ucgendat.c

ldap服务器源码

         * Convert the combining class code from decimal.
         */
        for (ccl_code = 0, e = s; *e && *e != ';'; e++)
          ccl_code = (ccl_code * 10) + (*e - '0');

        /*
         * Add the code if it not 0.
         */
        if (ccl_code != 0)
          ordered_ccl_insert(code, ccl_code);

        /*
         * Locate the second character property field.
         */
        for (s = e; *s != 0 && i < 4; s++) {
            if (*s == ';')
              i++;
        }
        for (e = s; *e && *e != ';'; e++) ;

        ordered_range_insert(code, s, e - s);

        /*
         * Check for a decomposition.
         */
        s = ++e;
        if (*s != ';') {
	    compat = *s == '<';
	    if (compat) {
		/*
		 * Skip compatibility formatting tag.
		 */
		while (*s++ != '>');
	    }
            /*
             * Collect the codes of the decomposition.
             */
            for (dectmp_size = 0; *s != ';'; ) {
                /*
                 * Skip all leading non-hex digits.
                 */
                while (!ishdigit(*s))
 		  s++;

                for (dectmp[dectmp_size] = 0; ishdigit(*s); s++) {
                    dectmp[dectmp_size] <<= 4;
                    if (*s >= '0' && *s <= '9')
                      dectmp[dectmp_size] += *s - '0';
                    else if (*s >= 'A' && *s <= 'F')
                      dectmp[dectmp_size] += (*s - 'A') + 10;
                    else if (*s >= 'a' && *s <= 'f')
                      dectmp[dectmp_size] += (*s - 'a') + 10;
                }
                dectmp_size++;
            }

            /*
             * If there are any codes in the temporary decomposition array,
             * then add the character with its decomposition.
             */
            if (dectmp_size > 0) {
		if (!compat) {
		    add_decomp(code, 0);
		}
		add_decomp(code, 1);
	    }
        }

        /*
         * Skip to the number field.
         */
        for (i = 0; i < 3 && *s; s++) {
            if (*s == ';')
              i++;
        }

        /*
         * Scan the number in.
         */
        number[0] = number[1] = 0;
        for (e = s, neg = wnum = 0; *e && *e != ';'; e++) {
            if (*e == '-') {
                neg = 1;
                continue;
            }

            if (*e == '/') {
                /*
                 * Move the the denominator of the fraction.
                 */
                if (neg)
                  number[wnum] *= -1;
                neg = 0;
                e++;
                wnum++;
            }
            number[wnum] = (number[wnum] * 10) + (*e - '0');
        }

        if (e > s) {
            /*
             * Adjust the denominator in case of integers and add the number.
             */
            if (wnum == 0)
              number[1] = 1;

            add_number(code, number[0], number[1]);
        }

        /*
         * Skip to the start of the possible case mappings.
         */
        for (s = e, i = 0; i < 4 && *s; s++) {
            if (*s == ';')
              i++;
        }

        /*
         * Collect the case mappings.
         */
        cases[0] = cases[1] = cases[2] = 0;
        for (i = 0; i < 3; i++) {
            while (ishdigit(*s)) {
                cases[i] <<= 4;
                if (*s >= '0' && *s <= '9')
                  cases[i] += *s - '0';
                else if (*s >= 'A' && *s <= 'F')
                  cases[i] += (*s - 'A') + 10;
                else if (*s >= 'a' && *s <= 'f')
                  cases[i] += (*s - 'a') + 10;
                s++;
            }
            if (*s == ';')
              s++;
        }
        if (cases[0] && cases[1])
          /*
           * Add the upper and lower mappings for a title case character.
           */
          add_title(code);
        else if (cases[1])
          /*
           * Add the lower and title case mappings for the upper case
           * character.
           */
          add_upper(code);
        else if (cases[0])
          /*
           * Add the upper and title case mappings for the lower case
           * character.
           */
          add_lower(code);
    }
}

static _decomp_t *
find_decomp(ac_uint4 code, short compat)
{
    long l, r, m;
    _decomp_t *decs;
    
    l = 0;
    r = (compat ? kdecomps_used : decomps_used) - 1;
    decs = compat ? kdecomps : decomps;
    while (l <= r) {
        m = (l + r) >> 1;
        if (code > decs[m].code)
          l = m + 1;
        else if (code < decs[m].code)
          r = m - 1;
        else
          return &decs[m];
    }
    return 0;
}

static void
decomp_it(_decomp_t *d, short compat)
{
    ac_uint4 i;
    _decomp_t *dp;

    for (i = 0; i < d->used; i++) {
        if ((dp = find_decomp(d->decomp[i], compat)) != 0)
          decomp_it(dp, compat);
        else
          dectmp[dectmp_size++] = d->decomp[i];
    }
}

