Test de Calgary sobre JACU vs GZIP vs BZIP2 done, me falta el pic en mi maquina que...

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

diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c
index 55346bd4bb04dcb32eeec11478f284630f5d8a1b..f339ef0558b67c93b54905a7514f223516a343a6 100644 (file)
--- a/src/statichuff/statichuff.c
+++ b/src/statichuff/statichuff.c
@@ -3,6 +3,7 @@
 #include <stdlib.h>
 #include <string.h>
 
 #include <stdlib.h>
 #include <string.h>
 

+/** Coloca un bit en un buffer statico */

 void putbit(char bit, char restart, char flush, VFILE *fp)
 {
        static unsigned long int bits_buffer = 0;
 void putbit(char bit, char restart, char flush, VFILE *fp)
 {
        static unsigned long int bits_buffer = 0;


@@ -35,6 +36,7 @@ void putbit(char bit, char restart, char flush, VFILE *fp)
        return;
 }
 
        return;
 }
 

+/** Realiza la copia de los datos de un nodo de huffman a otro */

 void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
 {
        node1->symbol = node2->symbol;
 void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
 {
        node1->symbol = node2->symbol;


@@ -43,6 +45,7 @@ void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
        node1->rchild = node2->rchild;  
 }
 
        node1->rchild = node2->rchild;  
 }
 

+/** Realiza una comparacion de dos nodos de huffman */

 int shuff_compnode(const void *node1, const void *node2)
 {      
        if (((SHUFFNODE*)node1)->freq < ((SHUFFNODE*)node2)->freq) return 1;
 int shuff_compnode(const void *node1, const void *node2)
 {      
        if (((SHUFFNODE*)node1)->freq < ((SHUFFNODE*)node2)->freq) return 1;


@@ -50,6 +53,23 @@ int shuff_compnode(const void *node1, const void *node2)
        return 0;
 }
 
        return 0;
 }
 

+/** Destruye un arbol de huffman recursivamente */
+void shuff_destroy_tree(SHUFFNODE *node) {
+       /* Si llegue a una hoja, destruyo y vuelvo */
+       if (node->symbol < 256) {
+               free(node);
+               return;
+       }
+       else {
+               /* Desciendo por izq, luego por derecha y luego libero */
+               shuff_destroy_tree(node->lchild);
+               shuff_destroy_tree(node->rchild);
+               free(node);
+               return;
+       }
+}
+
+/** Reescala las frecuencias de huffman a la mitad */

 int shuff_rescalefreq(t_freq *freqtable)
 { 
        int i;
 int shuff_rescalefreq(t_freq *freqtable)
 { 
        int i;


@@ -64,21 +84,24 @@ int shuff_rescalefreq(t_freq *freqtable)
        return totalfreq;
 }
 
        return totalfreq;
 }
 

+/** Escanea las frecuencias de un chunk de datos */

 int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
 {      
        /* Locals */                    
        int i = 0;
        unsigned char symbol = 0;       
                
 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;
+       /* Contamos las frecuencias del chunk a menos que se use un canonico */                 
+       if (!chunkshuff->canonic) {
+               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);
+                       /* 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 */
        }
        
        /* Dumpeamos el chunk en el temporal homero */


@@ -87,6 +110,7 @@ int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
        return 1;
 }
 
        return 1;
 }
 

+/** Escanea las frecuencias de un archivo y genera el modelo */

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


@@ -116,6 +140,7 @@ int shuff_scanfreq(char *inputfile, t_freq *freqtable)
        return 1;
 }
 
        return 1;
 }
 

+/** Genera un input list que sera utilizada para generar el arbol */

 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 {
        int i,j = 0,nonzero = 0;        
 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 {
        int i,j = 0,nonzero = 0;        


@@ -139,6 +164,7 @@ SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
        return inputlist;
 }
 
        return inputlist;
 }
 

+/** Genera el arbol de huffman en base a la tabla de frecuencias */

 SHUFFNODE *shuff_buildtree(t_freq *ftable)
 {
        SHUFFNODE *lastsymbol;
 SHUFFNODE *shuff_buildtree(t_freq *ftable)
 {
        SHUFFNODE *lastsymbol;


