michael@0: /*
michael@0:  * Copyright 1996, 1997, 1998 Computing Research Labs,
michael@0:  * New Mexico State University
michael@0:  *
michael@0:  * Permission is hereby granted, free of charge, to any person obtaining a
michael@0:  * copy of this software and associated documentation files (the "Software"),
michael@0:  * to deal in the Software without restriction, including without limitation
michael@0:  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
michael@0:  * and/or sell copies of the Software, and to permit persons to whom the
michael@0:  * Software is furnished to do so, subject to the following conditions:
michael@0:  *
michael@0:  * The above copyright notice and this permission notice shall be included in
michael@0:  * all copies or substantial portions of the Software.
michael@0:  *
michael@0:  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
michael@0:  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
michael@0:  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
michael@0:  * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY
michael@0:  * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT
michael@0:  * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
michael@0:  * THE USE OR OTHER DEALINGS IN THE SOFTWARE.
michael@0:  */
michael@0: #ifndef lint
michael@0: #ifdef __GNUC__
michael@0: static char rcsid[] __attribute__ ((unused)) = "$Id: ucdata.c,v 1.1 1999/01/08 00:19:11 ftang%netscape.com Exp $";
michael@0: #else
michael@0: static char rcsid[] = "$Id: ucdata.c,v 1.1 1999/01/08 00:19:11 ftang%netscape.com Exp $";
michael@0: #endif
michael@0: #endif
michael@0: 
michael@0: #include <stdio.h>
michael@0: #include <stdlib.h>
michael@0: #include <string.h>
michael@0: #ifndef WIN32
michael@0: #include <unistd.h>
michael@0: #endif
michael@0: 
michael@0: #include "ucdata.h"
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Miscellaneous types, data, and support functions.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: typedef struct {
michael@0:     unsigned short bom;
michael@0:     unsigned short cnt;
michael@0:     union {
michael@0:         unsigned long bytes;
michael@0:         unsigned short len[2];
michael@0:     } size;
michael@0: } _ucheader_t;
michael@0: 
michael@0: /*
michael@0:  * A simple array of 32-bit masks for lookup.
michael@0:  */
michael@0: static unsigned long masks32[32] = {
michael@0:     0x00000001, 0x00000002, 0x00000004, 0x00000008, 0x00000010, 0x00000020,
michael@0:     0x00000040, 0x00000080, 0x00000100, 0x00000200, 0x00000400, 0x00000800,
michael@0:     0x00001000, 0x00002000, 0x00004000, 0x00008000, 0x00010000, 0x00020000,
michael@0:     0x00040000, 0x00080000, 0x00100000, 0x00200000, 0x00400000, 0x00800000,
michael@0:     0x01000000, 0x02000000, 0x04000000, 0x08000000, 0x10000000, 0x20000000,
michael@0:     0x40000000, 0x80000000
michael@0: };
michael@0: 
michael@0: #define endian_short(cc) (((cc) >> 8) | (((cc) & 0xff) << 8))
michael@0: #define endian_long(cc) ((((cc) & 0xff) << 24)|((((cc) >> 8) & 0xff) << 16)|\
michael@0:                         ((((cc) >> 16) & 0xff) << 8)|((cc) >> 24))
michael@0: 
michael@0: static FILE *
michael@0: #ifdef __STDC__
michael@0: _ucopenfile(char *paths, char *filename, char *mode)
michael@0: #else
michael@0: _ucopenfile(paths, filename, mode)
michael@0: char *paths, *filename, *mode;
michael@0: #endif
michael@0: {
michael@0:     FILE *f;
michael@0:     char *fp, *dp, *pp, path[BUFSIZ];
michael@0: 
michael@0:     if (filename == 0 || *filename == 0)
michael@0:       return 0;
michael@0: 
michael@0:     dp = paths;
michael@0:     while (dp && *dp) {
michael@0:         pp = path;
michael@0:         while (*dp && *dp != ':')
michael@0:           *pp++ = *dp++;
michael@0:         *pp++ = '/';
michael@0: 
michael@0:         fp = filename;
michael@0:         while (*fp)
michael@0:           *pp++ = *fp++;
michael@0:         *pp = 0;
michael@0: 
michael@0:         if ((f = fopen(path, mode)) != 0)
michael@0:           return f;
michael@0: 
michael@0:         if (*dp == ':')
michael@0:           dp++;
michael@0:     }
michael@0: 
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Support for the character properties.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: static unsigned long  _ucprop_size;
michael@0: static unsigned short *_ucprop_offsets;
michael@0: static unsigned long  *_ucprop_ranges;
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _ucprop_load(char *paths, int reload)
michael@0: #else
michael@0: _ucprop_load(paths, reload)
michael@0: char *paths;
michael@0: int reload;
michael@0: #endif
michael@0: {
michael@0:     FILE *in;
michael@0:     unsigned long size, i;
michael@0:     _ucheader_t hdr;
michael@0: 
michael@0:     if (_ucprop_size > 0) {
michael@0:         if (!reload)
michael@0:           /*
michael@0:            * The character properties have already been loaded.
