Adding upstream version 2.4.57+dfsg.upstream/2.4.57+dfsgupstream

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 1501 insertions, 0 deletions

diff --git a/libraries/liblunicode/ucdata/ucdata.c b/libraries/liblunicode/ucdata/ucdata.c
new file mode 100644
index 0000000..abe70c2
--- /dev/null
+++ b/libraries/liblunicode/ucdata/ucdata.c
@@ -0,0 +1,1501 @@
+/* $OpenLDAP$ */
+/* This work is part of OpenLDAP Software <http://www.openldap.org/>.
+ *
+ * Copyright 1998-2021 The OpenLDAP Foundation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted only as authorized by the OpenLDAP
+ * Public License.
+ *
+ * A copy of this license is available in file LICENSE in the
+ * top-level directory of the distribution or, alternatively, at
+ * <http://www.OpenLDAP.org/license.html>.
+ */
+/* Copyright 2001 Computing Research Labs, New Mexico State University
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY
+ * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT
+ * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
+ * THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+/* $Id: ucdata.c,v 1.4 2001/01/02 18:46:20 mleisher Exp $" */
+
+#include "portable.h"
+#include "ldap_config.h"
+
+#include <stdio.h>
+#include <ac/stdlib.h>
+#include <ac/string.h>
+#include <ac/unistd.h>
+
+#include <ac/bytes.h>
+
+#include "lber_pvt.h"
+#include "ucdata.h"
+
+#ifndef HARDCODE_DATA
+#define	HARDCODE_DATA	1
+#endif
+
+#if HARDCODE_DATA
+#include "uctable.h"
+#endif
+
+/**************************************************************************
+ *
+ * Miscellaneous types, data, and support functions.
+ *
+ **************************************************************************/
+
+typedef struct {
+    ac_uint2 bom;
+    ac_uint2 cnt;
+    union {
+        ac_uint4 bytes;
+        ac_uint2 len[2]; 
+    } size;
+} _ucheader_t;
+
+/*
+ * A simple array of 32-bit masks for lookup.
+ */
+static ac_uint4 masks32[32] = {
+	0x00000001UL, 0x00000002UL, 0x00000004UL, 0x00000008UL,
+	0x00000010UL, 0x00000020UL, 0x00000040UL, 0x00000080UL,
+	0x00000100UL, 0x00000200UL, 0x00000400UL, 0x00000800UL,
+	0x00001000UL, 0x00002000UL, 0x00004000UL, 0x00008000UL,
+	0x00010000UL, 0x00020000UL, 0x00040000UL, 0x00080000UL,
+	0x00100000UL, 0x00200000UL, 0x00400000UL, 0x00800000UL,
+	0x01000000UL, 0x02000000UL, 0x04000000UL, 0x08000000UL,
+	0x10000000UL, 0x20000000UL, 0x40000000UL, 0x80000000UL
+};
+
+#define endian_short(cc) (((cc) >> 8) | (((cc) & 0xff) << 8))
+#define endian_long(cc) ((((cc) & 0xff) << 24)|((((cc) >> 8) & 0xff) << 16)|\
+                        ((((cc) >> 16) & 0xff) << 8)|((cc) >> 24))
+
+#if !HARDCODE_DATA
+static FILE *
+_ucopenfile(char *paths, char *filename, char *mode)
+{
+    FILE *f;
+    char *fp, *dp, *pp, path[BUFSIZ];
+
+    if (filename == 0 || *filename == 0)
+      return 0;
+
+    dp = paths;
+    while (dp && *dp) {
+        pp = path;
+        while (*dp && *dp != ':')
+          *pp++ = *dp++;
+        *pp++ = *LDAP_DIRSEP;
+
+        fp = filename;
+        while (*fp)
+          *pp++ = *fp++;
+        *pp = 0;
+
+        if ((f = fopen(path, mode)) != 0)
+          return f;
+
+        if (*dp == ':')
+          dp++;
+    }
+
+    return 0;
+}
+#endif
+
+/**************************************************************************
+ *
+ * Support for the character properties.
