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 #include "bidi.h"
  31
  32 /* #define _HEB_BUFFER_LENGTH (MAX_PATH + LCD_WIDTH/2 + 3 + 2 + 2) * 2 */
  33 #define _HEB_BLOCK_TYPE_ENG 1
  34 #define _HEB_BLOCK_TYPE_HEB 0
  35 #define _HEB_ORIENTATION_LTR 1
  36 #define _HEB_ORIENTATION_RTL 0
  37
  38 #define ischar(c) ((c > 0x0589 && c < 0x0700) || \
  39                    (c >= 0xfb50 && c <= 0xfefc) ? 1 : 0)
  40 #define _isblank(c) ((c==' ' || c=='\t') ? 1 : 0)
  41 #define _isnewline(c) ((c=='\n' || c=='\r') ? 1 : 0)
  42 #define XOR(a,b) ((a||b) && !(a&&b))
  43
  44 static const arab_t * arab_lookup(unsigned short uchar)
  45 {
  46     if (uchar >= 0x621 && uchar <= 0x63a)
  47         return &(jointable[uchar - 0x621]);
  48     if (uchar >= 0x640 && uchar <= 0x64a)
  49         return &(jointable[uchar - 0x621 - 5]);
  50     if (uchar >= 0x671 && uchar <= 0x6d5)
  51         return &(jointable[uchar - 0x621 - 5 - 38]);
  52     if (uchar == 0x200D) /* Support for the zero-width joiner */
  53         return &zwj;
  54     return 0;
  55 }
  56
  57 static void arabjoin(unsigned short * stringprt, int length)
  58 {
  59     bool connected = false;
  60     unsigned short * writeprt = stringprt;
  61
  62     const arab_t * prev = 0;
  63     const arab_t * cur;
  64     const arab_t * ligature = 0;
  65     short uchar;
  66
  67     int i;
  68     for (i = 0; i <= length; i++) {
  69         cur = arab_lookup(uchar = *stringprt++);
  70
  71         /* Skip non-arabic chars */
  72         if (cur == 0) {
  73             if (prev) {
  74                 /* Finish the last char */
  75                 if (connected) {
  76                     *writeprt++ = prev->final;
  77                     connected = false;
  78                 } else
  79                     *writeprt++ = prev->isolated;
  80                 prev = 0;
  81                 *writeprt++ = uchar;
  82             } else {
  83                 *writeprt++ = uchar;
  84             }
  85             continue;
  86         }
  87
  88         /* nothing to do for arabic char if the previous was non-arabic */
  89         if (prev == 0) {
  90             prev = cur;
  91             continue;
  92         }
  93
  94         /* if it's LAM, check for LAM+ALEPH ligatures */
  95         if (prev->isolated == 0xfedd) {
  96             switch (cur->isolated) {
  97                 case 0xfe8d:
  98                     ligature = &(lamaleph[0]);
  99                     break;
 100                 case 0xfe87:
 101                     ligature = &(lamaleph[1]);
 102                     break;
 103                 case 0xfe83:
 104                     ligature = &(lamaleph[2]);
 105                     break;
 106                 case 0xfe81:
 107                     ligature = &(lamaleph[3]);
 108             }
 109         }
 110
 111         if (ligature) { /* replace the 2 glyphs by their ligature */
 112             prev = ligature;
 113             ligature = 0;
 114         } else {
 115             if (connected) { /* previous char has something connected to it */
 116                 if (prev->medial && cur->final) /* Can we connect to it? */
 117                     *writeprt++ = prev->medial;
 118                 else {
 119                     *writeprt++ = prev->final;
 120                     connected = false;
 121                 }
 122             } else {
 123                 if (prev->initial && cur->final) { /* Can we connect to it? */
 124                     *writeprt++ = prev->initial;
 125                     connected = true;
 126                 } else
 127                     *writeprt++ = prev->isolated;
 128             }
 129             prev = cur;
 130         }
 131     }
 132 }
 133
 134 unsigned short *bidi_l2v(const unsigned char *str, int orientation)
 135 {
 136     int length = utf8length(str);
 137     static unsigned short  utf16_buf[SCROLL_LINE_SIZE];
 138     static unsigned short  bidi_buf[SCROLL_LINE_SIZE];
 139     unsigned short *heb_str, *target, *tmp; /* *broken_str */
 140     int block_start, block_end, block_type, block_length, i;
 141     /*
 142     long max_chars=0;
 143     int begin, end, char_count, orig_begin;
 144
 145     tmp = str;
 146     */
