/*
* int_stree_edge_merge
*
* When a node has no "leaves" and only one child, this function will
* remove that node and merge the edges from parent to node and node
* to child into a single edge from parent to child.
*
* Parameters: tree - A suffix tree
* node - The tree node to be removed
*
* Return: nothing.
*/
void int_stree_edge_merge(SUFFIX_TREE tree, STREE_NODE node)
{
int len;
STREE_NODE parent, child;
STREE_LEAF leaf;
if (node == stree_get_root(tree) || int_stree_isaleaf(tree, node) ||
int_stree_has_intleaves(tree, node))
return;
parent = stree_get_parent(tree, node);
child = stree_get_children(tree, node);
if (stree_get_next(tree, child) != NULL)
return;
len = stree_get_edgelen(tree, node);
if (int_stree_isaleaf(tree, child)) {
leaf = (STREE_LEAF) child;
leaf->pos -= len;
leaf->ch = stree_get_mapch(tree, node);
}
else {
child->edgestr -= len;
child->edgelen += len;
}
int_stree_reconnect(tree, parent, node, child);
tree->num_nodes--;
tree->idents_dirty = 1;
int_stree_free_node(tree, node);
}
/*
* int_stree_delete_subtree
*
* Free up all of the memory associated with the subtree rooted at node.
*
* Parameters: tree - a suffix tree
* node - a tree node
*
* Return: nothing.
*/
void int_stree_delete_subtree(SUFFIX_TREE tree, STREE_NODE node)
{
STREE_NODE child, next;
STREE_INTLEAF intleaf, intnext;
if (int_stree_isaleaf(tree, node))
int_stree_free_leaf(tree, (STREE_LEAF) node);
else {
child = stree_get_children(tree, node);
while (child != NULL) {
next = stree_get_next(tree, child);
int_stree_delete_subtree(tree, child);
child = next;
}
if (int_stree_has_intleaves(tree, node)) {
intleaf = (STREE_INTLEAF) node->children;
while (intleaf != NULL) {
intnext = intleaf->next;
int_stree_free_intleaf(tree, intleaf);
intleaf = intnext;
}
}
int_stree_free_node(tree, node);
}
tree->idents_dirty = 1;
}
/*
* int_stree_disc_from_parent
*
* Disconnect a node from its parent in the tree.
* NOTE: This procedure only does the link manipulation part of the
* disconnection process. int_stree_disconnect is the real
* disconnection function.
*
* Parameters: tree - A suffix tree
* parent - The parent node
* child - The child to be disconnected
*
* Return: nothing.
*/
void int_stree_disc_from_parent(SUFFIX_TREE tree, STREE_NODE parent,
STREE_NODE child)
{
STREE_NODE node, back;
node = stree_get_children(tree, parent);
back = NULL;
while (node != NULL && node != child) {
back = node;
node = stree_get_next(tree, node);
}
if (node == NULL)
return;
node = stree_get_next(tree, node);
if (back == NULL) {
if (int_stree_has_intleaves(tree, parent))
((STREE_INTLEAF) (parent->children))->nextchild = node;
else
parent->children = node;
}
else {
if (node == NULL) {
back->next = parent;
back->nextisparent = 1;
}
else
back->next = node;
}
}
/*
* int_stree_reconnect
*
* Replaces one node with another in the suffix tree, reconnecting
* the link from the parent to the new node.
*
* Parameters: tree - A suffix tree
* parent - The parent of the node being replaced
* oldchild - The child being replaced
* newchild - The new child
*
* Returns: nothing
*/
int int_stree_reconnect(SUFFIX_TREE tree, STREE_NODE parent,
STREE_NODE oldchild, STREE_NODE newchild)
{
STREE_NODE node, back;
node = stree_get_children(tree, parent);
back = NULL;
while (node != NULL && node != oldchild) {
back = node;
node = stree_get_next(tree, node);
}
if (node == NULL)
return 0;
newchild->next = oldchild->next;
newchild->nextisparent = oldchild->nextisparent;
if (back == NULL) {
if (int_stree_has_intleaves(tree, parent))
((STREE_INTLEAF) (parent->children))->nextchild = newchild;
else
parent->children = newchild;
}
else
back->next = newchild;
tree->idents_dirty = 1;
return 1;
}
/*
* compute_A
*
* Computes the A values for the LCA preprocessing.
* The A values are, for each node, the heights of the least significant
* bits of the ancestors of the node (where the bits of each A value are
* set to 1 for each such height of an ancestor).
*
* Parameters: lca - an LCA_STRUCT structure
* tree - a suffix tree
* node - a suffix tree node
* Amask - the bits set by the ancestors of node
*/
static void compute_A(LCA_STRUCT *lca, SUFFIX_TREE tree, STREE_NODE node, unsigned int Amask)
{
// Shift idents so that they go from 1..num_nodes.
unsigned int id = (unsigned int)stree_get_ident(tree, node) + 1;
Amask |= 1 << h(lca->I[id]);
lca->A[id] = Amask;
for (STREE_NODE child = stree_get_children(tree, node); child; child = stree_get_next(tree, child)) {
compute_A(lca, tree, child, Amask);
IF_STATS(lca->num_prep++);
}
}
/*
* stree_get_num_children
*
* Return the number of children of a node.
*
* Parameters: tree - a suffix tree
* node - a tree node
*
* Returns: the number of children.
