enum huffman_error huffman_compute_tree_from_histo(struct huffman_decoder* decoder)
{
/* compute the number of data items in the histogram */
int i;
uint32_t upperweight;
uint32_t lowerweight = 0;
uint32_t sdatacount = 0;
for (i = 0; i < decoder->numcodes; i++)
sdatacount += decoder->datahisto[i];
/* binary search to achieve the optimum encoding */
upperweight = sdatacount * 2;
while (1)
{
/* build a tree using the current weight */
uint32_t curweight = (upperweight + lowerweight) / 2;
int curmaxbits = huffman_build_tree(decoder, sdatacount, curweight);
/* apply binary search here */
if (curmaxbits <= decoder->maxbits)
{
lowerweight = curweight;
/* early out if it worked with the raw weights, or if we're done searching */
if (curweight == sdatacount || (upperweight - lowerweight) <= 1)
break;
}
else
upperweight = curweight;
}
/* assign canonical codes for all nodes based on their code lengths */
return huffman_assign_canonical_codes(decoder);
}
END_TEST
START_TEST(test_table_encode)
{
TreeNode *t = huffman_build_tree("books/holmes.txt");//holmes
ck_assert_int_eq(tree_size(t), 161);
EncodeTable *etab = table_build(t);
ck_assert_msg(etab != NULL, "Encode table should not be NULL.");
char* e_encode = table_bit_encode(etab, 'e');
ck_assert_msg(e_encode != NULL, "Problem finding the encoding for 'e'.");
ck_assert_int_eq(e_encode[0], 0);
ck_assert_int_eq(e_encode[1], 0);
ck_assert_int_eq(e_encode[2], 0);
ck_assert_int_eq(e_encode[3], 0);
ck_assert_int_eq(e_encode[4], -1);
free(e_encode);
char* space_encode = table_bit_encode(etab, ' ');
ck_assert_msg(space_encode != NULL, "Problem finding the encoding for ' '.");
ck_assert_int_eq(space_encode[0], 1);
ck_assert_int_eq(space_encode[1], 0);
ck_assert_int_eq(space_encode[2], -1);
free(space_encode);
free(t);
free(etab);
}
int main (int argc, char *argv[]) {
if (argc != 2) {
usage();
exit(1);
}
char *infile = argv[1];
TreeNode *tree = huffman_build_tree(infile);
if (tree == NULL) {
printf("Could not build the tree!");
usage();
exit(1);
}
EncodeTable *etab = table_build(tree);
if (etab == NULL) {
printf("Could not build the table!");
usage();
exit(1);
}
table_print(etab);
table_free(etab);
tree_free(tree);
return 0;
}
END_TEST
START_TEST(test_table_free)
{
TreeNode *t = huffman_build_tree("books/holmes.txt");//holmes
ck_assert_int_eq(tree_size(t), 161);
EncodeTable *etab = table_build(t);
ck_assert_msg(etab != NULL, "Encode table should not be NULL.");
free(t);
free(etab);
}
END_TEST
//////////////////////////////////////////////////////////////////////
///////////// huffman unit tests
//////////////////////////////////////////////////////////////////////
START_TEST(test_huffman_build_tree)
{
TreeNode *t;
t = huffman_build_tree("books/aladdin.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 115);
free(t);
t = huffman_build_tree("books/holmes.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 161);
free(t);
t = huffman_build_tree("books/iliad.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 131);
free(t);
t = huffman_build_tree("books/KJV.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 147);
free(t);
t = huffman_build_tree("books/newton.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 187);
free(t);
t = huffman_build_tree("books/odyssy.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 161);
free(t);
t = huffman_build_tree("books/poems.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 185);
free(t);
t = huffman_build_tree("books/shakespeare.txt");
ck_assert_msg(t != NULL, "Tree should not be NULL.");
ck_assert_int_eq(tree_size(t), 159);
free(t);
}
END_TEST
START_TEST(test_huffman_find)
{
TreeNode *t = huffman_build_tree("books/holmes.txt");//holmes
ck_assert_int_eq(tree_size(t), 161);
EncodeTable *etab = table_build(t);
ck_assert_msg(etab != NULL, "Encode table should not be NULL.");
char* e_encode = table_bit_encode(etab, 'e');
ck_assert_msg(e_encode != NULL, "Problem finding the encoding for 'e'.");
char c = huffman_find(t, e_encode);
ck_assert_int_eq(c, 'e');
char* b_encode = table_bit_encode(etab, 'b');
c = huffman_find(t, b_encode);
ck_assert_int_eq(c, 'b');
}
/**
* Program entrypoint.
