Se quita repeticion de pagesize en cada bloque comprimido, grabandose una sola vez...

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

diff --git a/src/jacu.c b/src/jacu.c
index fc4188f0f0b2916c7ab52734c0a35bc7bd4709cf..ab4c79d7e8330f74df6d111f0cb728bdb59d7732 100644 (file)
--- a/src/jacu.c
+++ b/src/jacu.c
@@ -1,26 +1,34 @@
 
 

+/* Jacu Team - GPL */

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

+#include "vfile/common.h"

 #include <stdlib.h>
 #include <stdio.h>
 #include <unistd.h>
 
 #include <stdlib.h>
 #include <stdio.h>
 #include <unistd.h>
 

+long get_file_size(const char* filename);
+

 int main(int argc, char* argv[])
 {      
        int cflag = 0;
        int dflag = 0;
        int zflag = 0;
        int tflag = 0;
 int main(int argc, char* argv[])
 {      
        int cflag = 0;
        int dflag = 0;
        int zflag = 0;
        int tflag = 0;

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

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


@@ -31,130 +39,206 @@ int main(int argc, char* argv[])
 
                        case 'z': zflag = 1; 
                                          break; 
 
                        case 'z': zflag = 1; 
                                          break; 

-
-                       case 't': tflag = 1; 
-                                         volumesize = atol(optarg);
-                                         break; 
-
-                       case 'p': pflag = 1; 
-                                         pagesize = atoi(optarg);
+                       
+                       case 'm': mflag = 1;
+                                         staticmodel = optarg;

                                          break; 
                                          break; 

+                       
+                       case 's': sflag = 1;                                      
+                                         break;

 
 

-                       default: fprintf(stderr, "Usage: %s [-cdpt] sourcefile targetfile\n", argv[0]); 
+                       case 't': tflag = 1; 
+                               volumesize = atol(optarg);
+                               break; 
+
+                       case 'q': qflag = 1; 
+                               switch (atoi(optarg))
+                               {
+                                       case 0: pagesize = 1024; /* 1K */
+                                               break;
+                                       case 1: pagesize = 2048; /* 2K */
+                                               break;
+                                       case 2: pagesize = 4096; /* 4K */
+                                               break;
+                                       case 3: pagesize = 8192; /* 8K */
+                                               break;
+                                       case 4: pagesize = 16384; /* 16K */
+                                               break;
+                                       case 5: pagesize = 32768; /* 32K */
+                                               break;
+                                       case 6: pagesize = 65536; /* 64K */
+                                               break;
+                                       case 7: pagesize = 131072; /* 128K */
+                                               break;
+                                       case 8: pagesize = 262144; /* 256K */
+                                               break;
+                                       case 9: pagesize = 524288; /* 512K */
+                                               break;
+                                       default: pagesize = 0; /* error */
+                               }
+                               break; 
+
+                       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 blksize][-t volsize][-m modeldumpfile] source target\n", argv[0]); 

                return (3);
        }
        if ((tflag) && (volumesize <= 0l)) {
                fprintf(stderr,"Error: The volume size must be a non-zero value\n");
                return (4);
        }
                return (3);
        }
        if ((tflag) && (volumesize <= 0l)) {
                fprintf(stderr,"Error: The volume size must be a non-zero value\n");
                return (4);
        }

-       if ((pflag) && (pagesize <= 1u)) {
-               fprintf(stderr,"Error: El tamaño de página debe ser mayor a 1 byte.\n");
+       if ((qflag) && (pagesize <= 1u)) {
+               fprintf(stderr,"Error: El nivel de compresión debe ser entre 0 (menor) y 9 (mayor).\n");

                return (5);
        }
                
        if (cflag == 1) {
                return (5);
        }
                
        if (cflag == 1) {

-               /* Comprimo */
-               /* No me gusta el tmpfile ... es para probar como anda todo junto */
-               FILE *fp, *fp_out;
-               unsigned long int i, j, total, k;
-               char *mtf;
-               char *salida, *data, c;
-               data = malloc(sizeof(char)*pagesize);
-               /* Reservo lugar tambien para guardar el k y el tamaño  */
-               salida = malloc(sizeof(char)*(pagesize)+sizeof(unsigned long int)*2);
+               /* Comprimo */          
+               FILE *fp;
+               Uint32 i, j, total, k;
+               unsigned char *mtf;
+               unsigned char *salida, *data;
+               unsigned char *z;
+               int z_len;
+               
+               /* 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));
+

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

-                               data[i++] = c;
-                               c = fgetc(fp);
+                               data[i++] = fgetc(fp);

                                total++;
                        }
 
                                total++;
                        }
 

-                       /* Hago el BS */
-                       bs_solve(data, salida, bs, &k, i);
+                       /* Saco un EOF que lee de mas */
+                       if (i<pagesize) i--;

 
 