+ *
+ **************************************************************************/
+
+#if !HARDCODE_DATA
+
+static ac_uint4 _ucprop_size;
+static ac_uint2 *_ucprop_offsets;
+static ac_uint4 *_ucprop_ranges;
+
+/*
+ * Return -1 on error, 0 if okay
+ */
+static int
+_ucprop_load(char *paths, int reload)
+{
+    FILE *in;
+    ac_uint4 size, i;
+    _ucheader_t hdr;
+
+    if (_ucprop_size > 0) {
+        if (!reload)
+          /*
+           * The character properties have already been loaded.
+           */
+          return 0;
+
+        /*
+         * Unload the current character property data in preparation for
+         * loading a new copy.  Only the first array has to be deallocated
+         * because all the memory for the arrays is allocated as a single
+         * block.
+         */
+        free((char *) _ucprop_offsets);
+        _ucprop_size = 0;
+    }
+
+    if ((in = _ucopenfile(paths, "ctype.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);
+    }
+
+    if ((_ucprop_size = hdr.cnt) == 0) {
+        fclose(in);
+        return -1;
+    }
+
+    /*
+     * Allocate all the storage needed for the lookup table.
+     */
+    _ucprop_offsets = (ac_uint2 *) malloc(hdr.size.bytes);
+
+    /*
+     * Calculate the offset into the storage for the ranges.  The offsets
+     * array is on a 4-byte boundary and one larger than the value provided in
+     * the header count field.  This means the offset to the ranges must be
+     * calculated after aligning the count to a 4-byte boundary.
+     */
+    if ((size = ((hdr.cnt + 1) * sizeof(ac_uint2))) & 3)
+      size += 4 - (size & 3);
+    size >>= 1;
+    _ucprop_ranges = (ac_uint4 *) (_ucprop_offsets + size);
+
+    /*
+     * Load the offset array.
+     */
+    fread((char *) _ucprop_offsets, sizeof(ac_uint2), size, in);
+
+    /*
+     * Do an endian swap if necessary.  Don't forget there is an extra node on
+     * the end with the final index.
+     */
+    if (hdr.bom == 0xfffe) {
+        for (i = 0; i <= _ucprop_size; i++)
+          _ucprop_offsets[i] = endian_short(_ucprop_offsets[i]);
+    }
+
+    /*
+     * Load the ranges.  The number of elements is in the last array position
+     * of the offsets.
+     */
+    fread((char *) _ucprop_ranges, sizeof(ac_uint4),
+          _ucprop_offsets[_ucprop_size], in);
+
+    fclose(in);
+
+    /*
+     * Do an endian swap if necessary.
+     */
+    if (hdr.bom == 0xfffe) {
+        for (i = 0; i < _ucprop_offsets[_ucprop_size]; i++)
+          _ucprop_ranges[i] = endian_long(_ucprop_ranges[i]);
+    }
+    return 0;
+}
+
+static void
+_ucprop_unload(void)
+{
+    if (_ucprop_size == 0)
+      return;
+
+    /*
+     * Only need to free the offsets because the memory is allocated as a
+     * single block.
+     */
+    free((char *) _ucprop_offsets);
+    _ucprop_size = 0;
+}
+#endif
+
+static int
+_ucprop_lookup(ac_uint4 code, ac_uint4 n)
+{
+    long l, r, m;
+
+    if (_ucprop_size == 0)
+      return 0;
+
+    /*
+     * There is an extra node on the end of the offsets to allow this routine
+     * to work right.  If the index is 0xffff, then there are no nodes for the
+     * property.
+     */
+    if ((l = _ucprop_offsets[n]) == 0xffff)
+      return 0;
+
+    /*
+     * Locate the next offset that is not 0xffff.  The sentinel at the end of
+     * the array is the max index value.
+     */
+    for (m = 1;
+         n + m < _ucprop_size && _ucprop_offsets[n + m] == 0xffff; m++) ;
+
+    r = _ucprop_offsets[n + m] - 1;
+
+    while (l <= r) {
+        /*
+         * Determine a "mid" point and adjust to make sure the mid point is at
+         * the beginning of a range pair.