michael@0:            */
michael@0:           return;
michael@0: 
michael@0:         /*
michael@0:          * Unload the current character property data in preparation for
michael@0:          * loading a new copy.  Only the first array has to be deallocated
michael@0:          * because all the memory for the arrays is allocated as a single
michael@0:          * block.
michael@0:          */
michael@0:         free((char *) _ucprop_offsets);
michael@0:         _ucprop_size = 0;
michael@0:     }
michael@0: 
michael@0:     if ((in = _ucopenfile(paths, "ctype.dat", "rb")) == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Load the header.
michael@0:      */
michael@0:     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
michael@0: 
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         hdr.cnt = endian_short(hdr.cnt);
michael@0:         hdr.size.bytes = endian_long(hdr.size.bytes);
michael@0:     }
michael@0: 
michael@0:     if ((_ucprop_size = hdr.cnt) == 0) {
michael@0:         fclose(in);
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * Allocate all the storage needed for the lookup table.
michael@0:      */
michael@0:     _ucprop_offsets = (unsigned short *) malloc(hdr.size.bytes);
michael@0: 
michael@0:     /*
michael@0:      * Calculate the offset into the storage for the ranges.  The offsets
michael@0:      * array is on a 4-byte boundary and one larger than the value provided in
michael@0:      * the header count field.  This means the offset to the ranges must be
michael@0:      * calculated after aligning the count to a 4-byte boundary.
michael@0:      */
michael@0:     if ((size = ((hdr.cnt + 1) * sizeof(unsigned short))) & 3)
michael@0:       size += 4 - (size & 3);
michael@0:     size >>= 1;
michael@0:     _ucprop_ranges = (unsigned long *) (_ucprop_offsets + size);
michael@0: 
michael@0:     /*
michael@0:      * Load the offset array.
michael@0:      */
michael@0:     fread((char *) _ucprop_offsets, sizeof(unsigned short), size, in);
michael@0: 
michael@0:     /*
michael@0:      * Do an endian swap if necessary.  Don't forget there is an extra node on
michael@0:      * the end with the final index.
michael@0:      */
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         for (i = 0; i <= _ucprop_size; i++)
michael@0:           _ucprop_offsets[i] = endian_short(_ucprop_offsets[i]);
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * Load the ranges.  The number of elements is in the last array position
michael@0:      * of the offsets.
michael@0:      */
michael@0:     fread((char *) _ucprop_ranges, sizeof(unsigned long),
michael@0:           _ucprop_offsets[_ucprop_size], in);
michael@0: 
michael@0:     fclose(in);
michael@0: 
michael@0:     /*
michael@0:      * Do an endian swap if necessary.
michael@0:      */
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         for (i = 0; i < _ucprop_offsets[_ucprop_size]; i++)
michael@0:           _ucprop_ranges[i] = endian_long(_ucprop_ranges[i]);
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _ucprop_unload(void)
michael@0: #else
michael@0: _ucprop_unload()
michael@0: #endif
michael@0: {
michael@0:     if (_ucprop_size == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Only need to free the offsets because the memory is allocated as a
michael@0:      * single block.
michael@0:      */
michael@0:     free((char *) _ucprop_offsets);
michael@0:     _ucprop_size = 0;
michael@0: }
michael@0: 
michael@0: static int
michael@0: #ifdef __STDC__
michael@0: _ucprop_lookup(unsigned long code, unsigned long n)
michael@0: #else
michael@0: _ucprop_lookup(code, n)
michael@0: unsigned long code, n;
michael@0: #endif
michael@0: {
michael@0:     long l, r, m;
michael@0: 
michael@0:     /*
michael@0:      * There is an extra node on the end of the offsets to allow this routine
michael@0:      * to work right.  If the index is 0xffff, then there are no nodes for the
michael@0:      * property.
michael@0:      */
michael@0:     if ((l = _ucprop_offsets[n]) == 0xffff)
michael@0:       return 0;
michael@0: 
michael@0:     /*
michael@0:      * Locate the next offset that is not 0xffff.  The sentinel at the end of
michael@0:      * the array is the max index value.
michael@0:      */
michael@0:     for (m = 1;
michael@0:          n + m < _ucprop_size && _ucprop_offsets[n + m] == 0xffff; m++) ;
michael@0: 
michael@0:     r = _ucprop_offsets[n + m] - 1;
michael@0: 
michael@0:     while (l <= r) {
michael@0:         /*
michael@0:          * Determine a "mid" point and adjust to make sure the mid point is at
michael@0:          * the beginning of a range pair.