-                       /* Le aplico el MTF, salteo el tamaño del bloque para que no se pierda. */
-                       mtf = jacu_mtf(salida+sizeof(unsigned long int), i+sizeof(unsigned long int));
+                       /* Aplico BS guardando su resultado + el K en salida */
+                       bs_solve(data, salida, bs, &k, i);

 
 

+                       /* Le aplico el MTF a salida */
+                       mtf = jacu_mtf(salida, i+sizeof(Uint32), &z, &z_len);
+                       

                        /* Si me lo piden, aplico ZG. */
                        /* Si me lo piden, aplico ZG. */

-                       if (zflag)
-                       {
+                       if (zflag) {

                                size_t len;
                                char buff[2];
                                ZG zg;
                                zg_init(&zg);
                                size_t len;
                                char buff[2];
                                ZG zg;
                                zg_init(&zg);

+                               /* TODO HACER LO MISMO QUE EN EL ELSE XXX */

                                for (j = 0; j < i; ++j)
                                for (j = 0; j < i; ++j)

-                                       if ((len = zg_group(&zg, buff, mtf[j]))) fwrite(buff, 1, len, fp_out);
-                       }
-                       else
-                       {
-                               for(j=0; j<i; j++)
-                                       fputc(mtf[j], fp_out);
+                                       if ((len = zg_group(&zg, buff, mtf[j]))) shuff_scanfreq_chunk(shuff,buff,len);
+                       } else {                        
+                               /* Comprimo el Z len y el Z */
+                               shuff_scanfreq_chunk(shuff,(char*)&z_len,sizeof(int));
+                               shuff_scanfreq_chunk(shuff,z,z_len);                                                            
+
+                               /* Comprimo la salida del MTF */                                
+                               shuff_scanfreq_chunk(shuff,mtf,i+sizeof(Uint32));                               

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

+                       free(z);

                }
 
                /* 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);
+               /* Comprimo con Huffman */              
+               shuff_encode_file(shuff);
+               if (sflag == 1) shuff_savemodel(shuff);
+               /* Shutdown Huffman */
+               shuff_deinit_encoder(shuff);
+               free(shuff);

 
 

-               /* borro el temporal */
-               unlink("tmp.comp");
-               return i;
+               /* Muestro bpb */
+               printf("Comprimido a %.04f bpb.\n", get_file_size(argv[optind+1])*8.0/get_file_size(argv[optind]));
+               return 0;

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

-               VFILE *fp_in;
-               unsigned long int block_size, k;
-               char *block, *mtf, *orig;
-               int *pos;
-
-               shuff_decode_file(argv[optind], "tmp.comp"); /*argv[optind+1]);*/
-               fp_in = vfopen("tmp.comp", "r", 0);
+               /*FILE *fp_in;*/
+               Uint32 block_size = 0, k;
+               unsigned char *block, *mtf, *orig;
+               unsigned char *z;
+               int z_len,moredata = 0,decoded = 0;
+
+               /* 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;
+               block = malloc(block_size*sizeof(unsigned char)+sizeof(Uint32));
+               orig = malloc(block_size*sizeof(unsigned char));

 
 

-               while (!vfeof(fp_in)) {
-                       block_size = 0;
-                       vfread(&block_size, sizeof(unsigned long int), 1, fp_in);
+               /* Descomprimimos de a chunks segun convenga */
+               do {                    

                        if (block_size > 0) {
                        if (block_size > 0) {

-                               block = malloc((block_size+sizeof(unsigned long int))*sizeof(char));
-                               orig = malloc(block_size*sizeof(char));
-                               vfread(block, block_size, sizeof(char), fp_in);
-
-                               mtf = jacu_mtf_inv(block, /*XXX NO LO TENGO XXX */pos, block_size);
-
-                               memcpy(&k, block, sizeof(unsigned long int));
-
-                               bs_restore(orig, block+sizeof(unsigned long int), k, block_size);
-
-                               fwrite(orig, block_size, sizeof(char), fp_out);
-                               free(block);
-                               free(orig);
+                               /* 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);
+
+                               /* Ya tengo la salida del BS, tonces levanto su K */
+                               memcpy(&k, mtf, sizeof(Uint32));
+
+                               /* Obtengo el chunk original aplicando BS Inverso */
+                               bs_restore(orig, mtf+sizeof(Uint32), k, decoded - sizeof(Uint32));
+
+                               fwrite(orig, decoded - sizeof(Uint32), sizeof(unsigned char), fp_out);

                                free(mtf);
                                free(mtf);

+                               free(z);

                        }
                        }

-               }
-               vfclose(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;
 }

+
+long get_file_size(const char* filename)
+{
+       FILE* file;
+       long  file_size;
+
+       if (!(file = fopen(filename, "ab"))) return -1;
+       file_size = ftell(file);
+       fclose(file);
+       return file_size;
+}