Listo en teoria Huffman Compresor/Descompresor ByFile y Bychunk. Maniana detallo...

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

diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c
index 6d6de7f613f078696870adb19d79fbbed25151ac..908729be341efc986cff022f2c6aa42f018543b9 100644 (file)
--- a/src/statichuff/statichuff.c
+++ b/src/statichuff/statichuff.c
@@ -2,7 +2,7 @@
 #include "statichuff.h"
 #include <stdlib.h>
 
 #include "statichuff.h"
 #include <stdlib.h>
 

-void putbit(char bit, char restart, char flush, FILE *fp)
+void putbit(char bit, char restart, char flush, VFILE *fp)

 {
        static unsigned long int bits_buffer = 0;
        static unsigned char bits_used = 0;     
 {
        static unsigned long int bits_buffer = 0;
        static unsigned char bits_used = 0;     


@@ -10,7 +10,7 @@ void putbit(char bit, char restart, char flush, FILE *fp)
        /* me obligan a emitir el output */
        if ((flush == 1) && (bits_used > 0)) {
                bits_buffer = bits_buffer << ((sizeof(unsigned long int)*8) - bits_used);
        /* me obligan a emitir el output */
        if ((flush == 1) && (bits_used > 0)) {
                bits_buffer = bits_buffer << ((sizeof(unsigned long int)*8) - bits_used);

-               fwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
+               vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp);

                bits_buffer = 0;
                bits_used = 0;
                return;
                bits_buffer = 0;
                bits_used = 0;
                return;


@@ -27,7 +27,7 @@ void putbit(char bit, char restart, char flush, FILE *fp)
        
        /* lleno el buffer, escribo */
        if (bits_used == 32) {
        
        /* lleno el buffer, escribo */
        if (bits_used == 32) {

-               fwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
+               vfwrite(&bits_buffer,sizeof(unsigned long int),1,fp);

                bits_buffer = 0;
                bits_used = 0;
        }       
                bits_buffer = 0;
                bits_used = 0;
        }       


@@ -55,15 +55,37 @@ int shuff_rescalefreq(t_freq *freqtable)
        t_freq totalfreq = 0;
        
        /* Divido por la mitad las frecuencias, asegurando de no perder */
        t_freq totalfreq = 0;
        
        /* Divido por la mitad las frecuencias, asegurando de no perder */

-       /* frequencias en 1, por ello le sumo 1 antes de partir */

        for (i = 0; i < 256; i++) {             
        for (i = 0; i < 256; i++) {             

-               freqtable[i] = (freqtable[i] << 2) | 1;
+               freqtable[i] = (freqtable[i] >> 2) | 1;

                totalfreq += freqtable[i];
        }
        
        return totalfreq;
 }
 
                totalfreq += freqtable[i];
        }
        
        return totalfreq;
 }
 

+int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
+{      
+       /* Locals */                    
+       int i = 0;
+       unsigned char symbol = 0;       
+               
+       /* Contamos las frecuencias del chunk*/ 
+       for (i = 0; i < chunksize; ++i) {                               
+               symbol = chunk[i];              
+               chunkshuff->freqtable[symbol] += 1;
+               chunkshuff->sumfreq += 1;
+                               
+               /* Si llegue al tope de freq acumulada, halve em */
+               if (chunkshuff->sumfreq == 14930352) 
+                       chunkshuff->sumfreq = shuff_rescalefreq(chunkshuff->freqtable);
+       }
+       
+       /* Dumpeamos el chunk en el temporal homero */
+       fwrite(chunk,chunksize,1,chunkshuff->coderfp);
+               
+       return 1;
+}
+

 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 {
        /* Locals */    
 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 {
        /* Locals */    


@@ -75,7 +97,7 @@ int shuff_scanfreq(char *inputfile, t_freq *freqtable)
        for (i = 0; i < 256; ++i) freqtable[i] = 0;
                
        /* Abrimos el file */
        for (i = 0; i < 256; ++i) freqtable[i] = 0;
                
        /* Abrimos el file */

-       if ((fp = fopen(inputfile,"rb")) == NULL) return 0;
+       if ((fp = fopen(inputfile,"r")) == NULL) return 0;

        while (!feof(fp)) {             
                /* Contamos las frecuencias */          
                symbol = fgetc(fp);
        while (!feof(fp)) {             
                /* Contamos las frecuencias */          
                symbol = fgetc(fp);


@@ -90,7 +112,7 @@ int shuff_scanfreq(char *inputfile, t_freq *freqtable)
        }
        
        fclose(fp);
        }
        
        fclose(fp);

-       return 1;       
+       return 1;

 }
 
 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 }
 
 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)