michael@0:          */
michael@0:         m = (l + r) >> 1;
michael@0:         m -= (m & 1);
michael@0:         if (code > _ucprop_ranges[m + 1])
michael@0:           l = m + 2;
michael@0:         else if (code < _ucprop_ranges[m])
michael@0:           r = m - 2;
michael@0:         else if (code >= _ucprop_ranges[m] && code <= _ucprop_ranges[m + 1])
michael@0:           return 1;
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: int
michael@0: #ifdef __STDC__
michael@0: ucisprop(unsigned long code, unsigned long mask1, unsigned long mask2)
michael@0: #else
michael@0: ucisprop(code, mask1, mask2)
michael@0: unsigned long code, mask1, mask2;
michael@0: #endif
michael@0: {
michael@0:     unsigned long i;
michael@0: 
michael@0:     if (mask1 == 0 && mask2 == 0)
michael@0:       return 0;
michael@0: 
michael@0:     for (i = 0; mask1 && i < 32; i++) {
michael@0:         if ((mask1 & masks32[i]) && _ucprop_lookup(code, i))
michael@0:           return 1;
michael@0:     }
michael@0: 
michael@0:     for (i = 32; mask2 && i < _ucprop_size; i++) {
michael@0:         if ((mask2 & masks32[i & 31]) && _ucprop_lookup(code, i))
michael@0:           return 1;
michael@0:     }
michael@0: 
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Support for case mapping.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: static unsigned long _uccase_size;
michael@0: static unsigned short _uccase_len[2];
michael@0: static unsigned long *_uccase_map;
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _uccase_load(char *paths, int reload)
michael@0: #else
michael@0: _uccase_load(paths, reload)
michael@0: char *paths;
michael@0: int reload;
michael@0: #endif
michael@0: {
michael@0:     FILE *in;
michael@0:     unsigned long i;
michael@0:     _ucheader_t hdr;
michael@0: 
michael@0:     if (_uccase_size > 0) {
michael@0:         if (!reload)
michael@0:           /*
michael@0:            * The case mappings have already been loaded.
michael@0:            */
michael@0:           return;
michael@0: 
michael@0:         free((char *) _uccase_map);
michael@0:         _uccase_size = 0;
michael@0:     }
michael@0: 
michael@0:     if ((in = _ucopenfile(paths, "case.dat", "rb")) == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Load the header.
michael@0:      */
michael@0:     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
michael@0: 
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         hdr.cnt = endian_short(hdr.cnt);
michael@0:         hdr.size.len[0] = endian_short(hdr.size.len[0]);
michael@0:         hdr.size.len[1] = endian_short(hdr.size.len[1]);
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * Set the node count and lengths of the upper and lower case mapping
michael@0:      * tables.
michael@0:      */
michael@0:     _uccase_size = hdr.cnt * 3;
michael@0:     _uccase_len[0] = hdr.size.len[0] * 3;
michael@0:     _uccase_len[1] = hdr.size.len[1] * 3;
michael@0: 
michael@0:     _uccase_map = (unsigned long *)
michael@0:         malloc(_uccase_size * sizeof(unsigned long));
michael@0: 
michael@0:     /*
michael@0:      * Load the case mapping table.
michael@0:      */
michael@0:     fread((char *) _uccase_map, sizeof(unsigned long), _uccase_size, in);
michael@0: 
michael@0:     /*
michael@0:      * Do an endian swap if necessary.
michael@0:      */
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         for (i = 0; i < _uccase_size; i++)
michael@0:           _uccase_map[i] = endian_long(_uccase_map[i]);
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _uccase_unload(void)
michael@0: #else
michael@0: _uccase_unload()
michael@0: #endif
michael@0: {
michael@0:     if (_uccase_size == 0)
michael@0:       return;
michael@0: 
michael@0:     free((char *) _uccase_map);
michael@0:     _uccase_size = 0;
michael@0: }
michael@0: 
michael@0: static unsigned long
michael@0: #ifdef __STDC__
michael@0: _uccase_lookup(unsigned long code, long l, long r, int field)
michael@0: #else
michael@0: _uccase_lookup(code, l, r, field)
michael@0: unsigned long code;
michael@0: long l, r;
michael@0: int field;
michael@0: #endif
michael@0: {
michael@0:     long m;
michael@0: 
michael@0:     /*
michael@0:      * Do the binary search.
michael@0:      */
michael@0:     while (l <= r) {
michael@0:         /*
michael@0:          * Determine a "mid" point and adjust to make sure the mid point is at
michael@0:          * the beginning of a case mapping triple.
michael@0:          */
michael@0:         m = (l + r) >> 1;
michael@0:         m -= (m % 3);
michael@0:         if (code > _uccase_map[m])
michael@0:           l = m + 3;
michael@0:         else if (code < _uccase_map[m])
michael@0:           r = m - 3;
michael@0:         else if (code == _uccase_map[m])
michael@0:           return _uccase_map[m + field];
michael@0:     }
michael@0: 
michael@0:     return code;
michael@0: }
michael@0: 
michael@0: unsigned long
michael@0: #ifdef __STDC__
michael@0: uctoupper(unsigned long code)
michael@0: #else
michael@0: uctoupper(code)
michael@0: unsigned long code;
michael@0: #endif
michael@0: {
michael@0:     int field;
michael@0:     long l, r;
michael@0: 
michael@0:     if (ucisupper(code))
michael@0:       return code;
michael@0: 
michael@0:     if (ucislower(code)) {
michael@0:         /*
michael@0:          * The character is lower case.
michael@0:          */
michael@0:         field = 1;
michael@0:         l = _uccase_len[0];
michael@0:         r = (l + _uccase_len[1]) - 1;
michael@0:     } else {
michael@0:         /*
michael@0:          * The character is title case.