@@ -178,6 +204,7 @@ SHUFFNODE *shuff_buildtree(t_freq *ftable)
        return root;
 }
 
        return root;
 }
 

+/** Imprime los codigos prefijos generados para los symbolos */

 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 {
        int i,j;
 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 {
        int i,j;


@@ -187,7 +214,7 @@ void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
        for (i = 0; i < 256; ++i) {
                if (codetable[i].codelength > 0) {
                        auxcode = codetable[i].code;                    
        for (i = 0; i < 256; ++i) {
                if (codetable[i].codelength > 0) {
                        auxcode = codetable[i].code;                    

-                       printf("Symbol:%i  Freq: %li  Code:",i,freqtable[i]);
+                       printf("Symbol:%i  Freq: %lu  Code:",i,freqtable[i]);

                        for (j = codetable[i].codelength-1; j >= 0; --j) {
                                auxcode = codetable[i].code;                    
                                auxcode = auxcode >> j;
                        for (j = codetable[i].codelength-1; j >= 0; --j) {
                                auxcode = codetable[i].code;                    
                                auxcode = auxcode >> j;


@@ -199,6 +226,7 @@ void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
        }
 }
 
        }
 }
 

+/** Inicializa la tabla de codigos prefijos */

 void shuff_zerocodes(SHUFFCODE *table)
 {
        int i;
 void shuff_zerocodes(SHUFFCODE *table)
 {
        int i;


@@ -210,6 +238,7 @@ void shuff_zerocodes(SHUFFCODE *table)
        }
 }
 
        }
 }
 

+/** Genera la tabla de codigos prefijos en base al árbol de huffman */

 void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 {
        if (node->symbol < 256) {
 void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 {
        if (node->symbol < 256) {


@@ -225,6 +254,7 @@ void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
        }
 }
 
        }
 }
 

+/** Realiza la compresion / encoding efectivo de un archivo */

 int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
 {
        FILE *fpsource;
 int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
 {
        FILE *fpsource;


@@ -235,7 +265,10 @@ int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
        SHUFFCODE symbolcode;
                
        /* Abrimos el source y el destino */
        SHUFFCODE symbolcode;
                
        /* Abrimos el source y el destino */

-       if (shuff->coderfp != NULL) fclose(shuff->coderfp); /* close bychunk temp file */
+       if (shuff->coderfp != NULL) {
+               fclose(shuff->coderfp); /* close bychunk temp file */
+               shuff->coderfp = NULL;
+       }

        if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0;
        if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0;
                
        if ((fpsource = fopen(shuff->sourcefile,"r")) == NULL) return 0;
        if ((fpdest = vfopen(shuff->targetfile,"w",shuff->volsize)) == NULL) return 0;
                


@@ -267,13 +300,15 @@ int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
        return 1;       
 }
 
        return 1;       
 }
 

+/** Prepara las estructuras de datos necesarias para una compresion */

 int shuff_encode_file(HUFF_STATE *shuff)
 {
        /* Locals */            
        SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
        
        /* Veo si debo armar una freqtable o si esta preloaded */
 int shuff_encode_file(HUFF_STATE *shuff)
 {
        /* Locals */            
        SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
        
        /* Veo si debo armar una freqtable o si esta preloaded */

-       if (!shuff->preloadfreq) if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0;
+       if ((!shuff->canonic) && (!shuff->bychunk)) 
+               if (!shuff_scanfreq(shuff->sourcefile,shuff->freqtable)) return 0;

        
        /* Genero el arbol de huffman */
        shuff->codetree = shuff_buildtree(shuff->freqtable);
        
        /* Genero el arbol de huffman */
        shuff->codetree = shuff_buildtree(shuff->freqtable);


@@ -292,6 +327,7 @@ int shuff_encode_file(HUFF_STATE *shuff)
        return 1;
 }
 
        return 1;
 }
 

+/** Decodifica una serie de bits en un symbolo y lo devuelve */

 SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer, 
                                                         int *bitsleft, unsigned short int *symbol)
 {
 SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer, 
                                                         int *bitsleft, unsigned short int *symbol)
 {


