Intento de poner a andar el ZG.

[z.facultad/75.06/jacu.git] / src / jacu.c

diff --git a/src/jacu.c b/src/jacu.c
index ca93ae6aca2d98747a698fdba327c2d3e6388c95..599d63630d8a06067a2c3d0eb47ee80bce4a7e9b 100644 (file)
--- a/src/jacu.c
+++ b/src/jacu.c
@@ -1,4 +1,5 @@
 
 

+/* Jacu Team - GPL */

 #include "blocksorting/bs.h"
 #include "mtf/mtf.h"
 #include "zerogrouping/zerogrouping.h"
 #include "blocksorting/bs.h"
 #include "mtf/mtf.h"
 #include "zerogrouping/zerogrouping.h"


@@ -18,12 +19,16 @@ int main(int argc, char* argv[])
        int zflag = 0;
        int tflag = 0;
        int qflag = 0;
        int zflag = 0;
        int tflag = 0;
        int qflag = 0;

+       int sflag = 0;
+       int mflag = 0;

        long int volumesize = 0;
        size_t pagesize = 32768; /* 32KB */
        int ch;
        t_BlockSort *bs;
        long int volumesize = 0;
        size_t pagesize = 32768; /* 32KB */
        int ch;
        t_BlockSort *bs;

+       HUFF_STATE *shuff;
+       char *staticmodel = NULL;

                        
                        

-       while ((ch = getopt(argc, argv, "cdzt:q:")) != -1) { 
+       while ((ch = getopt(argc, argv, "scdzm:t:q:")) != -1) { 

                 
                switch (ch) { 
                        case 'c': cflag = 1; 
                 
                switch (ch) { 
                        case 'c': cflag = 1; 


@@ -34,6 +39,13 @@ int main(int argc, char* argv[])
 
                        case 'z': zflag = 1; 
                                          break; 
 
                        case 'z': zflag = 1; 
                                          break; 

+                       
+                       case 'm': mflag = 1;
+                                         staticmodel = optarg;
+                                         break; 
+                       
+                       case 's': sflag = 1;                                      
+                                         break;

 
                        case 't': tflag = 1; 
                                volumesize = atol(optarg);
 
                        case 't': tflag = 1; 
                                volumesize = atol(optarg);


@@ -66,13 +78,13 @@ int main(int argc, char* argv[])
                                }
                                break; 
 
                                }
                                break; 
 

-                       default: fprintf(stderr, "Usage: %s [-cdpt] sourcefile targetfile\n", argv[0]); 
+                       default: fprintf(stderr, "Usage: %s [-cdzs][-q blksize][-t volsize][-m modeldumpfile] source target\n", argv[0]); 

                                         return(2);
                }
        }
                
                                         return(2);
                }
        }
                

-       if ( (argc == 1) || (cflag & dflag) || !(cflag | dflag) || ((argc - optind) < 2) ) {
-               fprintf(stderr, "Usage: %s [-cdt] sourcefile targetfile\n", argv[0]); 
+       if ( (argc == 1) || (cflag & dflag) || !(cflag | dflag) || ((argc - optind) < 2) || (mflag & sflag)) {
+               fprintf(stderr, "Usage: %s [-cdzs][-q compressionquality][-t volsize][-m modeldumpfile] source target\n", argv[0]); 

                return (3);
        }
        if ((tflag) && (volumesize <= 0l)) {
                return (3);
        }
        if ((tflag) && (volumesize <= 0l)) {


@@ -85,24 +97,33 @@ int main(int argc, char* argv[])
        }
                
        if (cflag == 1) {
        }
                
        if (cflag == 1) {

-               /* Comprimo */
-               /* No me gusta el tmpfile ... es para probar como anda todo junto */
-               FILE *fp, *fp_out;
+               /* Comprimo */          
+               FILE *fp;

                Uint32 i, j, total, k;
                Uint32 i, j, total, k;

-               char *mtf;
-               char *salida, *data, c;
-               char *z;
+               unsigned char *mtf;
+               unsigned char *salida, *data;
+               unsigned char *z;

                int z_len;
                int z_len;

