memmove((char *) (rp + 1), (char *) rp,
                   sizeof(_ure_range_t) * (ccl->ranges_used - i));

    ccl->ranges_used++;
    rp->min_code = r->min_code;
    rp->max_code = r->max_code;
}

#define _URE_ALPHA_MASK  (_URE_UPPER|_URE_LOWER|_URE_OTHERLETTER|\
_URE_MODIFIER|_URE_TITLE|_URE_NONSPACING|_URE_COMBINING)
#define _URE_ALNUM_MASK  (_URE_ALPHA_MASK|_URE_NUMDIGIT)
#define _URE_PUNCT_MASK  (_URE_DASHPUNCT|_URE_OPENPUNCT|_URE_CLOSEPUNCT|\
_URE_OTHERPUNCT)
#define _URE_GRAPH_MASK (_URE_NUMDIGIT|_URE_NUMOTHER|_URE_ALPHA_MASK|\
_URE_MATHSYM|_URE_CURRENCYSYM|_URE_OTHERSYM)
#define _URE_PRINT_MASK (_URE_GRAPH_MASK|_URE_SPACESEP)
#define _URE_SPACE_MASK  (_URE_SPACESEP|_URE_LINESEP|_URE_PARASEP)

typedef void (*_ure_cclsetup_t)(
    _ure_symtab_t *sym,
    unsigned long mask,
    _ure_buffer_t *b
);

typedef struct {
    ucs2_t key;
    unsigned long len;
    unsigned long next;
    _ure_cclsetup_t func;
    unsigned long mask;
} _ure_trie_t;

static void
_ure_ccl_setup(_ure_symtab_t *sym, unsigned long mask, _ure_buffer_t *b)
{
    sym->props |= mask;
}

static void
_ure_space_setup(_ure_symtab_t *sym, unsigned long mask, _ure_buffer_t *b)
{
    _ure_range_t range;

    sym->props |= mask;

    /*
     * Add the additional characters needed for handling isspace().
     */
    range.min_code = range.max_code = '\t';
    _ure_add_range(&sym->sym.ccl, &range, b);
    range.min_code = range.max_code = '\r';
    _ure_add_range(&sym->sym.ccl, &range, b);
    range.min_code = range.max_code = '\n';
    _ure_add_range(&sym->sym.ccl, &range, b);
    range.min_code = range.max_code = '\f';
    _ure_add_range(&sym->sym.ccl, &range, b);
    range.min_code = range.max_code = 0xfeff;
    _ure_add_range(&sym->sym.ccl, &range, b);
}

static void
_ure_xdigit_setup(_ure_symtab_t *sym, unsigned long mask, _ure_buffer_t *b)
{
    _ure_range_t range;

    /*
     * Add the additional characters needed for handling isxdigit().
     */
    range.min_code = '0';
    range.max_code = '9';
    _ure_add_range(&sym->sym.ccl, &range, b);
    range.min_code = 'A';
    range.max_code = 'F';
    _ure_add_range(&sym->sym.ccl, &range, b);
    range.min_code = 'a';
    range.max_code = 'f';
    _ure_add_range(&sym->sym.ccl, &range, b);
}

