package eu.dnetlib.dhp.collection.plugin.utils;

import java.util.HashMap;
import java.util.HashSet;
import java.util.Map;
import java.util.Set;
import java.util.regex.Pattern;

/** @author jochen, Andreas Czerniak */
public class XmlCleaner {

	/** Pattern for numeric entities. */
	private static final Pattern validCharacterEntityPattern = Pattern.compile("^&#x?\\d{2,4};"); // $NON-NLS-1$
	// private static Pattern validCharacterEntityPattern = Pattern.compile("^&#?\\d{2,4};");
	// //$NON-NLS-1$

	// see https://www.w3.org/TR/REC-xml/#charsets , not only limited to 
	private static final Pattern invalidControlCharPattern = Pattern.compile("&#x?1[0-9a-fA-F];");

	/**
	 * Pattern that negates the allowable XML 4 byte unicode characters. Valid are: #x9 | #xA | #xD | [#x20-#xD7FF] |
	 * [#xE000-#xFFFD] | [#x10000-#x10FFFF]
	 */
	private static final Pattern invalidCharacterPattern = Pattern.compile("[^\t\r\n\u0020-\uD7FF\uE000-\uFFFD]"); // $NON-NLS-1$

	// Map entities to their unicode equivalent
	private static final Set<String> goodEntities = new HashSet<>();
	private static final Map<String, String> badEntities = new HashMap<>();