*/
int main(int argc, char **argv)
{
int char_count;
int string_count;
int encoded_size;
unsigned char charmap[256];
int frequencies[256];
huffman_node_t *leaf_nodes[256];
huffman_node_t *code_nodes[256];
huffman_node_t *root;
int symbol_count;
string_list_t *strings;
FILE *input;
FILE *table_output;
FILE *data_output;
int append_byte = -1;
int ignore_case = 0;
const char *input_filename = 0;
const char *charmap_filename = 0;
const char *table_output_filename = 0;
const char *data_output_filename = 0;
const char *table_label = "";
const char *node_label_prefix = "";
const char *string_table_label = "";
const char *string_label_prefix = "";
int generate_string_table = 0;
int verbose = 0;
/* Process arguments. */
{
char *p;
while ((p = *(++argv))) {
if (!strncmp("--", p, 2)) {
const char *opt = &p[2];
if (!strncmp("character-map=", opt, 14)) {
charmap_filename = &opt[14];
} else if (!strncmp("table-output=", opt, 13)) {
table_output_filename = &opt[13];
} else if (!strncmp("data-output=", opt, 12)) {
data_output_filename = &opt[12];
} else if (!strncmp("table-label=", opt, 12)) {
table_label = &opt[12];
} else if (!strncmp("node-label-prefix=", opt, 18)) {
node_label_prefix = &opt[18];
} else if (!strncmp("string-label-prefix=", opt, 20)) {
string_label_prefix = &opt[20];
generate_string_table = 1;
} else if (!strcmp("generate-string-table", opt)) {
generate_string_table = 1;
} else if (!strncmp("string-table-label=", opt, 19)) {
string_table_label = &opt[19];
} else if (!strncmp("append-byte=", opt, 12)) {
append_byte = strtol(&opt[12], 0, 0);
if ((append_byte < 0) || (append_byte >= 256)) {
fprintf(stderr, "huffpuff: --append-byte: value must be in range 0..255\n");
return(-1);
}
} else if (!strcmp("ignore-case", opt)) {
ignore_case = 1;
} else if (!strcmp("verbose", opt)) {
verbose = 1;
} else if (!strcmp("help", opt)) {
help();
} else if (!strcmp("usage", opt)) {
usage();
} else if (!strcmp("version", opt)) {
version();
} else {
fprintf(stderr, "huffpuff: unrecognized option `%s'\n"
"Try `huffpuff --help' or `huffpuff --usage' for more information.\n", p);
return(-1);
}
} else {
input_filename = p;
}
}
}
/* Set default character mapping f(c)=c */
{
int i;
for (i=0; i<256; i++)
charmap[i] = (unsigned char)i;
}
if (charmap_filename) {
if (verbose)
fprintf(stdout, "reading character map\n");
if (!charmap_parse(charmap_filename, charmap)) {
fprintf(stderr, "error: failed to parse character map `%s'\n",
charmap_filename);
return(-1);
}
}
if (input_filename) {
input = fopen(input_filename, "rt");
if (!input) {
fprintf(stderr, "error: failed to open `%s' for reading\n",
input_filename);
return(-1);
}
} else {
input = stdin;
}
/* Read strings to encode. */
if (verbose)
fprintf(stdout, "reading strings\n");
strings = read_strings(input, ignore_case, frequencies, &char_count, &string_count);
fclose(input);
/* Create Huffman leaf nodes. */
if (verbose)
fprintf(stdout, "creating Huffman leaf nodes\n");
symbol_count = 0;
{
int i;
if (append_byte != -1)
frequencies[append_byte] += string_count;
for (i=0; i<256; i++) {
if (frequencies[i] > 0) {
huffman_node_t *node;
node = huffman_create_node(
/*symbol=*/i, /*weight=*/frequencies[i],
/*left=*/NULL, /*right=*/NULL);
leaf_nodes[symbol_count++] = node;
code_nodes[i] = node;
} else {
code_nodes[i] = 0;
}
}
}
if (verbose)
fprintf(stdout, " number of symbols: %d\n", symbol_count);
/* Build the Huffman tree. */
if (verbose)
fprintf(stdout, "Building the Huffman tree\n");
root = huffman_build_tree(leaf_nodes, symbol_count);
/* Huffman-encode strings. */
if (verbose)
fprintf(stdout, "encoding strings\n");
encoded_size = encode_strings(strings, code_nodes, append_byte);
/* Sanity check */
if (verbose)
fprintf(stdout, "verifying output integrity\n");
if (!verify_data_integrity(strings, root)) {
assert(0);
/* Cleanup */
huffman_delete_node(root);
destroy_string_list(strings);
return(-1);
}
/* Prepare output */
if (!table_output_filename) {
table_output_filename = "huffpuff.tab.asm";
}
table_output = fopen(table_output_filename, "wt");
if (!table_output) {
fprintf(stderr, "error: failed to open `%s' for writing\n",
table_output_filename);
/* Cleanup */
huffman_delete_node(root);
destroy_string_list(strings);
return(-1);
}
if (!data_output_filename) {
data_output_filename = "huffpuff.dat.asm";
}
data_output = fopen(data_output_filename, "wt");
if (!data_output) {
fprintf(stderr, "error: failed to open `%s' for writing\n",
data_output_filename);
/* Cleanup */
huffman_delete_node(root);
destroy_string_list(strings);
return(-1);
}
fprintf(data_output, "; Huffman-encoded string data automatically generated by huffpuff.\n");
/* Print the Huffman codes in code length order. */
if (verbose)
fprintf(stdout, "writing Huffman decoder table\n");
fprintf(table_output, "; Huffman decoder table automatically generated by huffpuff.\n");
if (table_label && strlen(table_label))
fprintf(table_output, "%s:\n", table_label);
write_huffman_codes(table_output, root, charmap, node_label_prefix);
fclose(table_output);
if (generate_string_table) {
/* Print string pointer table */
int i;
string_list_t *lst;
if (verbose)
fprintf(stdout, "writing string pointer table\n");
if (string_table_label && strlen(string_table_label))
fprintf(data_output, "%s:\n", string_table_label);
for (i = 0, lst = strings; lst != 0; lst = lst->next, ++i) {
fprintf(data_output, ".dw %sString%d\n",
string_label_prefix, i);
}
}
/* Write the Huffman-encoded strings. */
if (verbose)
fprintf(stdout, "writing encoded string data\n");
write_huffman_strings(data_output, strings, string_label_prefix);
fclose(data_output);
if (verbose)
fprintf(stdout, "compressed size: %d%%\n", (encoded_size*100) / char_count);
/* Cleanup */
huffman_delete_node(root);
destroy_string_list(strings);
return 0;
}