 147     target = tmp = utf16_buf;
 148     while (*str)
 149         str = utf8decode(str, target++);
 150     *target = 0;
 151
 152     if (target == utf16_buf) /* empty string */
 153         return target;
 154
 155     /* properly join any arabic chars */
 156     arabjoin(utf16_buf, length);
 157
 158     block_start=block_end=block_length=0;
 159
 160     heb_str = bidi_buf;
 161     if (orientation) {
 162         target = heb_str;
 163     } else {
 164         target = heb_str + length;
 165         *target = 0;
 166         target--;
 167     }
 168
 169     if (ischar(*tmp))
 170         block_type = _HEB_BLOCK_TYPE_HEB;
 171     else
 172         block_type = _HEB_BLOCK_TYPE_ENG;
 173
 174     do {
 175         while((XOR(ischar(*(tmp+1)),block_type)
 176                || _isblank(*(tmp+1)) || ispunct((int)*(tmp+1))
 177                || *(tmp+1)=='\n')
 178               && block_end < length-1) {
 179                 tmp++;
 180                 block_end++;
 181                 block_length++;
 182         }
 183
 184         if (block_type != orientation) {
 185             while ((_isblank(*tmp) || ispunct((int)*tmp))
 186                    && *tmp!='/' && *tmp!='-' && block_end>block_start) {
 187                 tmp--;
 188                 block_end--;
 189             }
 190         }
 191
 192         for (i=block_start; i<=block_end; i++) {
 193             *target = (block_type == orientation) ?
 194                       *(utf16_buf+i) : *(utf16_buf+block_end-i+block_start);
 195             if (block_type!=orientation) {
 196                 switch (*target) {
 197                 case '(':
 198                     *target = ')';
 199                     break;
 200                 case ')':
 201                     *target = '(';
 202                     break;
 203                 default:
 204                     break;
 205                 }
 206             }
 207             target += orientation ? 1 : -1;
 208         }
 209         block_type = !block_type;
 210         block_start=block_end+1;
 211     } while(block_end<length-1);
 212
 213     *target = 0;
 214
 215 #if 0 /* Is this code really necessary? */
 216     broken_str = utf16_buf;
 217     begin=end=length-1;
 218     target = broken_str;
 219
 220     while (1) {
 221         char_count=0;
 222         while ((!max_chars || char_count<max_chars) && begin>0) {
 223             char_count++;
 224             begin--;
 225             if (begin<=0 || _isnewline(heb_str[begin])) {
 226                 while(begin>0 && _isnewline(heb_str[begin-1])) {
 227                     begin--;
 228                     char_count++;
 229                 }
 230                 break;
 231             }
 232         }
 233         if (char_count==max_chars) { /* try to avoid breaking words */
 234             int new_char_count = char_count;
 235             int new_begin = begin;
 236
 237             while (new_char_count>0) {
 238                 if (_isblank(heb_str[new_begin]) ||
 239                     _isnewline(heb_str[new_begin])) {
 240                     break;
 241                 }
 242                 new_begin++;
 243                 new_char_count--;
 244             }
 245             if (new_char_count>0) {
 246                 char_count=new_char_count;
 247                 begin=new_begin;
 248             }
 249         }
 250         orig_begin=begin;
 251
 252         /* if (_isblank(heb_str[begin])) {
 253             heb_str[begin]='\n';
 254         } */
 255
 256         /* skip leading newlines */
 257         while (begin<=end && _isnewline(heb_str[begin])) {
 258             begin++;
 259         }
 260
 261         /* copy content */
 262         for (i=begin; i<=end; i++) {
 263             *target = heb_str[i];
 264             target++;
 265         }
 266
 267         for (i=orig_begin; i<=end && _isnewline(heb_str[i]); i++) {
 268             *target = heb_str[i];
 269             target++;
 270         }
 271         begin=orig_begin;
 272
 273         if (begin<=0) {
 274             *target = 0;
 275             break;
 276         }
 277         begin--;
 278         end=begin;
 279     }
 280     return broken_str;
 281 #endif
 282     return heb_str;
 283 }
 284