*/
int stree_get_num_children(SUFFIX_TREE tree, STREE_NODE node)
{
int count;
STREE_NODE child;
count = 0;
child = stree_get_children(tree, node);
while (child != NULL) {
count++;
child = stree_get_next(tree, child);
}
return count;
}
void stree_traverse_subtree(SUFFIX_TREE tree, STREE_NODE root,
int (*preorder_fn)(), int (*postorder_fn)())
{
STREE_NODE node, next;
/*
* Use a non-recursive traversal
*/
node = root;
while (1) {
/*
* Begin processing a node. If it has any children, then move down
* and process the children.
*/
if (preorder_fn != NULL)
(*preorder_fn)(tree, node);
next = stree_get_children(tree, node);
if (next != NULL) {
node = next;
continue;
}
/*
* We've finished processing the children (if any). Finish the
* processing of the node, then either move to the next child
* below the parent of node (accessed by the next field, instead
* of moving up the tree to the parent and then down), or move up
* to the parent if there is no next.
*
* If we've finished processing the root of the subtree, then return.
*/
while (1) {
if (postorder_fn != NULL)
(*postorder_fn)(tree, node);
if (node == root)
return;
if ((next = stree_get_next(tree, node)) != NULL)
break;
node = stree_get_parent(tree, node);
}
node = next;
}
}
/*
* int_stree_connect
*
* Connect a node as the child of another node.
*
* Parameters: tree - A suffix tree
* node - The node to get the new child.
* child - The child being added.
*
* Returns: The parent after the child has been connected (if the
* parent was originally a leaf, this may mean replacing
* the leaf with a node).
*/
STREE_NODE int_stree_connect(SUFFIX_TREE tree, STREE_NODE parent,
STREE_NODE newchild)
{
char ch;
STREE_NODE node, back;
if (int_stree_isaleaf(tree, parent) &&
(parent = int_stree_convert_leafnode(tree, parent)) == NULL)
return NULL;
ch = stree_get_mapch(tree, newchild);
node = stree_get_children(tree, parent);
back = NULL;
while (node != NULL && stree_get_mapch(tree, node) < ch) {
back = node;
node = stree_get_next(tree, node);
}
if (node != NULL) {
if (stree_get_mapch(tree, node) == ch)
return NULL;
newchild->next = node;
newchild->nextisparent = 0;
}
else {
newchild->next = parent;
newchild->nextisparent = 1;
}
if (back == NULL) {
if (int_stree_has_intleaves(tree, parent))
((STREE_INTLEAF) (parent->children))->nextchild = newchild;
else
parent->children = newchild;
}
else {
back->next = newchild;
back->nextisparent = 0;
}
tree->idents_dirty = 1;
return parent;
}
/*
* stree_find_child
*
* Find the child of a node whose edge label begins with the character given
* as a parameter.
*
* Parameters: tree - a suffix tree
* node - a tree node
* ch - a character
*
* Returns: a tree node or NULL.
*/
STREE_NODE stree_find_child(SUFFIX_TREE tree, STREE_NODE node, char ch)
{
char mapch;
STREE_NODE child;
if (ch < 0 || ch >= tree->alpha_size)
return NULL;
mapch = stree_mapch(tree, ch);
child = stree_get_children(tree, node);
while (child != NULL && stree_get_mapch(tree, child) < mapch)
child = stree_get_next(tree, child);
if (child != NULL && mapch == stree_get_mapch(tree, child))
return child;
else
return NULL;
}
/*
* compute_I_and_L
*
* Compute the I values and L values for the LCA preprocessing.
* The I values are, for each node, the identifier with the largest least
* significant 1 bit in the subtree rooted at the node.
* The L values are,for each node corresponding to an I value, the node
* at the head of each "run" in the tree.
*
* Parameters: lca - an LCA_STRUCT structure
* tree - a suffix tree
* node - a suffix tree node
*
* Returns: the identifier with the largest least significant 1 bit in
* the subtree rooted at node.
*/
static int compute_I_and_L(LCA_STRUCT *lca, SUFFIX_TREE tree, STREE_NODE node)
{
// Shift idents so that they go from 1..num_nodes.
unsigned int id = (unsigned int)stree_get_ident(tree, node) + 1;
// Find the node with the maximum I value in the subtree.
unsigned int Imax = id;
for (STREE_NODE child = stree_get_children(tree, node); child; child = stree_get_next(tree, child)) {
unsigned int Ival = compute_I_and_L(lca, tree, child);
if (h(Ival) > h(Imax))
Imax = Ival;
IF_STATS(lca->num_prep++);
}
lca->I[id] = Imax;
lca->L[Imax] = node; // will be overwritten by the highest node in run
return Imax;
}
/*
* int_stree_set_idents
*
* Uses the non-recursive traversal to set the identifiers for the current
* nodes of the suffix tree. The nodes are numbered in a depth-first
* manner, beginning from the root and taking the nodes in the order they
* appear in the children lists.
*
* Parameters: tree - A suffix tree
*
* Return: nothing.
*/
void int_stree_set_idents(SUFFIX_TREE tree)
{
int id;
STREE_NODE node, next;
if (!tree->idents_dirty)
return;
tree->idents_dirty = 0;
/*
* Use a non-recursive traversal. See stree_traverse_subtree for
* details.
*/
id = 0;
node = stree_get_root(tree);
while (1) {
node->id = id++;
next = stree_get_children(tree, node);
if (next != NULL) {
node = next;
continue;
}
while (1) {
if (node == stree_get_root(tree))
return;
if ((next = stree_get_next(tree, node)) != NULL)
break;
node = stree_get_parent(tree, node);
}
node = next;
}
}