static _ure_trie_t cclass_trie[] = {
    {0x003a, 1, 1, 0, 0},
    {0x0061, 9, 10, 0, 0},
    {0x0063, 8, 19, 0, 0},
    {0x0064, 7, 24, 0, 0},
    {0x0067, 6, 29, 0, 0},
    {0x006c, 5, 34, 0, 0},
    {0x0070, 4, 39, 0, 0},
    {0x0073, 3, 49, 0, 0},
    {0x0075, 2, 54, 0, 0},
    {0x0078, 1, 59, 0, 0},
    {0x006c, 1, 11, 0, 0},
    {0x006e, 2, 13, 0, 0},
    {0x0070, 1, 16, 0, 0},
    {0x0075, 1, 14, 0, 0},
    {0x006d, 1, 15, 0, 0},
    {0x003a, 1, 16, _ure_ccl_setup, _URE_ALNUM_MASK},
    {0x0068, 1, 17, 0, 0},
    {0x0061, 1, 18, 0, 0},
    {0x003a, 1, 19, _ure_ccl_setup, _URE_ALPHA_MASK},
    {0x006e, 1, 20, 0, 0},
    {0x0074, 1, 21, 0, 0},
    {0x0072, 1, 22, 0, 0},
    {0x006c, 1, 23, 0, 0},
    {0x003a, 1, 24, _ure_ccl_setup, _URE_CNTRL},
    {0x0069, 1, 25, 0, 0},
    {0x0067, 1, 26, 0, 0},
    {0x0069, 1, 27, 0, 0},
    {0x0074, 1, 28, 0, 0},
    {0x003a, 1, 29, _ure_ccl_setup, _URE_NUMDIGIT},
    {0x0072, 1, 30, 0, 0},
    {0x0061, 1, 31, 0, 0},
    {0x0070, 1, 32, 0, 0},
    {0x0068, 1, 33, 0, 0},
    {0x003a, 1, 34, _ure_ccl_setup, _URE_GRAPH_MASK},
    {0x006f, 1, 35, 0, 0},
    {0x0077, 1, 36, 0, 0},
    {0x0065, 1, 37, 0, 0},
    {0x0072, 1, 38, 0, 0},
    {0x003a, 1, 39, _ure_ccl_setup, _URE_LOWER},
    {0x0072, 2, 41, 0, 0},
    {0x0075, 1, 45, 0, 0},
    {0x0069, 1, 42, 0, 0},
    {0x006e, 1, 43, 0, 0},
    {0x0074, 1, 44, 0, 0},
    {0x003a, 1, 45, _ure_ccl_setup, _URE_PRINT_MASK},
    {0x006e, 1, 46, 0, 0},
    {0x0063, 1, 47, 0, 0},
    {0x0074, 1, 48, 0, 0},
    {0x003a, 1, 49, _ure_ccl_setup, _URE_PUNCT_MASK},
    {0x0070, 1, 50, 0, 0},
    {0x0061, 1, 51, 0, 0},
    {0x0063, 1, 52, 0, 0},
    {0x0065, 1, 53, 0, 0},
    {0x003a, 1, 54, _ure_space_setup, _URE_SPACE_MASK},
    {0x0070, 1, 55, 0, 0},
    {0x0070, 1, 56, 0, 0},
    {0x0065, 1, 57, 0, 0},
    {0x0072, 1, 58, 0, 0},
    {0x003a, 1, 59, _ure_ccl_setup, _URE_UPPER},
    {0x0064, 1, 60, 0, 0},
    {0x0069, 1, 61, 0, 0},
    {0x0067, 1, 62, 0, 0},
    {0x0069, 1, 63, 0, 0},
    {0x0074, 1, 64, 0, 0},
    {0x003a, 1, 65, _ure_xdigit_setup, 0},
};

/*
 * Probe for one of the POSIX colon delimited character classes in the static
 * trie.
 */
static unsigned long
_ure_posix_ccl(ucs2_t *cp, unsigned long limit, _ure_symtab_t *sym,
               _ure_buffer_t *b)
{
    int i;
    unsigned long n;
    _ure_trie_t *tp;
    ucs2_t *sp, *ep;

    /*
     * If the number of characters left is less than 7, then this cannot be
     * interpreted as one of the colon delimited classes.
     */
    if (limit < 7)
      return 0;

    sp = cp;
    ep = sp + limit;
    tp = cclass_trie;
    for (i = 0; sp < ep && i < 8; i++, sp++) {
        n = tp->len;

        for (; n > 0 && tp->key != *sp; tp++, n--) ;

        if (n == 0)
          return 0;

        if (*sp == ':' && (i == 6 || i == 7)) {
            sp++;
            break;
        }
        if (sp + 1 < ep)
          tp = cclass_trie + tp->next;
    }
    if (tp->func == 0)
      return 0;

