sys/src/cmd/tcs/tune.c

   1 #include <u.h>
   2 #include <libc.h>
   3 #include <bio.h>
   4 #include "hdr.h"
   5 #include "conv.h"
   6
   7 typedef struct Tmap Tmap;
   8 struct Tmap
   9 {
  10         Rune u;
  11         Rune t;
  12 };
  13
  14 static Tmap t1[] =
  15 {
  16         {0x0b85/*அ*/, 0xe201/**/},
  17         {0x0b86/*ஆ*/, 0xe202/**/},
  18         {0x0b87/*இ*/, 0xe203/**/},
  19         {0x0b88/*ஈ*/, 0xe204/**/},
  20         {0x0b89/*உ*/, 0xe205/**/},
  21         {0x0b8a/*ஊ*/, 0xe206/**/},
  22         {0x0b8e/*எ*/, 0xe207/**/},
  23         {0x0b8f/*ஏ*/, 0xe208/**/},
  24         {0x0b90/*ஐ*/, 0xe209/**/},
  25         {0x0b92/*ஒ*/, 0xe20a/**/},
  26         {0x0b93/*ஓ*/, 0xe20b/**/},
  27         {0x0b94/*ஔ*/, 0xe20c/**/},
  28         {0x0b83/*ஃ*/, 0xe20d/**/}
  29 };
  30
  31 static Rune t2[] =
  32 {
  33         0x0bcd/*்*/,
  34         0x0bcd/*்*/,  // filler
  35         0x0bbe/*ா*/,
  36         0x0bbf/*ி*/,
  37         0x0bc0/*ீ*/,
  38         0x0bc1/*ு*/,
  39         0x0bc2/*ூ*/,
  40         0x0bc6/*ெ*/,
  41         0x0bc7/*ே*/,
  42         0x0bc8/*ை*/,
  43         0x0bca/*ொ*/,
  44         0x0bcb/*ோ*/,
  45         0x0bcc/*ௌ*/
  46 };
  47
  48 static Tmap t3[] =
  49 {
  50         {0x0b95/*க*/, 0xe211/**/},
  51         {0x0b99/*ங*/, 0xe221/**/},
  52         {0x0b9a/*ச*/, 0xe231/**/},
  53         {0x0b9c/*ஜ*/, 0xe331/**/},
  54         {0x0b9e/*ஞ*/, 0xe241/**/},
  55         {0x0b9f/*ட*/, 0xe251/**/},
  56         {0x0ba3/*ண*/, 0xe261/**/},
  57         {0x0ba4/*த*/, 0xe271/**/},
  58         {0x0ba8/*ந*/, 0xe281/**/},
  59         {0x0ba9/*ன*/, 0xe321/**/},
  60         {0x0baa/*ப*/, 0xe291/**/},
  61         {0x0bae/*ம*/, 0xe2a1/**/},
  62         {0x0baf/*ய*/, 0xe2b1/**/},
  63         {0x0bb0/*ர*/, 0xe2c1/**/},
  64         {0x0bb1/*ற*/, 0xe311/**/},
  65         {0x0bb2/*ல*/, 0xe2d1/**/},
  66         {0x0bb3/*ள*/, 0xe301/**/},
  67         {0x0bb4/*ழ*/, 0xe2f1/**/},
  68         {0x0bb5/*வ*/, 0xe2e1/**/},
  69         {0x0bb6/*ஶ*/, 0xe341/**/},
  70         {0x0bb7/*ஷ*/, 0xe351/**/},
  71         {0x0bb8/*ஸ*/, 0xe361/**/},
  72         {0x0bb9/*ஹ*/, 0xe371/**/}
  73 };
  74
  75 static Rune
  76 findbytune(Tmap *tab, int size, Rune t)
  77 {
  78         int i;
  79
  80         for(i = 0; i < size; i++)
  81                 if(tab[i].t == t)
  82                         return tab[i].u;
  83         return Runeerror;
  84 }
  85
  86 static Rune
  87 findbyuni(Tmap *tab, int size, Rune u)
  88 {
  89         int i;
  90
  91         for(i = 0; i < size; i++)
  92                 if(tab[i].u == u)
  93                         return tab[i].t;
  94         return Runeerror;
  95 }
  96
  97 static int
  98 findindex(Rune *rstr, int size, Rune r)
  99 {
 100         int i;
 101
 102         for(i = 0; i < size; i++)
 103                 if(rstr[i] == r)
 104                         return i;
 105         return -1;
 106 }
 107
 108 void
 109 tune_in(int fd, long *x, struct convert *out)
 110 {
 111         Biobuf b;
 112         Rune rbuf[N];
 113         Rune *r, *er, tr;
 114         int c, i;
 115
 116         USED(x);
 117         r = rbuf;
 118         er = rbuf+N-3;
 119         Binit(&b, fd, OREAD);
 120         while((c = Bgetrune(&b)) != Beof){
 121                 ninput += b.runesize;
 122                 if(r >= er){
 123                         OUT(out, rbuf, r-rbuf);
 124                         r = rbuf;
 125                 }
 126                 if(c>=0xe210/**/ && c <= 0xe38c/**/ && (i = c%16) < nelem(t2)){
 127                         if(c >= 0xe380/**/){
 128                                 *r++ = 0x0b95/*க*/;
 129                                 *r++ = 0x0bcd/*்*/;
 130                                 *r++ = 0x0bb7/*ஷ*/;
 131                         }else
 132                                 *r++ = findbytune(t3, nelem(t3), c-i+1);
 133                         if(i != 1)
 134                                 *r++ = t2[i];
 135                 }else if((tr = findbytune(t1, nelem(t1), c)) != Runeerror)
 136                         *r++ = tr;