+         */
+        m = (l + r) >> 1;
+        m -= (m & 1);
+        if (code > _ucprop_ranges[m + 1])
+          l = m + 2;
+        else if (code < _ucprop_ranges[m])
+          r = m - 2;
+        else if (code >= _ucprop_ranges[m] && code <= _ucprop_ranges[m + 1])
+          return 1;
+    }
+    return 0;
+}
+
+int
+ucisprop(ac_uint4 code, ac_uint4 mask1, ac_uint4 mask2)
+{
+    ac_uint4 i;
+
+    if (mask1 == 0 && mask2 == 0)
+      return 0;
+
+    for (i = 0; mask1 && i < 32; i++) {
+        if ((mask1 & masks32[i]) && _ucprop_lookup(code, i))
+          return 1;
+    }
+
+    for (i = 32; mask2 && i < _ucprop_size; i++) {
+        if ((mask2 & masks32[i & 31]) && _ucprop_lookup(code, i))
+          return 1;
+    }
+
+    return 0;
+}
+
+/**************************************************************************
+ *
+ * Support for case mapping.
+ *
+ **************************************************************************/
+
+#if !HARDCODE_DATA
+
+/* These record the number of slots in the map.
+ * There are 3 words per slot.
+ */
+static ac_uint4 _uccase_size;
+static ac_uint2 _uccase_len[2];
+static ac_uint4 *_uccase_map;
+
+/*
+ * Return -1 on error, 0 if okay
+ */
+static int
+_uccase_load(char *paths, int reload)
+{
+    FILE *in;
+    ac_uint4 i;
+    _ucheader_t hdr;
+
+    if (_uccase_size > 0) {
+        if (!reload)
+          /*
+           * The case mappings have already been loaded.
+           */
+          return 0;
+
+        free((char *) _uccase_map);
+        _uccase_size = 0;
+    }
+
+    if ((in = _ucopenfile(paths, "case.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.len[0] = endian_short(hdr.size.len[0]);
+        hdr.size.len[1] = endian_short(hdr.size.len[1]);
+    }
+
+    /*
+     * Set the node count and lengths of the upper and lower case mapping
+     * tables.
+     */
+    _uccase_size = hdr.cnt;
+    _uccase_len[0] = hdr.size.len[0];
+    _uccase_len[1] = hdr.size.len[1];
+
+    _uccase_map = (ac_uint4 *)
+        malloc(_uccase_size * 3 * sizeof(ac_uint4));
+
+    /*
+     * Load the case mapping table.
+     */
+    fread((char *) _uccase_map, sizeof(ac_uint4), _uccase_size * 3, in);
+
+    /*
+     * Do an endian swap if necessary.
+     */
+    if (hdr.bom == 0xfffe) {
+        for (i = 0; i < _uccase_size * 3; i++)
+          _uccase_map[i] = endian_long(_uccase_map[i]);
+    }
+    fclose(in);
+    return 0;
+}
+
+static void
+_uccase_unload(void)
+{
+    if (_uccase_size == 0)
+      return;
+
+    free((char *) _uccase_map);
+    _uccase_size = 0;
+}
+#endif
+
+static ac_uint4
+_uccase_lookup(ac_uint4 code, long l, long r, int field)
+{
+    long m;
+	const ac_uint4 *tmp;
+
+    /*
+     * Do the binary search.
+     */
+    while (l <= r) {
+        /*
+         * Determine a "mid" point and adjust to make sure the mid point is at
+         * the beginning of a case mapping triple.
+         */
+        m = (l + r) >> 1;
+		tmp = &_uccase_map[m*3];
+        if (code > *tmp)
+          l = m + 1;
+        else if (code < *tmp)
+          r = m - 1;
+        else if (code == *tmp)
+          return tmp[field];
+    }
+
+    return code;
+}
+
+ac_uint4
+uctoupper(ac_uint4 code)
+{
+    int field;
+    long l, r;
+
+    if (ucisupper(code))
+      return code;
+
+    if (ucislower(code)) {
+        /*
+         * The character is lower case.