    (*tp->func)(sym, tp->mask, b);

    return sp - cp;
}

/*
 * Construct a list of ranges and return the number of characters consumed.
 */
static unsigned long
_ure_cclass(ucs2_t *cp, unsigned long limit, _ure_symtab_t *symp,
            _ure_buffer_t *b)
{
    int range_end;
    unsigned long n;
    ucs2_t *sp, *ep;
    ucs4_t c, last;
    _ure_ccl_t *cclp;
    _ure_range_t range;

    sp = cp;
    ep = sp + limit;

    if (*sp == '^') {
      symp->type = _URE_NCCLASS;
      sp++;
    } else
      symp->type = _URE_CCLASS;

    for (last = 0, range_end = 0;
         b->error == _URE_OK && sp < ep && *sp != ']'; ) {
        c = *sp++;
        if (c == '\\') {
            if (sp == ep) {
                /*
                 * The EOS was encountered when expecting the reverse solidus
                 * to be followed by the character it is escaping.  Set an
                 * error code and return the number of characters consumed up
                 * to this point.
                 */
                b->error = _URE_UNEXPECTED_EOS;
                return sp - cp;
            }

            c = *sp++;
            switch (c) {
              case 'a':
                c = 0x07;
                break;
              case 'b':
                c = 0x08;
                break;
              case 'f':
                c = 0x0c;
                break;
              case 'n':
                c = 0x0a;
                break;
              case 'r':
                c = 0x0d;
                break;
              case 't':
                c = 0x09;
                break;
              case 'v':
                c = 0x0b;
                break;
              case 'p':
              case 'P':
                sp += _ure_prop_list(sp, ep - sp, &symp->props, b);
                /*
                 * Invert the bit mask of the properties if this is a negated
                 * character class or if 'P' is used to specify a list of
                 * character properties that should *not* match in a
                 * character class.
                 */
                if (c == 'P')
                  symp->props = ~symp->props;
                continue;
                break;
              case 'x':
              case 'X':
              case 'u':
              case 'U':
                if (sp < ep &&
                    ((*sp >= '0' && *sp <= '9') ||
                     (*sp >= 'A' && *sp <= 'F') ||
                     (*sp >= 'a' && *sp <= 'f')))
                  sp += _ure_hex(sp, ep - sp, &c);
            }
        } else if (c == ':') {
            /*
             * Probe for a POSIX colon delimited character class.
             */
            sp--;
            if ((n = _ure_posix_ccl(sp, ep - sp, symp, b)) == 0)
              sp++;
            else {
                sp += n;
                continue;
            }
        }

        cclp = &symp->sym.ccl;

        /*
         * Check to see if the current character is a low surrogate that needs
         * to be combined with a preceding high surrogate.
         */
        if (last != 0) {
            if (c >= 0xdc00 && c <= 0xdfff)
              /*
               * Construct the UTF16 character code.
               */
              c = 0x10000 + (((last & 0x03ff) << 10) | (c & 0x03ff));
            else {
                /*
                 * Add the isolated high surrogate to the range.
                 */
                if (range_end == 1)
                  range.max_code = last & 0xffff;
                else
                  range.min_code = range.max_code = last & 0xffff;

                _ure_add_range(cclp, &range, b);
                range_end = 0;
            }
        }

        /*
         * Clear the last character code.
         */
        last = 0;

        /*
         * This slightly awkward code handles the different cases needed to
         * construct a range.
         */
        if (c >= 0xd800 && c <= 0xdbff) {
            /*
             * If the high surrogate is followed by a range indicator, simply
             * add it as the range start.  Otherwise, save it in case the next
             * character is a low surrogate.
             */
            if (*sp == '-') {
                sp++;