michael@0:          */
michael@0:         field = 2;
michael@0:         l = _uccase_len[0] + _uccase_len[1];
michael@0:         r = _uccase_size - 1;
michael@0:     }
michael@0:     return _uccase_lookup(code, l, r, field);
michael@0: }
michael@0: 
michael@0: unsigned long
michael@0: #ifdef __STDC__
michael@0: uctolower(unsigned long code)
michael@0: #else
michael@0: uctolower(code)
michael@0: unsigned long code;
michael@0: #endif
michael@0: {
michael@0:     int field;
michael@0:     long l, r;
michael@0: 
michael@0:     if (ucislower(code))
michael@0:       return code;
michael@0: 
michael@0:     if (ucisupper(code)) {
michael@0:         /*
michael@0:          * The character is upper case.
michael@0:          */
michael@0:         field = 1;
michael@0:         l = 0;
michael@0:         r = _uccase_len[0] - 1;
michael@0:     } else {
michael@0:         /*
michael@0:          * The character is title case.
michael@0:          */
michael@0:         field = 2;
michael@0:         l = _uccase_len[0] + _uccase_len[1];
michael@0:         r = _uccase_size - 1;
michael@0:     }
michael@0:     return _uccase_lookup(code, l, r, field);
michael@0: }
michael@0: 
michael@0: unsigned long
michael@0: #ifdef __STDC__
michael@0: uctotitle(unsigned long code)
michael@0: #else
michael@0: uctotitle(code)
michael@0: unsigned long code;
michael@0: #endif
michael@0: {
michael@0:     int field;
michael@0:     long l, r;
michael@0: 
michael@0:     if (ucistitle(code))
michael@0:       return code;
michael@0: 
michael@0:     /*
michael@0:      * The offset will always be the same for converting to title case.
michael@0:      */
michael@0:     field = 2;
michael@0: 
michael@0:     if (ucisupper(code)) {
michael@0:         /*
michael@0:          * The character is upper case.
michael@0:          */
michael@0:         l = 0;
michael@0:         r = _uccase_len[0] - 1;
michael@0:     } else {
michael@0:         /*
michael@0:          * The character is lower case.
michael@0:          */
michael@0:         l = _uccase_len[0];
michael@0:         r = (l + _uccase_len[1]) - 1;
michael@0:     }
michael@0:     return _uccase_lookup(code, l, r, field);
michael@0: }
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Support for decompositions.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: static unsigned long  _ucdcmp_size;
michael@0: static unsigned long *_ucdcmp_nodes;
michael@0: static unsigned long *_ucdcmp_decomp;
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _ucdcmp_load(char *paths, int reload)
michael@0: #else
michael@0: _ucdcmp_load(paths, reload)
michael@0: char *paths;
michael@0: int reload;
michael@0: #endif
michael@0: {
michael@0:     FILE *in;
michael@0:     unsigned long size, i;
michael@0:     _ucheader_t hdr;
michael@0: 
michael@0:     if (_ucdcmp_size > 0) {
michael@0:         if (!reload)
michael@0:           /*
michael@0:            * The decompositions have already been loaded.
michael@0:            */
michael@0:           return;
michael@0: 
michael@0:         free((char *) _ucdcmp_nodes);
michael@0:         _ucdcmp_size = 0;
michael@0:     }
michael@0: 
michael@0:     if ((in = _ucopenfile(paths, "decomp.dat", "rb")) == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Load the header.
michael@0:      */
michael@0:     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
michael@0: 
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         hdr.cnt = endian_short(hdr.cnt);
michael@0:         hdr.size.bytes = endian_long(hdr.size.bytes);
michael@0:     }
michael@0: 
michael@0:     _ucdcmp_size = hdr.cnt << 1;
michael@0:     _ucdcmp_nodes = (unsigned long *) malloc(hdr.size.bytes);
michael@0:     _ucdcmp_decomp = _ucdcmp_nodes + (_ucdcmp_size + 1);
michael@0: 
michael@0:     /*
michael@0:      * Read the decomposition data in.
michael@0:      */
michael@0:     size = hdr.size.bytes / sizeof(unsigned long);
michael@0:     fread((char *) _ucdcmp_nodes, sizeof(unsigned long), size, in);
michael@0: 
michael@0:     /*
michael@0:      * Do an endian swap if necessary.
michael@0:      */
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         for (i = 0; i < size; i++)
michael@0:           _ucdcmp_nodes[i] = endian_long(_ucdcmp_nodes[i]);
michael@0:     }        
michael@0: }
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _ucdcmp_unload(void)
michael@0: #else
michael@0: _ucdcmp_unload()
michael@0: #endif
michael@0: {
michael@0:     if (_ucdcmp_size == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Only need to free the offsets because the memory is allocated as a
michael@0:      * single block.
