ucdata.c

ldap服务器源码

        r = (l + _uccase_len[1]) - 1;
    }
    return _uccase_lookup(code, l, r, field);
}

/**************************************************************************
 *
 * Support for compositions.
 *
 **************************************************************************/

#if !HARDCODE_DATA

static ac_uint4  _uccomp_size;
static ac_uint4 *_uccomp_data;

/*
 * Return -1 on error, 0 if okay
 */
static int
_uccomp_load(char *paths, int reload)
{
    FILE *in;
    ac_uint4 size, i;
    _ucheader_t hdr;

    if (_uccomp_size > 0) {
        if (!reload)
            /*
             * The compositions have already been loaded.
             */
            return 0;

        free((char *) _uccomp_data);
        _uccomp_size = 0;
    }

    if ((in = _ucopenfile(paths, "comp.dat", "rb")) == 0)
        return -1;

    /*
     * Load the header.
     */
    fread((char *) &hdr, sizeof(_ucheader_t), 1, in);

    if (hdr.bom == 0xfffe) {
        hdr.cnt = endian_short(hdr.cnt);
        hdr.size.bytes = endian_long(hdr.size.bytes);
    }

    _uccomp_size = hdr.cnt;
    _uccomp_data = (ac_uint4 *) malloc(hdr.size.bytes);

    /*
     * Read the composition data in.
     */
    size = hdr.size.bytes / sizeof(ac_uint4);
    fread((char *) _uccomp_data, sizeof(ac_uint4), size, in);

    /*
     * Do an endian swap if necessary.
     */
    if (hdr.bom == 0xfffe) {
        for (i = 0; i < size; i++)
            _uccomp_data[i] = endian_long(_uccomp_data[i]);
    }

    /*
     * Assume that the data is ordered on count, so that all compositions
     * of length 2 come first. Only handling length 2 for now.
     */
    for (i = 1; i < size; i += 4)
      if (_uccomp_data[i] != 2)
        break;
    _uccomp_size = i - 1;

    fclose(in);
    return 0;
}

static void
_uccomp_unload(void)
{
    if (_uccomp_size == 0)
        return;

    free((char *) _uccomp_data);
    _uccomp_size = 0;
}
#endif

int
uccomp(ac_uint4 node1, ac_uint4 node2, ac_uint4 *comp)
{
    int l, r, m;

    l = 0;
    r = _uccomp_size - 1;

    while (l <= r) {
        m = ((r + l) >> 1);
        m -= m & 3;
        if (node1 > _uccomp_data[m+2])
          l = m + 4;
        else if (node1 < _uccomp_data[m+2])
          r = m - 4;
        else if (node2 > _uccomp_data[m+3])
          l = m + 4;
        else if (node2 < _uccomp_data[m+3])
          r = m - 4;
        else {
            *comp = _uccomp_data[m];
            return 1;
        }
    }
    return 0;
}

int
uccomp_hangul(ac_uint4 *str, int len)
{
    const int SBase = 0xAC00, LBase = 0x1100,
        VBase = 0x1161, TBase = 0x11A7,
        LCount = 19, VCount = 21, TCount = 28,
        NCount = VCount * TCount,   /* 588 */
        SCount = LCount * NCount;   /* 11172 */
    
    int i, rlen;
    ac_uint4 ch, last, lindex, sindex;

    last = str[0];
    rlen = 1;
    for ( i = 1; i < len; i++ ) {
        ch = str[i];

        /* check if two current characters are L and V */
        lindex = last - LBase;
        if (lindex < (ac_uint4) LCount) {
            ac_uint4 vindex = ch - VBase;
            if (vindex < (ac_uint4) VCount) {
                /* make syllable of form LV */
                last = SBase + (lindex * VCount + vindex) * TCount;
                str[rlen-1] = last; /* reset last */
                continue;
            }
        }
        
        /* check if two current characters are LV and T */
        sindex = last - SBase;
        if (sindex < (ac_uint4) SCount
			&& (sindex % TCount) == 0)
		{
            ac_uint4 tindex = ch - TBase;
            if (tindex <= (ac_uint4) TCount) {
                /* make syllable of form LVT */
                last += tindex;
                str[rlen-1] = last; /* reset last */
                continue;
            }
        }

        /* if neither case was true, just add the character */
        last = ch;
        str[rlen] = ch;
        rlen++;
    }
    return rlen;
}

int
uccanoncomp(ac_uint4 *str, int len)
{
    int i, stpos, copos;
    ac_uint4 cl, prevcl, st, ch, co;

    st = str[0];
    stpos = 0;
    copos = 1;
    prevcl = uccombining_class(st) == 0 ? 0 : 256;
        
    for (i = 1; i < len; i++) {
        ch = str[i];
        cl = uccombining_class(ch);
        if (uccomp(st, ch, &co) && (prevcl < cl || prevcl == 0))
          st = str[stpos] = co;
        else {
            if (cl == 0) {
                stpos = copos;
                st = ch;
            }
            prevcl = cl;
            str[copos++] = ch;
        }
    }

    return uccomp_hangul(str, copos);
}

/**************************************************************************
 *
 * Support for decompositions.
 *
 **************************************************************************/

