X-Git-Url: https://git.llucax.com/z.facultad/75.06/jacu.git/blobdiff_plain/69bddb46f46871d27bcf797cd6fae14430802380..0f208e595dd865267c9a4af16164c9874f8e334e:/src/statichuff/statichuff.c

diff --git a/src/statichuff/statichuff.c b/src/statichuff/statichuff.c
index e608bb2..32b3e8d 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>
 
-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;	
@@ -10,7 +11,7 @@ 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);
-		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;
@@ -27,7 +28,7 @@ void putbit(char bit, char restart, char flush, FILE *fp)
 	
 	/* 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;
 	}	
@@ -42,10 +43,10 @@ void shuff_cpynode(SHUFFNODE *node1, SHUFFNODE *node2)
 	node1->rchild = node2->rchild;	
 }
 
-int shuff_compnode(SHUFFNODE *node1, SHUFFNODE *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;
 }
 
@@ -55,15 +56,34 @@ int shuff_rescalefreq(t_freq *freqtable)
 	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;
 }
 
+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);
+	}		
+		
+	return 1;
+}
+
 int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 {
 	/* Locals */	
@@ -75,7 +95,7 @@ int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 	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);
@@ -90,7 +110,7 @@ int shuff_scanfreq(char *inputfile, t_freq *freqtable)
 	}
 	
 	fclose(fp);
-	return 1;	
+	return 1;
 }
 
 SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
@@ -116,15 +136,20 @@ SHUFFNODE *shuff_buildlist(t_freq *freqtable, int *nonzerofreqs)
 	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;
 
-	/* Ordenamos inicialmente la inputlist para tomar las dos freqs min */
-	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));
@@ -136,11 +161,18 @@ 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;
 }
 
 void shuff_printcodes(SHUFFCODE *codetable,t_freq *freqtable)
@@ -190,25 +222,24 @@ void shuff_buildcodes(SHUFFCODE *table, SHUFFNODE *node, int level, int code)
 	}
 }
 
-
-
-int shuff_encode_symbols(t_freq *ftable, SHUFFCODE *ctable, char* inputfile, char *outputfile) {
-
-	FILE *fpsource,*fpdest;
+int shuff_encode_symbols(HUFF_STATE *shuff, SHUFFCODE *ctable)
+{
+	FILE *fpsource;
+	VFILE *fpdest;
 	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;
+	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);
-	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);
@@ -228,33 +259,114 @@ 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);
-	fclose(fpdest);
+	vfclose(fpdest);
 	return 1;	
 }
 
-int shuff_encode_file(char *inputfile, char *outputfile)
+int shuff_encode_chunk(HUFF_STATE *chunkshuff, char *chunk, int chunksize, int process)
+{
+	int i;
+	
+	if (!process) {
+		/* Scaneo las frecuencias de este chunk */
+		shuff_scanfreq_chunk(chunkshuff,chunk,chunksize);
+	}
+	else {		
+		/* Ya tengo la tabla y el archivo temporal, paso al huffman by file */		
+		return 0;
+	}
+	
+	return 1; 
+}
+
+HUFF_STATE *shuff_init_static_byfile(char *inputfile, char *outputfile, long volsize)
 {
 	/* Locals */
-	t_freq *freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
-	SHUFFNODE *inputlist;
+	HUFF_STATE *fshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));			
+	int i;
+	
+	/* Inicializo la estructura para trabajar con Huff Static by File */
+	fshuff->tmpfp = 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->preloadfreq = 0;
+	fshuff->volsize = volsize;
+	fshuff->sumfreq = 0;
+	fshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
+	for (i = 0; i < 256; ++i) fshuff->freqtable[i] = 0;		
+	
+	return fshuff;
+}
+
+HUFF_STATE *shuff_init_static_bychunk(char *outputfile, long volsize)
+{
+	/* Locals */
+	HUFF_STATE *cshuff = (HUFF_STATE*)malloc(sizeof(HUFF_STATE));		
+	FILE *fp;
+	int i;
+	
+	/* Inicializo la estructura para trabajar con Huff Static by Chunks */
+	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->sumfreq = 0;
+	cshuff->freqtable = (t_freq*)malloc(sizeof(t_freq)*256);
+	for (i = 0; i < 256; ++i) cshuff->freqtable[i] = 0;		
+	
+	/* Abrimos un archivo temporal para ir tirando los chunks */	
+	if ((cshuff->tmpfp = fopen(cshuff->sourcefile,"w")) == NULL) return NULL;	
+
+	return cshuff;
+}
+
+void shuff_deinit_static_byfile(HUFF_STATE *fshuff)
+{
+	/* Libero mallocs y cierro archivos */
+	free(fshuff->freqtable);
+	free(fshuff->sourcefile);
+	free(fshuff->targetfile);		
+}
+
+void shuff_deinit_static_bychunk(HUFF_STATE *cshuff)
+{
+	/* Libero mallocs y cierro archivos */
+	free(cshuff->freqtable);
+	free(cshuff->sourcefile);
+	free(cshuff->targetfile);
+	fclose(cshuff->tmpfp);
+}
+
+int shuff_encode_file(HUFF_STATE *shuff)
+{
+	/* Locals */	
 	SHUFFNODE *codetree;
 	SHUFFCODE *codetable = (SHUFFCODE*)malloc(sizeof(SHUFFCODE)*256);
-	int freqcount = 0;
+	int i;
 	
