很多時候叢網頁中抽取的內容中都含有很多特殊的轉義字元，網頁中顯示的要麼是實體名稱，要麼是實體編碼，如

原理是根據實體編碼中的數字，得到對應的字元。但上述方法不支援將實體名稱轉化為可顯示字元，如能將"""轉化為雙引號，卻不能識別"""。這次沒有固定的規律，只能自己作映射了。比較折中的做法是轉化如下幾個常見的：

/**
 * Convert HTML character enitities(Unicode) to part of a Java String
 */
import java.util.regex.*;

public class UnicodeCeToJavaString {
	static final String mbs = "&#(\\d+);"; // like "ロ"

	public static String EncodeCesToChars(String paramStr) {
		paramStr = paramStr.replace("&","&")
		        .replace("<","<")
		        .replace(">",">")
		        .replace(""","\"")
		        .replace(" "," ")
		        .replace("'","'");
		
		String mbChar;
		StringBuffer sb = new StringBuffer();
		Pattern pat = Pattern.compile(mbs);
		Matcher mat = pat.matcher(paramStr);
		while (mat.find()) {
			mbChar = getMbCharStr(mat.group(1)); // pass the digit part
			mat.appendReplacement(sb, mbChar);
		}
		mat.appendTail(sb);
		return new String(sb);
	}

	/* worker method */
	static String getMbCharStr(String digits) { // handle "12525" part which is
												// a
		char[] cha = new char[1]; // Unicode value stringnized
		try {
			int val = Integer.parseInt(digits);
			char ch = (char) val;
			cha[0] = ch;
		} catch (Exception e) {
			System.err.println("Error from getMbCharStr:");
			e.printStackTrace(System.err);
		}
		return new String(cha); // easy!, because Java uses Unicode
	}
	
	public static void main(String[] args) {
		System.out.println(UnicodeCeToJavaString.EncodeCesToChars("George&#39;s War in North America"));
	}
}