+         */
+        field = 2;
+        l = _uccase_len[0];
+        r = (l + _uccase_len[1]) - 1;
+    } else {
+        /*
+         * The character is title case.
+         */
+        field = 1;
+        l = _uccase_len[0] + _uccase_len[1];
+        r = _uccase_size - 1;
+    }
+    return _uccase_lookup(code, l, r, field);
+}
+
+ac_uint4
+uctolower(ac_uint4 code)
+{
+    int field;
+    long l, r;
+
+    if (ucislower(code))
+      return code;
+
+    if (ucisupper(code)) {
+        /*
+         * The character is upper case.
+         */
+        field = 1;
+        l = 0;
+        r = _uccase_len[0] - 1;
+    } else {
+        /*
+         * The character is title case.
+         */
+        field = 2;
+        l = _uccase_len[0] + _uccase_len[1];
+        r = _uccase_size - 1;
+    }
+    return _uccase_lookup(code, l, r, field);
+}
+
+ac_uint4
+uctotitle(ac_uint4 code)
+{
+    int field;
+    long l, r;
+
+    if (ucistitle(code))
+      return code;
+
+    /*
+     * The offset will always be the same for converting to title case.
+     */
+    field = 2;
+
+    if (ucisupper(code)) {
+        /*
+         * The character is upper case.
+         */
+        l = 0;
+        r = _uccase_len[0] - 1;
+    } else {
+        /*
+         * The character is lower case.
+         */
+        l = _uccase_len[0];
+        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;
+}
+
+int
+uccanondecomp(const ac_uint4 *in, int inlen,
+              ac_uint4 **out, int *outlen, void *ctx)
+{
+    return uccanoncompatdecomp(in, inlen, out, outlen, 0, ctx);
+}
+
+int
+uccompatdecomp(const ac_uint4 *in, int inlen,
+	       ac_uint4 **out, int *outlen, void *ctx)
+{
+    return uccanoncompatdecomp(in, inlen, out, outlen, 1, ctx);
+}
+
+/**************************************************************************
+ *
+ * Support for combining classes.
+ *
+ **************************************************************************/
+
+#if !HARDCODE_DATA
+static ac_uint4  _uccmcl_size;
+static ac_uint4 *_uccmcl_nodes;
+
+/*
+ * Return -1 on error, 0 if okay
+ */
+static int
+_uccmcl_load(char *paths, int reload)
+{
+    FILE *in;
+    ac_uint4 i;
+    _ucheader_t hdr;
+
+    if (_uccmcl_size > 0) {
+        if (!reload)
+            /*
+             * The combining classes have already been loaded.
+             */
+            return 0;
+
+        free((char *) _uccmcl_nodes);
+        _uccmcl_size = 0;
+    }
+
+    if ((in = _ucopenfile(paths, "cmbcl.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);
+    }
+
+    _uccmcl_size = hdr.cnt * 3;
+    _uccmcl_nodes = (ac_uint4 *) malloc(hdr.size.bytes);
+
+    /*
+     * Read the combining classes in.
+     */
+    fread((char *) _uccmcl_nodes, sizeof(ac_uint4), _uccmcl_size, in);
+
+    /*
+     * Do an endian swap if necessary.
+     */
+    if (hdr.bom == 0xfffe) {
+        for (i = 0; i < _uccmcl_size; i++)
+            _uccmcl_nodes[i] = endian_long(_uccmcl_nodes[i]);
+    }
+    fclose(in);
+    return 0;
+}
+
+static void
+_uccmcl_unload(void)
+{
+    if (_uccmcl_size == 0)
+      return;
+
+    free((char *) _uccmcl_nodes);
+    _uccmcl_size = 0;
+}
+#endif
+
+ac_uint4
+uccombining_class(ac_uint4 code)
+{
+    long l, r, m;
+
+    l = 0;
+    r = _uccmcl_size - 1;
+
+    while (l <= r) {
+        m = (l + r) >> 1;
+        m -= (m % 3);
+        if (code > _uccmcl_nodes[m + 1])
+          l = m + 3;
+        else if (code < _uccmcl_nodes[m])
+          r = m - 3;
+        else if (code >= _uccmcl_nodes[m] && code <= _uccmcl_nodes[m + 1])
+          return _uccmcl_nodes[m + 2];
+    }
+    return 0;
+}
+
+/**************************************************************************
+ *
+ * Support for numeric values.
