java 去掉._java去除html tags

public static String removeHTML(String htmlString)

{

// Remove HTML tag from java String

String noHTMLString = htmlString.replaceAll("//<.>", "");

// Remove Carriage return from java String

noHTMLString = noHTMLString.replaceAll("/r", "
");

// Remove New line from java string and replace html break

noHTMLString = noHTMLString.replaceAll("/n", " ");

noHTMLString = noHTMLString.replaceAll("/'", "'");

noHTMLString = noHTMLString.replaceAll("/"", """);

return noHTMLString;

}

public static void main(String[] args) {

String strHTML= ""+

"

"+

"

Convert HTML to Text String"+

""+

"

"+

"This is HTML String of java's source code  /"my program/""+

""+

"";

String stringWithoutHTML=removeHTML(strHTML);

System.out.println(stringWithoutHTML);

}

publicstaticString regEx_script ="

publicstaticString regEx_style ="

publicstaticString regEx_html ="<[^>]+>";

publicstaticPattern p_style = Pattern.compile(regEx_style, Pattern.CASE_INSENSITIVE);

publicstaticPattern p_script = Pattern.compile(regEx_script, Pattern.CASE_INSENSITIVE);

publicstaticPattern p_html = Pattern.compile(regEx_html, Pattern.CASE_INSENSITIVE);

publicstaticString getOptimizedData(String inputString) {

if(inputString ==null) {

returninputString;

}

//stripping script tags whether thetagcontains "/n" or "/r" or not.

Matcher m_script = p_script.matcher(inputString);

String htmlStr = m_script.replaceAll("");

//stripping style tags whether thetagcontains "/n" or "/r" or not.

Matcher m_style = p_style.matcher(htmlStr);

htmlStr = m_style.replaceAll("");

//strippinghtmltags but continue to have the "/n" and "/r" in right place.

Matcher m_html = p_html.matcher(htmlStr);

htmlStr = m_html.replaceAll("");

returnhtmlStr;

}