michael@0:      */
michael@0:     free((char *) _ucdcmp_nodes);
michael@0:     _ucdcmp_size = 0;
michael@0: }
michael@0: 
michael@0: int
michael@0: #ifdef __STDC__
michael@0: ucdecomp(unsigned long code, unsigned long *num, unsigned long **decomp)
michael@0: #else
michael@0: ucdecomp(code, num, decomp)
michael@0: unsigned long code, *num, **decomp;
michael@0: #endif
michael@0: {
michael@0:     long l, r, m;
michael@0: 
michael@0:     l = 0;
michael@0:     r = _ucdcmp_nodes[_ucdcmp_size] - 1;
michael@0: 
michael@0:     while (l <= r) {
michael@0:         /*
michael@0:          * Determine a "mid" point and adjust to make sure the mid point is at
michael@0:          * the beginning of a code+offset pair.
michael@0:          */
michael@0:         m = (l + r) >> 1;
michael@0:         m -= (m & 1);
michael@0:         if (code > _ucdcmp_nodes[m])
michael@0:           l = m + 2;
michael@0:         else if (code < _ucdcmp_nodes[m])
michael@0:           r = m - 2;
michael@0:         else if (code == _ucdcmp_nodes[m]) {
michael@0:             *num = _ucdcmp_nodes[m + 3] - _ucdcmp_nodes[m + 1];
michael@0:             *decomp = &_ucdcmp_decomp[_ucdcmp_nodes[m + 1]];
michael@0:             return 1;
michael@0:         }
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: int
michael@0: #ifdef __STDC__
michael@0: ucdecomp_hangul(unsigned long code, unsigned long *num, unsigned long decomp[])
michael@0: #else
michael@0: ucdecomp_hangul(code, num, decomp)
michael@0: unsigned long code, *num, decomp[];
michael@0: #endif
michael@0: {
michael@0:     if (!ucishangul(code))
michael@0:       return 0;
michael@0: 
michael@0:     code -= 0xac00;
michael@0:     decomp[0] = 0x1100 + (unsigned long) (code / 588);
michael@0:     decomp[1] = 0x1161 + (unsigned long) ((code % 588) / 28);
michael@0:     decomp[2] = 0x11a7 + (unsigned long) (code % 28);
michael@0:     *num = (decomp[2] != 0x11a7) ? 3 : 2;
michael@0: 
michael@0:     return 1;
michael@0: }
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Support for combining classes.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: static unsigned long  _uccmcl_size;
michael@0: static unsigned long *_uccmcl_nodes;
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _uccmcl_load(char *paths, int reload)
michael@0: #else
michael@0: _uccmcl_load(paths, reload)
michael@0: char *paths;
michael@0: int reload;
michael@0: #endif
michael@0: {
michael@0:     FILE *in;
michael@0:     unsigned long i;
michael@0:     _ucheader_t hdr;
michael@0: 
michael@0:     if (_uccmcl_size > 0) {
michael@0:         if (!reload)
michael@0:           /*
michael@0:            * The combining classes have already been loaded.
michael@0:            */
michael@0:           return;
michael@0: 
michael@0:         free((char *) _uccmcl_nodes);
michael@0:         _uccmcl_size = 0;
michael@0:     }
michael@0: 
michael@0:     if ((in = _ucopenfile(paths, "cmbcl.dat", "rb")) == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Load the header.
michael@0:      */
michael@0:     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
michael@0: 
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         hdr.cnt = endian_short(hdr.cnt);
michael@0:         hdr.size.bytes = endian_long(hdr.size.bytes);
michael@0:     }
michael@0: 
michael@0:     _uccmcl_size = hdr.cnt * 3;
michael@0:     _uccmcl_nodes = (unsigned long *) malloc(hdr.size.bytes);
michael@0: 
michael@0:     /*
michael@0:      * Read the combining classes in.
michael@0:      */
michael@0:     fread((char *) _uccmcl_nodes, sizeof(unsigned long), _uccmcl_size, in);
michael@0: 
michael@0:     /*
michael@0:      * Do an endian swap if necessary.
michael@0:      */
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         for (i = 0; i < _uccmcl_size; i++)
michael@0:           _uccmcl_nodes[i] = endian_long(_uccmcl_nodes[i]);
michael@0:     }        
michael@0: }
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _uccmcl_unload(void)
michael@0: #else
michael@0: _uccmcl_unload()
michael@0: #endif
michael@0: {
michael@0:     if (_uccmcl_size == 0)
michael@0:       return;
michael@0: 
michael@0:     free((char *) _uccmcl_nodes);
michael@0:     _uccmcl_size = 0;
michael@0: }
michael@0: 
michael@0: unsigned long
michael@0: #ifdef __STDC__
michael@0: uccombining_class(unsigned long code)
michael@0: #else
michael@0: uccombining_class(code)
michael@0: unsigned long code;
michael@0: #endif
michael@0: {
michael@0:     long l, r, m;
michael@0: 
michael@0:     l = 0;
michael@0:     r = _uccmcl_size - 1;
michael@0: 
michael@0:     while (l <= r) {
michael@0:         m = (l + r) >> 1;
michael@0:         m -= (m % 3);
michael@0:         if (code > _uccmcl_nodes[m + 1])
michael@0:           l = m + 3;
michael@0:         else if (code < _uccmcl_nodes[m])
michael@0:           r = m - 3;
michael@0:         else if (code >= _uccmcl_nodes[m] && code <= _uccmcl_nodes[m + 1])
michael@0:           return _uccmcl_nodes[m + 2];
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Support for numeric values.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: static unsigned long *_ucnum_nodes;
michael@0: static unsigned long _ucnum_size;
michael@0: static short *_ucnum_vals;
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _ucnumb_load(char *paths, int reload)
michael@0: #else
michael@0: _ucnumb_load(paths, reload)
michael@0: char *paths;
michael@0: int reload;
michael@0: #endif
michael@0: {
michael@0:     FILE *in;
michael@0:     unsigned long size, i;
michael@0:     _ucheader_t hdr;
michael@0: 
michael@0:     if (_ucnum_size > 0) {
michael@0:         if (!reload)
michael@0:           /*
michael@0:            * The numbers have already been loaded.