+ *
+ **************************************************************************/
+
+#if !HARDCODE_DATA
+static ac_uint4 *_ucnum_nodes;
+static ac_uint4 _ucnum_size;
+static short *_ucnum_vals;
+
+/*
+ * Return -1 on error, 0 if okay
+ */
+static int
+_ucnumb_load(char *paths, int reload)
+{
+    FILE *in;
+    ac_uint4 size, i;
+    _ucheader_t hdr;
+
+    if (_ucnum_size > 0) {
+        if (!reload)
+          /*
+           * The numbers have already been loaded.
+           */
+          return 0;
+
+        free((char *) _ucnum_nodes);
+        _ucnum_size = 0;
+    }
+
+    if ((in = _ucopenfile(paths, "num.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);
+    }
+
+    _ucnum_size = hdr.cnt;
+    _ucnum_nodes = (ac_uint4 *) malloc(hdr.size.bytes);
+    _ucnum_vals = (short *) (_ucnum_nodes + _ucnum_size);
+
+    /*
+     * Read the combining classes in.
+     */
+    fread((char *) _ucnum_nodes, sizeof(unsigned char), hdr.size.bytes, in);
+
+    /*
+     * Do an endian swap if necessary.
+     */
+    if (hdr.bom == 0xfffe) {
+        for (i = 0; i < _ucnum_size; i++)
+          _ucnum_nodes[i] = endian_long(_ucnum_nodes[i]);
+
+        /*
+         * Determine the number of values that have to be adjusted.
+         */
+        size = (hdr.size.bytes -
+                (_ucnum_size * (sizeof(ac_uint4) << 1))) /
+            sizeof(short);
+
+        for (i = 0; i < size; i++)
+          _ucnum_vals[i] = endian_short(_ucnum_vals[i]);
+    }
+    fclose(in);
+    return 0;
+}
+
+static void
+_ucnumb_unload(void)
+{
+    if (_ucnum_size == 0)
+      return;
+
+    free((char *) _ucnum_nodes);
+    _ucnum_size = 0;
+}
+#endif
+
+int
+ucnumber_lookup(ac_uint4 code, struct ucnumber *num)
+{
+    long l, r, m;
+    short *vp;
+
+    l = 0;
+    r = _ucnum_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 > _ucnum_nodes[m])
+          l = m + 2;
+        else if (code < _ucnum_nodes[m])
+          r = m - 2;
+        else {
+            vp = (short *)_ucnum_vals + _ucnum_nodes[m + 1];
+            num->numerator = (int) *vp++;
+            num->denominator = (int) *vp;
+            return 1;
+        }
+    }
+    return 0;
+}
+
+int
+ucdigit_lookup(ac_uint4 code, int *digit)
+{
+    long l, r, m;
+    short *vp;
+
+    l = 0;
+    r = _ucnum_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 > _ucnum_nodes[m])
+          l = m + 2;
+        else if (code < _ucnum_nodes[m])
+          r = m - 2;
+        else {
+            vp = (short *)_ucnum_vals + _ucnum_nodes[m + 1];
+            if (*vp == *(vp + 1)) {
+              *digit = *vp;
+              return 1;
+            }
+            return 0;
+        }
+    }
+    return 0;
+}
+
+struct ucnumber
+ucgetnumber(ac_uint4 code)
+{
+    struct ucnumber num;
+
+    /*
+     * Initialize with some arbitrary value, because the caller simply cannot
+     * tell for sure if the code is a number without calling the ucisnumber()
+     * macro before calling this function.
+     */
+    num.numerator = num.denominator = -111;
+
+    (void) ucnumber_lookup(code, &num);
+
+    return num;
+}
+
+int
+ucgetdigit(ac_uint4 code)
+{
+    int dig;
+
+    /*
+     * Initialize with some arbitrary value, because the caller simply cannot
+     * tell for sure if the code is a number without calling the ucisdigit()
+     * macro before calling this function.
