/**
* Converts appropriate Unicode characters to their HTML character entity counterparts (c.f., <a
* href="http://www.w3.org/TR/REC-html40/sgml/entities.html">http://www.w3.org/TR/REC-html40/sgml/entities.html</a>).
*
* @param s the string to convert
* @return the resulting, possibly modified, string
* @see #stripHTMLTags
* @see #convertCharacterEntities
*/
public static String makeCharacterEntities(String s) {
// First, make a character-to-entity-name map from the resource bundle.
ResourceBundle bundle = getResourceBundle();
Map<Character, String> charToEntityName = new HashMap<Character, String>();
Enumeration<String> keys = bundle.getKeys();
XStringBuffer buf = new XStringBuffer();
while (keys.hasMoreElements()) {
String key = keys.nextElement();
String sChar = bundle.getString(key);
char c = sChar.charAt(0);
// Transform the bundle key into an entity name by removing the
// "html_" prefix.
buf.clear();
buf.append(key);
buf.delete("html_");
charToEntityName.put(c, buf.toString());
}
char[] chars = s.toCharArray();
buf.clear();
for (int i = 0; i < chars.length; i++) {
char c = chars[i];
String entity = charToEntityName.get(c);
if (entity == null) {
if (!TextUtil.isPrintable(c)) {
buf.append("&#");
buf.append(Integer.valueOf(c));
buf.append(';');
} else {
buf.append(c);
}
} else {
buf.append('&');
buf.append(entity);
buf.append(';');
}
}
return buf.toString();
}
/**
* Converts all inline HTML character entities (c.f., <a
* href="http://www.w3.org/TR/REC-html40/sgml/entities.html">http://www.w3.org/TR/REC-html40/sgml/entities.html</a>)
* to their Unicode character counterparts, if possible.
*
* @param s the string to convert
* @return the resulting, possibly modified, string
* @see #stripHTMLTags
* @see #makeCharacterEntities
*/
public static String convertCharacterEntities(String s) {
// The resource bundle contains the mappings for symbolic entity
// names like "amp". Note: Must protect matching and MatchResult in
// a critical section, for thread-safety. See javadocs for
// Perl5Util.
synchronized (HTMLUtil.class) {
try {
if (entityPattern == null) entityPattern = Pattern.compile("&(#?[^; \t]+);");
} catch (PatternSyntaxException ex) {
// Should not happen unless I've screwed up the pattern.
// Throw a runtime error.
assert (false);
}
}
ResourceBundle bundle = getResourceBundle();
XStringBuffer buf = new XStringBuffer();
Matcher matcher = null;
synchronized (HTMLUtil.class) {
matcher = entityPattern.matcher(s);
}
for (; ; ) {
String match = null;
String preMatch = null;
String postMatch = null;
if (!matcher.find()) break;
match = matcher.group(1);
preMatch = s.substring(0, matcher.start(1) - 1);
postMatch = s.substring(matcher.end(1) + 1);
if (preMatch != null) buf.append(preMatch);
if (match.charAt(0) == '#') {
if (match.length() == 1) buf.append('#');
else {
// It might be a numeric entity code. Try to parse it
// as a number. If the parse fails, just put the whole
// string in the result, as is. Be sure to handle both
// the decimal form (e.g., ™) and the hexadecimal
// form (e.g., ™).
int cc;
boolean isHex = (match.length() > 2) && (match.charAt(1) == 'x');
boolean isLegal = false;
try {
if (isHex) cc = Integer.parseInt(match.substring(2), 16);
else cc = Integer.parseInt(match.substring(1));
// It parsed. Is it a valid Unicode character?
if (Character.isDefined((char) cc)) {
buf.append((char) cc);
isLegal = true;
}
} catch (NumberFormatException ex) {
}
if (!isLegal) {
buf.append("&#");
if (isHex) buf.append('x');
buf.append(match + ";");
}
}
} else {
// Not a numeric entity. Try to find a matching symbolic
// entity.
try {
String rep = bundle.getString("html_" + match);
buf.append(rep);
} catch (MissingResourceException ex) {
buf.append("&" + match + ";");
}
}
if (postMatch == null) break;
s = postMatch;
matcher.reset(s);
}
if (s.length() > 0) buf.append(s);
return buf.toString();
}