Initial commit of proper tree huffman

header.c

@@ -6,24 +6,30 @@
 static int write_entries(struct BIT_BUFFER *bitbuf, struct tree *parent)
 {
 	int bit;
-	if (parent->left) {
+
+	struct tree *left = parent->left;
+	struct tree *right = parent->right;
+
+	if (left && right) {
 		bit = 1;
 		bb_write(bitbuf, &bit, 1);
-		write_entries(bitbuf, parent->left);
+		printf(" 1");
 	}
 
-	if (!parent->left) {
+	if (left)
+		write_entries(bitbuf, left);
+
+	if (right)
+		write_entries(bitbuf, right);
+
+	if (!left && !right) {
 		bit = 0;
-		bb_write(bitbuf, &bit, 1); /* This means now we have 8 bits of char. */
+		bb_write(bitbuf, &bit, 1);
 		bb_writebyte(bitbuf, parent->ch);
+		printf(" 0");
+		printf("%c", parent->ch);
 	}
 
-	if (parent->right) {
-		/* Why should we write this if we can save more bits. */
-		//bit = 0;
-		//bb_write(bitbuf, &bit, 1);
-		bb_writebyte(bitbuf, parent->right->ch);
-	}
 	return 0;
 }
 
@@ -33,37 +39,31 @@ int encode_header(struct BIT_BUFFER *bitbuf, struct tree *parent)
 	return 0;
 }
 
-static int read_entries(struct BIT_BUFFER *bitbuf, struct tree *parent)
+static int read_entry(struct BIT_BUFFER *bitbuf, struct tree *parent)
 {
-	int *bit = bb_read(bitbuf, 1);
+	int *bit;
 
-	if (bit == NULL)
-		return 0;
+	bit = bb_read(bitbuf, 1);
 
 	if (*bit == 1) {
 		parent->left = insert_simple_tree(0, 0);
 		parent->right = insert_simple_tree(0, 0);
-		read_entries(bitbuf, parent->left);
-	}
-
-	/* Reads last left character (the inner left). */
-	if (*bit == 0) {
+		read_entry(bitbuf, parent->left);
+		read_entry(bitbuf, parent->right);
+	} else {
 		int *c = bb_readbyte(bitbuf);
 		parent->ch = *c;
-		free(c);
+		printf("Found from header %c\n", *c);
 	}
 
-	/* Reads right character. */
-	if (*bit == 1) {
-		int *c = bb_readbyte(bitbuf);
-		parent->right->ch = *c;
-		free(c);
-	}
-
-	free(bit);
 	return 0;
 }
 
+static int read_entries(struct BIT_BUFFER *bitbuf, struct tree *parent)
+{
+	read_entry(bitbuf, parent);
+}
+
 int decode_header(struct BIT_BUFFER *bitbuf, struct tree *parent)
 {
 	read_entries(bitbuf, parent);

print_helper.c

@@ -48,7 +48,7 @@ void print_char(unsigned char c)
 
 void print_tree(struct tree *parent)
 {
-	int level = 1;
+	/*int level = 1;
 	printf("Showing tree\n");
 	while (parent->left != NULL ) {
 		if (parent->right) {
@@ -70,5 +70,5 @@ void print_tree(struct tree *parent)
 
 		level += 1;
 		parent = parent->left;
-	}
+	}*/
 }

tree.c

 
+#include <stdbool.h>
 #include <stdlib.h>
 
 #include "huffman.h"
@@ -27,6 +28,49 @@ struct tree *insert_proper_tree(struct tree *left, struct tree *right)
 	return new;
 }
 
