firmware/bidi.c

   1 /***************************************************************************
   2  *             __________               __   ___.
   3  *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
   4  *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
   5  *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
   6  *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
   7  *                     \/            \/     \/    \/            \/
   8  * $Id$
   9  *
  10  * Copyright (C) 2005 by Gadi Cohen
  11  *
  12  * Largely based on php_hebrev by Zeev Suraski <zeev@php.net>
  13  * Heavily modified by Gadi Cohen aka Kinslayer <dragon@wastelands.net>
  14  *
  15  * All files in this archive are subject to the GNU General Public License.
  16  * See the file COPYING in the source tree root for full license agreement.
  17  *
  18  * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
  19  * KIND, either express or implied.
  20  *
  21  ****************************************************************************/
  22 #include <stdio.h>
  23 #include <string.h>
  24 #include <ctype.h>
  25 #include "file.h"
  26 #include "lcd.h"
  27 #include "rbunicode.h"
  28 #include "arabjoin.h"
  29
  30 //#define _HEB_BUFFER_LENGTH (MAX_PATH + LCD_WIDTH/2 + 3 + 2 + 2) * 2
  31 #define _HEB_BLOCK_TYPE_ENG 1
  32 #define _HEB_BLOCK_TYPE_HEB 0
  33 #define _HEB_ORIENTATION_LTR 1
  34 #define _HEB_ORIENTATION_RTL 0
  35
  36 #define ischar(c) ((c > 0x0589 && c < 0x0700) || (c >= 0xfb50 && c <= 0xfefc) ? 1 : 0)
  37 #define _isblank(c) ((c==' ' || c=='\t') ? 1 : 0)
  38 #define _isnewline(c) ((c=='\n' || c=='\r') ? 1 : 0)
  39 #define XOR(a,b) ((a||b) && !(a&&b))
  40
  41 const arab_t * arab_lookup(unsigned short uchar)
  42 {
  43     if (uchar >= 0x621 && uchar <= 0x63a)
  44         return &(jointable[uchar - 0x621]);
  45     if (uchar >= 0x640 && uchar <= 0x64a)
  46         return &(jointable[uchar - 0x621 - 5]);
  47     if (uchar >= 0x671 && uchar <= 0x6d5)
  48         return &(jointable[uchar - 0x621 - 5 - 38]);
  49     if (uchar == 0x200D) /* Support for the zero-width joiner */
  50         return &zwj;
  51     return 0;
  52 }
  53
  54 void arabjoin(unsigned short * stringprt, int length){
  55
  56     bool connected = false;
  57     unsigned short * writeprt = stringprt;
  58
  59     const arab_t * prev = 0;
  60     const arab_t * cur;
  61     const arab_t * ligature = 0;
  62     short uchar;
  63
  64     int i;
  65     for (i = 0; i <= length; i++) {
  66         cur = arab_lookup(uchar = *stringprt++);
  67
  68         /* Skip non-arabic chars */
  69         if (cur == 0) {
  70             if (prev) {
  71                 /* Finish the last char */
  72                 if (connected) {
  73                     *writeprt++ = prev->final;
  74                     connected = false;
  75                 } else
  76                     *writeprt++ = prev->isolated;
  77                 prev = 0;
  78                 *writeprt++ = uchar;
  79             } else {
  80                 *writeprt++ = uchar;
  81             }
  82             continue;
  83         }
  84
  85         /* nothing to do for arabic char if the previous was non-arabic */
  86         if (prev == 0) {
  87             prev = cur;
  88             continue;
  89         }
  90
  91         /* if it's LAM, check for LAM+ALEPH ligatures */
  92         if (prev->isolated == 0xfedd) {
  93             switch (cur->isolated) {
  94                 case 0xfe8d:
  95                     ligature = &(lamaleph[0]);
  96                     break;
  97                 case 0xfe87:
  98                     ligature = &(lamaleph[1]);
  99                     break;
 100                 case 0xfe83:
 101                     ligature = &(lamaleph[2]);
 102                     break;
 103                 case 0xfe81:
 104                     ligature = &(lamaleph[3]);
 105             }
 106         }
 107
 108         if (ligature) { /* replace the 2 glyphs by their ligature */
 109             prev = ligature;
 110             ligature = 0;
 111         } else {
 112             if (connected) { /* previous char has something connected to it */
 113                 if (prev->medial && cur->final) /* Can we connect to it? */
 114                     *writeprt++ = prev->medial;
 115                 else {
 116                     *writeprt++ = prev->final;
 117                     connected = false;
 118                 }
 119             } else {
 120                 if (prev->initial && cur->final) { /* Can we connect to it? */
 121                     *writeprt++ = prev->initial;
 122                     connected = true;
 123                 } else
 124                     *writeprt++ = prev->isolated;
