src/statichuff/statichuff.c

   1
   2 #include "statichuff.h"
   3 #include <stdlib.h>
   4 #include <string.h>
   5
   6 void putbit(char bit, char restart, char flush, VFILE *fp)
   7 {
   8         static unsigned long int bits_buffer = 0;
   9         static unsigned char bits_used = 0;
  10
  11         /* me obligan a emitir el output */
  12         if ((flush == 1) && (bits_used > 0)) {
  13                 bits_buffer = bits_buffer << ((sizeof(unsigned long int)*8) - bits_used);
  14                 vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
  15                 bits_buffer = 0;
  16                 bits_used = 0;
  17                 return;
  18         }
  19         /* me indican que comienza un nuevo output */
  20         if (restart) {
  21                 bits_buffer = 0;
  22                 bits_used = 0;
  23         }
  24         /* inserto el bit en el buffer */
  25         bits_buffer = bits_buffer << 1;
  26         bits_buffer |= bit;
  27         bits_used++;
  28
  29         /* lleno el buffer, escribo */
  30         if (bits_used == 32) {
  31                 vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
  32                 bits_buffer = 0;
  33                 bits_used = 0;
  34         }
  35         return;
  36 }
  37
  38 void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
  39 {
  40         node1->symbol = node2->symbol;
  41         node1->freq = node2->freq;
  42         node1->lchild = node2->lchild;
  43         node1->rchild = node2->rchild;
  44 }
  45
  46 int shuff_compnode(const void *node1, const void *node2)
  47 {
  48         if (((SHUFFNODE*)node1)->freq < ((SHUFFNODE*)node2)->freq) return 1;
  49         if (((SHUFFNODE*)node1)->freq > ((SHUFFNODE*)node2)->freq) return -1;
  50         return 0;
  51 }
  52
  53 int shuff_rescalefreq(t_freq *freqtable)
  54 {
  55         int i;
  56         t_freq totalfreq = 0;
  57
  58         /* Divido por la mitad las frecuencias, asegurando de no perder */
  59         for (i = 0; i < 256; i++) {
  60                 freqtable[i] = (freqtable[i] >> 2) | 1;
  61                 totalfreq += freqtable[i];
  62         }
  63
  64         return totalfreq;
  65 }
  66
  67 int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
  68 {
  69         /* Locals */
  70         int i = 0;
  71         unsigned char symbol = 0;
  72
  73         /* Contamos las frecuencias del chunk a menos que se use un canonico */
  74         if (!chunkshuff->canonic) {
  75                 for (i = 0; i < chunksize; ++i) {
  76                         symbol = chunk[i];
  77                         chunkshuff->freqtable[symbol] += 1;
  78                         chunkshuff->sumfreq += 1;
  79
  80                         /* Si llegue al tope de freq acumulada, halve em */
  81                         if (chunkshuff->sumfreq == 14930352)
  82                                 chunkshuff->sumfreq = shuff_rescalefreq(chunkshuff->freqtable);
  83                 }
  84         }
  85
  86         /* Dumpeamos el chunk en el temporal homero */
  87         fwrite(chunk,chunksize,1,chunkshuff->coderfp);
  88
  89         return 1;
  90 }
  91
  92 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
  93 {
  94         /* Locals */
  95         FILE *fp;
  96         t_freq sumfreq = 0;
  97         int i,symbol;
  98
  99         /* Inicializamos la tabla de frecuencias */
 100         for (i = 0; i < 256; ++i) freqtable[i] = 0;
 101
 102         /* Abrimos el file */
 103         if ((fp = fopen(inputfile,"r")) == NULL) return 0;
 104         while (!feof(fp)) {
 105                 /* Contamos las frecuencias */
 106                 symbol = fgetc(fp);
 107                 if (symbol == EOF) continue;
 108
 109                 freqtable[symbol]++;
 110                 ++sumfreq;
 111
 112                 /* Si llegue al tope de freq acumulada, halve em */
 113                 if (sumfreq == 14930352)
 114                         sumfreq = shuff_rescalefreq(freqtable);
 115         }
 116
 117         fclose(fp);
 118         return 1;
 119 }
 120