-
-               data = malloc(sizeof(char)*pagesize);
-               /* Reservo lugar tambien para guardar el k y el tamaño  */
-               salida = malloc(sizeof(char)*pagesize+sizeof(Uint32)*2);
+               
+               /* Preparo el compresor huffman */
+               if ((shuff = shuff_init_encoder_bychunk(argv[optind+1],volumesize*1024)) == NULL) return 1;
+               if (mflag == 1) shuff_loadmodel(shuff,staticmodel);
+               
+               /* Preparo el BS alocando mem para la Salida: V(vector) + K(colnum) */
+               data = malloc(sizeof(unsigned char)*pagesize);
+               salida = malloc(sizeof(unsigned char)*pagesize+sizeof(Uint32));

                bs = bs_create(pagesize);
 
                bs = bs_create(pagesize);
 

+               /* Abrimos el archivo a comprimir y encodeamos bloques */

                fp = fopen(argv[optind], "rb");
                fp = fopen(argv[optind], "rb");

-               fp_out = fopen("tmp.comp", "wb");

 
 

-               /*c = fgetc(fp);*/
+               /* Guardamos el pagesize como header (huffencoded) */
+               shuff_scanfreq_chunk(shuff,(char*)&pagesize,sizeof(size_t));
+
+               /* Guardamos cabecera para indicar si usamos ZG (huffencoded) */
+               if (zflag) shuff_scanfreq_chunk(shuff, "\001", 1);
+               else       shuff_scanfreq_chunk(shuff, "\000", 1);
+

                total = 0;
                while (!feof(fp)) {
                        i = 0;
                total = 0;
                while (!feof(fp)) {
                        i = 0;


@@ -114,145 +135,110 @@ int main(int argc, char* argv[])
                        /* Saco un EOF que lee de mas */
                        if (i<pagesize) i--;
 
                        /* Saco un EOF que lee de mas */
                        if (i<pagesize) i--;
 

-                       /* Hago el BS */
+                       /* Aplico BS guardando su resultado + el K en salida */

                        bs_solve(data, salida, bs, &k, i);
 
                        bs_solve(data, salida, bs, &k, i);
 

-                       /*printf("BS k=%ld\n", *(Uint32 *)(salida+sizeof(Uint32)));
-                       printf("PageSize = %ld\n", *(Uint32 *)salida);
+                       /* Le aplico el MTF a salida */
+                       mtf = jacu_mtf(salida, i+sizeof(Uint32), &z, &z_len);

 
 

-                       printf("Antes de MTF = %ld [", i);
-                       {
-                               int ii;
-                               for(ii=0; ii<(i+sizeof(Uint32)); ii++)
-                                       printf("(%c)", salida[ii+sizeof(Uint32)]);
-                               printf("]\n");
-                       }*/
-                       /* Le aplico el MTF, salteo el tamaño del bloque para que no se pierda. */
-                       mtf = jacu_mtf(salida+sizeof(Uint32), i+sizeof(Uint32), &z, &z_len);
-/*
-                       printf("MTF Z (len=%d) = [", z_len);
-                       {
-                               int ii;
-                               for(ii=0; ii<z_len; ii++)
-                                       printf("(%c)", z[ii]);
-                               printf("]\n");
-
-                       }*/
+                       /* Guardo el z_len y el Z */
+                       shuff_scanfreq_chunk(shuff,(char*)&z_len,sizeof(int));
+                       shuff_scanfreq_chunk(shuff,z,z_len);
+                       

                        /* Si me lo piden, aplico ZG. */
                        if (zflag) {
                                size_t len;
                                char buff[2];
                        /* Si me lo piden, aplico ZG. */
                        if (zflag) {
                                size_t len;
                                char buff[2];

+                               Uint32 total_size = i + sizeof(Uint32);

                                ZG zg;
                                zg_init(&zg);
                                ZG zg;
                                zg_init(&zg);

