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 #include "scroll_engine.h"
  30
  31 //#define _HEB_BUFFER_LENGTH (MAX_PATH + LCD_WIDTH/2 + 3 + 2 + 2) * 2
  32 #define _HEB_BLOCK_TYPE_ENG 1
  33 #define _HEB_BLOCK_TYPE_HEB 0
  34 #define _HEB_ORIENTATION_LTR 1
  35 #define _HEB_ORIENTATION_RTL 0
  36
  37 #define ischar(c) ((c > 0x0589 && c < 0x0700) || (c >= 0xfb50 && c <= 0xfefc) ? 1 : 0)
  38 #define _isblank(c) ((c==' ' || c=='\t') ? 1 : 0)
  39 #define _isnewline(c) ((c=='\n' || c=='\r') ? 1 : 0)
  40 #define XOR(a,b) ((a||b) && !(a&&b))
  41
  42 const arab_t * arab_lookup(unsigned short uchar)
  43 {
  44     if (uchar >= 0x621 && uchar <= 0x63a)
  45         return &(jointable[uchar - 0x621]);
  46     if (uchar >= 0x640 && uchar <= 0x64a)
  47         return &(jointable[uchar - 0x621 - 5]);
  48     if (uchar >= 0x671 && uchar <= 0x6d5)
  49         return &(jointable[uchar - 0x621 - 5 - 38]);
  50     if (uchar == 0x200D) /* Support for the zero-width joiner */
  51         return &zwj;
  52     return 0;
  53 }
  54
  55 void arabjoin(unsigned short * stringprt, int length){
  56
  57     bool connected = false;
  58     unsigned short * writeprt = stringprt;
  59
  60     const arab_t * prev = 0;
  61     const arab_t * cur;
  62     const arab_t * ligature = 0;
  63     short uchar;
  64
  65     int i;
  66     for (i = 0; i <= length; i++) {
  67         cur = arab_lookup(uchar = *stringprt++);
  68
  69         /* Skip non-arabic chars */
  70         if (cur == 0) {
  71             if (prev) {
  72                 /* Finish the last char */
  73                 if (connected) {
  74                     *writeprt++ = prev->final;
  75                     connected = false;
  76                 } else
  77                     *writeprt++ = prev->isolated;
  78                 prev = 0;
  79                 *writeprt++ = uchar;
  80             } else {
  81                 *writeprt++ = uchar;
  82             }
  83             continue;
  84         }
  85
  86         /* nothing to do for arabic char if the previous was non-arabic */
  87         if (prev == 0) {
  88             prev = cur;
  89             continue;
  90         }
  91
  92         /* if it's LAM, check for LAM+ALEPH ligatures */
  93         if (prev->isolated == 0xfedd) {
  94             switch (cur->isolated) {
  95                 case 0xfe8d:
  96                     ligature = &(lamaleph[0]);
  97                     break;
  98                 case 0xfe87:
  99                     ligature = &(lamaleph[1]);
 100                     break;
 101                 case 0xfe83:
 102                     ligature = &(lamaleph[2]);
 103                     break;
 104                 case 0xfe81:
 105                     ligature = &(lamaleph[3]);
 106             }
 107         }
 108
 109         if (ligature) { /* replace the 2 glyphs by their ligature */
 110             prev = ligature;
 111             ligature = 0;
 112         } else {
 113             if (connected) { /* previous char has something connected to it */
 114                 if (prev->medial && cur->final) /* Can we connect to it? */
 115                     *writeprt++ = prev->medial;
 116                 else {
 117                     *writeprt++ = prev->final;
 118                     connected = false;
 119                 }
 120             } else {
 121                 if (prev->initial && cur->final) { /* Can we connect to it? */
 122                     *writeprt++ = prev->initial;
 123                     connected = true;
 124                 } else
 125                     *writeprt++ = prev->isolated;
 126             }
 127             prev = cur;
 128         }
 129     }
 130 }
 131
 132 unsigned short *bidi_l2v(const unsigned char *str, int orientation)
 133 {
 134     int length = utf8length(str);
 135     static unsigned short  utf16_buf[SCROLL_LINE_SIZE];
 136     static unsigned short  bidi_buf[SCROLL_LINE_SIZE];
 137     unsigned short *heb_str, *target, *tmp; // *broken_str
 138     int block_start, block_end, block_type, block_length, i;
 139     //long max_chars=0;
 140     //int begin, end, char_count, orig_begin;
 141
 142     //tmp = str;
