int main(int argc, char ** argv) {
unsigned char c;
FILE * file;
int freqs[CHARACTER_SET_SIZE] = {};
if (argc != 2) {
fprintf(stderr, "Usage: huffman <filename>\n");
exit(1); // exit with error code
}
file = fopen(argv[1], "r");
assert(file != NULL);
c = fgetc(file);
while (!feof(file)) {
freqs[c]++;
c = fgetc(file);
}
fclose(file);
huffman_tree * tree = huffman_tree_new(CHARACTER_SET_SIZE);
int i;
for (i = 0; i < CHARACTER_SET_SIZE; i++) {
if (freqs[i] == 0) {
freqs[i] = 1;
}
symbol * char_symbol = symbol_new_char(freqs[i], i);
huffman_tree_insert(tree, i, char_symbol);
}
huffman_tree_merge(tree);
huffman_tree_walk(tree);
huffman_tree_free(tree);
return 0;
}
static void initialise_rbt(GtHuffman *huffman,
const void *distr,
GtDistrFunc distrfunc)
{
GtHuffmanTree *huffptr, GT_UNUSED *huffptr2;
unsigned long i;
bool nodecreated = false;
huffman->numofsymbols = 0;
huffman->rbt_root = gt_rbtree_new(huffman_tree_cmp,
huffman_tree_delete,
NULL);
gt_assert(huffman->rbt_root);
for (i = 0; i < huffman->totalnumofsymbols; i++) {
if (distrfunc(distr, i) > 0) {
huffptr = huffman_tree_new(i, distrfunc(distr, i));
huffptr2 = (GtHuffmanTree*)gt_rbtree_search(huffman->rbt_root,
huffptr,
&nodecreated);
gt_assert(nodecreated && huffptr2);
huffman->numofsymbols++;
}
}
}
static int make_huffman_tree(GtHuffman *huffman)
{
GtHuffmanTree *n1 = NULL,
*n2 = NULL,
GT_UNUSED *n3 = NULL,
*newnode = NULL;
unsigned long i,
symbol;
unsigned long long freq;
int GT_UNUSED deleted;
bool nodecreated = false;
if (huffman->numofsymbols == 0)
huffman->roothuffmantree = NULL;
else if (huffman->numofsymbols == 1) {
huffman->roothuffmantree =
gt_rbtree_root_key(huffman->rbt_root);
huffman->roothuffmantree->code.code = 0;
huffman->roothuffmantree->code.numofbits = 1;
}
else {
for (i = 0; i < huffman->numofsymbols - 1; i++) {
n1 = gt_rbtree_minimum_key(huffman->rbt_root);
n1 = huffman_tree_ref(n1);
deleted = gt_rbtree_erase(huffman->rbt_root,
n1);
gt_assert(deleted == 0);
n2 = gt_rbtree_minimum_key(huffman->rbt_root);
n2 = huffman_tree_ref(n2);
deleted = gt_rbtree_erase(huffman->rbt_root,
n2);
gt_assert(deleted == 0);
symbol = n1->symbol.symbol
< n2->symbol.symbol ? n2->symbol.symbol : n1->symbol.symbol;
freq = n1->symbol.freq + n2->symbol.freq;
newnode = huffman_tree_new(symbol, freq);
if (n1->symbol.freq < n2->symbol.freq) {
newnode->leftchild = n2;
newnode->rightchild = n1;
}
else {
newnode->leftchild = n1;
newnode->rightchild = n2;
}
gt_assert(huffman->rbt_root);
n3 = gt_rbtree_search(huffman->rbt_root, newnode, &nodecreated);
gt_assert(nodecreated && n3);
}
huffman->roothuffmantree = (GtHuffmanTree*)newnode;
huffman->roothuffmantree->code.code = 0;
huffman->roothuffmantree->code.numofbits = 0;
}
return 0;
}
void huffman_tree_init(void)
{
huffman_tree *t0 = huffman_tree_new('H', NULL, NULL);
huffman_tree *t100 = huffman_tree_new('G', NULL, NULL);
huffman_tree *t10100 = huffman_tree_new('C', NULL, NULL);
huffman_tree *t101010 = huffman_tree_new('A', NULL, NULL);
huffman_tree *t1010110 = huffman_tree_new('<', NULL, NULL);
huffman_tree *t10101110 = huffman_tree_new('9', NULL, NULL);
huffman_tree *t10101111 = huffman_tree_new(':', NULL, NULL);
huffman_tree *t1010111 = huffman_tree_new('\0', t10101110, t10101111);
huffman_tree *t101011 = huffman_tree_new('\0', t1010110, t1010111);
huffman_tree *t10101 = huffman_tree_new('\0', t101010, t101011);
huffman_tree *t1010 = huffman_tree_new('\0', t10100, t10101);
huffman_tree *t10110 = huffman_tree_new('B', NULL, NULL);
huffman_tree *t1011100 = huffman_tree_new('>', NULL, NULL);
huffman_tree *t1011101000 = huffman_tree_new('.', NULL, NULL);
huffman_tree *t101110100100 = huffman_tree_new(')', NULL, NULL);
huffman_tree *t101110100101 = huffman_tree_new('I', NULL, NULL);
huffman_tree *t10111010010 = huffman_tree_new('\0', t101110100100, t101110100101);
huffman_tree *t10111010011 = huffman_tree_new('+', NULL, NULL);