-                               /* TODO HACER LO MISMO QUE EN EL ELSE XXX */
-                               for (j = 0; j < i; ++j)
-                                       if ((len = zg_group(&zg, buff, mtf[j]))) fwrite(buff, 1, len, fp_out);
+                               /* Guardo la salida del MTF con ceros agrupados (ZG) */
+                               for (j = 0; j < total_size; ++j)
+                                       if ((len = zg_group(&zg, buff, mtf[j])))
+                                               shuff_scanfreq_chunk(shuff, buff, len);

                        } else {
                        } else {

-                               /* Guardo el PageSize */
-                               fwrite(salida, sizeof(Uint32), 1, fp_out);
-
-                               /* Guardo el Z len y el Z */
-                               fwrite(&z_len, sizeof(int), 1, fp_out);
-                               fwrite(z, z_len, sizeof(char), fp_out);
-
-                               /* Guardo la salida del MTF */
-                               /*printf("Despues de MTF : [");*/
-                               for(j=0; j<(i+sizeof(Uint32)); j++) {
-                                       fputc(mtf[j], fp_out);
-                       /*              putchar('(');
-                                       fputc(mtf[j], stdout);
-                                       putchar(')');*/
-                               }
-                               /*printf("]\n");*/
+                               /* Comprimo la salida del MTF */                                
+                               shuff_scanfreq_chunk(shuff,mtf,i+sizeof(Uint32));

                        }
                        free(mtf);
                        free(z);
                }
 
                        }
                        free(mtf);
                        free(z);
                }
 

-               printf("Total Bytes read = %ld\n", total);

                /* Limpiando */
                /* Limpiando */

-               fclose(fp);
-               fclose(fp_out);
+               fclose(fp);             

                bs_destroy(bs);
 
                bs_destroy(bs);
 

-               /* Comprimo con huffman */
-               i = shuff_encode_file("tmp.comp", argv[optind+1], volumesize);
-
-               /* borro el temporal */
-               remove("tmp.comp");
+               /* Comprimo con Huffman */              
+               shuff_encode_file(shuff);
+               if (sflag == 1) shuff_savemodel(shuff);
+               /* Shutdown Huffman */
+               shuff_deinit_encoder(shuff);
+               free(shuff);

 
                /* Muestro bpb */
                printf("Comprimido a %.04f bpb.\n", get_file_size(argv[optind+1])*8.0/get_file_size(argv[optind]));
 
                /* Muestro bpb */
                printf("Comprimido a %.04f bpb.\n", get_file_size(argv[optind+1])*8.0/get_file_size(argv[optind]));

-               return i;
+               return 0;

        }
        
        if (dflag == 1) { 
                /* Descomprimo */
                FILE *fp_out;
        }
        
        if (dflag == 1) { 
                /* Descomprimo */
                FILE *fp_out;

-               FILE *fp_in;
-               Uint32 block_size, k;
-               char *block, *mtf, *orig;
-               char *z;
-               int z_len;
-
-               shuff_decode_file(argv[optind], "tmp.comp"); /*argv[optind+1]);*/
-               fp_in = fopen("tmp.comp", "rb");
+               /*FILE *fp_in;*/
+               Uint32 block_size = 0, k;
+               unsigned char *block, *mtf, *orig;
+               unsigned char *z;
+               int z_len,moredata = 0,decoded = 0;
+               char use_zg;
+
+               /* Inicializo el descompresor */
+               if ((shuff = shuff_init_decoder(argv[optind],NULL)) == NULL) return 1;
+                               
+               /* Abrimos el archivo de salida */

                fp_out = fopen(argv[optind+1], "wb");
                fp_out = fopen(argv[optind+1], "wb");

+               
+               /* Descomprimo primero que nada el pagesize utilizado para comprimir */
+               if (!(moredata = shuff_decode_chunk(shuff,(char*)&block_size,sizeof(Uint32),&decoded))) return 1;

 
 

