int
main(int argc, char **argv)
{
LST_STree *tree;
LST_String *string = NULL;
LST_StringSet *set;
int i;
if (argc < 2)
test_usage(argv[0]);
/* Create a string set to conveniently hold all our strings: */
set = lst_stringset_new();
/* Add all strings passed on the command line to the set.
* Note that we pass the string length excluding the null
* terminator, libstree handles that generically.
*/
for (i = 1; i < argc; i++)
{
string = lst_string_new(argv[i], 1, strlen(argv[i]));
lst_stringset_add(set, string);
}
/* Create a suffix tree for all strings in the set: */
tree = lst_stree_new(set);
fprintf(stderr, "Tree after all insertions:\n");
lst_debug_print_tree(tree);
/* Free suffix tree: */
lst_stree_free(tree);
return 0;
}
int main(void){
char * str1 = "HTTP/1.";
char * str2 = "HTTP/1.0";
char * str3 = "HTTP/1.1";
char * str4 = "HTTP/1.1 20";
char * str5 = "HTTP/1.1 30";
char * str6 = "HTTP/1.1 40";
char * str7 = "HTTP/1.0 20";
char * str8 = "HTTP/1.0 30";
LST_String * substring1 = lst_string_new(str1, 1, strlen(str1));
LST_String * substring2 = lst_string_new(str2, 1, strlen(str2));
LST_String * substring3 = lst_string_new(str3, 1, strlen(str3));
LST_String * substring4 = lst_string_new(str4, 1, strlen(str4));
LST_String * substring5 = lst_string_new(str5, 1, strlen(str5));
LST_String * substring6 = lst_string_new(str6, 1, strlen(str6));
LST_String * substring7 = lst_string_new(str7, 1, strlen(str7));
LST_String * substring8 = lst_string_new(str8, 1, strlen(str8));
LST_StringSet *substrings = lst_stringset_new();
lst_stringset_add(substrings, substring1);
lst_stringset_add(substrings, substring2);
lst_stringset_add(substrings, substring3);
lst_stringset_add(substrings, substring4);
lst_stringset_add(substrings, substring5);
lst_stringset_add(substrings, substring6);
lst_stringset_add(substrings, substring7);
lst_stringset_add(substrings, substring8);
merge_common_prefix_main(substrings);
if(set) {
printf("set->size : %d\n", set->size);
lst_stringset_foreach(set, string_cb, "\n");
}
return 0;
}
static PyObject*
st_create(PyObject* self, PyObject* args)
{
PyObject *pylist = NULL, *pystring = NULL;
char *cstring = NULL;
int length;
tree_handle_t *handle;
LST_String *string;
/* parse arguments */
if (!PyArg_ParseTuple(args, "O:st_create", &pylist)) return NULL;
/* create the tree */
handle = malloc(sizeof(tree_handle_t));
if (handle == NULL) {
return PyErr_NoMemory();
}
handle->tree = lst_stree_new(NULL);
assert(handle != NULL);
/* fill in the tree and stringset */
PyObject *iterator = PyObject_GetIter(pylist);
handle->string_set = lst_stringset_new();
assert(handle->string_set != NULL);
handle->numstrings = 0;
while((pystring = PyIter_Next(iterator)) != NULL) {
handle->numstrings++;
if (PyString_AsStringAndSize(pystring,&cstring,&length)) {
Py_DECREF(iterator);
lst_stree_free(handle->tree);
lst_stringset_free(handle->string_set);
free(handle);
return NULL;
}
/* LST_String is pointing back at the python string's data.
* python wrapper class keeps a reference to each string added,
* so don't keep the reference here.
*/
Py_DECREF(pystring);
/* this is freed when stringset is freed */
string = (LST_String*)malloc(sizeof(LST_String));
if (string == NULL) {
Py_DECREF(iterator);
lst_stree_free(handle->tree);
lst_stringset_free(handle->string_set);
free(handle);
return PyErr_NoMemory();
}
lst_string_init(string, cstring, 1, length);
lst_stringset_add(handle->string_set, string);
lst_stree_add_string(handle->tree, string);
}
Py_DECREF(iterator);
iterator = NULL;
/* not annotated */
handle->annotated = False;
// lst_debug_print_tree(handle->tree);
/* return pointer to the handle */
return Py_BuildValue("l", (long)handle);
}
int main(int arg, char *argv) {
char * str1 = "HTTP/1.1 index.html";
char * str2 = "HTTP/1.0 aaaaa.html";
char * str3 = "HTTP/1.0 bbbbb.html";
char * str4 = "HTTP/1.1 ccccc.html";
char * str5 = "HTTP/0.9 ddddd.html";
char * str6 = "HTTP/1.1 fffff.html";
char * str7 = "HTTP/0.9 eeeee.html";
LST_String * nbytes1 = lst_string_new(str1, 1, strlen(str1));
LST_String * nbytes2 = lst_string_new(str2, 1, strlen(str2));
LST_String * nbytes3 = lst_string_new(str3, 1, strlen(str3));
LST_String * nbytes4 = lst_string_new(str4, 1, strlen(str4));
LST_String * nbytes5 = lst_string_new(str5, 1, strlen(str5));
LST_String * nbytes6 = lst_string_new(str6, 1, strlen(str6));
LST_String * nbytes7 = lst_string_new(str7, 1, strlen(str7));
LST_StringSet * set = lst_stringset_new();
lst_stringset_add(set, nbytes1);
lst_stringset_add(set, nbytes2);
lst_stringset_add(set, nbytes3);
lst_stringset_add(set, nbytes4);
lst_stringset_add(set, nbytes5);
lst_stringset_add(set, nbytes6);
lst_stringset_add(set, nbytes7);
int first_bytes = 8;
int last_bytes = 10;
int num_bytes = 19;
int gamma_merge = 2;
product_distribution_t * pd = product_distribution_new(set, first_bytes, last_bytes, num_bytes);
/* print pd */
HashTableIterator iterator1;
hash_table_iterate(pd->offset_distribution, &iterator1);
while (hash_table_iter_has_more(&iterator1)){
HashTablePair pair1 = hash_table_iter_next(&iterator1);
int *key1 = (int *) pair1.key;
byte_distribution_t *value1 = (byte_distribution_t *) pair1.value;
HashTableIterator iterator2;
hash_table_iterate(value1->value_frequency, &iterator2);
if (hash_table_num_entries(value1->value_frequency) > gamma_merge) {
continue;
}
printf("offset %d : ", *key1);
while(hash_table_iter_has_more(&iterator2)){
HashTablePair pair2 = hash_table_iter_next(&iterator2);
char *key2 = (char *) pair2.key;
int *value2 = (int *) pair2.value;
//printf("<%c, %d>\t", key2[0], *value2);
if (0 == *key1) {
printf("^%s\t", key2);
} else if (num_bytes - 1 == *key1) {
printf("%s$\t", key2);
} else {
printf("%s\t", key2);
}
}
printf("\n");
}
product_distribution_free(pd);
return 0;
}