+struct tree *merge_smallest(struct tree **forest, int length)
+{
+	struct tree *smallest1 = NULL;
+	struct tree *smallest2 = NULL;
+
+	printf("--\n");
+	int i1 = 0, i2 = 0;
+
+	for (int i = 0; i < length; i++) {
+		if (forest[i] == NULL)
+			continue;
+
+		if (smallest1 == NULL) {
+			smallest1 = forest[i];
+			i1 = i;
+			printf("init Smallest1 %c:%d\n", smallest1->ch, smallest1->freq);
+			continue;
+		} else if (smallest2 == NULL) {
+			smallest2 = forest[i];
+			i2 = i;
+			printf("init Smallest2 %c:%d\n", smallest2->ch, smallest2->freq);
+			continue;
+		}
+
+		printf("comp %c:%d\n", forest[i]->ch, forest[i]->freq);
+		if (forest[i]->freq < smallest1->freq) {
+			smallest1 = forest[i];
+			i1 = i;
+		} else if (forest[i]->freq < smallest2->freq) {
+			smallest2 = forest[i];
+			i2 = i;
+		}
+	}
+
+	forest[i1] = NULL;
+	forest[i2] = insert_proper_tree(smallest1, smallest2);
+	printf("Smallest1 %c:%d\n", smallest1->ch, smallest1->freq);
+	printf("Smallest2 %c:%d\n", smallest2->ch, smallest2->freq);
+	printf("Merged    %c:%d\n", forest[i2]->ch, forest[i2]->freq);
+
+	return forest[i2];
+}
+
 struct tree *create_tree(FILE *file)
 {
 	int freq[MY_BUF] = { };
@@ -58,6 +102,8 @@ struct tree *create_tree(FILE *file)
 		}
 	}
 
+	printf("Created initial tree with %d trees\n", ntrees);
+
 	/* Sort ascending */
 	for (int i = 0; i < ntrees; i++) {
 		for (int j = i+1; j < ntrees; j++) {
@@ -69,20 +115,24 @@ struct tree *create_tree(FILE *file)
 		}
 	}
 
-	int newforestcount = 0;
-	for (int i = 0; i < ntrees-1; i++) {
-		struct tree *small1 = forest[i];
-		struct tree *small2 = forest[i+1];
+	for (int i = 0; i < ntrees; i++) {
+		printf("ch %c freq %d\n", forest[i]->ch, forest[i]->freq);
+	}
 
-		forest[newforestcount] = insert_proper_tree(small1, small2);
-		forest[i+1] = forest[newforestcount];
-		newforestcount++;
+	struct tree *parent = forest[0];
+	for (int i = 0; i < ntrees-1; i++) {
+		parent = merge_smallest(forest, ntrees);
 	}
 
-	/* We have our leader. */
-	struct tree *parent = forest[newforestcount-1];
+	/*
+	 * At this step, we have the leader.
+	 */
+
 	/* This is not needed anymore */
 	free(forest);
+
+
+	printf("Returning tree\n");
 	return parent;
 }
 
@@ -90,41 +140,69 @@ struct tree *create_tree(FILE *file)
  * Encoding/Decodíng
  */
 
-static int write_entry(struct BIT_BUFFER *bitbuf, struct tree *parent,
-					   unsigned char c)
+static unsigned int get_max_height(struct tree *parent)
+{
+	if (!parent)
+		return 0;
+
+	int left_height = get_max_height(parent->left);
+	int right_height = get_max_height(parent->right);
+	return (left_height > right_height) ? left_height + 1 : right_height + 1;
+}
+
+static int get_rev_path(struct tree *parent, unsigned char c, int *buf,
+						unsigned int index)
 {
 	int bit = 0;
-	while (parent != NULL) {
-		struct tree *left, *right;
+	int ret = -1;
 
-		left = parent->left;
-		right = parent->right;
+	if (parent == NULL)
+		return -1;
 
-		if (right->ch == c) {
-			bit = 1;
-			bb_write(bitbuf, &bit, 1);
-			break;
-		} else if (left->left == NULL) {
-			bit = 0;
-			bb_write(bitbuf, &bit, 1);
-			break;
-		}
+	if (parent->ch == c)
+		return index;
 
