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