-               while (!feof(fp_in)) {
-                       block_size = 0;
-                       fread(&block_size, sizeof(Uint32), 1, fp_in);
-                       if (block_size > 0) {
-                               printf("PageSize = %ld\n", block_size);
-                               fread(&z_len, sizeof(int), 1, fp_in);
-                               z = malloc(sizeof(char)*z_len);
-                               fread(z, z_len, sizeof(char), fp_in);
-
-                               /*printf("MTF Z (len=%d) = [", z_len);
-                               {
-                                       int ii;
-                                       for(ii=0; ii<z_len; ii++)
-                                               printf("(%c)", z[ii]);
-                                       printf("]\n");
-       
-                               }*/
-                               
-                               block = malloc(block_size*sizeof(char)+sizeof(Uint32));
-                               orig = malloc(block_size*sizeof(char));
-                               fread(block, block_size+sizeof(Uint32), sizeof(char), fp_in);
+               /* Descomprimo byte que indica si se usa ZG */
+               if (!(moredata = shuff_decode_chunk(shuff, &use_zg, 1, &decoded))) return 1;

 
 

-                               /*printf("Antes MTF_inv = [");
-                               {
-                                       int ii;
-                                       for(ii=0; ii<block_size+sizeof(Uint32); ii++)
-                                               printf("(%c)", block[ii]);
-                                       printf("]\n");
-                               }*/
-                               /* Hago el MTF inverso */
-                               mtf = jacu_mtf_inv(z, block, block_size*sizeof(char)+sizeof(Uint32));
+               /* Creo buffers */
+               block = malloc(block_size*sizeof(unsigned char)+sizeof(Uint32));
+               orig = malloc(block_size*sizeof(unsigned char));

 
 

-                               /*printf("Luego de MTF Inv= [");
-                               {
-                                       int ii;
-                                       for(ii=0; ii<block_size+sizeof(Uint32); ii++)
-                                               printf("(%c)", mtf[ii]);
-                                       printf("]\n");
-                               }*/
+               /* Descomprimimos de a chunks segun convenga */
+               do {                    
+                       if (block_size > 0) {
+                               /* Descomprimo el Zlen y el Z del MTF */
+                               moredata = shuff_decode_chunk(shuff,(char*)&z_len,sizeof(int),&decoded);                                
+                               z = malloc(sizeof(unsigned char)*z_len);
+                               moredata = shuff_decode_chunk(shuff,z,z_len,&decoded);
+                               
+                               /* Levanto una salida de MTF y le aplico MTF Inverso */
+                               moredata = shuff_decode_chunk(shuff,block,block_size+sizeof(Uint32),&decoded);                          
+                               mtf = jacu_mtf_inv(z, block, decoded);

 
 

-                               /* Luego de hacer el MTF inverso ya puedo recuperar el k */
+                               /* Ya tengo la salida del BS, tonces levanto su K */

                                memcpy(&k, mtf, sizeof(Uint32));
 
                                memcpy(&k, mtf, sizeof(Uint32));
 

-                               /*printf("Restored : k=%ld\n", k);*/
-                               bs_restore(orig, mtf+sizeof(Uint32), k, block_size);
+                               /* Obtengo el chunk original aplicando BS Inverso */
+                               bs_restore(orig, mtf+sizeof(Uint32), k, decoded - sizeof(Uint32));

 
 

-                               fwrite(orig, block_size, sizeof(char), fp_out);
-                               free(block);
-                               free(orig);
+                               fwrite(orig, decoded - sizeof(Uint32), sizeof(unsigned char), fp_out);

                                free(mtf);
                                free(mtf);

+                               free(z);

                        }
                        }

-               }
-               fclose(fp_in);
+                       else return 1;
+               } while (moredata);             
+               
+               /* Close up files and free mem */

                fclose(fp_out);
                fclose(fp_out);

+               free(block);
+               free(orig);
+               
+               /* Shutdown Huffman */
+               shuff_deinit_decoder(shuff);
+               free(shuff);

        }
 
        return 0;
        }
 
        return 0;


@@ -268,4 +254,3 @@ long get_file_size(const char* filename)
        fclose(file);
        return file_size;
 }
        fclose(file);
        return file_size;
 }

-