yahoo.java

利用多线程从搜索引擎下载网页并提取数据到数据库。

class Yahoo implements SearchEngine {
	final String sampleSearchUrl = "http://siteexplorer.search.yahoo.com/advsearch?ei=UTF-8&p=http%3A%2F%2Fwww.linkedin.com%2Fpub%2F2%2F234&searchbwm=Explore+URL&.crumb=jkhe6XXCLUO&.scrumb=0&txtSiteUrl=http%3A%2F%2Fwww.linkedin.com%2Fpub%2F2%2F999";
	final String sampleUrl = "http://66.218.69.11/search/cache?ipc=1&.intl=&u=www.linkedin.com/pub/2/234/003&d=GG7zLHDuQMIP&p=http%3A%2F%2Fwww.linkedin.com%2Fpub%2F2%2F234"
			.replaceAll("&", "&");
final String sampleNextPage="https://siteexplorer.search.yahoo.com/advsearch?p=http%3A%2F%2Fwww.linkedin.com%2Fpub%2F2%2F1&b=601&bwm=p&bwmo=&bwmf=";
	public final static int ID = 1;

	public final int id() {
		return ID;
	}

	final static String sample = "http://66.218.69.11/search/cache?ei=UTF-8&p=http%3A%2F%2Fwww.linkedin.com%2Fin%2Fdanzeo&u=www.linkedin.com/in/danzeo&d=B6eVPPL9PuQA&icp=1&.intl=us";

	public final int minLength() {
		return 146 - 40;
	}

	public final int maxLength() {
		return minLength() + 80;
	}

	public final String searchResultQueryString() {
		return "http://siteexplorer.search.yahoo.com/advsearch?ei=UTF-8&p=";
	}
	
	final public String searchResultQuesryStringAppend() {
		return "&b=601&bwm=p&bwmo=&bwmf=";
	}


	public final String quesryString() {
		return "http://siteexplorer.search.yahoo.com/advsearch?ei=UTF-8&p=";
	}

	public final String pattern() {
		return "cache?";
	}

	public int hitCount = 0;

	public void hit() {
		hitCount += 1;

	}
	
	
	

}