+	if ((ret = get_rev_path(parent->left, c, buf, index)) >= 0)
 		bit = 0;
-		bb_write(bitbuf, &bit, 1);
+	else if ((ret = get_rev_path(parent->right, c, buf, index)) >= 0)
+		bit = 1;
 
-		parent = left;
+	if (ret >= 0) {
+		index = ret;
+		buf[index] = bit;
+		return index+1;
+	}
+
+	return -1;
+}
+
+static int write_entry(struct BIT_BUFFER *bitbuf, struct tree *parent,
+					   unsigned char c, int *pathbuf)
+{
+	int length = get_rev_path(parent, c, pathbuf, 0);
+	printf("Length %d\n", length);
+
+	/* The path is in reverse. Now write it out in correct order. */
+	for (int i = length-1; i >= 0; i--) {
+		bb_write(bitbuf, &pathbuf[i], 1);
+		printf("i %d Bit %d\n", i, pathbuf[i]);
 	}
-	return 0;
 }
 
 int encode_tree(struct BIT_BUFFER *bitbuf, struct tree *parent, char *buf,
 				size_t length)
 {
+	printf("Bit pos at %d %d\n", ftell(bitbuf->fp), bitbuf->pos);
+
+	int maxheight = get_max_height(parent) - 1;
+	int *pathbuf = calloc(maxheight, sizeof(int));
+
 	for (int i = 0; i < length; i++) {
-		write_entry(bitbuf, parent, buf[i]);
+		write_entry(bitbuf, parent, buf[i], pathbuf);
 	}
 
+	free(pathbuf);
+
 	insert_throwaways(bitbuf->fp, 8 - bitbuf->pos);
 
 	return 0;
@@ -134,34 +212,45 @@ static int read_entries(struct BIT_BUFFER *bitbuf, FILE *out,
 						struct tree *parent)
 {
 	int *bit;
-	int bitsread = 0;
-
-	int totalbits = get_file_size(bitbuf->fp) * 8;
-	int huffmanbits = totalbits - read_throwaways(bitbuf->fp);
 
-	printf("Total bits to parse %d\n", huffmanbits);
+	int lastbyte = get_file_size(bitbuf->fp);
+	int throwaways = read_throwaways(bitbuf->fp);
 
 	struct tree *leaf = parent;
+
+	/* If we compress only one letter, the root node is the node containing
+	 * character. So this needs special care. */
+	#if 0
+	if (!parent->left && !parent->right) {
+		while (ftell(bitbuf->fp) != lastbyte && bitbuf->pos + throwaways < 8) {
+			bit = bb_read(bitbuf, 1);
+			free(bit);
+			fwrite(&parent->ch, 1, 1, out);
+		}
+		return 0;
+	}
+	#endif
+
 	while ((bit = bb_read(bitbuf, 1)) != NULL) {
+		printf("Reading bit %d\n", *bit);
 
-		if (*bit == 1) {
-			fwrite(&leaf->right->ch, 1, 1, out);
-			leaf->right->freq++;
-			leaf = parent; /* Reset */
-		} else {
+		if (*bit == 0)
 			leaf = leaf->left;
-			if (!leaf->left) {
-				fwrite(&leaf->ch, 1, 1, out);
-				leaf->freq++;
-				leaf = parent; /* Reset */
-			}
+		else
+			leaf = leaf->right;
+
+		if (!leaf->left && !leaf->right) {
+			printf("Found from encoded %c\n", leaf->ch);
+			fwrite(&leaf->ch, 1, 1, out);
+			leaf = parent;
 		}
+
 		free(bit);
 
-		if (bitsread == huffmanbits)
+		if (ftell(bitbuf->fp) == lastbyte && bitbuf->pos + throwaways >= 8) {
+			printf("Stopping\n");
 			break;
-
-		bitsread++;
+		}
 	}
 
 	return 0;