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 void shuff_destroy_tree(SHUFFNODE *node) {
  54         /* Si llegue a una hoja, destruyo y vuelvo */
  55         if (node->symbol < 256) {
  56                 free(node);
  57                 return;
  58         }
  59         else {
  60                 /* Desciendo por izq, luego por derecha y luego libero */
  61                 shuff_destroy_tree(node->lchild);
  62                 shuff_destroy_tree(node->rchild);
  63                 free(node);
  64                 return;
  65         }
  66 }
  67
  68 int shuff_rescalefreq(t_freq *freqtable)
  69 {
  70         int i;
  71         t_freq totalfreq = 0;
  72
  73         /* Divido por la mitad las frecuencias, asegurando de no perder */
  74         for (i = 0; i < 256; i++) {
  75                 freqtable[i] = (freqtable[i] >> 2) | 1;
  76                 totalfreq += freqtable[i];
  77         }
  78
  79         return totalfreq;
  80 }
  81
  82 int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
  83 {
  84         /* Locals */
  85         int i = 0;
  86         unsigned char symbol = 0;
  87
  88         /* Contamos las frecuencias del chunk a menos que se use un canonico */
  89         if (!chunkshuff->canonic) {
  90                 for (i = 0; i < chunksize; ++i) {
  91                         symbol = chunk[i];
  92                         chunkshuff->freqtable[symbol] += 1;
  93                         chunkshuff->sumfreq += 1;
  94
  95                         /* Si llegue al tope de freq acumulada, halve em */
  96                         if (chunkshuff->sumfreq == 14930352)
  97                                 chunkshuff->sumfreq = shuff_rescalefreq(chunkshuff->freqtable);
  98                 }
  99         }
 100
 101         /* Dumpeamos el chunk en el temporal homero */
 102         fwrite(chunk,chunksize,1,chunkshuff->coderfp);
 103
 104         return 1;
 105 }
 106
 107 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 108 {
 109         /* Locals */
 110         FILE *fp;
 111         t_freq sumfreq = 0;
 112         int i,symbol;
 113
 114         /* Inicializamos la tabla de frecuencias */
 115         for (i = 0; i < 256; ++i) freqtable[i] = 0;
 116
 117         /* Abrimos el file */
 118         if ((fp = fopen(inputfile,"r")) == NULL) return 0;
 119         while (!feof(fp)) {
 120                 /* Contamos las frecuencias */
 121                 symbol = fgetc(fp);
 122                 if (symbol == EOF) continue;
 123
 124                 freqtable[symbol]++;
 125                 ++sumfreq;
 126
 127                 /* Si llegue al tope de freq acumulada, halve em */
 128                 if (sumfreq == 14930352)
 129                         sumfreq = shuff_rescalefreq(freqtable);
 130         }
 131
 132         fclose(fp);
 133         return 1;
 134 }
 135
 136 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 137 {
 138         int i,j = 0,nonzero = 0;
 139         SHUFFNODE *inputlist;
 140
 141         /* Calculo cuantas frequencias > 0 hay y creo la tabla */
 142         for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++;
 143         inputlist = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)*nonzero);
 144
 145         /* Cargo la inputlist del huffman solo con freqs > 0 */
 146         for (i = 0; i < 256; ++i)
 147                 if (freqtable[i] > 0) {
 148                         inputlist[j].symbol = i;
 149                         inputlist[j].freq = freqtable[i];
 150                         inputlist[j].lchild = NULL;
 151                         inputlist[j].rchild = NULL;
 152                         j++;
 153                 }
 154
 155         *nonzerofreqs = nonzero;
 156         return inputlist;
 157 }
 158
 159 SHUFFNODE *shuff_buildtree(t_freq *ftable)
 160 {
 161         SHUFFNODE *lastsymbol;
 162         SHUFFNODE *node1,*node2,*root;
 163         SHUFFNODE *inputlist;
 164         int freqcount = 0;
 165
 166         /* Genero la input list en base a la cual genera el arbol */
 167         inputlist = shuff_buildlist(ftable, &freqcount);
 168         lastsymbol = inputlist+(freqcount-1);
 169
 170         while (lastsymbol > inputlist) {
 171                 /* Ordeno la lista por frecuencia descendente */
 172                 qsort(inputlist,freqcount,sizeof(SHUFFNODE),shuff_compnode);
 173                 /* Tomo los ultimos dos elementos, generando dos nodos del arbol */
 174                 node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 175                 node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 176                 shuff_cpynode(node1,lastsymbol-1);
 177                 shuff_cpynode(node2,lastsymbol);
 178                 lastsymbol -= 1;
 179                 /* Nodo ficticio con la suma de las probs y los ptros a childs */
 180                 lastsymbol->symbol = 256;
