X-Git-Url: https://git.llucax.com/z.facultad/75.06/jacu.git/blobdiff_plain/f3fc0719a4ec62ae33d20f694ee563400b91eb3d..d53cc94464bba53e0be1297ca0cc746901ec113b:/src/statichuff/statichuff.c

diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c
index 79fcdce..f339ef0 100644
--- a/src/statichuff/statichuff.c
+++ b/src/statichuff/statichuff.c
@@ -1,8 +1,9 @@
 
 #include "statichuff.h"
-#include "../vfile/vfile.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;
@@ -35,6 +36,7 @@ void putbit(char bit, char restart, char flush, VFILE *fp)
 	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;
@@ -43,6 +45,7 @@ void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
 	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;
@@ -50,21 +53,64 @@ int shuff_compnode(const void *node1, const void *node2)
 	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;
 	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++) {		
-		freqtable[i] = (freqtable[i] << 2) | 1;
+		freqtable[i] = (freqtable[i] >> 2) | 1;
 		totalfreq += freqtable[i];
 	}
   	
 	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;	
+		
+	/* 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);
+		}
+	}
+	
+	/* Dumpeamos el chunk en el temporal homero */
+	fwrite(chunk,chunksize,1,chunkshuff->coderfp);
+		
+	return 1;
+}
+
+/** Escanea las frecuencias de un archivo y genera el modelo */
 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 {
 	/* Locals */	
@@ -94,6 +140,7 @@ int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 	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;	
@@ -117,14 +164,21 @@ SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 	return inputlist;
 }
 
-SHUFFNODE *shuff_buildtree(SHUFFNODE *list, int listcount)
+/** Genera el arbol de huffman en base a la tabla de frecuencias */
+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 */
-		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));
@@ -136,13 +190,21 @@ SHUFFNODE *shuff_buildtree(SHUFFNODE *list, int listcount)
 		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 */
-	return lastsymbol;
+	return root;
 }
 
+/** Imprime los codigos prefijos generados para los symbolos */
 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 {
 	int i,j;
@@ -152,7 +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;			
-			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;
@@ -164,6 +226,7 @@ void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 	}
 }
 
+/** Inicializa la tabla de codigos prefijos */
 void shuff_zerocodes(SHUFFCODE *table)
 {
 	int i;
@@ -175,6 +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) {
@@ -190,7 +254,8 @@ void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 	}
 }
 
-int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, char *outputfile, long volsize)
+/** Realiza la compresion / encoding efectivo de un archivo */
+int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
 {
 	FILE *fpsource;
 	VFILE *fpdest;
@@ -199,15 +264,19 @@ int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, cha
 	char bit;
 	SHUFFCODE symbolcode;
 		
-	/* Abrimos el file */
-	if ((fpsource = fopen(inputfile,"r")) == NULL) return 0;
-	if ((fpdest = vfopen(outputfile,"w",volsize)) == NULL) return 0;
+	/* Abrimos el source y el destino */
+	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;
 		
 	/* Guardamos el size el archivo original e inputlist como header */
 	fseek(fpsource,0,SEEK_END);
 	sourcesize = ftell(fpsource);
 	vfwrite(&sourcesize,sizeof(unsigned long int),1,fpdest);
-	vfwrite(ftable,sizeof(t_freq),256,fpdest);
+	vfwrite(shuff->freqtable,sizeof(t_freq),256,fpdest);
 	
 	/* Encodeo */
 	fseek(fpsource,0,SEEK_SET);
@@ -231,38 +300,34 @@ int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, cha
 	return 1;	
 }
 
-int shuff_encode_file(char *inputfile, char *outputfile, long volsize)
+/** Prepara las estructuras de datos necesarias para una compresion */
+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);
-	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->canonic) && (!shuff->bychunk)) 
+		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);
-	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 */
-	shuff_encode_symbols(freqtable,codetable,inputfile,outputfile,volsize);
+	shuff_encode_symbols(shuff,codetable);
 	
-	/* Free up memory baby yeah */
-	free(freqtable);
-	free(inputlist);	
+	/* Free up memory baby yeah */	
 	free(codetable);
 	
 	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)
 {
@@ -280,50 +345,220 @@ SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer,
 	else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 }
 
-int shuff_decode_file(char *inputfile, char *outputfile)
+/** 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 *inputlist;
-	SHUFFNODE *codetree,*currnode;
-	t_freq *ftable = (t_freq*)malloc(sizeof(t_freq)*256);
-	unsigned long int bytesleft,codebuffer;
-	VFILE *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;
+}
+
+/** Realiza la descompresión de un archivo comprimido */
+int shuff_decode_file(HUFF_STATE *shuff)
+{	
+	SHUFFNODE *currnode;	
+	unsigned long int codebuffer;	
 	FILE *fpdest;
 	unsigned short int decoded_symbol;
-	int bitsleft,freqcount = 0;	
-	
-	/* Levanto cuantos bytes decodeo y la freq table */
-	if ((fpsource = vfopen(inputfile,"r",0)) == NULL) return 0;
-	if ((fpdest = fopen(outputfile,"w")) == NULL) return 0;
-	vfread(&bytesleft,sizeof(unsigned long int),1,fpsource);
-	vfread(ftable,sizeof(unsigned long int),256,fpsource);	
-	inputlist = shuff_buildlist(ftable, &freqcount);
-	codetree = shuff_buildtree(inputlist,freqcount);
-	currnode = codetree;
-	
-	while (!vfeof(fpsource) && (bytesleft > 0)) {
+	int bitsleft;	
+	
+	/* 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 (!vfeof(shuff->decoderfp) && (shuff->bytesleft > 0)) {
 		
 		/* Leo un buffer de 32 bits */
-		if (vfread(&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 */
-		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 = codetree;
-				--bytesleft;				
+				currnode = shuff->codetree;
+				--(shuff->bytesleft);
 			}						
 		}		
 	}
-		
-	vfclose(fpsource);
-	fclose(fpdest);
-	
-	/* Free up memory baby yeah */
-	free(ftable);
-	free(inputlist);
+			
+	/* Close destination */
+	fclose(fpdest);	
 	
 	return 1;
 }
+
+/** Inicializa un descompresor de huffman */
+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 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(t_freq),256,shuff->decoderfp);		
+	/* Armo el arbol de huffman que uso para decodificar */
+	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 *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->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->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;
+	
+	/* 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->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->codetree = NULL;	
+	
+	/* Abrimos un archivo temporal para ir tirando los chunks */	
+	if ((cshuff->coderfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;	
+
+	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 */
+	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 */
+	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 */	
+	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->codetree) shuff_destroy_tree(shuff->codetree);
+}