	static {
		// pre-defined XML entities
		goodEntities.add("&quot;"); // $NON-NLS-1$ // quotation mark
		goodEntities.add("&amp;"); // $NON-NLS-1$ // ampersand
		goodEntities.add("&lt;"); // $NON-NLS-1$ // less-than sign
		goodEntities.add("&gt;"); // $NON-NLS-1$ // greater-than sign
		// control entities
		// badEntities.put("&#11;", "");
		badEntities.put("&#127;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#128;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#129;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#130;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#131;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#132;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#133;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#134;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#135;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#136;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#137;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#138;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#139;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#140;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#141;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#142;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#143;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#144;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#145;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#146;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#147;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#148;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#149;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#150;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#151;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#152;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#153;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#154;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#155;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#156;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#157;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#158;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		badEntities.put("&#159;", " "); // $NON-NLS-1$ //$NON-NLS-2$ // illegal HTML character
		// misc entities
		badEntities.put("&euro;", "\u20AC"); // $NON-NLS-1$ //$NON-NLS-2$ // euro
		badEntities.put("&lsquo;", "\u2018"); // $NON-NLS-1$ //$NON-NLS-2$ // left single quotation
		// mark
		badEntities.put("&rsquo;", "\u2019"); // $NON-NLS-1$ //$NON-NLS-2$ // right single quotation
		// mark
		// Latin 1 entities
		badEntities.put("&nbsp;", "\u00A0"); // $NON-NLS-1$ //$NON-NLS-2$ // no-break space
		badEntities.put("&iexcl;", "\u00A1"); // $NON-NLS-1$ //$NON-NLS-2$ // inverted exclamation
		// mark
		badEntities.put("&cent;", "\u00A2"); // $NON-NLS-1$ //$NON-NLS-2$ // cent sign
		badEntities.put("&pound;", "\u00A3"); // $NON-NLS-1$ //$NON-NLS-2$ // pound sign
		badEntities.put("&curren;", "\u00A4"); // $NON-NLS-1$ //$NON-NLS-2$ // currency sign
		badEntities.put("&yen;", "\u00A5"); // $NON-NLS-1$ //$NON-NLS-2$ // yen sign
		badEntities.put("&brvbar;", "\u00A6"); // $NON-NLS-1$ //$NON-NLS-2$ // broken vertical bar
		badEntities.put("&sect;", "\u00A7"); // $NON-NLS-1$ //$NON-NLS-2$ // section sign
		badEntities.put("&uml;", "\u00A8"); // $NON-NLS-1$ //$NON-NLS-2$ // diaeresis
		badEntities.put("&copy;", "\u00A9"); // $NON-NLS-1$ //$NON-NLS-2$ // copyright sign
		badEntities.put("&ordf;", "\u00AA"); // $NON-NLS-1$ //$NON-NLS-2$ // feminine ordinal
		// indicator
		badEntities.put("&laquo;", "\u00AB"); // $NON-NLS-2$ // left-pointing double angle
		// quotation mark
		badEntities.put("&not;", "\u00AC"); // $NON-NLS-1$ //$NON-NLS-2$ // not sign
		badEntities.put("&shy;", "\u00AD"); // $NON-NLS-1$ //$NON-NLS-2$ // soft hyphen
		badEntities.put("&reg;", "\u00AE"); // $NON-NLS-1$ //$NON-NLS-2$ // registered sign
		badEntities.put("&macr;", "\u00AF"); // $NON-NLS-1$ //$NON-NLS-2$ // macron
		badEntities.put("&deg;", "\u00B0"); // $NON-NLS-1$ //$NON-NLS-2$ // degree sign
		badEntities.put("&plusmn;", "\u00B1"); // $NON-NLS-1$ //$NON-NLS-2$ // plus-minus sign
		badEntities.put("&sup2;", "\u00B2"); // $NON-NLS-1$ //$NON-NLS-2$ // superscript two
		badEntities.put("&sup3;", "\u00B3"); // $NON-NLS-1$ //$NON-NLS-2$ // superscript three
		badEntities.put("&acute;", "\u00B4"); // $NON-NLS-1$ //$NON-NLS-2$ // acute accent
		badEntities.put("&micro;", "\u00B5"); // $NON-NLS-1$ //$NON-NLS-2$ // micro sign
		badEntities.put("&para;", "\u00B6"); // $NON-NLS-1$ //$NON-NLS-2$ // pilcrow sign
		badEntities.put("&middot;", "\u00B7"); // $NON-NLS-1$ //$NON-NLS-2$ // middle dot
		badEntities.put("&cedil;", "\u00B8"); // $NON-NLS-1$ //$NON-NLS-2$ // cedilla
		badEntities.put("&sup1;", "\u00B9"); // $NON-NLS-1$ //$NON-NLS-2$ // superscript one
		badEntities.put("&ordm;", "\u00BA"); // $NON-NLS-1$ //$NON-NLS-2$ // masculine ordinal
		// indicator
		badEntities.put("&raquo;", "\u00BB"); // $NON-NLS-2$ // right-pointing double
		// angle quotation
		// mark
		badEntities.put("&frac14;", "\u00BC"); // $NON-NLS-1$ //$NON-NLS-2$ // vulgar fraction one
		// quarter
		badEntities.put("&frac12;", "\u00BD"); // $NON-NLS-1$ //$NON-NLS-2$ // vulgar fraction one
		// half
		badEntities.put("&frac34;", "\u00BE"); // $NON-NLS-1$ //$NON-NLS-2$ // vulgar fraction three
		// quarters
		badEntities.put("&iquest;", "\u00BF"); // $NON-NLS-1$ //$NON-NLS-2$ // inverted question
		// mark
		badEntities.put("&Agrave;", "\u00C0"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// A
		// with grave
		badEntities.put("&Aacute;", "\u00C1"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// A
		// with acute
		badEntities.put("&Acirc;", "\u00C2"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter A
		// with circumflex
		badEntities.put("&Atilde;", "\u00C3"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// A
		// with tilde
		badEntities.put("&Auml;", "\u00C4"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter A
		// with
		// diaeresis
		badEntities.put("&Aring;", "\u00C5"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter A
		// with ring above
		badEntities.put("&AElig;", "\u00C6"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// AE
		badEntities.put("&Ccedil;", "\u00C7"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// C
		// with cedilla
