src/jacu.c

   1
   2 /* Jacu Team - GPL */
   3 #include "blocksorting/bs.h"
   4 #include "mtf/mtf.h"
   5 #include "zerogrouping/zerogrouping.h"
   6 #include "statichuff/statichuff.h"
   7 #include "vfile/vfile.h"
   8 #include "vfile/common.h"
   9 #include <stdlib.h>
  10 #include <stdio.h>
  11 #include <unistd.h>
  12
  13 long fsize(const char* filename);
  14
  15 typedef struct _flags_ {
  16         int cflag;
  17         int dflag;
  18         int zflag;
  19         int tflag;
  20         int qflag;
  21         int sflag;
  22         int mflag;
  23         int rflag; /* Richard Dictionary :-) */
  24 } t_Flags;
  25
  26 int comprimir(char *src, char *dst, Uint32 pagesize, Uint32 volumesize, t_Flags *flags, char *staticmodel);
  27 int descomprimir(char *src, char *dst);
  28
  29 int main(int argc, char* argv[])
  30 {
  31         long int volumesize = 0;
  32         Uint32 pagesize = 32768; /* 32KB */
  33         int ch;
  34         t_Flags flags;
  35         char *staticmodel = NULL;
  36
  37         memset(&flags, 0, sizeof(t_Flags));
  38
  39         while ((ch = getopt(argc, argv, "rscdzm:t:q:")) != -1) {
  40
  41                 switch (ch) {
  42                         case 'c': flags.cflag = 1;
  43                                           break;
  44
  45                         case 'd': flags.dflag = 1;
  46                                           break;
  47
  48                         case 'z': flags.zflag = 1;
  49                                           break;
  50
  51                         case 'm': flags.mflag = 1;
  52                                           staticmodel = optarg;
  53                                           break;
  54
  55                         case 's': flags.sflag = 1;
  56                                           break;
  57
  58                         case 't': flags.tflag = 1;
  59                                 volumesize = atol(optarg);
  60                                 break;
  61
  62                         case 'r': flags.rflag = 1;
  63                                 break;
  64                         case 'q': flags.qflag = 1;
  65                                 switch (atoi(optarg))
  66                                 {
  67                                         case 0: pagesize = 1024; /* 1K */
  68                                                 break;
  69                                         case 1: pagesize = 2048; /* 2K */
  70                                                 break;
  71                                         case 2: pagesize = 4096; /* 4K */
  72                                                 break;
  73                                         case 3: pagesize = 8192; /* 8K */
  74                                                 break;
  75                                         case 4: pagesize = 16384; /* 16K */
  76                                                 break;
  77                                         case 5: pagesize = 32768; /* 32K */
  78                                                 break;
  79                                         case 6: pagesize = 65536; /* 64K */
  80                                                 break;
  81                                         case 7: pagesize = 131072; /* 128K */
  82                                                 break;
  83                                         case 8: pagesize = 262144; /* 256K */
  84                                                 break;
  85                                         case 9: pagesize = 524288; /* 512K */
  86                                                 break;
  87                                         default: pagesize = 0; /* error */
  88                                 }
  89                                 break;
  90
  91                         default: fprintf(stderr, "Usage: %s [-cdzsr][-q blksize][-t volsize][-m modeldumpfile] source target\n", argv[0]);
  92                                          return(2);
  93                 }
  94         }
  95
  96         if ( (argc == 1) || (flags.cflag & flags.dflag) || !(flags.cflag | flags.dflag) || ((argc - optind) < 2) || (flags.mflag & flags.sflag)) {
  97                 fprintf(stderr, "Usage: %s [-cdzsr][-q compressionquality][-t volsize][-m modeldumpfile] source target\n", argv[0]);
  98                 return (3);
  99         }
 100         if ((flags.tflag) && (volumesize <= 0l)) {
 101                 fprintf(stderr,"Error: The volume size must be a non-zero value\n");
 102                 return (4);
 103         }
 104         if ((flags.qflag) && (pagesize <= 1u)) {
 105                 fprintf(stderr,"Error: El nivel de compresión debe ser entre 0 (menor) y 9 (mayor).\n");