@@ -116,14 +138,20 @@ SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
        return inputlist;
 }
 
        return inputlist;
 }
 

-SHUFFNODE *shuff_buildtree(SHUFFNODE *list, int listcount)
+SHUFFNODE *shuff_buildtree(t_freq *ftable)

 {
 {

-       SHUFFNODE *lastsymbol = list+(listcount-1);
-       SHUFFNODE *node1,*node2;
+       SHUFFNODE *lastsymbol;
+       SHUFFNODE *node1,*node2,*root;
+       SHUFFNODE *inputlist;
+       int freqcount = 0;

 
 

-       while (lastsymbol > list) {             
+       /* Genero la input list en base a la cual genera el arbol */
+       inputlist = shuff_buildlist(ftable, &freqcount);        
+       lastsymbol = inputlist+(freqcount-1);
+       
+       while (lastsymbol > inputlist) {                

                /* Ordeno la lista por frecuencia descendente */
                /* Ordeno la lista por frecuencia descendente */

-               qsort(list,listcount,sizeof(SHUFFNODE),shuff_compnode);                         
+               qsort(inputlist,freqcount,sizeof(SHUFFNODE),shuff_compnode);                            

                /* Tomo los ultimos dos elementos, generando dos nodos del arbol */
                node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
                node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
                /* Tomo los ultimos dos elementos, generando dos nodos del arbol */
                node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
                node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));


@@ -135,11 +163,18 @@ SHUFFNODE *shuff_buildtree(SHUFFNODE *list, int listcount)
                lastsymbol->freq = node1->freq + node2->freq;
                lastsymbol->lchild = node1;
                lastsymbol->rchild = node2;
                lastsymbol->freq = node1->freq + node2->freq;
                lastsymbol->lchild = node1;
                lastsymbol->rchild = node2;

-               --listcount;
+               --freqcount;

        }
        }

-               
+       
+       /* Copio la raiz para poder liberar la lista sin perderla */
+       root = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
+       shuff_cpynode(root,lastsymbol);
+       
+       /* Free up mem */
+       free(inputlist);
+       

        /* Devuelvo el puntero a la raiz del arbol de huffman */
        /* Devuelvo el puntero a la raiz del arbol de huffman */

-       return lastsymbol;
+       return root;

 }
 
 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 }
 
 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)


@@ -189,25 +224,25 @@ void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
        }
 }
 
        }
 }
 

-
-
-int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, char *outputfile)
+int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)

 {
 {

-       FILE *fpsource,*fpdest;
+       FILE *fpsource;
+       VFILE *fpdest;

        int symbol,i;
        unsigned long int sourcesize;
        char bit;
        SHUFFCODE symbolcode;
                
        int symbol,i;
        unsigned long int sourcesize;
        char bit;
        SHUFFCODE symbolcode;
                

-       /* Abrimos el file */
-       if ((fpsource = fopen(inputfile,"rb")) == NULL) return 0;
-       if ((fpdest = fopen(outputfile,"wb")) == NULL) return 0;
+       /* Abrimos el source y el destino */
+       if (shuff->coderfp != NULL) fclose(shuff->coderfp); /* close bychunk temp file */
+       if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0;
+       if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0;

                
        /* Guardamos el size el archivo original e inputlist como header */
        fseek(fpsource,0,SEEK_END);
        sourcesize = ftell(fpsource);
                
        /* Guardamos el size el archivo original e inputlist como header */
        fseek(fpsource,0,SEEK_END);
        sourcesize = ftell(fpsource);

-       fwrite(&sourcesize,sizeof(unsigned long int),1,fpdest);
-       fwrite(ftable,sizeof(t_freq),256,fpdest);
+       vfwrite(&sourcesize,sizeof(unsigned long int),1,fpdest);
+       vfwrite(shuff->freqtable,sizeof(t_freq),256,fpdest);

        
        /* Encodeo */
        fseek(fpsource,0,SEEK_SET);
        
        /* Encodeo */
        fseek(fpsource,0,SEEK_SET);


@@ -227,37 +262,30 @@ int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, cha
        /* Hacemos un flush de lo que haya quedado en el buffer de salida */
        putbit(0,0,1,fpdest);
        fclose(fpsource);
        /* Hacemos un flush de lo que haya quedado en el buffer de salida */
        putbit(0,0,1,fpdest);
        fclose(fpsource);

-       fclose(fpdest);
+       vfclose(fpdest);

        return 1;       
 }
 
        return 1;       
 }
 

-int shuff_encode_file(char *inputfile, char *outputfile)
+int shuff_encode_file(HUFF_STATE *shuff)

 {
 {

-       /* Locals */
-       t_freq *freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
-       SHUFFNODE *inputlist;
-       SHUFFNODE *codetree;
+       /* Locals */            

        SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
        SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);

-       int freqcount = 0;

        
        

