src/statichuff/statichuff.c

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