X-Git-Url: https://git.llucax.com/z.facultad/75.06/jacu.git/blobdiff_plain/c85503d1a07d56f61d4fb85696212118c0c470cd..a4b2f0e262d30c15eeb4f94a23b199a7996979c9:/src/statichuff/statichuff.c?ds=inline

diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c
index 7503511..6d6de7f 100644
--- a/src/statichuff/statichuff.c
+++ b/src/statichuff/statichuff.c
@@ -1,21 +1,40 @@
 
-#include <stdio.h>
+#include "statichuff.h"
+#include <stdlib.h>
 
-typedef unsigned short int t_freq;
-
-typedef struct t_freqnode {
-	unsigned short int symbol;
-	unsigned short int freq;
-	struct t_freqnode *lchild;
-	struct t_freqnode *rchild;
-} HUFFNODE;
+void putbit(char bit, char restart, char flush, FILE *fp)
+{
+	static unsigned long int bits_buffer = 0;
+	static unsigned char bits_used = 0;	
 
-typedef struct t_code {
-	unsigned short int code;
-	unsigned char codelength;
-} CODE;
+	/* 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);
+		bits_buffer = 0;
+		bits_used = 0;
+		return;
+	}	
+	/* me indican que comienza un nuevo output */
+	if (restart) {
+		bits_buffer = 0;
+		bits_used = 0;
+	}			
+	/* inserto el bit en el buffer */	
+	bits_buffer = bits_buffer << 1;
+	bits_buffer |= bit;
+	bits_used++;
+	
+	/* lleno el buffer, escribo */
+	if (bits_used == 32) {
+		fwrite(&bits_buffer,sizeof(unsigned long int),1,fp);
+		bits_buffer = 0;
+		bits_used = 0;
+	}	
+	return;
+}
 
-void cpynode(HUFFNODE *node1, HUFFNODE *node2)
+void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
 {
 	node1->symbol = node2->symbol;
 	node1->freq = node2->freq;
@@ -23,59 +42,36 @@ void cpynode(HUFFNODE *node1, HUFFNODE *node2)
 	node1->rchild = node2->rchild;	
 }
 
-int compnode(HUFFNODE *node1, HUFFNODE *node2)
+int shuff_compnode(const void *node1, const void *node2)
 {	
-	if (node1->freq < node2->freq) return 1;
-	if (node1->freq > node2->freq) return -1;
+	if (((SHUFFNODE*)node1)->freq < ((SHUFFNODE*)node2)->freq) return 1;
+	if (((SHUFFNODE*)node1)->freq > ((SHUFFNODE*)node2)->freq) return -1;
 	return 0;
 }
 
-HUFFNODE *buildlist(t_freq *freqtable, int *nonzerofreqs)
-{
- 	int i,j = 0,nonzero = 0;	
-	HUFFNODE *inputlist;
-	
-	/* Calculo cuantas frequencias > 0 hay y creo la tabla */
-	for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++;
-	inputlist = (HUFFNODE*)malloc(sizeof(HUFFNODE)*nonzero);
-		
-	/* Cargo la inputlist del huffman solo con freqs > 0 */
-	for (i = 0; i < 256; ++i)
-		if (freqtable[i] > 0) {			
-			inputlist[j].symbol = i;
-			inputlist[j].freq = freqtable[i];
-			inputlist[j].lchild = NULL;
-			inputlist[j].rchild = NULL;			
-			j++;
-		}
-		
-	*nonzerofreqs = nonzero;
-	return inputlist;
-}
-
-int rescalefreq(t_freq *freqtable)
+int shuff_rescalefreq(t_freq *freqtable)
 { 
 	int i;
-	int totalfreq = 0;
+	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]+1)/2;
+		freqtable[i] = (freqtable[i] << 2) | 1;
 		totalfreq += freqtable[i];
 	}
   	
 	return totalfreq;
 }
 
-int scanfreq(char *inputfile, t_freq *freqtable)
+int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 {
 	/* Locals */	
 	FILE *fp;
-	int sumfreq = 0,auxsum = 0;
+	t_freq sumfreq = 0;
 	int i,symbol;
 	
-	/* Inicializamos la tabla de frecuencias */
+	/* Inicializamos la tabla de frecuencias */	
 	for (i = 0; i < 256; ++i) freqtable[i] = 0;
 		
 	/* Abrimos el file */
@@ -89,15 +85,64 @@ int scanfreq(char *inputfile, t_freq *freqtable)
 		++sumfreq;
 				