 125             }
 126             prev = cur;
 127         }
 128     }
 129 }
 130
 131 unsigned short *bidi_l2v(const unsigned char *str, int orientation)
 132 {
 133     int length = utf8length(str);
 134     static unsigned short  utf16_buf[SCROLL_LINE_SIZE];
 135     static unsigned short  bidi_buf[SCROLL_LINE_SIZE];
 136     unsigned short *heb_str, *target, *tmp; // *broken_str
 137     int block_start, block_end, block_type, block_length, i;
 138     //long max_chars=0;
 139     //int begin, end, char_count, orig_begin;
 140
 141     //tmp = str;
 142     target = tmp = utf16_buf;
 143     while (*str)
 144         str = utf8decode(str, target++);
 145     *target = 0;
 146
 147     if (target == utf16_buf) /* empty string */
 148         return target;
 149
 150     /* properly join any arabic chars */
 151     arabjoin(utf16_buf, length);
 152
 153     block_start=block_end=block_length=0;
 154
 155     heb_str = bidi_buf;
 156     if (orientation) {
 157         target = heb_str;
 158     } else {
 159         target = heb_str + length;
 160         *target = 0;
 161         target--;
 162     }
 163
 164     if (ischar(*tmp))
 165         block_type = _HEB_BLOCK_TYPE_HEB;
 166     else
 167         block_type = _HEB_BLOCK_TYPE_ENG;
 168
 169     do {
 170         while((XOR(ischar(*(tmp+1)),block_type)
 171                || _isblank(*(tmp+1)) || ispunct((int)*(tmp+1))
 172                || *(tmp+1)=='\n')
 173               && block_end < length-1) {
 174                 tmp++;
 175                 block_end++;
 176                 block_length++;
 177         }
 178
 179         if (block_type != orientation) {
 180             while ((_isblank(*tmp) || ispunct((int)*tmp))
 181                    && *tmp!='/' && *tmp!='-' && block_end>block_start) {
 182                 tmp--;
 183                 block_end--;
 184             }
 185         }
 186
 187         for (i=block_start; i<=block_end; i++) {
 188             *target = (block_type == orientation) ? *(utf16_buf+i) : *(utf16_buf+block_end-i+block_start);
 189             if (block_type!=orientation) {
 190                 switch (*target) {
 191                 case '(':
 192                     *target = ')';
 193                     break;
 194                 case ')':
 195                     *target = '(';
 196                     break;
 197                 default:
 198                     break;
 199                 }
 200             }
 201             target += orientation ? 1 : -1;
 202         }
 203         block_type = !block_type;
 204         block_start=block_end+1;
 205     } while(block_end<length-1);
 206
 207     *target = 0;
 208
 209 #if 0 /* Is this code really necessary? */
 210     broken_str = utf16_buf;
 211     begin=end=length-1;
 212     target = broken_str;
 213
 214     while (1) {
 215         char_count=0;
 216         while ((!max_chars || char_count<max_chars) && begin>0) {
 217             char_count++;
 218             begin--;
 219             if (begin<=0 || _isnewline(heb_str[begin])) {
 220                 while(begin>0 && _isnewline(heb_str[begin-1])) {
 221                     begin--;
 222                     char_count++;
 223                 }
 224                 break;
 225             }
 226         }
 227         if (char_count==max_chars) { /* try to avoid breaking words */
 228             int new_char_count = char_count;
 229             int new_begin = begin;
 230
 231             while (new_char_count>0) {
 232                 if (_isblank(heb_str[new_begin]) ||
 233                     _isnewline(heb_str[new_begin])) {
 234                     break;
 235                 }
 236                 new_begin++;
 237                 new_char_count--;
 238             }
 239             if (new_char_count>0) {
 240                 char_count=new_char_count;
 241                 begin=new_begin;
 242             }
 243         }
 244         orig_begin=begin;
 245
 246         /* if (_isblank(heb_str[begin])) {
 247             heb_str[begin]='\n';
 248         } */
 249
 250         /* skip leading newlines */
 251         while (begin<=end && _isnewline(heb_str[begin])) {
 252             begin++;
 253         }
 254
 255         /* copy content */
 256         for (i=begin; i<=end; i++) {
 257             *target = heb_str[i];
 258             target++;
 259         }
 260
 261         for (i=orig_begin; i<=end && _isnewline(heb_str[i]); i++) {
 262             *target = heb_str[i];
 263             target++;
 264         }
 265         begin=orig_begin;
 266
 267         if (begin<=0) {
 268             *target = 0;
 269             break;
 270         }
 271         begin--;
 272         end=begin;
 273     }
 274     return broken_str;
 275 #endif
 276     return heb_str;
 277 }
 278