 137                 else switch(c){
 138                         case 0xe3d0/**/:
 139                                 *r++ = 0x0ba3/*ண*/; *r++ = 0x0bbe/*ா*/;
 140                                 break;
 141                         case 0xe3d1/**/:
 142                                 *r++ = 0x0bb1/*ற*/; *r++ = 0x0bbe/*ா*/;
 143                                 break;
 144                         case 0xe3d2/**/:
 145                                 *r++ = 0x0ba9/*ன*/; *r++ = 0x0bbe/*ா*/;
 146                                 break;
 147                         case 0xe3d4/**/:
 148                                 *r++ = 0x0ba3/*ண*/; *r++ = 0x0bc8/*ை*/;
 149                                 break;
 150                         case 0xe3d5/**/:
 151                                 *r++ = 0x0bb2/*ல*/; *r++ = 0x0bc8/*ை*/;
 152                                 break;
 153                         case 0xe3d6/**/:
 154                                 *r++ = 0x0bb3/*ள*/; *r++ = 0x0bc8/*ை*/;
 155                                 break;
 156                         case 0xe3d7/**/:
 157                                 *r++ = 0x0ba9/*ன*/; *r++ = 0x0bc8/*ை*/;
 158                                 break;
 159                         case 0xe38d/**/:
 160                                 *r++ = 0x0bb6/*ஶ*/; *r++ = 0x0bcd/*்*/; *r++ = 0x0bb0/*ர*/; *r++ = 0x0bc0/*ீ*/;
 161                                 break;
 162                         default:
 163                                 if(c >= 0xe200 && c <= 0xe3ff){
 164                                         if(squawk)
 165                                                 EPR "%s: rune 0x%x not in output cs\n", argv0, c);
 166                                         nerrors++;
 167                                         if(clean)
 168                                                 break;
 169                                         c = BADMAP;
 170                                 }
 171                                 *r++ = c;
 172                                 break;
 173                 }
 174         }
 175         if(r > rbuf)
 176                 OUT(out, rbuf, r-rbuf);
 177         OUT(out, rbuf, 0);
 178 }
 179
 180 void
 181 tune_out(Rune *r, int n, long *x)
 182 {
 183         static int state = 0;
 184         static Rune lastr;
 185         Rune *er, tr, rr;
 186         char *p;
 187         int i;
 188
 189         USED(x);
 190         nrunes += n;
 191         er = r+n;
 192         for(p = obuf; r < er; r++){
 193                 switch(state){
 194                 case 0:
 195                 case0:
 196                         if((tr = findbyuni(t3, nelem(t3), *r)) != Runeerror){
 197                                 lastr = tr;
 198                                 state = 1;
 199                         }else if(*r == 0x0b92/*ஒ*/){
 200                                 lastr = 0xe20a/**/;
 201                                 state = 3;
 202                         }else if((tr = findbyuni(t1, nelem(t1), *r)) != Runeerror)
 203                                 p += runetochar(p, &tr);
 204                         else
 205                                 p += runetochar(p, r);
 206                         break;
 207                 case 1:
 208                 case1:
 209                         if((i = findindex(t2, nelem(t2), *r)) != -1){
 210                                 if(lastr && lastr != Runeerror)
 211                                         lastr += i-1;
 212                                 if(*r ==0x0bc6/*ெ*/)
 213                                         state = 5;
 214                                 else if(*r ==0x0bc7/*ே*/)
 215                                         state = 4;
 216                                 else if(lastr == 0xe210/**/)
 217                                         state = 2;
 218                                 else if(lastr == 0xe340/**/)
 219                                         state = 6;
 220                                 else{
 221                                         if(lastr)
 222                                                 p += runetochar(p, &lastr);
