google.java

利用多线程从搜索引擎下载网页并提取数据到数据库。

public class Google implements SearchEngine {
	// <a class=fl
	// href="http://203.208.33.101/search?q=cache:NENQqU49HYIJ:www.linkedin.com/in/jorgena+http://www.linkedin.com/in/jorgena&hl=zh-CN&ct=clnk&cd=1&gl=cn&st_usg=ALhdy2-A33BH6PaqW7J4-jvlM5X8joC4Fw"
	// target=_blank>网页快照</a>
	final static String sample = "http://203.208.33.101/search?q=cache:NENQqU49HYIJ:www.linkedin.com/in/jorgena+http://www.linkedin.com/in/jorgena&hl=zh-CN&ct=clnk&cd=1&gl=cn&st_usg=ALhdy2-A33BH6PaqW7J4-jvlM5X8joC4Fw";
	final static String sample2 = "http://203.208.33.101/search?q=cache:vtVgmw4u_9sJ:www.linkedin.com/in/jessicaagunsday+http://www.linkedin.com/in/jessicaagunsday&hl=zh-CN&ct=clnk&cd=1&gl=cn&st_usg=ALhdy2-fBNwMl46vCt1GTwr3IbOFWascgQ";
	public final static int ID = 2;

	public final int id() {
		return ID;
	}

	public final int minLength() {
		return 182 - 40;
	}

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

	public final String host() {
		return "http://www.google.cn";
	}
	public final String quesryString() {
		return "http://www.google.cn/search?q=";
	}

	final public String quesryStringAppend() {
		return "&num=100&complete=1&hl=zh-CN&filter=0";
	}

	public final String pattern() {
		return "search?q=cache:";
	}

	public int hitCount = 0;

	public void hit() {
		hitCount += 1;

	}
}