huffman_tree *t1011101001 = huffman_tree_new('\0', t10111010010, t10111010011);
huffman_tree *t101110100 = huffman_tree_new('\0', t1011101000, t1011101001);
huffman_tree *t101110101 = huffman_tree_new('6', NULL, NULL);
huffman_tree *t10111010 = huffman_tree_new('\0', t101110100, t101110101);
huffman_tree *t10111011 = huffman_tree_new(';', NULL, NULL);
huffman_tree *t1011101 = huffman_tree_new('\0', t10111010, t10111011);
huffman_tree *t101110 = huffman_tree_new('\0', t1011100, t1011101);
huffman_tree *t101111 = huffman_tree_new('#', NULL, NULL);
huffman_tree *t10111 = huffman_tree_new('\0', t101110, t101111);
huffman_tree *t1011 = huffman_tree_new('\0', t10110, t10111);
huffman_tree *t101 = huffman_tree_new('\0', t1010, t1011);
huffman_tree *t10 = huffman_tree_new('\0', t100, t101);
huffman_tree *t110 = huffman_tree_new('F', NULL, NULL);
huffman_tree *t1110 = huffman_tree_new('E', NULL, NULL);
huffman_tree *t11110 = huffman_tree_new('D', NULL, NULL);
huffman_tree *t1111100000 = huffman_tree_new('2', NULL, NULL);
huffman_tree *t11111000010 = huffman_tree_new(',', NULL, NULL);
huffman_tree *t11111000011 = huffman_tree_new('/', NULL, NULL);
huffman_tree *t1111100001 = huffman_tree_new('\0', t11111000010, t11111000011);
huffman_tree *t111110000 = huffman_tree_new('\0', t1111100000, t1111100001);
huffman_tree *t111110001 = huffman_tree_new('8', NULL, NULL);
huffman_tree *t11111000 = huffman_tree_new('\0', t111110000, t111110001);
huffman_tree *t111110010 = huffman_tree_new('5', NULL, NULL);
huffman_tree *t111110011 = huffman_tree_new('7', NULL, NULL);
huffman_tree *t11111001 = huffman_tree_new('\0', t111110010, t111110011);
huffman_tree *t1111100 = huffman_tree_new('\0', t11111000, t11111001);
huffman_tree *t1111101 = huffman_tree_new('?', NULL, NULL);
huffman_tree *t111110 = huffman_tree_new('\0', t1111100, t1111101);
huffman_tree *t11111100 = huffman_tree_new('=', NULL, NULL);
huffman_tree *t1111110100 = huffman_tree_new('3', NULL, NULL);
huffman_tree *t111111010100 = huffman_tree_new('(', NULL, NULL);
huffman_tree *t111111010101 = huffman_tree_new('*', NULL, NULL);
huffman_tree *t11111101010 = huffman_tree_new('\0', t111111010100, t111111010101);
huffman_tree *t11111101011000 = huffman_tree_new('%', NULL, NULL);
huffman_tree *t11111101011001 = huffman_tree_new('&', NULL, NULL);
huffman_tree *t1111110101100 = huffman_tree_new('\0', t11111101011000, t11111101011001);
huffman_tree *t1111110101101 = huffman_tree_new('\'', NULL, NULL);
huffman_tree *t111111010110 = huffman_tree_new('\0', t1111110101100, t1111110101101);
huffman_tree *t111111010111 = huffman_tree_new('-', NULL, NULL);
huffman_tree *t11111101011 = huffman_tree_new('\0', t111111010110, t111111010111);
huffman_tree *t1111110101 = huffman_tree_new('\0', t11111101010, t11111101011);
huffman_tree *t111111010 = huffman_tree_new('\0', t1111110100, t1111110101);
huffman_tree *t1111110110 = huffman_tree_new('4', NULL, NULL);
huffman_tree *t11111101110 = huffman_tree_new('0', NULL, NULL);
huffman_tree *t11111101111 = huffman_tree_new('1', NULL, NULL);
huffman_tree *t1111110111 = huffman_tree_new('\0', t11111101110, t11111101111);
huffman_tree *t111111011 = huffman_tree_new('\0', t1111110110, t1111110111);
huffman_tree *t11111101 = huffman_tree_new('\0', t111111010, t111111011);
huffman_tree *t1111110 = huffman_tree_new('\0', t11111100, t11111101);
huffman_tree *t1111111 = huffman_tree_new('@', NULL, NULL);
huffman_tree *t111111 = huffman_tree_new('\0', t1111110, t1111111);
huffman_tree *t11111 = huffman_tree_new('\0', t111110, t111111);
huffman_tree *t1111 = huffman_tree_new('\0', t11110, t11111);
huffman_tree *t111 = huffman_tree_new('\0', t1110, t1111);
huffman_tree *t11 = huffman_tree_new('\0', t110, t111);
huffman_tree *t1 = huffman_tree_new('\0', t10, t11);
TREE = huffman_tree_new('\0', t0, t1);
}