void
lst_stringset_free(LST_StringSet *set)
{
LST_String *string;
if (!set)
return;
while (set->members.lh_first)
{
string = set->members.lh_first;
LIST_REMOVE(set->members.lh_first, set);
lst_string_free(string);
}
free(set);
}
/* callback for tree traversal.
* annotates nodes to keep track of which strings
* have leaves in its subtree.
*/
int annotate(LST_Node *node, params_t *p)
{
annotation_t *annotation;
annotation_t *child_annotation;
LST_Edge *edge;
LST_Node *child;
int i;
/* allocate annotation for this node */
if (node->annotation == NULL) {
node->annotation = malloc(sizeof(annotation_t));
assert(node->annotation != NULL);
annotation = (annotation_t*)node->annotation;
annotation->strings = (int*)calloc((p->tree->numstrings), sizeof(int));
assert(annotation->strings != NULL);
} else {
/* reallocate in case number of strings has changed */
annotation = (annotation_t*)node->annotation;
annotation->strings = realloc(annotation->strings,
p->tree->numstrings * sizeof(int));
assert(annotation->strings != NULL);
}
/* initialize the strings annotation */
bzero(annotation->strings, (p->tree->numstrings)*sizeof(int));
if (lst_node_is_leaf(node)) {
annotation->strings[lst_stree_get_string_index(p->tree->tree, node->up_edge->range.string)] = 1;
return True;
}
/* incoorporate children's annotations */
for (edge = node->kids.lh_first; edge; edge = edge->siblings.le_next) {
child = edge->dst_node;
assert(child->annotation != NULL);
child_annotation = (annotation_t*)child->annotation;
for(i=0; i<p->tree->numstrings; i++) {
// annotation->strings[i] = annotation->strings[i] || child_annotation->strings[i];
annotation->strings[i] += child_annotation->strings[i];
}
}
/* how many strings have this substring? */
int string_count = 0;
int occ_count = 0;
// LST_String *s = lst_node_get_string(node, 1000);
// printf("%s: ", lst_string_print(s));
for(i=0; i<p->tree->numstrings; i++) {
// printf("%d, ", annotation->strings[i]);
if(annotation->strings[i]) {
occ_count += annotation->strings[i];
string_count++;
}
}
// printf(": %d %d\n", occ_count, string_count);
/* if this node's substring occurs in min_occ strings,
* and is at least min_len long, report it.
*/
if (string_count >= p->min_occ) {
int len = lst_node_get_string_length(node);
if (len >= p->min_len) {
/* add a dictionary mapping the strings that this
* substring occurred in to the number of times
* it occurred in each string.
*/
PyObject *pydict = subtree_occurrences(node,p->tree->numstrings);
LST_String *s = lst_node_get_string(node, len);
PyObject *pystring =
PyString_FromStringAndSize(s->data, s->num_items-1);
PyDict_SetItem(p->substrings, pystring, pydict);
Py_DECREF(pystring);
Py_DECREF(pydict);
// printf("%s\n", lst_string_print(s));
lst_string_free(s);
}
}
return True;
}