-       /* Armamos la tabla de frecuencias */
-       if (!shuff_scanfreq(inputfile,freqtable)) return 0;
+       /* Veo si debo armar una freqtable o si esta preloaded */
+       if (!shuff->preloadfreq) if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0;

        
        

-       /* Armo el input list y genero el arbol de huffman */
-       inputlist = shuff_buildlist(freqtable, &freqcount);
-       codetree = shuff_buildtree(inputlist,freqcount);
+       /* Genero el arbol de huffman */
+       shuff->codetree = shuff_buildtree(shuff->freqtable);

 
        /* Armo la tabla de codigos prefijos para el encoder */
        shuff_zerocodes(codetable);
 
        /* Armo la tabla de codigos prefijos para el encoder */
        shuff_zerocodes(codetable);

-       shuff_buildcodes(codetable,codetree,0,0);
-       /*shuff_printcodes(codetable,freqtable);*/
+       shuff_buildcodes(codetable,shuff->codetree,0,0);
+       /*shuff_printcodes(codetable,shuff->freqtable);*/

 
        /* Encodeo byte per byte */
 
        /* Encodeo byte per byte */

-       shuff_encode_symbols(freqtable,codetable,inputfile,outputfile);
+       shuff_encode_symbols(shuff,codetable);

        
        

-       /* Free up memory baby yeah */
-       free(freqtable);
-       free(inputlist);        
+       /* Free up memory baby yeah */  

        free(codetable);
        
        return 1;
        free(codetable);
        
        return 1;


@@ -280,51 +308,169 @@ SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer,
        else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 }
 
        else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 }
 

-int shuff_decode_file(char *inputfile, char *outputfile)
+int shuff_decode_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, int *decodedbytes)

 {
 {

-       SHUFFNODE *inputlist;
-       SHUFFNODE *codetree,*currnode;
-       t_freq *ftable = (t_freq*)malloc(sizeof(t_freq)*256);
-       unsigned long int bytesleft,codebuffer;
-       FILE *fpsource;
+       SHUFFNODE *currnode = shuff->codetree;  
+       unsigned short int decoded_symbol;      
+       *decodedbytes = 0;
+       
+       while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) {
+               
+               /* Leo un buffer de 32 bits si es que quedo vacio el anterior */
+               if (shuff->bitsleft == 0) {
+                       if (vfread(&(shuff->codebuffer),sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;
+                       shuff->bitsleft = sizeof(unsigned long int) * 8;
+               }
+               
+               /* Proceso el buffer sacando simbolos till se me agote el buffer, file o chunk */
+               while ((shuff->bitsleft > 0) && (shuff->bytesleft > 0) && (*decodedbytes < chunksize)) {        
+                       currnode = shuff_decode_symbols(currnode,shuff->codebuffer,&(shuff->bitsleft),&decoded_symbol);
+                       /* Si obtuve un symbolo valido lo emito*/
+                       if (decoded_symbol != 256) {                            
+                               chunk[(*decodedbytes)++] = decoded_symbol;
+                               currnode = shuff->codetree;                             
+                               --(shuff->bytesleft);
+                       }                                               
+               }               
+       }
+       
+       if (shuff->bytesleft == 0) return 0;
+       else return 1;
+}
+
+int shuff_decode_file(HUFF_STATE *shuff)
+{      
+       SHUFFNODE *currnode;    
+       unsigned long int codebuffer;   

        FILE *fpdest;
        unsigned short int decoded_symbol;
        FILE *fpdest;
        unsigned short int decoded_symbol;

-       int bitsleft,freqcount = 0;     
+       int bitsleft;   

        
        

-       /* Levanto cuantos bytes decodeo y la freq table */
-       if ((fpsource = fopen(inputfile,"rb")) == NULL) return 0;
-       if ((fpdest = fopen(outputfile,"wb")) == NULL) return 0;
-       fread(&bytesleft,sizeof(unsigned long int),1,fpsource);
-       fread(ftable,sizeof(unsigned long int),256,fpsource);   
-       inputlist = shuff_buildlist(ftable, &freqcount);
-       codetree = shuff_buildtree(inputlist,freqcount);
-       currnode = codetree;
+       /* Comienzo a decodificar, pues la tabla ya la levante en el decinit */
+       if ((fpdest = fopen(shuff->targetfile,"w")) == NULL) return 0;  
+       currnode = shuff->codetree;

        
        

-       while (!feof(fpsource) && (bytesleft > 0)) {
+       while (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0)) {

                
                /* Leo un buffer de 32 bits */
                
                /* Leo un buffer de 32 bits */

-               if (fread(&codebuffer,sizeof(unsigned long int),1,fpsource) != 1) continue;
+               if (vfread(&codebuffer,sizeof(unsigned long int),1,shuff->decoderfp) != 1) continue;

                bitsleft = sizeof(unsigned long int) * 8;
                