+     */
+    dig = -111;
+
+    (void) ucdigit_lookup(code, &dig);
+
+    return dig;
+}
+
+/**************************************************************************
+ *
+ * Setup and cleanup routines.
+ *
+ **************************************************************************/
+
+#if HARDCODE_DATA
+int ucdata_load(char *paths, int masks) { return 0; }
+void ucdata_unload(int masks) { }
+int ucdata_reload(char *paths, int masks) { return 0; }
+#else
+/*
+ * Return 0 if okay, negative on error
+ */
+int
+ucdata_load(char *paths, int masks)
+{
+    int error = 0;
+
+    if (masks & UCDATA_CTYPE)
+      error |= _ucprop_load(paths, 0) < 0 ? UCDATA_CTYPE : 0;
+    if (masks & UCDATA_CASE)
+      error |= _uccase_load(paths, 0) < 0 ? UCDATA_CASE : 0;
+    if (masks & UCDATA_DECOMP)
+      error |= _ucdcmp_load(paths, 0) < 0 ? UCDATA_DECOMP : 0;
+    if (masks & UCDATA_CMBCL)
+      error |= _uccmcl_load(paths, 0) < 0 ? UCDATA_CMBCL : 0;
+    if (masks & UCDATA_NUM)
+      error |= _ucnumb_load(paths, 0) < 0 ? UCDATA_NUM : 0;
+    if (masks & UCDATA_COMP)
+      error |= _uccomp_load(paths, 0) < 0 ? UCDATA_COMP : 0;
+    if (masks & UCDATA_KDECOMP)
+      error |= _uckdcmp_load(paths, 0) < 0 ? UCDATA_KDECOMP : 0;
+
+    return -error;
+}
+
+void
+ucdata_unload(int masks)
+{
+    if (masks & UCDATA_CTYPE)
+      _ucprop_unload();
+    if (masks & UCDATA_CASE)
+      _uccase_unload();
+    if (masks & UCDATA_DECOMP)
+      _ucdcmp_unload();
+    if (masks & UCDATA_CMBCL)
+      _uccmcl_unload();
+    if (masks & UCDATA_NUM)
+      _ucnumb_unload();
+    if (masks & UCDATA_COMP)
+      _uccomp_unload();
+    if (masks & UCDATA_KDECOMP)
+      _uckdcmp_unload();
+}
+
+/*
+ * Return 0 if okay, negative on error
+ */
+int
+ucdata_reload(char *paths, int masks)
+{
+    int error = 0;
+
+    if (masks & UCDATA_CTYPE)
+        error |= _ucprop_load(paths, 1) < 0 ? UCDATA_CTYPE : 0;
+    if (masks & UCDATA_CASE)
+        error |= _uccase_load(paths, 1) < 0 ? UCDATA_CASE : 0;
+    if (masks & UCDATA_DECOMP)
+        error |= _ucdcmp_load(paths, 1) < 0 ? UCDATA_DECOMP : 0;
+    if (masks & UCDATA_CMBCL)
+        error |= _uccmcl_load(paths, 1) < 0 ? UCDATA_CMBCL : 0;
+    if (masks & UCDATA_NUM)
+        error |= _ucnumb_load(paths, 1) < 0 ? UCDATA_NUM : 0;
+    if (masks & UCDATA_COMP)
+        error |= _uccomp_load(paths, 1) < 0 ? UCDATA_COMP : 0;
+    if (masks & UCDATA_KDECOMP)
+        error |= _uckdcmp_load(paths, 1) < 0 ? UCDATA_KDECOMP : 0;
+
+    return -error;
+}
+#endif
+
+#ifdef TEST
+
+void
+main(void)
+{
+    int dig;
+    ac_uint4 i, lo, *dec;
+    struct ucnumber num;
+
+/*    ucdata_setup("."); */
+
+    if (ucisweak(0x30))
+      printf("WEAK\n");
+    else
+      printf("NOT WEAK\n");
+
+    printf("LOWER 0x%04lX\n", uctolower(0xff3a));
+    printf("UPPER 0x%04lX\n", uctoupper(0xff5a));
+
+    if (ucisalpha(0x1d5))
+      printf("ALPHA\n");
+    else
+      printf("NOT ALPHA\n");
+
+    if (ucisupper(0x1d5)) {