 121 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 122 {
 123         int i,j = 0,nonzero = 0;
 124         SHUFFNODE *inputlist;
 125
 126         /* Calculo cuantas frequencias > 0 hay y creo la tabla */
 127         for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++;
 128         inputlist = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)*nonzero);
 129
 130         /* Cargo la inputlist del huffman solo con freqs > 0 */
 131         for (i = 0; i < 256; ++i)
 132                 if (freqtable[i] > 0) {
 133                         inputlist[j].symbol = i;
 134                         inputlist[j].freq = freqtable[i];
 135                         inputlist[j].lchild = NULL;
 136                         inputlist[j].rchild = NULL;
 137                         j++;
 138                 }
 139
 140         *nonzerofreqs = nonzero;
 141         return inputlist;
 142 }
 143
 144 SHUFFNODE *shuff_buildtree(t_freq *ftable)
 145 {
 146         SHUFFNODE *lastsymbol;
 147         SHUFFNODE *node1,*node2,*root;
 148         SHUFFNODE *inputlist;
 149         int freqcount = 0;
 150
 151         /* Genero la input list en base a la cual genera el arbol */
 152         inputlist = shuff_buildlist(ftable, &freqcount);
 153         lastsymbol = inputlist+(freqcount-1);
 154
 155         while (lastsymbol > inputlist) {
 156                 /* Ordeno la lista por frecuencia descendente */
 157                 qsort(inputlist,freqcount,sizeof(SHUFFNODE),shuff_compnode);
 158                 /* Tomo los ultimos dos elementos, generando dos nodos del arbol */
 159                 node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 160                 node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 161                 shuff_cpynode(node1,lastsymbol-1);
 162                 shuff_cpynode(node2,lastsymbol);
 163                 lastsymbol -= 1;
 164                 /* Nodo ficticio con la suma de las probs y los ptros a childs */
 165                 lastsymbol->symbol = 256;
 166                 lastsymbol->freq = node1->freq + node2->freq;
 167                 lastsymbol->lchild = node1;
 168                 lastsymbol->rchild = node2;
 169                 --freqcount;
 170         }
 171
 172         /* Copio la raiz para poder liberar la lista sin perderla */
 173         root = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 174         shuff_cpynode(root,lastsymbol);
 175
 176         /* Free up mem */
 177         free(inputlist);
 178
 179         /* Devuelvo el puntero a la raiz del arbol de huffman */
 180         return root;
 181 }
 182
 183 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 184 {
 185         int i,j;
 186         unsigned short int auxcode;
 187         unsigned char bit;
 188
 189         for (i = 0; i < 256; ++i) {
 190                 if (codetable[i].codelength > 0) {
 191                         auxcode = codetable[i].code;
 192                         printf("Symbol:%i  Freq: %lu  Code:",i,freqtable[i]);
 193                         for (j = codetable[i].codelength-1; j >= 0; --j) {
 194                                 auxcode = codetable[i].code;
 195                                 auxcode = auxcode >> j;
 196                                 bit = auxcode & 1;
 197                                 printf("%i",bit);
 198                         }
 199                         printf("  Length:%i\n",codetable[i].codelength);
 200                 }
 201         }
 202 }
 203
 204 void shuff_zerocodes(SHUFFCODE *table)
 205 {
 206         int i;
 207
 208         /* Inicializo los codigos prefijos */
 209         for (i = 0; i < 256; ++i) {
 210                 table[i].code = 0;
 211                 table[i].codelength = 0;
 212         }
 213 }
 214
 215 void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 216 {
 217         if (node->symbol < 256) {
 218                 /* Guardo el codigo en la tabla */
 219                 table[node->symbol].code = code;
 220                 table[node->symbol].codelength = level;
 221         }
 222         else {
 223                 code = code << 1;