                /* Proceso el buffer sacando simbolos hasta que se me agote */
                bitsleft = sizeof(unsigned long int) * 8;
                
                /* Proceso el buffer sacando simbolos hasta que se me agote */

-               while ((bitsleft > 0) && (bytesleft > 0)) {     
+               while ((bitsleft > 0) && (shuff->bytesleft > 0)) {      

                        currnode = shuff_decode_symbols(currnode,codebuffer,&bitsleft,&decoded_symbol);
                        /* Si obtuve un symbolo valido lo emito*/
                        if (decoded_symbol != 256) {
                                fputc(decoded_symbol,fpdest);
                        currnode = shuff_decode_symbols(currnode,codebuffer,&bitsleft,&decoded_symbol);
                        /* Si obtuve un symbolo valido lo emito*/
                        if (decoded_symbol != 256) {
                                fputc(decoded_symbol,fpdest);

-                               currnode = codetree;
-                               --bytesleft;                            
+                               currnode = shuff->codetree;
+                               --(shuff->bytesleft);

                        }                                               
                }               
        }
                        }                                               
                }               
        }

-               
-       fclose(fpsource);
-       fclose(fpdest);
-       
-       /* Free up memory baby yeah */
-       free(ftable);
-       free(inputlist);
+                       
+       /* Close destination */
+       fclose(fpdest); 

        
        return 1;
 }
 
        
        return 1;
 }
 

+HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
+{
+       /* Locals */
+       HUFF_STATE *shuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));                    
+       shuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256); 
+       
+       /* Init fields */
+       shuff->codebuffer = 0;
+       shuff->bitsleft = 0;
+       shuff->coderfp = NULL;
+       shuff->targetfile = NULL;
+       shuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
+       strcpy(shuff->sourcefile,inputfile);
+       if (outputfile != NULL) {
+               shuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
+               strcpy(shuff->targetfile,outputfile);
+       }       
+       
+       /* Levanto cuantos bytes decodeo y la freq table */
+       if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return 0;     
+       vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp);
+       vfread(shuff->freqtable,sizeof(unsigned long int),256,shuff->decoderfp);                
+       shuff->codetree = shuff_buildtree(shuff->freqtable);
+       
+       return shuff;
+}
+
+HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize)
+{
+       /* Locals */
+       HUFF_STATE *fshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));                   
+       int i;
+       
+       /* Inicializo la estructura para trabajar con Huff Static by File */
+       fshuff->coderfp = NULL; 
+       fshuff->decoderfp = NULL;       
+       fshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(inputfile)+1));
+       fshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
+       strcpy(fshuff->sourcefile,inputfile);   
+       strcpy(fshuff->targetfile,outputfile);
+       fshuff->volsize = volsize;
+       fshuff->preloadfreq = 0;        
+       fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
+       for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0;     
+       fshuff->sumfreq = 0;            
+       fshuff->bytesleft = 0;
+       fshuff->codetree = NULL;
+       
+       return fshuff;
+}
+
+HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
+{
+       /* Locals */
+       HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));                   
+       int i;
+       
+       /* Inicializo la estructura para trabajar con Huff Static by Chunks */  
+       cshuff->decoderfp = NULL;
+       cshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(outputfile)+2));
+       cshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));
+       strcpy(cshuff->targetfile,outputfile);  
+       strcpy(cshuff->sourcefile,outputfile);
+       strcat(cshuff->sourcefile,"~"); 
+       cshuff->volsize = volsize;
+       cshuff->preloadfreq = 1;                
+       cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);        
+       for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0;     
+       cshuff->sumfreq = 0;
+       cshuff->bytesleft = 0;
+       cshuff->codetree = NULL;
+       
+       /* Abrimos un archivo temporal para ir tirando los chunks */    
+       if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;     
+
+       return cshuff;
+}
+
+void shuff_deinit_encoder(HUFF_STATE *shuff)
+{
+       /* Libero mallocs y cierro archivos */
+       free(shuff->freqtable); 
+       if (shuff->coderfp != NULL) fclose(shuff->coderfp);
+       if (shuff->preloadfreq == 1) unlink(shuff->sourcefile);
+       if (shuff->targetfile != NULL) free(shuff->targetfile);                 
+       free(shuff->sourcefile);        
+       
+       /* Destruyo recursivamente el arbol de codigos */
+}
+
+void shuff_deinit_decoder(HUFF_STATE *shuff)
+{
+       /* Libero mallocs y cierro archivos */  
+       free(shuff->freqtable);
+       free(shuff->sourcefile);
+       if (shuff->targetfile != NULL) free(shuff->targetfile);
+       if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
+               
+       /* Destruyo recursivamente el arbol de codigos */
+}