 		/* Si llegue al tope de freq acumulada, halve em */
-		if (sumfreq == 4181) 
-			sumfreq = rescalefreq(freqtable);
+		if (sumfreq == 14930352) 
+			sumfreq = shuff_rescalefreq(freqtable);
 	}
 	
 	fclose(fp);
 	return 1;	
 }
 
-void printcodes(CODE *codetable,t_freq *freqtable)
+SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
+{
+ 	int i,j = 0,nonzero = 0;	
+	SHUFFNODE *inputlist;
+	
+	/* Calculo cuantas frequencias > 0 hay y creo la tabla */
+	for (i = 0; i < 256; ++i) if (freqtable[i] > 0) nonzero++;
+	inputlist = (SHUFFNODE*)malloc(sizeof(SHUFFNODE)*nonzero);
+		
+	/* Cargo la inputlist del huffman solo con freqs > 0 */
+	for (i = 0; i < 256; ++i)
+		if (freqtable[i] > 0) {			
+			inputlist[j].symbol = i;
+			inputlist[j].freq = freqtable[i];
+			inputlist[j].lchild = NULL;
+			inputlist[j].rchild = NULL;			
+			j++;
+		}
+		
+	*nonzerofreqs = nonzero;
+	return inputlist;
+}
+
+SHUFFNODE *shuff_buildtree(SHUFFNODE *list, int listcount)
+{
+	SHUFFNODE *lastsymbol = list+(listcount-1);
+	SHUFFNODE *node1,*node2;
+
+	while (lastsymbol > list) {		
+		/* Ordeno la lista por frecuencia descendente */
+		qsort(list,listcount,sizeof(SHUFFNODE),shuff_compnode);				
+		/* Tomo los ultimos dos elementos, generando dos nodos del arbol */
+		node1 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
+		node2 = (SHUFFNODE*)malloc(sizeof(SHUFFNODE));
+		shuff_cpynode(node1,lastsymbol-1);
+		shuff_cpynode(node2,lastsymbol);		
+		lastsymbol -= 1;
+		/* Nodo ficticio con la suma de las probs y los ptros a childs */
+		lastsymbol->symbol = 256;
+		lastsymbol->freq = node1->freq + node2->freq;
+		lastsymbol->lchild = node1;
+		lastsymbol->rchild = node2;
+		--listcount;
+	}
+		
+	/* Devuelvo el puntero a la raiz del arbol de huffman */
+	return lastsymbol;
+}
+
+void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
 {
 	int i,j;
 	unsigned short int auxcode;
@@ -106,7 +151,7 @@ void printcodes(CODE *codetable,t_freq *freqtable)
 	for (i = 0; i < 256; ++i) {
 		if (codetable[i].codelength > 0) {
 			auxcode = codetable[i].code;			
-			printf("Symbol:%i  Freq: %i  Code:",i,freqtable[i]);
+			printf("Symbol:%i  Freq: %li  Code:",i,freqtable[i]);
 			for (j = codetable[i].codelength-1; j >= 0; --j) {
 				auxcode = codetable[i].code;			
 				auxcode = auxcode >> j;
@@ -118,83 +163,168 @@ void printcodes(CODE *codetable,t_freq *freqtable)
 	}
 }
 
-void zerocodes(CODE *table)
+void shuff_zerocodes(SHUFFCODE *table)
 {
 	int i;
 	
-	/* Inicializo los codigos prefijos */
+	/* Inicializo los codigos prefijos */	
 	for (i = 0; i < 256; ++i) {
 		table[i].code = 0;
 		table[i].codelength = 0;
 	}
 }
 
-void buildcodes(CODE *table, HUFFNODE *node, int level, int code)
+void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 {
 	if (node->symbol < 256) {
 		/* Guardo el codigo en la tabla */
 		table[node->symbol].code = code;
-		table[node->symbol].codelength = level;
-		/*printf("Found symbol %i with freq %i at depth %i\n",node->symbol,node->freq,level);*/
+		table[node->symbol].codelength = level;		
 	}
 	else {
 		code = code << 1;
-		buildcodes(table,node->lchild,level+1,code);
+		shuff_buildcodes(table,node->lchild,level+1,code);
 		code |= 1;
-		buildcodes(table,node->rchild,level+1,code);
+		shuff_buildcodes(table,node->rchild,level+1,code);
 	}
 }
 
-HUFFNODE *buildtree(HUFFNODE *list, int listcount)
-{
-	HUFFNODE *lastsymbol = list+(listcount-1);
-	HUFFNODE *node1,*node2,*fictnode;
 
-	/* Ordenamos inicialmente la inputlist para tomar las dos freqs min */
-	while (lastsymbol > list) {		
-		/* Ordeno la lista por frecuencia descendente */
-		qsort(list,listcount,sizeof(HUFFNODE),compnode);				
-		/* Tomo los ultimos dos elementos, generando dos nodos del arbol */
-		node1 = (HUFFNODE*)malloc(sizeof(HUFFNODE));
-		node2 = (HUFFNODE*)malloc(sizeof(HUFFNODE));
-		cpynode(node1,lastsymbol-1);
-		cpynode(node2,lastsymbol);		
-		lastsymbol -= 1;
-		/* Nodo ficticio con la suma de las probs y los ptros a childs */
-		lastsymbol->symbol = 256;
-		lastsymbol->freq = node1->freq + node2->freq;
-		lastsymbol->lchild = node1;
-		lastsymbol->rchild = node2;
-		--listcount;
-	}
+
+int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, char *outputfile)
+{
+	FILE *fpsource,*fpdest;
+	int symbol,i;
+	unsigned long int sourcesize;
+	char bit;
+	SHUFFCODE symbolcode;
 		