 224                 shuff_buildcodes(table,node->lchild,level+1,code);
 225                 code |= 1;
 226                 shuff_buildcodes(table,node->rchild,level+1,code);
 227         }
 228 }
 229
 230 int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
 231 {
 232         FILE *fpsource;
 233         VFILE *fpdest;
 234         int symbol,i;
 235         unsigned long int sourcesize;
 236         char bit;
 237         SHUFFCODE symbolcode;
 238
 239         /* Abrimos el source y el destino */
 240         if (shuff->coderfp != NULL) {
 241                 fclose(shuff->coderfp); /* close bychunk temp file */
 242                 shuff->coderfp = NULL;
 243         }
 244         if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0;
 245         if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0;
 246
 247         /* Guardamos el size el archivo original e inputlist como header */
 248         fseek(fpsource,0,SEEK_END);
 249         sourcesize = ftell(fpsource);
 250         vfwrite(&sourcesize,sizeof(unsigned long int),1,fpdest);
 251         vfwrite(shuff->freqtable,sizeof(t_freq),256,fpdest);
 252
 253         /* Encodeo */
 254         fseek(fpsource,0,SEEK_SET);
 255         while (!feof(fpsource)) {
 256                 /* Levanto un symbolo (byte) */
 257                 symbol = fgetc(fpsource);
 258                 if (symbol == EOF) continue;
 259
 260                 /* Cargamos el codigo y lo emitimos */
 261                 symbolcode = ctable[symbol];
 262                 for (i = symbolcode.codelength; i > 0; --i) {
 263                         bit = (symbolcode.code >> (i-1)) & 1;
 264                         putbit(bit,0,0,fpdest);
 265                 }
 266         }
 267
 268         /* Hacemos un flush de lo que haya quedado en el buffer de salida */
 269         putbit(0,0,1,fpdest);
 270         fclose(fpsource);
 271         vfclose(fpdest);
 272         return 1;
 273 }
 274
 275 int shuff_encode_file(HUFF_STATE *shuff)
 276 {
 277         /* Locals */
 278         SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
 279
 280         /* Veo si debo armar una freqtable o si esta preloaded */
 281         if ((!shuff->canonic) && (!shuff->bychunk))
 282                 if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0;
 283
 284         /* Genero el arbol de huffman */
 285         shuff->codetree = shuff_buildtree(shuff->freqtable);
 286
 287         /* Armo la tabla de codigos prefijos para el encoder */
 288         shuff_zerocodes(codetable);
 289         shuff_buildcodes(codetable,shuff->codetree,0,0);
 290         /*shuff_printcodes(codetable,shuff->freqtable);*/
 291
 292         /* Encodeo byte per byte */
 293         shuff_encode_symbols(shuff,codetable);
 294
 295         /* Free up memory baby yeah */
 296         free(codetable);
 297
 298         return 1;
 299 }
 300
 301 SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer,
 302                                                          int *bitsleft, unsigned short int *symbol)
 303 {
 304         char bit = 0;
 305
 306         /* Levanto el symbolo y si es uno valido, devuelvo */
 307         *symbol = entrynode->symbol;
 308         if (*symbol != 256) return entrynode;
 309         if (*bitsleft == 0) return entrynode;
 310
 311         /* Obtengo otro bit a procesar y me muevo en el arbol */
 312         bit = (buffer >> ((*bitsleft)-1)) & 1;
 313         --(*bitsleft);
 314         if (bit == 0) return shuff_decode_symbols(entrynode->lchild,buffer,bitsleft,symbol);
 315         else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 316 }
 317
 318 int shuff_decode_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, int *decodedbytes)
 319 {
 320         SHUFFNODE *currnode = shuff->codetree;
 321         unsigned short int decoded_symbol;
 322         *decodedbytes = 0;
 323
 324         while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) {
 325
 326                 /* Leo un buffer de 32 bits si es que quedo vacio el anterior */
 327                 if (shuff->bitsleft == 0) {
 328                         if (vfread(&(shuff->codebuffer),sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;