 106                 return (5);
 107         }
 108
 109         if (flags.cflag == 1) {
 110                 return comprimir(argv[optind], argv[optind+1], pagesize, volumesize, &flags, staticmodel);
 111         }
 112
 113         if (flags.dflag == 1) {
 114                 return descomprimir(argv[optind], argv[optind+1]);
 115         }
 116
 117         return 0;
 118 }
 119
 120 long fsize(const char* filename)
 121 {
 122         FILE* file;
 123         long  file_size;
 124
 125         if (!(file = fopen(filename, "ab"))) return -1;
 126         file_size = ftell(file);
 127         fclose(file);
 128         return file_size;
 129 }
 130
 131 int comprimir(char *src, char *dst, Uint32 pagesize, Uint32 volumesize, t_Flags *flags, char *staticmodel)
 132 {
 133         /* Comprimo */
 134         t_BlockSort *bs;
 135         HUFF_STATE *shuff;
 136         FILE *fp;
 137         Uint32 i, j, total, k;
 138         unsigned char *mtf;
 139         unsigned char *salida, *data;
 140         unsigned char *z;
 141         int z_len;
 142
 143         /* Abrimos el archivo a comprimir y encodeamos bloques */
 144         if ((fp = fopen(src, "rb")) == NULL) return 1;
 145
 146         /* Preparo el compresor huffman */
 147         if ((shuff = shuff_init_encoder_bychunk(dst, volumesize*1024)) == NULL) return 1;
 148         if (flags->mflag == 1) shuff_loadmodel(shuff, staticmodel);
 149
 150         /* Preparo el BS alocando mem para la Salida: V(vector) + K(colnum) */
 151         data = malloc(sizeof(unsigned char)*pagesize);
 152         salida = malloc(sizeof(unsigned char)*pagesize+sizeof(Uint32));
 153         bs = bs_create(pagesize);
 154
 155         /* Guardamos el pagesize como header (huffencoded) */
 156         shuff_scanfreq_chunk(shuff,(char*)&pagesize,sizeof(Uint32));
 157
 158         /* Guardamos cabecera para indicar si usamos ZG (huffencoded) */
 159         if (flags->zflag)
 160                 shuff_scanfreq_chunk(shuff, "\001", 1);
 161         else
 162                 shuff_scanfreq_chunk(shuff, "\000", 1);
 163
 164         total = 0;
 165         while (!feof(fp)) {
 166                 i = 0;
 167                 i = bs_readblock(fp, data, pagesize, flags->rflag);
 168                 total += i;
 169
 170
 171                 /* Aplico BS guardando su resultado + el K en salida */
 172                 bs_solve(data, salida, bs, &k, i);
 173
 174                 /* Le aplico el MTF a salida */
 175                 mtf = jacu_mtf(salida, i+sizeof(Uint32), &z, &z_len);
 176
 177                 /* Guardo el z_len y el Z */
 178                 shuff_scanfreq_chunk(shuff,(char*)&z_len,sizeof(int));
 179                 shuff_scanfreq_chunk(shuff,z,z_len);
 180
 181                 /* Si me lo piden, aplico ZG. */
 182                 if (flags->zflag) {
 183                         Uint32 len;
 184                         unsigned char buff[2];
 185                         Uint32 total_size = i + sizeof(Uint32);
 186                         ZG zg;
 187                         /* Guardo la salida del MTF con ceros agrupados (ZG) */
 188                         zg_init(&zg);
 189                         for (j = 0; j < total_size; ++j)
 190                                 if ((len = zg_group(&zg, buff, mtf[j])))
 191                                         shuff_scanfreq_chunk(shuff, buff, len);
 192
 193                                 /* Flusheo ultimo zgrouping */
 194                                 if ((len = zg_group_finish(&zg,buff)))
 195                                         shuff_scanfreq_chunk(shuff, buff, len);
 196                 } else {
 197                         /* Comprimo la salida del MTF */
 198                         shuff_scanfreq_chunk(shuff,mtf,i+sizeof(Uint32));
 199                 }
 200                 free(mtf);
 201                 free(z);
 202         }
 203
 204         /* Limpiando */
 205         if (fclose(fp)) fprintf(stderr, "Error al cerrar archivo de entrada!\n");
 206         bs_destroy(bs);
 207         free(data);
 208         free(salida);
 209
 210         /* Comprimo con Huffman */
 211         shuff_encode_file(shuff);
 212         if (flags->sflag == 1) shuff_savemodel(shuff);
 213         /* Shutdown Huffman */
 214         shuff_deinit_encoder(shuff);
 215         free(shuff);
 216
 217         /* Muestro bpb */
 218         printf("%s: %.04f bits/byte.\n", dst, vfsize(dst)*8.0f/fsize(src));
 219         return 0;
 220 }
 221
 222 int descomprimir(char *src, char *dst)
 223 {
 224         /* Descomprimo */
 225         FILE *fp_out;
 226         Uint32 block_size = 0, k;
 227         unsigned char *block, *mtf, *orig;
 228         unsigned char *z;
