//Creates an occurrence, or increments the
//count of an occurrence on the class BST
//@par url - the url that the occurrence is on
void OccurrenceSet::push(string url){
BSTNode<string,Occurrence*> * node = occurrences.Find(url);
if(node==NULL){
Occurrence * occurrence = new Occurrence(url);
occurrence->addOccurrence();
occurrences.Insert(url,occurrence);
}else{
node->GetValue()->addOccurrence();
}
}
void Crawler::addWords(BST < Pair < string,int > >* newOccurrences, string url){
BSTIterator<Pair <string,int> > iter = newOccurrences->Iterator();
BSTNode<Pair<string,int> > newNode(Pair<string,int>("",-1));
BSTNode<Word>* oldNode;
Occurrence occ;
occ.setURL(url);
while(iter.hasNext()){
newNode = iter.next();
//is either a new node or an old node
oldNode = words->Insert(Word(newNode.GetValue().getFirst()));
occ.setOccurrences(newNode.GetValue().getSecond());
oldNode->GetValue().addOccurrence(occ);
}
}
bool WordIndex::Test (ostream & os) {
bool success = true;
const int PAGES = 3;
const int WORDS = 30;
string urlStrs[PAGES] = {
"http://www.google.com/index.html"
, "file:///home/file.txt"
, "http://www.msn.com/sports.html"
};
URL urls[PAGES] = {
URL(urlStrs[0])
, URL(urlStrs[1])
, URL(urlStrs[2])
};
Word words[WORDS] = {
"and", "the", "a", "wood", "couch", "potato", "Henry", "the", "a", "and"
, "a", "house", "dog", "wood", "couch", "frisbee", "green", "then", "why", "how"
, "a", "a", "yes", "no", "maybe", "Henry", "the", "frisbee", "green", "couch"
};
WordIndex wordIndex;
for (int i = 0; i < PAGES; i++) {
for (int j = 0; j < WORDS; j++) {
wordIndex.Insert(words[j], urls[i]);
}
}
OccurrenceSet set = wordIndex.GetValue("a");
BSTNode<Occurrence>* node = set.Find(Occurrence(urls[1]));
TEST (NULL != node);
Occurrence current = node->GetValue();
TEST(current.getURL().getFullURL() == urls[1].getFullURL());
TEST(current.getCount() == 5);
return success;
}
void OcrBST :: recursive_insert(OcrBSTNode * current_node, const Occurrence & v)
{
int compare_result;
compare_result = v.compare(current_node->GetValue());
if(compare_result < 0)
{
//go left
if(current_node->GetLeft() == NULL)
{
current_node->left = new OcrBSTNode(v);
size++;
returnValue = current_node->left;
}
else
{
recursive_insert(current_node->GetLeft(), v);
}
}
else if(compare_result > 0)
{
//go right
if(current_node->GetRight() == NULL)
{
current_node->right = new OcrBSTNode(v);
size++;
returnValue = current_node->right;
}
else
{
recursive_insert(current_node->GetRight(), v);
}
}
else
{
returnValue = NULL;
}
}
int Occurrence::compare(const Occurrence & other)const{
return url.compare(other.getURL().getURLstring());
}
