// search for the finishing website
// inputs: strings for start and end, ArticleCollection to store webpages in, ForbiddenLinks object
void searchForWebsite(string StartPageName, string FinishPageName, ArticleCollection Webpages, ForbiddenLinks Forbidden) {
// get data for first webpage
WikipediaWebRequest request = new WikipediaWebRequest("wiki/"+StartPageName, Webpages, Forbidden, null);
// successful request, start program
if (request.SuccessfulWebRequest()) {
WikipediaSearcher s = null;
// create article crawler
ArticleCrawler c = new ArticleCrawler(Webpages, Forbidden);
Thread crawler = new Thread(() => c.Start());
crawler.Start();
crawler.Priority = ThreadPriority.Normal;
// create searcher
Console.WriteLine("Creating Searcher....");
Thread searcher = new Thread(() => {s = new WikipediaSearcher(Webpages, StartPageName, FinishPageName, Forbidden, "BFS");});
searcher.Start();
searcher.Priority = ThreadPriority.AboveNormal;
// wait for searcher to complete
searcher.Join();
// try to stop crawler
try {
crawler.Abort();
}
catch (PlatformNotSupportedException) {
// TODO: investigate this further
Console.WriteLine("System does not support abort of threads. Please terminate manually");
}
}
}
// creates request to get webpage
// note: only to be used within this class, use GetWebpage external to class
private bool TryToGetWebpage(string requestedPage, Webpage parent)
{
WikipediaWebRequest r = new WikipediaWebRequest(requestedPage, this, ForbiddenLinksCollection, parent);
if (r == null)
{
return(false);
}
else
{
return(true);
}
}
// depth first search
// slow
// may contain bugs... abandoned when realized BFS was better
private void SearchDFS()
{
Webpage currentPage, nextPage;
string nextPageString;
bool successfulTake, answerFound = false;
currentPage = Articles.GetWebpage(StartPage, null);
if (currentPage != null)
{
PathTaken.Push(currentPage);
}
while (PathTaken.Count > 0)
{
currentPage = PathTaken.Peek();
// get next page to be searched
try {
successfulTake = currentPage.WebpagesToBeSearched.TryDequeue(out nextPageString);
if (successfulTake)
{
// get next page down
nextPage = Articles.GetWebpage(nextPageString, currentPage);
// don't search pages that are not in database or that are the page I am currently on
if (nextPage == null || nextPage.Title == currentPage.Title)
{
continue;
}
PathTaken.Push(nextPage);
answerFound = checkIfAnswerFound(nextPage);
if (answerFound)
{
break;
}
}
else
{
PathTaken.Pop();
}
if (PathTaken.Count >= 7)
{
PathTaken.Pop();
}
} catch (NullReferenceException e) {
Console.WriteLine("No webpages have been located yet. " + e);
WikipediaWebRequest r = new WikipediaWebRequest(StartPage, Articles, ForbiddenLinksCollection, currentPage);
currentPage = Articles.GetWebpage(StartPage, null);
}
}
PrintResultsDFS(answerFound);
}
// function given to thread
// input: GetPageObject
private void GetPage(object o)
{
// cast generic object back to GetPageObject
GetPageObject obj = (GetPageObject)o;
string _str = obj.Str;
Webpage parent = obj.Page;
// check again to see if webpage alread in dictionary
Webpage w = Articles.WebpageInDictionary(_str);
if (w == null)
{
// download webpage
Console.WriteLine("requesting: " + _str);
WikipediaWebRequest r = new WikipediaWebRequest(_str, Articles, ForbiddenLinksCollection, parent);
threadCount--;
}
}