		badEntities.put("&Egrave;", "\u00C8"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// E
		// with grave
		badEntities.put("&Eacute;", "\u00C9"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// E
		// with acute
		badEntities.put("&Ecirc;", "\u00CA"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter E
		// with circumflex
		badEntities.put("&Euml;", "\u00CB"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter E
		// with
		// diaeresis
		badEntities.put("&Igrave;", "\u00CC"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// I
		// with grave
		badEntities.put("&Iacute;", "\u00CD"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// I
		// with acute
		badEntities.put("&Icirc;", "\u00CE"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter I
		// with circumflex
		badEntities.put("&Iuml;", "\u00CF"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter I
		// with
		// diaeresis
		badEntities.put("&ETH;", "\u00D0"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter ETH
		badEntities.put("&Ntilde;", "\u00D1"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// N
		// with tilde
		badEntities.put("&Ograve;", "\u00D2"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// O
		// with grave
		badEntities.put("&Oacute;", "\u00D3"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// O
		// with acute
		badEntities.put("&Ocirc;", "\u00D4"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter O
		// with circumflex
		badEntities.put("&Otilde;", "\u00D5"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// O
		// with tilde
		badEntities.put("&Ouml;", "\u00D6"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter O
		// with
		// diaeresis
		badEntities.put("&times;", "\u00D7"); // $NON-NLS-1$ //$NON-NLS-2$ // multiplication sign
		badEntities.put("&Oslash;", "\u00D8"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// O
		// with stroke
		badEntities.put("&Ugrave;", "\u00D9"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// U
		// with grave
		badEntities.put("&Uacute;", "\u00DA"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// U
		// with acute
		badEntities.put("&Ucirc;", "\u00DB"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter U
		// with circumflex
		badEntities.put("&Uuml;", "\u00DC"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter U
		// with
		// diaeresis
		badEntities.put("&Yacute;", "\u00DD"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// Y
		// with acute
		badEntities.put("&THORN;", "\u00DE"); // $NON-NLS-1$ //$NON-NLS-2$ // latin capital letter
		// THORN
		badEntities.put("&szlig;", "\u00DF"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter
		// sharp s
		badEntities.put("&agrave;", "\u00E0"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter a
		// with
		// grave
		badEntities.put("&aacute;", "\u00E1"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter a
		// with
		// acute
		badEntities.put("&acirc;", "\u00E2"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter a
		// with
		// circumflex
		badEntities.put("&atilde;", "\u00E3"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter a
		// with
		// tilde
		badEntities.put("&auml;", "\u00E4"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter a
		// with
		// diaeresis
		badEntities.put("&aring;", "\u00E5"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter a
		// with
		// ring above
		badEntities.put("&aelig;", "\u00E6"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter ae
		badEntities.put("&ccedil;", "\u00E7"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter c
		// with
		// cedilla
		badEntities.put("&egrave;", "\u00E8"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter e
		// with
		// grave
		badEntities.put("&eacute;", "\u00E9"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter e
		// with
		// acute
		badEntities.put("&ecirc;", "\u00EA"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter e
		// with
		// circumflex
		badEntities.put("&euml;", "\u00EB"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter e
		// with
		// diaeresis
		badEntities.put("&igrave;", "\u00EC"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter i
		// with
		// grave
		badEntities.put("&iacute;", "\u00ED"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter i
		// with
		// acute
		badEntities.put("&icirc;", "\u00EE"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter i
		// with
		// circumflex
		badEntities.put("&iuml;", "\u00EF"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter i
		// with
		// diaeresis
		badEntities.put("&eth;", "\u00F0"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter eth
		badEntities.put("&ntilde;", "\u00F1"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter n
		// with
		// tilde
		badEntities.put("&ograve;", "\u00F2"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter o
		// with
		// grave
		badEntities.put("&oacute;", "\u00F3"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter o
		// with
		// acute
		badEntities.put("&ocirc;", "\u00F4"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter o
		// with
		// circumflex
		badEntities.put("&otilde;", "\u00F5"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter o
		// with
		// tilde
		badEntities.put("&ouml;", "\u00F6"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter o
		// with
		// diaeresis
		badEntities.put("&divide;", "\u00F7"); // $NON-NLS-1$ //$NON-NLS-2$ // division sign
		badEntities.put("&oslash;", "\u00F8"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter o
		// with
		// stroke
		badEntities.put("&ugrave;", "\u00F9"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter u
		// with
		// grave
		badEntities.put("&uacute;", "\u00FA"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter u
		// with
		// acute
		badEntities.put("&ucirc;", "\u00FB"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter u
		// with
		// circumflex
		badEntities.put("&uuml;", "\u00FC"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter u
		// with
		// diaeresis
		badEntities.put("&yacute;", "\u00FD"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter y
		// with
		// acute
		badEntities.put("&thorn;", "\u00FE"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter
		// thorn
		badEntities.put("&yuml;", "\u00FF"); // $NON-NLS-1$ //$NON-NLS-2$ // latin small letter y
		// with
		// diaeresis
	}