 181                 lastsymbol->freq = node1->freq + node2->freq;
 182                 lastsymbol->lchild = node1;
 183                 lastsymbol->rchild = node2;
 184                 --freqcount;
 185         }
 186
 187         /* Copio la raiz para poder liberar la lista sin perderla */
 188         root = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
 189         shuff_cpynode(root,lastsymbol);
 190
 191         /* Free up mem */
 192         free(inputlist);
 193
 194         /* Devuelvo el puntero a la raiz del arbol de huffman */
 195         return root;
 196 }
 197
 198 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 199 {
 200         int i,j;
 201         unsigned short int auxcode;
 202         unsigned char bit;
 203
 204         for (i = 0; i < 256; ++i) {
 205                 if (codetable[i].codelength > 0) {
 206                         auxcode = codetable[i].code;
 207                         printf("Symbol:%i  Freq: %lu  Code:",i,freqtable[i]);
 208                         for (j = codetable[i].codelength-1; j >= 0; --j) {
 209                                 auxcode = codetable[i].code;
 210                                 auxcode = auxcode >> j;
 211                                 bit = auxcode & 1;
 212                                 printf("%i",bit);
 213                         }
 214                         printf("  Length:%i\n",codetable[i].codelength);
 215                 }
 216         }
 217 }
 218
 219 void shuff_zerocodes(SHUFFCODE *table)
 220 {
 221         int i;
 222
 223         /* Inicializo los codigos prefijos */
 224         for (i = 0; i < 256; ++i) {
 225                 table[i].code = 0;
 226                 table[i].codelength = 0;
 227         }
 228 }
 229
 230 void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 231 {
 232         if (node->symbol < 256) {
 233                 /* Guardo el codigo en la tabla */
 234                 table[node->symbol].code = code;
 235                 table[node->symbol].codelength = level;
 236         }
 237         else {
 238                 code = code << 1;
 239                 shuff_buildcodes(table,node->lchild,level+1,code);
 240                 code |= 1;
 241                 shuff_buildcodes(table,node->rchild,level+1,code);
 242         }
 243 }
 244
 245 int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
 246 {
 247         FILE *fpsource;
 248         VFILE *fpdest;
 249         int symbol,i;
 250         unsigned long int sourcesize;
 251         char bit;
 252         SHUFFCODE symbolcode;
 253
 254         /* Abrimos el source y el destino */
 255         if (shuff->coderfp != NULL) {
 256                 fclose(shuff->coderfp); /* close bychunk temp file */
 257                 shuff->coderfp = NULL;
 258         }
 259         if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0;
 260         if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0;
 261
 262         /* Guardamos el size el archivo original e inputlist como header */
 263         fseek(fpsource,0,SEEK_END);
 264         sourcesize = ftell(fpsource);
 265         vfwrite(&sourcesize,sizeof(unsigned long int),1,fpdest);
 266         vfwrite(shuff->freqtable,sizeof(t_freq),256,fpdest);
 267
 268         /* Encodeo */
 269         fseek(fpsource,0,SEEK_SET);
 270         while (!feof(fpsource)) {
 271                 /* Levanto un symbolo (byte) */
 272                 symbol = fgetc(fpsource);
 273                 if (symbol == EOF) continue;
 274
 275                 /* Cargamos el codigo y lo emitimos */
 276                 symbolcode = ctable[symbol];
 277                 for (i = symbolcode.codelength; i > 0; --i) {
 278                         bit = (symbolcode.code >> (i-1)) & 1;
 279                         putbit(bit,0,0,fpdest);
 280                 }
 281         }
 282
 283         /* Hacemos un flush de lo que haya quedado en el buffer de salida */
 284         putbit(0,0,1,fpdest);
 285         fclose(fpsource);
 286         vfclose(fpdest);
 287         return 1;
 288 }
 289
 290 int shuff_encode_file(HUFF_STATE *shuff)
 291 {
 292         /* Locals */
 293         SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
 294
 295         /* Veo si debo armar una freqtable o si esta preloaded */
 296         if ((!shuff->canonic) && (!shuff->bychunk))
 297                 if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0;
 298
 299         /* Genero el arbol de huffman */
 300         shuff->codetree = shuff_buildtree(shuff->freqtable);
 301
 302         /* Armo la tabla de codigos prefijos para el encoder */
 303         shuff_zerocodes(codetable);
 304         shuff_buildcodes(codetable,shuff->codetree,0,0);
 305         /*shuff_printcodes(codetable,shuff->freqtable);*/
 306
 307         /* Encodeo byte per byte */
 308         shuff_encode_symbols(shuff,codetable);
 309