 329                         shuff->bitsleft = sizeof(unsigned long int) * 8;
 330                 }
 331
 332                 /* Proceso el buffer sacando simbolos till se me agote el buffer, file o chunk */
 333                 while ((shuff->bitsleft > 0) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) {
 334                         currnode = shuff_decode_symbols(currnode,shuff->codebuffer,&(shuff->bitsleft),&decoded_symbol);
 335                         /* Si obtuve un symbolo valido lo emito*/
 336                         if (decoded_symbol != 256) {
 337                                 chunk[(*decodedbytes)++] = decoded_symbol;
 338                                 currnode = shuff->codetree;
 339                                 --(shuff->bytesleft);
 340                         }
 341                 }
 342         }
 343
 344         if (shuff->bytesleft == 0) return 0;
 345         else return 1;
 346 }
 347
 348 int shuff_decode_file(HUFF_STATE *shuff)
 349 {
 350         SHUFFNODE *currnode;
 351         unsigned long int codebuffer;
 352         FILE *fpdest;
 353         unsigned short int decoded_symbol;
 354         int bitsleft;
 355
 356         /* Comienzo a decodificar, pues la tabla ya la levante en el decinit */
 357         if ((fpdest = fopen(shuff->targetfile,"w")) == NULL) return 0;
 358         currnode = shuff->codetree;
 359
 360         while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0)) {
 361
 362                 /* Leo un buffer de 32 bits */
 363                 if (vfread(&codebuffer,sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;
 364                 bitsleft = sizeof(unsigned long int) * 8;
 365
 366                 /* Proceso el buffer sacando simbolos hasta que se me agote */
 367                 while ((bitsleft > 0) && (shuff->bytesleft > 0)) {
 368                         currnode = shuff_decode_symbols(currnode,codebuffer,&bitsleft,&decoded_symbol);
 369                         /* Si obtuve un symbolo valido lo emito*/
 370                         if (decoded_symbol != 256) {
 371                                 fputc(decoded_symbol,fpdest);
 372                                 currnode = shuff->codetree;
 373                                 --(shuff->bytesleft);
 374                         }
 375                 }
 376         }
 377
 378         /* Close destination */
 379         fclose(fpdest);
 380
 381         return 1;
 382 }
 383
 384 HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
 385 {
 386         /* Locals */
 387         HUFF_STATE *shuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 388         shuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 389
 390         /* Init fields */
 391         shuff->codebuffer = 0;
 392         shuff->bitsleft = 0;
 393         shuff->coderfp = NULL;
 394         shuff->targetfile = NULL;
 395         shuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
 396         strcpy(shuff->sourcefile,inputfile);
 397         if (outputfile != NULL) {
 398                 shuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 399                 strcpy(shuff->targetfile,outputfile);
 400         }
 401
 402         /* Levanto cuantos bytes debo decodificar y la freqtable */
 403         if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return NULL;
 404         vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp);
 405         vfread(shuff->freqtable,sizeof(t_freq),256,shuff->decoderfp);
 406         /* Armo el arbol de huffman que uso para decodificar */
 407         shuff->codetree = shuff_buildtree(shuff->freqtable);
 408
 409         return shuff;
 410 }
 411
 412 HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize)
 413 {
 414         /* Locals */
 415         HUFF_STATE *fshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 416         int i;
 417
 418         /* Inicializo la estructura para trabajar con Huff Static by File */
 419         fshuff->coderfp = NULL;
 420         fshuff->decoderfp = NULL;
 421         fshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