	/**
	 * For each entity in the input that is not allowed in XML, replace the entity with its unicode equivalent or remove
	 * it. For each instance of a bare {@literal &}, replace it with {@literal
	 * &amp;<br/>
	 } XML only allows 4 entities: {@literal &amp;amp;}, {@literal &amp;quot;}, {@literal
	 * &amp;lt;} and {@literal &amp;gt;}.
	 *
	 * @param broken the string to handle entities
	 * @return the string with entities appropriately fixed up
	 */
	public static String cleanAllEntities(final String broken) {
		if (broken == null) {
			return null;
		}

		String working = invalidControlCharPattern.matcher(broken).replaceAll("");
		working = invalidCharacterPattern.matcher(working).replaceAll("");

		int cleanfrom = 0;

		while (true) {
			int amp = working.indexOf('&', cleanfrom);
			// If there are no more amps then we are done
			if (amp == -1) {
				break;
			}
			// Skip references of the kind &#ddd;
			if (validCharacterEntityPattern.matcher(working.substring(amp)).find()) {
				cleanfrom = working.indexOf(';', amp) + 1;
				continue;
			}
			int i = amp + 1;
			while (true) {
				// if we are at the end of the string then just escape the '&';
				if (i >= working.length()) {
					return working.substring(0, amp) + "&amp;" + working.substring(amp + 1); // $NON-NLS-1$
				}
				// if we have come to a ; then we have an entity
				// If it is something that xml can't handle then replace it.
				final char c = working.charAt(i);
				if (c == ';') {
					final String entity = working.substring(amp, i + 1);
					final String replace = handleEntity(entity);
					working = working.substring(0, amp) + replace + working.substring(i + 1);
					break;
				}
				// Did we end an entity without finding a closing ;
				// Then treat it as an '&' that needs to be replaced with &amp;
				if (!Character.isLetterOrDigit(c)) {
					working = working.substring(0, amp) + "&amp;" + working.substring(amp + 1); // $NON-NLS-1$
					amp = i + 4; // account for the 4 extra characters
					break;
				}
				i++;
			}
			cleanfrom = amp + 1;
		}

		if (Pattern.compile("<<").matcher(working).find()) {
			working = working.replaceAll("<<", "&lt;&lt;");
		}

		if (Pattern.compile(">>").matcher(working).find()) {
			working = working.replaceAll(">>", "&gt;&gt;");
		}

		return working;
	}

	/**
	 * Replace entity with its unicode equivalent, if it is not a valid XML entity. Otherwise strip it out. XML only
	 * allows 4 entities: &amp;amp;, &amp;quot;, &amp;lt; and &amp;gt;.
	 *
	 * @param entity the entity to be replaced
	 * @return the substitution for the entity, either itself, the unicode equivalent or an empty string.
	 */
	private static String handleEntity(final String entity) {
		if (goodEntities.contains(entity)) {
			return entity;
		}

		final String replace = badEntities.get(entity);
		if (replace != null) {
			return replace;
		}

		return replace != null ? replace : "";
	}
}