 310         /* Free up memory baby yeah */
 311         free(codetable);
 312
 313         return 1;
 314 }
 315
 316 SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer,
 317                                                          int *bitsleft, unsigned short int *symbol)
 318 {
 319         char bit = 0;
 320
 321         /* Levanto el symbolo y si es uno valido, devuelvo */
 322         *symbol = entrynode->symbol;
 323         if (*symbol != 256) return entrynode;
 324         if (*bitsleft == 0) return entrynode;
 325
 326         /* Obtengo otro bit a procesar y me muevo en el arbol */
 327         bit = (buffer >> ((*bitsleft)-1)) & 1;
 328         --(*bitsleft);
 329         if (bit == 0) return shuff_decode_symbols(entrynode->lchild,buffer,bitsleft,symbol);
 330         else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 331 }
 332
 333 int shuff_decode_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, int *decodedbytes)
 334 {
 335         SHUFFNODE *currnode = shuff->codetree;
 336         unsigned short int decoded_symbol;
 337         *decodedbytes = 0;
 338
 339         while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) {
 340
 341                 /* Leo un buffer de 32 bits si es que quedo vacio el anterior */
 342                 if (shuff->bitsleft == 0) {
 343                         if (vfread(&(shuff->codebuffer),sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;
 344                         shuff->bitsleft = sizeof(unsigned long int) * 8;
 345                 }
 346
 347                 /* Proceso el buffer sacando simbolos till se me agote el buffer, file o chunk */
 348                 while ((shuff->bitsleft > 0) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) {
 349                         currnode = shuff_decode_symbols(currnode,shuff->codebuffer,&(shuff->bitsleft),&decoded_symbol);
 350                         /* Si obtuve un symbolo valido lo emito*/
 351                         if (decoded_symbol != 256) {
 352                                 chunk[(*decodedbytes)++] = decoded_symbol;
 353                                 currnode = shuff->codetree;
 354                                 --(shuff->bytesleft);
 355                         }
 356                 }
 357         }
 358
 359         if (shuff->bytesleft == 0) return 0;
 360         else return 1;
 361 }
 362
 363 int shuff_decode_file(HUFF_STATE *shuff)
 364 {
 365         SHUFFNODE *currnode;
 366         unsigned long int codebuffer;
 367         FILE *fpdest;
 368         unsigned short int decoded_symbol;
 369         int bitsleft;
 370
 371         /* Comienzo a decodificar, pues la tabla ya la levante en el decinit */
 372         if ((fpdest = fopen(shuff->targetfile,"w")) == NULL) return 0;
 373         currnode = shuff->codetree;
 374
 375         while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0)) {
 376
 377                 /* Leo un buffer de 32 bits */
 378                 if (vfread(&codebuffer,sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;
 379                 bitsleft = sizeof(unsigned long int) * 8;
 380
 381                 /* Proceso el buffer sacando simbolos hasta que se me agote */
 382                 while ((bitsleft > 0) && (shuff->bytesleft > 0)) {
 383                         currnode = shuff_decode_symbols(currnode,codebuffer,&bitsleft,&decoded_symbol);
 384                         /* Si obtuve un symbolo valido lo emito*/
 385                         if (decoded_symbol != 256) {
 386                                 fputc(decoded_symbol,fpdest);
 387                                 currnode = shuff->codetree;
 388                                 --(shuff->bytesleft);
 389                         }
 390                 }
 391         }
 392
 393         /* Close destination */
 394         fclose(fpdest);
 395
 396         return 1;
 397 }
 398
 399 HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
 400 {
 401         /* Locals */
 402         HUFF_STATE *shuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 403         shuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 404
 405         /* Init fields */
 406         shuff->codebuffer = 0;
 407         shuff->bitsleft = 0;
 408         shuff->coderfp = NULL;
 409         shuff->targetfile = NULL;
 410         shuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
 411         strcpy(shuff->sourcefile,inputfile);
 412         if (outputfile != NULL) {
 413                 shuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 414                 strcpy(shuff->targetfile,outputfile);
 415         }
 416
 417         /* Levanto cuantos bytes debo decodificar y la freqtable */
 418         if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return NULL;
 419         vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp);
 420         vfread(shuff->freqtable,sizeof(t_freq),256,shuff->decoderfp);