 223                                         state = 0;
 224                                 }
 225                         }else if(lastr && lastr != Runeerror && (*r == 0x00b2/*²*/ || *r == 0x00b3/*³*/ || *r == 0x2074/*⁴*/)){
 226                                 if(squawk)
 227                                         EPR "%s: character <U+%.4X, U+%.4X> not in output cs\n", argv0, lastr, *r);
 228                                 lastr = clean ? 0 : Runeerror;
 229                                 nerrors++;
 230                         }else{
 231                                 if(lastr)
 232                                         p += runetochar(p, &lastr);
 233                                 state = 0;
 234                                 goto case0;
 235                         }
 236                         break;
 237                 case 2:
 238                         if(*r == 0x0bb7/*ஷ*/){
 239                                 lastr = 0xe381/**/;
 240                                 state = 1;
 241                                 break;
 242                         }
 243                         p += runetochar(p, &lastr);
 244                         state = 0;
 245                         goto case0;
 246                 case 3:
 247                         state = 0;
 248                         if(*r == 0x0bd7/*ௗ*/){
 249                                 rr = 0xe20c/**/;
 250                                 p += runetochar(p, &rr);
 251                                 break;
 252                         }
 253                         p += runetochar(p, &lastr);
 254                         goto case0;
 255                 case 4:
 256                         state = 0;
 257                         if(*r == 0x0bbe/*ா*/){
 258                                 if(lastr){
 259                                         if(lastr != Runeerror)
 260                                                 lastr += 3;
 261                                         p += runetochar(p, &lastr);
 262                                 }
 263                                 break;
 264                         }
 265                         if(lastr)
 266                                 p += runetochar(p, &lastr);
 267                         goto case0;
 268                 case 5:
 269                         state = 0;
 270                         if(*r == 0x0bbe/*ா*/ || *r == 0x0bd7/*ௗ*/){
 271                                 if(lastr){
 272                                         if(lastr != Runeerror)
 273                                                 lastr += *r == 0x0bbe/*ா*/ ? 3 : 5;
 274                                         p += runetochar(p, &lastr);
 275                                 }
 276                                 break;
 277                         }
 278                         if(lastr)
 279                                 p += runetochar(p, &lastr);
 280                         goto case0;
 281                 case 6:
 282                         if(*r == 0x0bb0/*ர*/){
 283                                 state = 7;
 284                                 break;
 285                         }
 286                         p += runetochar(p, &lastr);
 287                         state = 0;
 288                         goto case0;
 289                 case 7:
 290                         if(*r == 0x0bc0/*ீ*/){
 291                                 rr = 0xe38d/**/;
 292                                 p += runetochar(p, &rr);
 293                                 state = 0;
 294                                 break;
 295                         }
 296                         p += runetochar(p, &lastr);
 297                         lastr = 0xe2c1/**/;
 298                         state = 1;
 299                         goto case1;
 300                 }
 301         }
 302         if(n == 0 && state != 0){
 303                 if(lastr)
 304                         p += runetochar(p, &lastr);
 305                 state = 0;
 306         }
 307         noutput += p-obuf;
 308         write(1, obuf, p-obuf);
 309 }