michael@0:            */
michael@0:           return;
michael@0: 
michael@0:         free((char *) _ucnum_nodes);
michael@0:         _ucnum_size = 0;
michael@0:     }
michael@0: 
michael@0:     if ((in = _ucopenfile(paths, "num.dat", "rb")) == 0)
michael@0:       return;
michael@0: 
michael@0:     /*
michael@0:      * Load the header.
michael@0:      */
michael@0:     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
michael@0: 
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         hdr.cnt = endian_short(hdr.cnt);
michael@0:         hdr.size.bytes = endian_long(hdr.size.bytes);
michael@0:     }
michael@0: 
michael@0:     _ucnum_size = hdr.cnt;
michael@0:     _ucnum_nodes = (unsigned long *) malloc(hdr.size.bytes);
michael@0:     _ucnum_vals = (short *) (_ucnum_nodes + _ucnum_size);
michael@0: 
michael@0:     /*
michael@0:      * Read the combining classes in.
michael@0:      */
michael@0:     fread((char *) _ucnum_nodes, sizeof(unsigned char), hdr.size.bytes, in);
michael@0: 
michael@0:     /*
michael@0:      * Do an endian swap if necessary.
michael@0:      */
michael@0:     if (hdr.bom == 0xfffe) {
michael@0:         for (i = 0; i < _ucnum_size; i++)
michael@0:           _ucnum_nodes[i] = endian_long(_ucnum_nodes[i]);
michael@0: 
michael@0:         /*
michael@0:          * Determine the number of values that have to be adjusted.
michael@0:          */
michael@0:         size = (hdr.size.bytes -
michael@0:                 (_ucnum_size * (sizeof(unsigned long) << 1))) /
michael@0:             sizeof(short);
michael@0: 
michael@0:         for (i = 0; i < size; i++)
michael@0:           _ucnum_vals[i] = endian_short(_ucnum_vals[i]);
michael@0:     }        
michael@0: }
michael@0: 
michael@0: static void
michael@0: #ifdef __STDC__
michael@0: _ucnumb_unload(void)
michael@0: #else
michael@0: _ucnumb_unload()
michael@0: #endif
michael@0: {
michael@0:     if (_ucnum_size == 0)
michael@0:       return;
michael@0: 
michael@0:     free((char *) _ucnum_nodes);
michael@0:     _ucnum_size = 0;
michael@0: }
michael@0: 
michael@0: int
michael@0: #ifdef __STDC__
michael@0: ucnumber_lookup(unsigned long code, struct ucnumber *num)
michael@0: #else
michael@0: ucnumber_lookup(code, num)
michael@0: unsigned long code;
michael@0: struct ucnumber *num;
michael@0: #endif
michael@0: {
michael@0:     long l, r, m;
michael@0:     short *vp;
michael@0: 
michael@0:     l = 0;
michael@0:     r = _ucnum_size - 1;
michael@0:     while (l <= r) {
michael@0:         /*
michael@0:          * Determine a "mid" point and adjust to make sure the mid point is at
michael@0:          * the beginning of a code+offset pair.
michael@0:          */
michael@0:         m = (l + r) >> 1;
michael@0:         m -= (m & 1);
michael@0:         if (code > _ucnum_nodes[m])
michael@0:           l = m + 2;
michael@0:         else if (code < _ucnum_nodes[m])
michael@0:           r = m - 2;
michael@0:         else {
michael@0:             vp = _ucnum_vals + _ucnum_nodes[m + 1];
michael@0:             num->numerator = (int) *vp++;
michael@0:             num->denominator = (int) *vp;
michael@0:             return 1;
michael@0:         }
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: int
michael@0: #ifdef __STDC__
michael@0: ucdigit_lookup(unsigned long code, int *digit)
michael@0: #else
michael@0: ucdigit_lookup(code, digit)
michael@0: unsigned long code;
michael@0: int *digit;
michael@0: #endif
michael@0: {
michael@0:     long l, r, m;
michael@0:     short *vp;
michael@0: 
michael@0:     l = 0;
michael@0:     r = _ucnum_size - 1;
michael@0:     while (l <= r) {
michael@0:         /*
michael@0:          * Determine a "mid" point and adjust to make sure the mid point is at
michael@0:          * the beginning of a code+offset pair.