-	/* Devuelvo el puntero a la raiz del arbol de huffman */
-	return lastsymbol;
+	/* Abrimos el file */
+	if ((fpsource = fopen(inputfile,"rb")) == NULL) return 0;
+	if ((fpdest = fopen(outputfile,"wb")) == NULL) return 0;
+		
+	/* 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);
+	
+	/* Encodeo */
+	fseek(fpsource,0,SEEK_SET);
+	while (!feof(fpsource)) {
+		/* Levanto un symbolo (byte) */		
+		symbol = fgetc(fpsource);
+		if (symbol == EOF) continue;
+		
+		/* Cargamos el codigo y lo emitimos */
+		symbolcode = ctable[symbol];
+		for (i = symbolcode.codelength; i > 0; --i) {
+			bit = (symbolcode.code >> (i-1)) & 1;
+			putbit(bit,0,0,fpdest);
+		}		
+	}
+	
+	/* Hacemos un flush de lo que haya quedado en el buffer de salida */
+	putbit(0,0,1,fpdest);
+	fclose(fpsource);
+	fclose(fpdest);
+	return 1;	
 }
 
-int main(int argc, char* argv[])
+int shuff_encode_file(char *inputfile, char *outputfile)
 {
 	/* Locals */
 	t_freq *freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
-	HUFFNODE *inputlist;
-	HUFFNODE *codetree;
-	CODE *codetable = (CODE*)malloc(sizeof(CODE)*256);
-	int freqcount = 0,i;
-	
-	if (argc == 1) return -1;
+	SHUFFNODE *inputlist;
+	SHUFFNODE *codetree;
+	SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
+	int freqcount = 0;
 	
 	/* Armamos la tabla de frecuencias */
-	if (!scanfreq(argv[1],freqtable)) return -1;
+	if (!shuff_scanfreq(inputfile,freqtable)) return 0;
 	
 	/* Armo el input list y genero el arbol de huffman */
-	inputlist = buildlist(freqtable, &freqcount);	
-	codetree = buildtree(inputlist,freqcount);
+	inputlist = shuff_buildlist(freqtable, &freqcount);
+	codetree = shuff_buildtree(inputlist,freqcount);
+
 	/* Armo la tabla de codigos prefijos para el encoder */
-	zerocodes(codetable);
-	buildcodes(codetable,codetree,0,0);
-	printcodes(codetable,freqtable);
-	/*encode(codetable)*/
+	shuff_zerocodes(codetable);
+	shuff_buildcodes(codetable,codetree,0,0);
+	/*shuff_printcodes(codetable,freqtable);*/
+
+	/* Encodeo byte per byte */
+	shuff_encode_symbols(freqtable,codetable,inputfile,outputfile);
 	
-	return 0;
+	/* Free up memory baby yeah */
+	free(freqtable);
+	free(inputlist);	
+	free(codetable);
+	
+	return 1;
+}
 
+SHUFFNODE *shuff_decode_symbols(SHUFFNODE *entrynode, unsigned long int buffer, 
+							 int *bitsleft, unsigned short int *symbol)
+{
+	char bit = 0;
+		
+	/* Levanto el symbolo y si es uno valido, devuelvo */
+	*symbol = entrynode->symbol;
+	if (*symbol != 256) return entrynode;		
+	if (*bitsleft == 0) return entrynode;
+		
+	/* Obtengo otro bit a procesar y me muevo en el arbol */
+	bit = (buffer >> ((*bitsleft)-1)) & 1;	
+	--(*bitsleft);
+	if (bit == 0) return shuff_decode_symbols(entrynode->lchild,buffer,bitsleft,symbol);
+	else return shuff_decode_symbols(entrynode->rchild,buffer,bitsleft,symbol);
 }
+
+int shuff_decode_file(char *inputfile, char *outputfile)
+{
+	SHUFFNODE *inputlist;
+	SHUFFNODE *codetree,*currnode;
+	t_freq *ftable = (t_freq*)malloc(sizeof(t_freq)*256);
+	unsigned long int bytesleft,codebuffer;
+	FILE *fpsource;
+	FILE *fpdest;
+	unsigned short int decoded_symbol;
+	int bitsleft,freqcount = 0;	
+	
+	/* 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;
+	
+	while (!feof(fpsource) && (bytesleft > 0)) {
+		
+		/* Leo un buffer de 32 bits */
+		if (fread(&codebuffer,sizeof(unsigned long int),1,fpsource) != 1) continue;
+		bitsleft = sizeof(unsigned long int) * 8;
+		
+		/* Proceso el buffer sacando simbolos hasta que se me agote */
+		while ((bitsleft > 0) && (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;				
+			}						
+		}		
+	}
+		
+	fclose(fpsource);
+	fclose(fpdest);
+	
+	/* Free up memory baby yeah */
+	free(ftable);
+	free(inputlist);
+	
+	return 1;
+}
+