#if !HARDCODE_DATA

static ac_uint4  _ucdcmp_size;
static ac_uint4 *_ucdcmp_nodes;
static ac_uint4 *_ucdcmp_decomp;

static ac_uint4  _uckdcmp_size;
static ac_uint4 *_uckdcmp_nodes;
static ac_uint4 *_uckdcmp_decomp;

/*
 * Return -1 on error, 0 if okay
 */
static int
_ucdcmp_load(char *paths, int reload)
{
    FILE *in;
    ac_uint4 size, i;
    _ucheader_t hdr;

    if (_ucdcmp_size > 0) {
        if (!reload)
            /*
             * The decompositions have already been loaded.
             */
          return 0;

        free((char *) _ucdcmp_nodes);
        _ucdcmp_size = 0;
    }

    if ((in = _ucopenfile(paths, "decomp.dat", "rb")) == 0)
        return -1;

    /*
     * Load the header.
     */
    fread((char *) &hdr, sizeof(_ucheader_t), 1, in);

    if (hdr.bom == 0xfffe) {
        hdr.cnt = endian_short(hdr.cnt);
        hdr.size.bytes = endian_long(hdr.size.bytes);
    }

    _ucdcmp_size = hdr.cnt << 1;
    _ucdcmp_nodes = (ac_uint4 *) malloc(hdr.size.bytes);
    _ucdcmp_decomp = _ucdcmp_nodes + (_ucdcmp_size + 1);

    /*
     * Read the decomposition data in.
     */
    size = hdr.size.bytes / sizeof(ac_uint4);
    fread((char *) _ucdcmp_nodes, sizeof(ac_uint4), size, in);

    /*
     * Do an endian swap if necessary.
     */
    if (hdr.bom == 0xfffe) {
        for (i = 0; i < size; i++)
            _ucdcmp_nodes[i] = endian_long(_ucdcmp_nodes[i]);
    }
    fclose(in);
    return 0;
}

/*
 * Return -1 on error, 0 if okay
 */
static int
_uckdcmp_load(char *paths, int reload)
{
    FILE *in;
    ac_uint4 size, i;
    _ucheader_t hdr;

    if (_uckdcmp_size > 0) {
        if (!reload)
            /*
             * The decompositions have already been loaded.
             */
          return 0;

        free((char *) _uckdcmp_nodes);
        _uckdcmp_size = 0;
    }

    if ((in = _ucopenfile(paths, "kdecomp.dat", "rb")) == 0)
        return -1;

    /*
     * Load the header.
     */
    fread((char *) &hdr, sizeof(_ucheader_t), 1, in);

    if (hdr.bom == 0xfffe) {
        hdr.cnt = endian_short(hdr.cnt);
        hdr.size.bytes = endian_long(hdr.size.bytes);
    }

    _uckdcmp_size = hdr.cnt << 1;
    _uckdcmp_nodes = (ac_uint4 *) malloc(hdr.size.bytes);
    _uckdcmp_decomp = _uckdcmp_nodes + (_uckdcmp_size + 1);

    /*
     * Read the decomposition data in.
     */
    size = hdr.size.bytes / sizeof(ac_uint4);
    fread((char *) _uckdcmp_nodes, sizeof(ac_uint4), size, in);

    /*
     * Do an endian swap if necessary.
     */
    if (hdr.bom == 0xfffe) {
        for (i = 0; i < size; i++)
            _uckdcmp_nodes[i] = endian_long(_uckdcmp_nodes[i]);
    }
    fclose(in);
    return 0;
}

static void
_ucdcmp_unload(void)
{
    if (_ucdcmp_size == 0)
      return;

    /*
     * Only need to free the offsets because the memory is allocated as a
     * single block.
     */
    free((char *) _ucdcmp_nodes);
    _ucdcmp_size = 0;
}

static void
_uckdcmp_unload(void)
{
    if (_uckdcmp_size == 0)
      return;

