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