Huffman con capacidad de comprimir chunks y presentando nueva API. El descompresor...

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

diff --git a/src/jacu.c b/src/jacu.c
index 3d18ed3be44af85c1c97ae75db13ec99fe6283ff..ca93ae6aca2d98747a698fdba327c2d3e6388c95 100644 (file)
--- a/src/jacu.c
+++ b/src/jacu.c
@@ -1,24 +1,29 @@
 
 

-#include "statichuff/statichuff.h"

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

+#include "zerogrouping/zerogrouping.h"
+#include "statichuff/statichuff.h"

 #include "vfile/vfile.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 main(int argc, char* argv[])
 {      
        int cflag = 0;
        int dflag = 0;

+       int zflag = 0;

        int tflag = 0;
        int tflag = 0;

-       int pflag = 0;
+       int qflag = 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;
                        

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

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


@@ -27,13 +32,39 @@ int main(int argc, char* argv[])
                        case 'd': dflag = 1; 
                                          break; 
 
                        case 'd': dflag = 1; 
                                          break; 
 

-                       case 't': tflag = 1; 
-                                         volumesize = atol(optarg);
+                       case 'z': zflag = 1; 

                                          break; 
 
                                          break; 
 

-                       case 'p': pflag = 1; 
-                                         pagesize = atoi(optarg);
-                                         break; 
+                       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 [-cdpt] sourcefile targetfile\n", argv[0]); 
                                         return(2);
 
                        default: fprintf(stderr, "Usage: %s [-cdpt] sourcefile targetfile\n", argv[0]); 
                                         return(2);


@@ -48,8 +79,8 @@ int main(int argc, char* argv[])
                fprintf(stderr,"Error: The volume size must be a non-zero value\n");
                return (4);
        }
                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);
        }
                
                return (5);
        }
                


@@ -57,75 +88,162 @@ int main(int argc, char* argv[])
                /* Comprimo */
                /* No me gusta el tmpfile ... es para probar como anda todo junto */
                FILE *fp, *fp_out;
                /* Comprimo */
                /* No me gusta el tmpfile ... es para probar como anda todo junto */
                FILE *fp, *fp_out;

-               unsigned long int i, j, total, k;
+               Uint32 i, j, total, k;

                char *mtf;
                char *salida, *data, c;
                char *mtf;
                char *salida, *data, c;

+               char *z;
+               int z_len;
+

                data = malloc(sizeof(char)*pagesize);
                /* Reservo lugar tambien para guardar el k y el tamaño  */
                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);
+               salida = malloc(sizeof(char)*pagesize+sizeof(Uint32)*2);

                bs = bs_create(pagesize);
 
                fp = fopen(argv[optind], "rb");
                fp_out = fopen("tmp.comp", "wb");
 
                bs = bs_create(pagesize);
 
                fp = fopen(argv[optind], "rb");
                fp_out = fopen("tmp.comp", "wb");
 

-               c = fgetc(fp);
+               /*c = fgetc(fp);*/

                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++;
                        }
 

+                       /* Saco un EOF que lee de mas */
+                       if (i<pagesize) i--;
+

                        /* Hago el BS */
                        bs_solve(data, salida, bs, &k, i);
 
                        /* Hago el BS */
                        bs_solve(data, salida, bs, &k, i);
 

+                       /*printf("BS k=%ld\n", *(Uint32 *)(salida+sizeof(Uint32)));
+                       printf("PageSize = %ld\n", *(Uint32 *)salida);
+
+                       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. */
                        /* 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));
-                       for(j=0; j<i; j++)
-                               fputc(mtf[j], fp_out);
+                       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");
+
+                       }*/
+                       /* Si me lo piden, aplico ZG. */
+                       if (zflag) {
+                               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)
+                                       if ((len = zg_group(&zg, buff, mtf[j]))) fwrite(buff, 1, len, fp_out);
+                       } 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");*/
+                       }

                        free(mtf);
                        free(mtf);

+                       free(z);

                }
 
                }
 

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

                /* Limpiando */
                fclose(fp);
                fclose(fp_out);
                bs_destroy(bs);
 
                /* Comprimo con huffman */
                /* Limpiando */
                fclose(fp);
                fclose(fp_out);
                bs_destroy(bs);
 
                /* Comprimo con huffman */

-               i = shuff_encode_file("tmp.comp",argv[optind+1], volumesize);
+               i = shuff_encode_file("tmp.comp", argv[optind+1], volumesize);

 
                /* borro el temporal */
 
                /* borro el temporal */

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

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

-               VFILE *fp_in;
-               unsigned long int block_size, k;
+               FILE *fp_in;
+               Uint32 block_size, k;

                char *block, *mtf, *orig;
                char *block, *mtf, *orig;

-               int *pos;
+               char *z;
+               int z_len;

 
                shuff_decode_file(argv[optind], "tmp.comp"); /*argv[optind+1]);*/
 
                shuff_decode_file(argv[optind], "tmp.comp"); /*argv[optind+1]);*/

-               fp_in = vfopen("tmp.comp", "r", 0);
+               fp_in = fopen("tmp.comp", "rb");

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

-               while (!vfeof(fp_in)) {
+               while (!feof(fp_in)) {

                        block_size = 0;
                        block_size = 0;

-                       vfread(&block_size, sizeof(unsigned long int), 1, fp_in);
+                       fread(&block_size, sizeof(Uint32), 1, fp_in);

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

-                               block = malloc((block_size+sizeof(unsigned long int))*sizeof(char));
+                               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));
                                orig = malloc(block_size*sizeof(char));

-                               vfread(block, block_size, sizeof(char), fp_in);
+                               fread(block, block_size+sizeof(Uint32), sizeof(char), fp_in);

 
 

-                               mtf = jacu_mtf_inv(block, /*XXX NO LO TENGO XXX */pos, block_size);
+                               /*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));

 
 

-                               memcpy(&k, block, sizeof(unsigned long int));
+                               /*printf("Luego de MTF Inv= [");
+                               {
+                                       int ii;
+                                       for(ii=0; ii<block_size+sizeof(Uint32); ii++)
+                                               printf("(%c)", mtf[ii]);
+                                       printf("]\n");
+                               }*/

 
 

-                               bs_restore(orig, block+sizeof(unsigned long int), k, block_size);
+                               /* Luego de hacer el MTF inverso ya puedo recuperar el k */
+                               memcpy(&k, mtf, sizeof(Uint32));
+
+                               /*printf("Restored : k=%ld\n", k);*/
+                               bs_restore(orig, mtf+sizeof(Uint32), k, block_size);

 
                                fwrite(orig, block_size, sizeof(char), fp_out);
                                free(block);
 
                                fwrite(orig, block_size, sizeof(char), fp_out);
                                free(block);


@@ -133,9 +251,21 @@ int main(int argc, char* argv[])
                                free(mtf);
                        }
                }
                                free(mtf);
                        }
                }

-               vfclose(fp_in);
+               fclose(fp_in);

                fclose(fp_out);
        }
 
        return 0;
 }
                fclose(fp_out);
        }
 
        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;
+}
+