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