    /*
     * Only need to free the offsets because the memory is allocated as a
     * single block.
     */
    free((char *) _uckdcmp_nodes);
    _uckdcmp_size = 0;
}
#endif

int
ucdecomp(ac_uint4 code, ac_uint4 *num, ac_uint4 **decomp)
{
    long l, r, m;

    if (code < _ucdcmp_nodes[0]) {
	return 0;
    }

    l = 0;
    r = _ucdcmp_nodes[_ucdcmp_size] - 1;

    while (l <= r) {
        /*
         * Determine a "mid" point and adjust to make sure the mid point is at
         * the beginning of a code+offset pair.
         */
        m = (l + r) >> 1;
        m -= (m & 1);
        if (code > _ucdcmp_nodes[m])
          l = m + 2;
        else if (code < _ucdcmp_nodes[m])
          r = m - 2;
        else if (code == _ucdcmp_nodes[m]) {
            *num = _ucdcmp_nodes[m + 3] - _ucdcmp_nodes[m + 1];
            *decomp = (ac_uint4*)&_ucdcmp_decomp[_ucdcmp_nodes[m + 1]];
            return 1;
        }
    }
    return 0;
}

int
uckdecomp(ac_uint4 code, ac_uint4 *num, ac_uint4 **decomp)
{
    long l, r, m;

    if (code < _uckdcmp_nodes[0]) {
	return 0;
    }
    
    l = 0;
    r = _uckdcmp_nodes[_uckdcmp_size] - 1;

    while (l <= r) {
        /*
         * Determine a "mid" point and adjust to make sure the mid point is at
         * the beginning of a code+offset pair.
         */
        m = (l + r) >> 1;
        m -= (m & 1);
        if (code > _uckdcmp_nodes[m])
          l = m + 2;
        else if (code < _uckdcmp_nodes[m])
          r = m - 2;
        else if (code == _uckdcmp_nodes[m]) {
            *num = _uckdcmp_nodes[m + 3] - _uckdcmp_nodes[m + 1];
            *decomp = (ac_uint4*)&_uckdcmp_decomp[_uckdcmp_nodes[m + 1]];
            return 1;
        }
    }
    return 0;
}

int
ucdecomp_hangul(ac_uint4 code, ac_uint4 *num, ac_uint4 decomp[])
{
    if (!ucishangul(code))
      return 0;

    code -= 0xac00;
    decomp[0] = 0x1100 + (ac_uint4) (code / 588);
    decomp[1] = 0x1161 + (ac_uint4) ((code % 588) / 28);
    decomp[2] = 0x11a7 + (ac_uint4) (code % 28);
    *num = (decomp[2] != 0x11a7) ? 3 : 2;

    return 1;
}

/* mode == 0 for canonical, mode == 1 for compatibility */
static int
uccanoncompatdecomp(const ac_uint4 *in, int inlen,
		    ac_uint4 **out, int *outlen, short mode, void *ctx)
{
    int l, size;
	unsigned i, j, k;
    ac_uint4 num, class, *decomp, hangdecomp[3];

    size = inlen * 2;
    *out = (ac_uint4 *) ber_memalloc_x(size * sizeof(**out), ctx);
    if (*out == NULL)
        return *outlen = -1;

    i = 0;
    for (j = 0; j < (unsigned) inlen; j++) {
	if (mode ? uckdecomp(in[j], &num, &decomp) : ucdecomp(in[j], &num, &decomp)) {
            if ( size - i < num) {
                size = inlen + i - j + num - 1;
                *out = (ac_uint4 *) ber_memrealloc_x(*out, size * sizeof(**out), ctx );
                if (*out == NULL)
                    return *outlen = -1;
            }
            for (k = 0; k < num; k++) {
                class = uccombining_class(decomp[k]);
                if (class == 0) {
                    (*out)[i] = decomp[k];
                } else {
                    for (l = i; l > 0; l--)
                        if (class >= uccombining_class((*out)[l-1]))
                            break;
                    AC_MEMCPY(*out + l + 1, *out + l, (i - l) * sizeof(**out));
                    (*out)[l] = decomp[k];
                }
                i++;
            }
        } else if (ucdecomp_hangul(in[j], &num, hangdecomp)) {
            if (size - i < num) {
                size = inlen + i - j + num - 1;
                *out = (ac_uint4 *) ber_memrealloc_x(*out, size * sizeof(**out), ctx);
                if (*out == NULL)
                    return *outlen = -1;
            }
            for (k = 0; k < num; k++) {
                (*out)[i] = hangdecomp[k];
                i++;
            }
        } else {
            if (size - i < 1) {
                size = inlen + i - j;
                *out = (ac_uint4 *) ber_memrealloc_x(*out, size * sizeof(**out), ctx);
                if (*out == NULL)
                    return *outlen = -1;
            }
            class = uccombining_class(in[j]);
            if (class == 0) {
                (*out)[i] = in[j];
            } else {
                for (l = i; l > 0; l--)
                    if (class >= uccombining_class((*out)[l-1]))
                        break;
                AC_MEMCPY(*out + l + 1, *out + l, (i - l) * sizeof(**out));
                (*out)[l] = in[j];
            }
            i++;
        }
    }
    return *outlen = i;
}