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

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

diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c
index 55346bd4bb04dcb32eeec11478f284630f5d8a1b..000f0bab7f1c0820f1555f75e420c2d172e19d32 100644 (file)
--- a/src/statichuff/statichuff.c
+++ b/src/statichuff/statichuff.c
@@ -70,15 +70,17 @@ int shuff_scanfreq_chunk(HUFF_STATE *chunkshuff, char* chunk, int chunksize)
        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 */
@@ -273,7 +275,8 @@ int shuff_encode_file(HUFF_STATE *shuff)
        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);
@@ -393,10 +396,11 @@ HUFF_STATE *shuff_init_decoder(char *inputfile, char *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);
-       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;
@@ -416,11 +420,11 @@ HUFF_STATE *shuff_init_encoder_byfile(char *inputfile, char *outputfile, long vo
        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->bytesleft = 0;
        fshuff->codetree = NULL;
        
        return fshuff;
@@ -432,7 +436,7 @@ HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
        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));
@@ -440,12 +444,12 @@ HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long 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->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;     
@@ -453,14 +457,53 @@ HUFF_STATE *shuff_init_encoder_bychunk(char *outputfile, long volsize)
        return cshuff;
 }
 
+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;       
+}
+
+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;
+}
+
 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 */
 }
@@ -468,8 +511,8 @@ void shuff_deinit_encoder(HUFF_STATE *shuff)
 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);