michael@0:          */
michael@0:         m = (l + r) >> 1;
michael@0:         m -= (m & 1);
michael@0:         if (code > _ucnum_nodes[m])
michael@0:           l = m + 2;
michael@0:         else if (code < _ucnum_nodes[m])
michael@0:           r = m - 2;
michael@0:         else {
michael@0:             vp = _ucnum_vals + _ucnum_nodes[m + 1];
michael@0:             if (*vp == *(vp + 1)) {
michael@0:               *digit = *vp;
michael@0:               return 1;
michael@0:             }
michael@0:             return 0;
michael@0:         }
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: struct ucnumber
michael@0: #ifdef __STDC__
michael@0: ucgetnumber(unsigned long code)
michael@0: #else
michael@0: ucgetnumber(code)
michael@0: unsigned long code;
michael@0: #endif
michael@0: {
michael@0:     struct ucnumber num;
michael@0: 
michael@0:     /*
michael@0:      * Initialize with some arbitrary value, because the caller simply cannot
michael@0:      * tell for sure if the code is a number without calling the ucisnumber()
michael@0:      * macro before calling this function.
michael@0:      */
michael@0:     num.numerator = num.denominator = -111;
michael@0: 
michael@0:     (void) ucnumber_lookup(code, &num);
michael@0: 
michael@0:     return num;
michael@0: }
michael@0: 
michael@0: int
michael@0: #ifdef __STDC__
michael@0: ucgetdigit(unsigned long code)
michael@0: #else
michael@0: ucgetdigit(code)
michael@0: unsigned long code;
michael@0: #endif
michael@0: {
michael@0:     int dig;
michael@0: 
michael@0:     /*
michael@0:      * Initialize with some arbitrary value, because the caller simply cannot
michael@0:      * tell for sure if the code is a number without calling the ucisdigit()
michael@0:      * macro before calling this function.
michael@0:      */
michael@0:     dig = -111;
michael@0: 
michael@0:     (void) ucdigit_lookup(code, &dig);
michael@0: 
michael@0:     return dig;
michael@0: }
michael@0: 
michael@0: /**************************************************************************
michael@0:  *
michael@0:  * Setup and cleanup routines.
michael@0:  *
michael@0:  **************************************************************************/
michael@0: 
michael@0: void
michael@0: #ifdef __STDC__
michael@0: ucdata_load(char *paths, int masks)
michael@0: #else
michael@0: ucdata_load(paths, masks)
michael@0: char *paths;
michael@0: int masks;
michael@0: #endif
michael@0: {
michael@0:     if (masks & UCDATA_CTYPE)
michael@0:       _ucprop_load(paths, 0);
michael@0:     if (masks & UCDATA_CASE)
michael@0:       _uccase_load(paths, 0);
michael@0:     if (masks & UCDATA_DECOMP)
michael@0:       _ucdcmp_load(paths, 0);
michael@0:     if (masks & UCDATA_CMBCL)
michael@0:       _uccmcl_load(paths, 0);
michael@0:     if (masks & UCDATA_NUM)
michael@0:       _ucnumb_load(paths, 0);
michael@0: }
michael@0: 
michael@0: void
michael@0: #ifdef __STDC__
michael@0: ucdata_unload(int masks)
michael@0: #else
michael@0: ucdata_unload(masks)
michael@0: int masks;
michael@0: #endif
michael@0: {
michael@0:     if (masks & UCDATA_CTYPE)
michael@0:       _ucprop_unload();
michael@0:     if (masks & UCDATA_CASE)
michael@0:       _uccase_unload();
michael@0:     if (masks & UCDATA_DECOMP)
michael@0:       _ucdcmp_unload();
michael@0:     if (masks & UCDATA_CMBCL)
michael@0:       _uccmcl_unload();
michael@0:     if (masks & UCDATA_NUM)
michael@0:       _ucnumb_unload();
michael@0: }
michael@0: 
michael@0: void
michael@0: #ifdef __STDC__
michael@0: ucdata_reload(char *paths, int masks)
michael@0: #else
michael@0: ucdata_reload(paths, masks)
michael@0: char *paths;
michael@0: int masks;
michael@0: #endif
michael@0: {
michael@0:     if (masks & UCDATA_CTYPE)
michael@0:       _ucprop_load(paths, 1);
michael@0:     if (masks & UCDATA_CASE)
michael@0:       _uccase_load(paths, 1);
michael@0:     if (masks & UCDATA_DECOMP)
michael@0:       _ucdcmp_load(paths, 1);
michael@0:     if (masks & UCDATA_CMBCL)
michael@0:       _uccmcl_load(paths, 1);
michael@0:     if (masks & UCDATA_NUM)
michael@0:       _ucnumb_load(paths, 1);
michael@0: }
michael@0: 
michael@0: #ifdef TEST
michael@0: 
michael@0: void
michael@0: #ifdef __STDC__
michael@0: main(void)
michael@0: #else
michael@0: main()
michael@0: #endif
michael@0: {
michael@0:     int dig;
michael@0:     unsigned long i, lo, *dec;
michael@0:     struct ucnumber num;
michael@0: 
michael@0:     ucdata_setup(".");
michael@0: 
michael@0:     if (ucisweak(0x30))
