static void Main(string[] args) { DI.Initialize(); DI.webCrawler.Initialize(); DI.DPC.Initialize(); if (crawl) { WebCrawler.WebCrawler webCrawler = DI.webCrawler; webCrawler.Run(); } Indexer.Indexer indexer = DI.indexer; DI.indexer.Initialize(arePagesInMemory, numberOfPagesToLoad); DI.ranker.Initialize(); indexer.Run(numberOfPagesToLoad); List <string> urlsOfMatchedDocuments = QueryPages(indexer); if (urlsOfMatchedDocuments.Count == 0) { Console.WriteLine("No matching pages"); } else { Console.WriteLine("Url of matching pages:"); foreach (string url in urlsOfMatchedDocuments) { Console.WriteLine(url); } } Console.ReadLine(); }
public static void Initialize() { ranker = new Ranker.Ranker(); indexer = new Indexer.Indexer(); urlFilter = new UrlFilter(); urlFrontier = new UrlFrontier(); indexCreator = new IndexCreator(); pageRetriever = new PageRetriever(); termContructor = new TermConstructor(); tokenizer = new Tokenizer(); webCrawler = new WebCrawler.WebCrawler(); pageDB = new PageDB(); DPC = new DuplicatePageChecker(); DUC = new DuplicateURLChecker(); pageFetcher = new PageFetcher(); pageParser = new PageParser(); }
public static List <string> QueryPages(Indexer.Indexer indexer) { string query = "Jyllands, jyllandsposten, Amerika, Øl, Cirkus, Fastelavn"; return(indexer.ProcessQuery(query)); }