/*
 * Expand all decompositions by recursively decomposing each character
 * in the decomposition.
 */
static void
expand_decomp(void)
{
    ac_uint4 i;

    for (i = 0; i < decomps_used; i++) {
        dectmp_size = 0;
        decomp_it(&decomps[i], 0);
        if (dectmp_size > 0)
          add_decomp(decomps[i].code, 0);
    }

    for (i = 0; i < kdecomps_used; i++) {
        dectmp_size = 0;
        decomp_it(&kdecomps[i], 1);
        if (dectmp_size > 0)
          add_decomp(kdecomps[i].code, 1);
    }
}

static int
cmpcomps(const void *v_comp1, const void *v_comp2)
{
	const _comp_t *comp1 = v_comp1, *comp2 = v_comp2;
    long diff = comp1->code1 - comp2->code1;

    if (!diff)
	diff = comp1->code2 - comp2->code2;
    return (int) diff;
}

/*
 * Load composition exclusion data
 */
static void
read_compexdata(FILE *in)
{
    ac_uint2 i;
    ac_uint4 code;
    char line[512], *s;

    (void) memset((char *) compexs, 0, sizeof(compexs));

    while (fgets(line, sizeof(line), in)) {
	if( (s=strchr(line, '\n')) ) *s = '\0';
        /*
         * Skip blank lines and lines that start with a '#'.
         */
        if (line[0] == 0 || line[0] == '#')
	    continue;

	/*
         * Collect the code.  Assume max 6 digits
         */

	for (s = line, i = code = 0; *s != '#' && i < 6; i++, s++) {
	    if (isspace((unsigned char)*s)) break;
            code <<= 4;
            if (*s >= '0' && *s <= '9')
		code += *s - '0';
            else if (*s >= 'A' && *s <= 'F')
		code += (*s - 'A') + 10;
            else if (*s >= 'a' && *s <= 'f')
		code += (*s - 'a') + 10;
        }
        COMPEX_SET(code);
    }
}

/*
 * Creates array of compositions from decomposition array
 */
static void
create_comps(void)
{
    ac_uint4 i, cu;

    comps = (_comp_t *) malloc(comps_used * sizeof(_comp_t));

    for (i = cu = 0; i < decomps_used; i++) {
	if (decomps[i].used != 2 || COMPEX_TEST(decomps[i].code))
	    continue;
	comps[cu].comp = decomps[i].code;
	comps[cu].count = 2;
	comps[cu].code1 = decomps[i].decomp[0];
	comps[cu].code2 = decomps[i].decomp[1];
	cu++;
    }
    comps_used = cu;
    qsort(comps, comps_used, sizeof(_comp_t), cmpcomps);
}

#if HARDCODE_DATA
static void
write_case(FILE *out, _case_t *tab, int num, int first)
{
    int i;

    for (i=0; i<num; i++) {
	if (first) first = 0;
	else fprintf(out, ",");
	fprintf(out, "\n\t0x%08lx, 0x%08lx, 0x%08lx",
		(unsigned long) tab[i].key, (unsigned long) tab[i].other1,
		(unsigned long) tab[i].other2);
    }
}

#define PREF "static const "

#endif

static void
write_cdata(char *opath)
{
    FILE *out;
	ac_uint4 bytes;
    ac_uint4 i, idx, nprops;
#if !(HARDCODE_DATA)
    ac_uint2 casecnt[2];
#endif
    char path[BUFSIZ];
#if HARDCODE_DATA
    int j, k;

    /*****************************************************************
     *
     * Generate the ctype data.
     *
     *****************************************************************/

    /*
     * Open the output file.
     */
    snprintf(path, sizeof path, "%s" LDAP_DIRSEP "uctable.h", opath);
    if ((out = fopen(path, "w")) == 0)
      return;
#else
    /*
     * Open the ctype.dat file.
     */
    snprintf(path, sizeof path, "%s" LDAP_DIRSEP "ctype.dat", opath);
    if ((out = fopen(path, "wb")) == 0)
      return;
#endif