+        printf("UPPER\n");
+        lo = uctolower(0x1d5);
+        printf("0x%04lx\n", lo);
+        lo = uctotitle(0x1d5);
+        printf("0x%04lx\n", lo);
+    } else
+      printf("NOT UPPER\n");
+
+    if (ucistitle(0x1d5))
+      printf("TITLE\n");
+    else
+      printf("NOT TITLE\n");
+
+    if (uciscomposite(0x1d5))
+      printf("COMPOSITE\n");
+    else
+      printf("NOT COMPOSITE\n");
+
+    if (ucdecomp(0x1d5, &lo, &dec)) {
+        for (i = 0; i < lo; i++)
+          printf("0x%04lx ", dec[i]);
+        putchar('\n');
+    }
+
+    if ((lo = uccombining_class(0x41)) != 0)
+      printf("0x41 CCL %ld\n", lo);
+
+    if (ucisxdigit(0xfeff))
+      printf("0xFEFF HEX DIGIT\n");
+    else
+      printf("0xFEFF NOT HEX DIGIT\n");
+
+    if (ucisdefined(0x10000))
+      printf("0x10000 DEFINED\n");
+    else
+      printf("0x10000 NOT DEFINED\n");
+
+    if (ucnumber_lookup(0x30, &num)) {
+        if (num.denominator != 1)
+          printf("UCNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
+        else
+          printf("UCNUMBER: 0x30 = %d\n", num.numerator);
+    } else
+      printf("UCNUMBER: 0x30 NOT A NUMBER\n");
+
+    if (ucnumber_lookup(0xbc, &num)) {
+        if (num.denominator != 1)
+          printf("UCNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
+        else
+          printf("UCNUMBER: 0xbc = %d\n", num.numerator);
+    } else
+      printf("UCNUMBER: 0xbc NOT A NUMBER\n");
+
+
+    if (ucnumber_lookup(0xff19, &num)) {
+        if (num.denominator != 1)
+          printf("UCNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
+        else
+          printf("UCNUMBER: 0xff19 = %d\n", num.numerator);
+    } else
+      printf("UCNUMBER: 0xff19 NOT A NUMBER\n");
+
+    if (ucnumber_lookup(0x4e00, &num)) {
+        if (num.denominator != 1)
+          printf("UCNUMBER: 0x4e00 = %d/%d\n", num.numerator, num.denominator);
+        else
+          printf("UCNUMBER: 0x4e00 = %d\n", num.numerator);
+    } else
+      printf("UCNUMBER: 0x4e00 NOT A NUMBER\n");
+
+    if (ucdigit_lookup(0x06f9, &dig))
+      printf("UCDIGIT: 0x6f9 = %d\n", dig);
+    else
+      printf("UCDIGIT: 0x6f9 NOT A NUMBER\n");
+
+    dig = ucgetdigit(0x0969);
+    printf("UCGETDIGIT: 0x969 = %d\n", dig);
+
+    num = ucgetnumber(0x30);
+    if (num.denominator != 1)
+      printf("UCGETNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
+    else
+      printf("UCGETNUMBER: 0x30 = %d\n", num.numerator);
+
+    num = ucgetnumber(0xbc);
+    if (num.denominator != 1)
+      printf("UCGETNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
+    else
+      printf("UCGETNUMBER: 0xbc = %d\n", num.numerator);
+
+    num = ucgetnumber(0xff19);
+    if (num.denominator != 1)
+      printf("UCGETNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
+    else
+      printf("UCGETNUMBER: 0xff19 = %d\n", num.numerator);
+
+/*    ucdata_cleanup(); */
+    exit(0);
+}
+
+#endif /* TEST */