 229         int z_len=0,moredata = 0,decoded = 0;
 230         unsigned char use_zg = 0,retbytes = 0;
 231         HUFF_STATE *shuff;
 232
 233         /* Inicializo el descompresor */
 234         if ((shuff = shuff_init_decoder(src, NULL)) == NULL) return 1;
 235
 236         /* Abrimos el archivo de salida */
 237         fp_out = fopen(dst, "wb");
 238
 239         /* Descomprimo primero que nada el pagesize utilizado para comprimir */
 240         if (!(moredata = shuff_decode_chunk(shuff,(char*)&block_size,sizeof(Uint32),&decoded))) return 1;
 241
 242         /* Descomprimo byte que indica si se usa ZG */
 243         if (!(moredata = shuff_decode_chunk(shuff, &use_zg, 1, &decoded))) return 1;
 244
 245         /* Creo buffers */
 246         block = malloc(block_size*sizeof(unsigned char)+sizeof(Uint32));
 247         orig = malloc(block_size*sizeof(unsigned char));
 248
 249         /* Descomprimimos de a chunks segun convenga */
 250         do {
 251                 if (block_size > 0) {
 252                         /* Descomprimo el Zlen y el Z del MTF*/
 253                         moredata = shuff_decode_chunk(shuff,(char*)&z_len,sizeof(int),&decoded);
 254                         z = malloc(sizeof(unsigned char)*z_len);
 255                         moredata = shuff_decode_chunk(shuff,z,z_len,&decoded);
 256
 257                         /* Veo si se uso Zero Grouping para comprimir */
 258                         if (use_zg) {
 259                                 ZG zg;
 260                                 unsigned char zgbuffer[255];
 261                                 unsigned char zgbyte = 0;
 262                                 int zgmoved = 0;
 263                                 Uint32 zgungrouped = 0;
 264                                 /* Desagrupo bytes hasta completar la pagina or End of Source File */
 265                                 zg_init(&zg);
 266                                 do {
 267                                         /* Levanto un byte zerogrouped y lo paso por el zg_ungroup */
 268                                         zgmoved = 0;
 269                                         moredata = shuff_decode_chunk(shuff,&zgbyte,1,&decoded);
 270                                         retbytes = zg_ungroup(&zg,zgbuffer,zgbyte);
 271                                         /* Muevo del zgbuffer a mi bloque lo que corresponda */
 272                                         while ((zgmoved < retbytes) && (zgungrouped < block_size+sizeof(Uint32))) {
 273                                                 block[zgungrouped++] = zgbuffer[zgmoved++];
 274                                         }
 275                                 } while ((moredata) && (zgungrouped < block_size+sizeof(Uint32)));
 276
 277                                 /* Me fijo si el ultimo byte procesado que me completo la pagina fue un 0 */
 278                                 if (zgbyte == 0) {
 279                                         /* Leo un byte mas (un 0 seguro) y zg_ungroup cambiara su estado */
 280                                         moredata = shuff_decode_chunk(shuff,&zgbyte,1,&decoded);
 281                                         zg_ungroup(&zg,zgbuffer,zgbyte);
 282                                 }
 283
 284                                 /* Normalizo variables para continuar en common code */
 285                                 decoded = zgungrouped;
 286                         }
 287                         else {
 288                                 /* Levanto una salida de MTF */
 289                                 moredata = shuff_decode_chunk(shuff,block,block_size+sizeof(Uint32),&decoded);
 290                         }
 291
 292                         /* Le aplico MTF inverso a la salida de MTF levantada previamente */
 293                         mtf = jacu_mtf_inv(z, block, decoded);
 294
 295                         /* Ya tengo la salida del BS, tonces levanto su K */
 296                         memcpy(&k, mtf, sizeof(Uint32));
 297
 298                         /* Obtengo el chunk original aplicando BS Inverso */
 299                         bs_restore(orig, mtf+sizeof(Uint32), k, decoded - sizeof(Uint32));
 300
 301                         fwrite(orig, decoded - sizeof(Uint32), sizeof(unsigned char), fp_out);
 302                         free(mtf);
 303                         free(z);
 304                 }
 305                 else return 1;
 306         } while (moredata);
 307
 308         /* Close up files and free mem */
 309         fclose(fp_out);
 310         free(block);
 311         free(orig);
 312
 313         /* Shutdown Huffman */
 314         shuff_deinit_decoder(shuff);
 315         free(shuff);
 316         return 0;
 317 }