    /*
     * Collect the offsets for the properties.  The offsets array is
     * on a 4-byte boundary to keep things efficient for architectures
     * that need such a thing.
     */
    for (i = idx = 0; i < NUMPROPS; i++) {
        propcnt[i] = (proptbl[i].used != 0) ? idx : 0xffff;
        idx += proptbl[i].used;
    }

    /*
     * Add the sentinel index which is used by the binary search as the upper
     * bound for a search.
     */
    propcnt[i] = idx;

    /*
     * Record the actual number of property lists.  This may be different than
     * the number of offsets actually written because of aligning on a 4-byte
     * boundary.
     */
    hdr[1] = NUMPROPS;

    /*
     * Calculate the byte count needed and pad the property counts array to a
     * 4-byte boundary.
     */
    if ((bytes = sizeof(ac_uint2) * (NUMPROPS + 1)) & 3)
      bytes += 4 - (bytes & 3);
    nprops = bytes / sizeof(ac_uint2);
    bytes += sizeof(ac_uint4) * idx;

#if HARDCODE_DATA
    fprintf(out, PREF "ac_uint4 _ucprop_size = %d;\n\n", NUMPROPS);

    fprintf(out, PREF "ac_uint2 _ucprop_offsets[] = {");

    for (i = 0; i<nprops; i++) {
       if (i) fprintf(out, ",");
       if (!(i&7)) fprintf(out, "\n\t");
       else fprintf(out, " ");
       fprintf(out, "0x%04x", propcnt[i]);
    }
    fprintf(out, "\n};\n\n");

    fprintf(out, PREF "ac_uint4 _ucprop_ranges[] = {");

    k = 0;
    for (i = 0; i < NUMPROPS; i++) {
	if (proptbl[i].used > 0) {
	  for (j=0; j<proptbl[i].used; j++) {
	    if (k) fprintf(out, ",");
	    if (!(k&3)) fprintf(out,"\n\t");
	    else fprintf(out, " ");
	    k++;
	    fprintf(out, "0x%08lx", (unsigned long) proptbl[i].ranges[j]);
	  }
	}
    }
    fprintf(out, "\n};\n\n");
#else
    /*
     * Write the header.
     */
    fwrite((char *) hdr, sizeof(ac_uint2), 2, out);

    /*
     * Write the byte count.
     */
    fwrite((char *) &bytes, sizeof(ac_uint4), 1, out);

    /*
     * Write the property list counts.
     */
    fwrite((char *) propcnt, sizeof(ac_uint2), nprops, out);

    /*
     * Write the property lists.
     */
    for (i = 0; i < NUMPROPS; i++) {
        if (proptbl[i].used > 0)
          fwrite((char *) proptbl[i].ranges, sizeof(ac_uint4),
                 proptbl[i].used, out);
    }

    fclose(out);
#endif

    /*****************************************************************
     *
     * Generate the case mapping data.
     *
     *****************************************************************/

#if HARDCODE_DATA
    fprintf(out, PREF "ac_uint4 _uccase_size = %ld;\n\n",
        (long) (upper_used + lower_used + title_used));

    fprintf(out, PREF "ac_uint2 _uccase_len[2] = {%ld, %ld};\n\n",
        (long) upper_used, (long) lower_used);
    fprintf(out, PREF "ac_uint4 _uccase_map[] = {");

    if (upper_used > 0)
      /*
       * Write the upper case table.
       */
      write_case(out, upper, upper_used, 1);

    if (lower_used > 0)
      /*
       * Write the lower case table.
       */
      write_case(out, lower, lower_used, !upper_used);

    if (title_used > 0)
      /*
       * Write the title case table.
       */
      write_case(out, title, title_used, !(upper_used||lower_used));

    if (!(upper_used || lower_used || title_used))
	fprintf(out, "\t0");

    fprintf(out, "\n};\n\n");
#else
    /*
     * Open the case.dat file.
     */
    snprintf(path, sizeof path, "%s" LDAP_DIRSEP "case.dat", opath);
    if ((out = fopen(path, "wb")) == 0)
      return;

    /*
     * Write the case mapping tables.
     */
    hdr[1] = upper_used + lower_used + title_used;
    casecnt[0] = upper_used;
    casecnt[1] = lower_used;

    /*
     * Write the header.
     */
    fwrite((char *) hdr, sizeof(ac_uint2), 2, out);

    /*
     * Write the upper and lower case table sizes.
     */
    fwrite((char *) casecnt, sizeof(ac_uint2), 2, out);