@@ -309,7 +345,8 @@ 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_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, int *decodedbytes)
+/** Decodifica chunksize symbolos y los devuelve en un chunk de datos */
+int shuff_decode_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, unsigned long int *decodedbytes)

 {
        SHUFFNODE *currnode = shuff->codetree;  
        unsigned short int decoded_symbol;      
 {
        SHUFFNODE *currnode = shuff->codetree;  
        unsigned short int decoded_symbol;      


@@ -339,6 +376,7 @@ int shuff_decode_chunk(HUFF_STATE *shuff, char *chunk, int chunksize, int *decod
        else return 1;
 }
 
        else return 1;
 }
 

+/** Realiza la descompresión de un archivo comprimido */

 int shuff_decode_file(HUFF_STATE *shuff)
 {      
        SHUFFNODE *currnode;    
 int shuff_decode_file(HUFF_STATE *shuff)
 {      
        SHUFFNODE *currnode;    


@@ -375,6 +413,7 @@ int shuff_decode_file(HUFF_STATE *shuff)
        return 1;
 }
 
        return 1;
 }
 

+/** Inicializa un descompresor de huffman */

 HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
 {
        /* Locals */
 HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
 {
        /* Locals */


@@ -393,15 +432,17 @@ HUFF_STATE *shuff_init_decoder(char *inputfile, char *outputfile)
                strcpy(shuff->targetfile,outputfile);
        }       
        
                strcpy(shuff->targetfile,outputfile);
        }       
        

-       /* Levanto cuantos bytes decodeo y la freq table */
+       /* Levanto cuantos bytes debo decodificar y la freqtable */

        if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return NULL;  
        vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp);
        if ((shuff->decoderfp = vfopen(shuff->sourcefile,"r",0)) == NULL) return NULL;  
        vfread(&(shuff->bytesleft),sizeof(unsigned long int),1,shuff->decoderfp);

-       vfread(shuff->freqtable,sizeof(unsigned long int),256,shuff->decoderfp);                
+       vfread(shuff->freqtable,sizeof(t_freq),256,shuff->decoderfp);           
+       /* Armo el arbol de huffman que uso para decodificar */

        shuff->codetree = shuff_buildtree(shuff->freqtable);
        
        return shuff;
 }
 
        shuff->codetree = shuff_buildtree(shuff->freqtable);
        
        return shuff;
 }
 

+/** Inicializa compresor de huffman por archivo */

 HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize)
 {
        /* Locals */
 HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long volsize)
 {
        /* Locals */


@@ -416,23 +457,24 @@ HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long vo
        strcpy(fshuff->sourcefile,inputfile);   
        strcpy(fshuff->targetfile,outputfile);
        fshuff->volsize = volsize;
        strcpy(fshuff->sourcefile,inputfile);   
        strcpy(fshuff->targetfile,outputfile);
        fshuff->volsize = volsize;

-       fshuff->preloadfreq = 0;        
+       fshuff->bychunk = 0;
+       fshuff->canonic = 0;    

        fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
        for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0;     
        fshuff->sumfreq = 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;
 }
 
        fshuff->codetree = NULL;
        
        return fshuff;
 }
 

+/** Inicializa compresor de huffman de a chunks */

 HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
 {
        /* Locals */
        HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));                   
        int i;
        
 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 */  
+       /* 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));
        cshuff->decoderfp = NULL;
        cshuff->sourcefile = (char*)malloc(sizeof(char)*(strlen(outputfile)+2));
        cshuff->targetfile = (char*)malloc(sizeof(char)*(strlen(outputfile)+1));


@@ -440,12 +482,12 @@ HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
        strcpy(cshuff->sourcefile,outputfile);
        strcat(cshuff->sourcefile,"~"); 
        cshuff->volsize = volsize;
        strcpy(cshuff->sourcefile,outputfile);
        strcat(cshuff->sourcefile,"~"); 
        cshuff->volsize = volsize;