-	/* 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 */
+	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_printcodes(codetable,shuff->freqtable);*/
 
 	/* Encodeo byte per byte */
-	shuff_encode_symbols(freqtable,codetable,inputfile,outputfile);
+	shuff_encode_symbols(shuff,codetable);
+	
+	/* Free up memory baby yeah */	
+	free(codetable);
+	
+	/* Destruyo el arbol recursivamente TODO */
 	
 	return 1;
 }
@@ -276,30 +388,27 @@ 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)
-{
-	SHUFFNODE *inputlist;
-	SHUFFNODE *codetree,*currnode;
-	t_freq *ftable = (t_freq*)malloc(sizeof(t_freq)*256);
+int shuff_decode_file(HUFF_STATE *shuff)
+{	
+	SHUFFNODE *codetree,*currnode;	
 	unsigned long int bytesleft,codebuffer;
-	FILE *fpsource;
+	VFILE *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);
+	if ((fpsource = vfopen(shuff->sourcefile,"r",0)) == NULL) return 0;
+	if ((fpdest = fopen(shuff->targetfile,"w")) == NULL) return 0;
+	vfread(&bytesleft,sizeof(unsigned long int),1,fpsource);
+	vfread(shuff->freqtable,sizeof(unsigned long int),256,fpsource);		
+	codetree = shuff_buildtree(shuff->freqtable);
 	currnode = codetree;
 	
-	while (!feof(fpsource) && (bytesleft > 0)) {
+	while (!vfeof(fpsource) && (bytesleft > 0)) {
 		
 		/* 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,fpsource) != 1) continue;
 		bitsleft = sizeof(unsigned long int) * 8;
 		
 		/* Proceso el buffer sacando simbolos hasta que se me agote */
@@ -314,53 +423,10 @@ int shuff_decode_file(char *inputfile, char *outputfile)
 		}		
 	}
 		
-	fclose(fpsource);
+	vfclose(fpsource);
 	fclose(fpdest);
 	
-	return 1;
-}
-
-int main(int argc, char* argv[])
-{	
-	int cflag = 0;
-	int dflag = 0;
-	int tflag = 0;
-	long int volumesize = 0;
-	int ch;
-			
-	while ((ch = getopt(argc, argv, "cdt:")) != -1) { 
-		 
-		switch (ch) { 
-			case 'c': cflag = 1; 
-					  break;
-			
-			case 'd': dflag = 1; 
-					  break; 
-			
-			case 't': tflag = 1; 
-					  volumesize = atoi(optarg);					  
-					  break; 
-			
-			default: fprintf(stderr, "Usage: %s [-cdt] sourcefile targetfile\n", argv[0]); 
-					 return(2);
-		}
-	}
-		
-	if ( (argc == 1) || (cflag & dflag) || !(cflag | dflag) || ((argc - optind) < 2) ) {
-		fprintf(stderr, "Usage: %s [-cdt] sourcefile targetfile\n", argv[0]); 
-		if ((tflag == 1) && (volumesize <= 0)) fprintf(stderr,"Error: The volume size must be a non-zero value\n");
-		return (2);		
-	}
-		
-	if (cflag == 1) {
-		/* Comprimo */
-	    return shuff_encode_file(argv[optind],argv[optind+1]);
-	}
+	/* Libero el arbol recursivamente TODO */	
 	
-	if (dflag == 1) { 
-		/* Descomprimo */
-		return shuff_decode_file(argv[optind],argv[optind+1]);
-	}
-		
-	return 0;
+	return 1;
 }