 422         fshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 423         strcpy(fshuff->sourcefile,inputfile);
 424         strcpy(fshuff->targetfile,outputfile);
 425         fshuff->volsize = volsize;
 426         fshuff->bychunk = 0;
 427         fshuff->canonic = 0;
 428         fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 429         for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0;
 430         fshuff->sumfreq = 0;
 431         fshuff->codetree = NULL;
 432
 433         return fshuff;
 434 }
 435
 436 HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
 437 {
 438         /* Locals */
 439         HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 440         int i;
 441
 442         /* Inicializo la estructura para trabajar con Huff Static by Chunks */
 443         cshuff->decoderfp = NULL;
 444         cshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(outputfile)+2));
 445         cshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 446         strcpy(cshuff->targetfile,outputfile);
 447         strcpy(cshuff->sourcefile,outputfile);
 448         strcat(cshuff->sourcefile,"~");
 449         cshuff->volsize = volsize;
 450         cshuff->bychunk = 1;
 451         cshuff->canonic = 0;
 452         cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 453         for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0;
 454         cshuff->sumfreq = 0;
 455         cshuff->codetree = NULL;
 456
 457         /* Abrimos un archivo temporal para ir tirando los chunks */
 458         if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;
 459
 460         return cshuff;
 461 }
 462
 463 int shuff_loadmodel(HUFF_STATE *shuff, char *modelfile) {
 464
 465         FILE *fp;
 466
 467         if ((shuff) && (shuff->freqtable) && (modelfile)) {
 468                 /* Cargo el modelo de disco */
 469                 if ((fp = fopen(modelfile,"r")) == NULL) return 0;
 470                 if (fread(shuff->freqtable,sizeof(t_freq),256,fp) != 256) return 0;
 471                 shuff->canonic = 1;
 472                 if (fp) fclose(fp);
 473                 return 1;
 474         }
 475         return 0;
 476 }
 477
 478 int shuff_savemodel(HUFF_STATE *shuff) {
 479
 480         FILE *fp;
 481         char *auxfilename;
 482         char *stopchar;
 483
 484         if ((shuff) && (shuff->targetfile) && (shuff->freqtable)) {
 485                 /* Preparo el nombre del archivo con la tabla */
 486                 auxfilename = (char*)malloc(strlen(shuff->targetfile)+1);
 487                 stopchar = strrchr(shuff->targetfile,'.');
 488                 strncpy(auxfilename,shuff->targetfile,stopchar - shuff->targetfile);
 489                 auxfilename[stopchar - shuff->targetfile] = 0;
 490                 strcat(auxfilename,".ftb");
 491
 492                 /* Lo creamos y dumpeamos la tabla de frecuencias (modelo) */
 493                 if ((fp = fopen(auxfilename,"w")) == NULL) return 0;
 494                 fwrite(shuff->freqtable,sizeof(t_freq),256,fp);
 495                 if (fp) fclose(fp);
 496
 497                 return 1;
 498         }
 499         return 0;
 500 }
 501
 502 void shuff_deinit_encoder(HUFF_STATE *shuff)
 503 {
 504         /* Libero mallocs y cierro archivos */
 505         if (shuff->freqtable) free(shuff->freqtable);
 506         if (shuff->coderfp) fclose(shuff->coderfp);
 507         if (shuff->bychunk) unlink(shuff->sourcefile);
 508         if (shuff->sourcefile) free(shuff->sourcefile);
 509         if (shuff->targetfile) free(shuff->targetfile);
 510
 511         /* Destruyo recursivamente el arbol de codigos */
 512 }
 513
 514 void shuff_deinit_decoder(HUFF_STATE *shuff)
 515 {
 516         /* Libero mallocs y cierro archivos */
 517         if (shuff->freqtable) free(shuff->freqtable);
 518         if (shuff->sourcefile != NULL) free(shuff->sourcefile);
 519         if (shuff->targetfile != NULL) free(shuff->targetfile);
 520         if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
 521
 522         /* Destruyo recursivamente el arbol de codigos */
 523 }