-       cshuff->preloadfreq = 1;                
+       cshuff->bychunk = 1;
+       cshuff->canonic = 0;

        cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);        
        for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0;     
        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;
+       cshuff->sumfreq = 0;    
+       cshuff->codetree = NULL;        

        
        /* Abrimos un archivo temporal para ir tirando los chunks */    
        if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;     
        
        /* Abrimos un archivo temporal para ir tirando los chunks */    
        if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;     


@@ -453,25 +495,70 @@ HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
        return cshuff;
 }
 
        return cshuff;
 }
 

+/** Carga un modelo estadistico para huffman */
+int shuff_loadmodel(HUFF_STATE *shuff, char *modelfile) {
+
+       FILE *fp;
+       
+       if ((shuff) && (shuff->freqtable) && (modelfile)) {
+               /* Cargo el modelo de disco */
+               if ((fp = fopen(modelfile,"r")) == NULL) return 0;
+               if (fread(shuff->freqtable,sizeof(t_freq),256,fp) != 256) return 0;
+               shuff->canonic = 1;
+               if (fp) fclose(fp);             
+               return 1;
+       }       
+       return 0;       
+}
+
+/** Graba un modelo estadístico de huffman */
+int shuff_savemodel(HUFF_STATE *shuff) {
+
+       FILE *fp;
+       char *auxfilename;
+       char *stopchar;
+       
+       if ((shuff) && (shuff->targetfile) && (shuff->freqtable)) {
+               /* Preparo el nombre del archivo con la tabla */
+               auxfilename = (char*)malloc(strlen(shuff->targetfile)+1);               
+               stopchar = strrchr(shuff->targetfile,'.');              
+               strncpy(auxfilename,shuff->targetfile,stopchar - shuff->targetfile);
+               auxfilename[stopchar - shuff->targetfile] = 0;
+               strcat(auxfilename,".ftb");
+               
+               /* Lo creamos y dumpeamos la tabla de frecuencias (modelo) */
+               if ((fp = fopen(auxfilename,"w")) == NULL) return 0;
+               fwrite(shuff->freqtable,sizeof(t_freq),256,fp);
+               if (fp) fclose(fp);
+                               
+               return 1;
+       }       
+       return 0;
+}
+
+/** Desinicializa un compresor de huffman */

 void shuff_deinit_encoder(HUFF_STATE *shuff)
 {
        /* Libero mallocs y cierro archivos */
 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);        
+       if (shuff->freqtable) free(shuff->freqtable);
+       if (shuff->coderfp) fclose(shuff->coderfp);
+       if (shuff->bychunk) unlink(shuff->sourcefile);
+       if (shuff->sourcefile) free(shuff->sourcefile); 
+       if (shuff->targetfile) free(shuff->targetfile);                         

        
        /* Destruyo recursivamente el arbol de codigos */
        
        /* Destruyo recursivamente el arbol de codigos */

+       if (shuff->codetree) shuff_destroy_tree(shuff->codetree);

 }
 
 }
 

+/** Desinicializa un descompresor de huffman */

 void shuff_deinit_decoder(HUFF_STATE *shuff)
 {
        /* Libero mallocs y cierro archivos */  
 void shuff_deinit_decoder(HUFF_STATE *shuff)
 {
        /* Libero mallocs y cierro archivos */  

-       free(shuff->freqtable);
-       free(shuff->sourcefile);
+       if (shuff->freqtable) free(shuff->freqtable);
+       if (shuff->sourcefile != NULL) free(shuff->sourcefile);

        if (shuff->targetfile != NULL) free(shuff->targetfile);
        if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
                
        /* Destruyo recursivamente el arbol de codigos */
        if (shuff->targetfile != NULL) free(shuff->targetfile);
        if (shuff->decoderfp != NULL) vfclose(shuff->decoderfp);
                
        /* Destruyo recursivamente el arbol de codigos */

+       if (shuff->codetree) shuff_destroy_tree(shuff->codetree);

 }
 }