 421         /* Armo el arbol de huffman que uso para decodificar */
 422         shuff->codetree = shuff_buildtree(shuff->freqtable);
 423
 424         return shuff;
 425 }
 426
 427 HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize)
 428 {
 429         /* Locals */
 430         HUFF_STATE *fshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 431         int i;
 432
 433         /* Inicializo la estructura para trabajar con Huff Static by File */
 434         fshuff->coderfp = NULL;
 435         fshuff->decoderfp = NULL;
 436         fshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
 437         fshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 438         strcpy(fshuff->sourcefile,inputfile);
 439         strcpy(fshuff->targetfile,outputfile);
 440         fshuff->volsize = volsize;
 441         fshuff->bychunk = 0;
 442         fshuff->canonic = 0;
 443         fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 444         for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0;
 445         fshuff->sumfreq = 0;
 446         fshuff->codetree = NULL;
 447
 448         return fshuff;
 449 }
 450
 451 HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
 452 {
 453         /* Locals */
 454         HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));
 455         int i;
 456
 457         /* Inicializo la estructura para trabajar con Huff Static by Chunks */
 458         cshuff->decoderfp = NULL;
 459         cshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(outputfile)+2));
 460         cshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
 461         strcpy(cshuff->targetfile,outputfile);
 462         strcpy(cshuff->sourcefile,outputfile);
 463         strcat(cshuff->sourcefile,"~");
 464         cshuff->volsize = volsize;
 465         cshuff->bychunk = 1;
 466         cshuff->canonic = 0;
 467         cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
 468         for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0;
 469         cshuff->sumfreq = 0;
 470         cshuff->codetree = NULL;
 471
 472         /* Abrimos un archivo temporal para ir tirando los chunks */
 473         if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;
 474
 475         return cshuff;
 476 }
 477
 478 int shuff_loadmodel(HUFF_STATE *shuff, char *modelfile) {
 479
 480         FILE *fp;
 481
 482         if ((shuff) && (shuff->freqtable) && (modelfile)) {
 483                 /* Cargo el modelo de disco */
 484                 if ((fp = fopen(modelfile,"r")) == NULL) return 0;
 485                 if (fread(shuff->freqtable,sizeof(t_freq),256,fp) != 256) return 0;
 486                 shuff->canonic = 1;
 487                 if (fp) fclose(fp);
 488                 return 1;
 489         }
 490         return 0;
 491 }
 492
 493 int shuff_savemodel(HUFF_STATE *shuff) {
 494
 495         FILE *fp;
 496         char *auxfilename;
 497         char *stopchar;
 498
 499         if ((shuff) && (shuff->targetfile) && (shuff->freqtable)) {
 500                 /* Preparo el nombre del archivo con la tabla */
 501                 auxfilename = (char*)malloc(strlen(shuff->targetfile)+1);
 502                 stopchar = strrchr(shuff->targetfile,'.');
 503                 strncpy(auxfilename,shuff->targetfile,stopchar - shuff->targetfile);
 504                 auxfilename[stopchar - shuff->targetfile] = 0;
 505                 strcat(auxfilename,".ftb");
 506
 507                 /* Lo creamos y dumpeamos la tabla de frecuencias (modelo) */
 508                 if ((fp = fopen(auxfilename,"w")) == NULL) return 0;
 509                 fwrite(shuff->freqtable,sizeof(t_freq),256,fp);
 510                 if (fp) fclose(fp);
 511
 512                 return 1;
 513         }
 514         return 0;
 515 }
 516
 517 void shuff_deinit_encoder(HUFF_STATE *shuff)
 518 {
 519         /* Libero mallocs y cierro archivos */
 520         if (shuff->freqtable) free(shuff->freqtable);
 521         if (shuff->coderfp) fclose(shuff->coderfp);
 522         if (shuff->bychunk) unlink(shuff->sourcefile);
 523         if (shuff->sourcefile) free(shuff->sourcefile);
 524         if (shuff->targetfile) free(shuff->targetfile);
 525
 526         /* Destruyo recursivamente el arbol de codigos */
 527         if (shuff->codetree) shuff_destroy_tree(shuff->codetree);
 528 }
 529
 530 void shuff_deinit_decoder(HUFF_STATE *shuff)
 531 {
 532         /* Libero mallocs y cierro archivos */
 533         if (shuff->freqtable) free(shuff->freqtable);
 534         if (shuff->sourcefile != NULL) free(shuff->sourcefile);
 535         if (shuff->targetfile != NULL) free(shuff->targetfile);
 536         if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
 537
 538         /* Destruyo recursivamente el arbol de codigos */
 539         if (shuff->codetree) shuff_destroy_tree(shuff->codetree);
 540 }