 143     target = tmp = utf16_buf;
 144     while (*str)
 145         str = utf8decode(str, target++);
 146     *target = 0;
 147
 148     if (target == utf16_buf) /* empty string */
 149         return target;
 150
 151     /* properly join any arabic chars */
 152     arabjoin(utf16_buf, length);
 153
 154     block_start=block_end=block_length=0;
 155
 156     heb_str = bidi_buf;
 157     if (orientation) {
 158         target = heb_str;
 159     } else {
 160         target = heb_str + length;
 161         *target = 0;
 162         target--;
 163     }
 164
 165     if (ischar(*tmp))
 166         block_type = _HEB_BLOCK_TYPE_HEB;
 167     else
 168         block_type = _HEB_BLOCK_TYPE_ENG;
 169
 170     do {
 171         while((XOR(ischar(*(tmp+1)),block_type)
 172                || _isblank(*(tmp+1)) || ispunct((int)*(tmp+1))
 173                || *(tmp+1)=='\n')
 174               && block_end < length-1) {
 175                 tmp++;
 176                 block_end++;
 177                 block_length++;
 178         }
 179
 180         if (block_type != orientation) {
 181             while ((_isblank(*tmp) || ispunct((int)*tmp))
 182                    && *tmp!='/' && *tmp!='-' && block_end>block_start) {
 183                 tmp--;
 184                 block_end--;
 185             }
 186         }
 187
 188         for (i=block_start; i<=block_end; i++) {
 189             *target = (block_type == orientation) ? *(utf16_buf+i) : *(utf16_buf+block_end-i+block_start);
 190             if (block_type!=orientation) {
 191                 switch (*target) {
 192                 case '(':
 193                     *target = ')';
 194                     break;
 195                 case ')':
 196                     *target = '(';
 197                     break;
 198                 default:
 199                     break;
 200                 }
 201             }
 202             target += orientation ? 1 : -1;
 203         }
 204         block_type = !block_type;
 205         block_start=block_end+1;
 206     } while(block_end<length-1);
 207
 208     *target = 0;
 209
 210 #if 0 /* Is this code really necessary? */
 211     broken_str = utf16_buf;
 212     begin=end=length-1;
 213     target = broken_str;
 214
 215     while (1) {
 216         char_count=0;
 217         while ((!max_chars || char_count<max_chars) && begin>0) {
 218             char_count++;
 219             begin--;
 220             if (begin<=0 || _isnewline(heb_str[begin])) {
 221                 while(begin>0 && _isnewline(heb_str[begin-1])) {
 222                     begin--;
 223                     char_count++;
 224                 }
 225                 break;
 226             }
 227         }
 228         if (char_count==max_chars) { /* try to avoid breaking words */
 229             int new_char_count = char_count;
 230             int new_begin = begin;
 231
 232             while (new_char_count>0) {
 233                 if (_isblank(heb_str[new_begin]) ||
 234                     _isnewline(heb_str[new_begin])) {
 235                     break;
 236                 }
 237                 new_begin++;
 238                 new_char_count--;
 239             }
 240             if (new_char_count>0) {
 241                 char_count=new_char_count;
 242                 begin=new_begin;
 243             }
 244         }
 245         orig_begin=begin;
 246
 247         /* if (_isblank(heb_str[begin])) {
 248             heb_str[begin]='\n';
 249         } */
 250
 251         /* skip leading newlines */
 252         while (begin<=end && _isnewline(heb_str[begin])) {
 253             begin++;
 254         }
 255
 256         /* copy content */
 257         for (i=begin; i<=end; i++) {
 258             *target = heb_str[i];
 259             target++;
 260         }
 261
 262         for (i=orig_begin; i<=end && _isnewline(heb_str[i]); i++) {
 263             *target = heb_str[i];
 264             target++;
 265         }
 266         begin=orig_begin;
 267
 268         if (begin<=0) {
 269             *target = 0;
 270             break;
 271         }
 272         begin--;
 273         end=begin;
 274     }
 275     return broken_str;
 276 #endif
 277     return heb_str;
 278 }
 279