michael@0:       printf("WEAK\n");
michael@0:     else
michael@0:       printf("NOT WEAK\n");
michael@0: 
michael@0:     printf("LOWER 0x%04lX\n", uctolower(0xff3a));
michael@0:     printf("UPPER 0x%04lX\n", uctoupper(0xff5a));
michael@0: 
michael@0:     if (ucisalpha(0x1d5))
michael@0:       printf("ALPHA\n");
michael@0:     else
michael@0:       printf("NOT ALPHA\n");
michael@0: 
michael@0:     if (ucisupper(0x1d5)) {
michael@0:         printf("UPPER\n");
michael@0:         lo = uctolower(0x1d5);
michael@0:         printf("0x%04lx\n", lo);
michael@0:         lo = uctotitle(0x1d5);
michael@0:         printf("0x%04lx\n", lo);
michael@0:     } else
michael@0:       printf("NOT UPPER\n");
michael@0: 
michael@0:     if (ucistitle(0x1d5))
michael@0:       printf("TITLE\n");
michael@0:     else
michael@0:       printf("NOT TITLE\n");
michael@0: 
michael@0:     if (uciscomposite(0x1d5))
michael@0:       printf("COMPOSITE\n");
michael@0:     else
michael@0:       printf("NOT COMPOSITE\n");
michael@0: 
michael@0:     if (ucdecomp(0x1d5, &lo, &dec)) {
michael@0:         for (i = 0; i < lo; i++)
michael@0:           printf("0x%04lx ", dec[i]);
michael@0:         putchar('\n');
michael@0:     }
michael@0: 
michael@0:     if ((lo = uccombining_class(0x41)) != 0)
michael@0:       printf("0x41 CCL %ld\n", lo);
michael@0: 
michael@0:     if (ucisxdigit(0xfeff))
michael@0:       printf("0xFEFF HEX DIGIT\n");
michael@0:     else
michael@0:       printf("0xFEFF NOT HEX DIGIT\n");
michael@0: 
michael@0:     if (ucisdefined(0x10000))
michael@0:       printf("0x10000 DEFINED\n");
michael@0:     else
michael@0:       printf("0x10000 NOT DEFINED\n");
michael@0: 
michael@0:     if (ucnumber_lookup(0x30, &num)) {
michael@0:         if (num.numerator != num.denominator)
michael@0:           printf("UCNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
michael@0:         else
michael@0:           printf("UCNUMBER: 0x30 = %d\n", num.numerator);
michael@0:     } else
michael@0:       printf("UCNUMBER: 0x30 NOT A NUMBER\n");
michael@0: 
michael@0:     if (ucnumber_lookup(0xbc, &num)) {
michael@0:         if (num.numerator != num.denominator)
michael@0:           printf("UCNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
michael@0:         else
michael@0:           printf("UCNUMBER: 0xbc = %d\n", num.numerator);
michael@0:     } else
michael@0:       printf("UCNUMBER: 0xbc NOT A NUMBER\n");
michael@0: 
michael@0: 
michael@0:     if (ucnumber_lookup(0xff19, &num)) {
michael@0:         if (num.numerator != num.denominator)
michael@0:           printf("UCNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
michael@0:         else
michael@0:           printf("UCNUMBER: 0xff19 = %d\n", num.numerator);
michael@0:     } else
michael@0:       printf("UCNUMBER: 0xff19 NOT A NUMBER\n");
michael@0: 
michael@0:     if (ucnumber_lookup(0x4e00, &num)) {
michael@0:         if (num.numerator != num.denominator)
michael@0:           printf("UCNUMBER: 0x4e00 = %d/%d\n", num.numerator, num.denominator);
michael@0:         else
michael@0:           printf("UCNUMBER: 0x4e00 = %d\n", num.numerator);
michael@0:     } else
michael@0:       printf("UCNUMBER: 0x4e00 NOT A NUMBER\n");
michael@0: 
michael@0:     if (ucdigit_lookup(0x06f9, &dig))
michael@0:       printf("UCDIGIT: 0x6f9 = %d\n", dig);
michael@0:     else
michael@0:       printf("UCDIGIT: 0x6f9 NOT A NUMBER\n");
michael@0: 
michael@0:     dig = ucgetdigit(0x0969);
michael@0:     printf("UCGETDIGIT: 0x969 = %d\n", dig);
michael@0: 
michael@0:     num = ucgetnumber(0x30);
michael@0:     if (num.numerator != num.denominator)
michael@0:       printf("UCGETNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
michael@0:     else
michael@0:       printf("UCGETNUMBER: 0x30 = %d\n", num.numerator);
michael@0: 
michael@0:     num = ucgetnumber(0xbc);
michael@0:     if (num.numerator != num.denominator)
michael@0:       printf("UCGETNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
michael@0:     else
michael@0:       printf("UCGETNUMBER: 0xbc = %d\n", num.numerator);
michael@0: 
michael@0:     num = ucgetnumber(0xff19);
michael@0:     if (num.numerator != num.denominator)
michael@0:       printf("UCGETNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
michael@0:     else
michael@0:       printf("UCGETNUMBER: 0xff19 = %d\n", num.numerator);
michael@0: 
michael@0:     ucdata_cleanup();
michael@0:     exit(0);
michael@0: }
michael@0: 
michael@0: #endif /* TEST */