Solved nasty problem with 0x0085 char

Author: egulias

src/main/java/emailvalidator4j/lexer/EmailLexer.java

@@ -16,7 +16,7 @@ public class EmailLexer {
 
     public void lex(String input) {
         Pattern pattern = Pattern.compile(
-                "([a-zA-Z_]+[46]?)|([0-9]+)|(\r\n)|(::)|(\\s+?)|(.)", Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE
+                "([a-zA-Z_]+[46]?)|([0-9]+)|(\r\n)|(::)|(\\s+?)|(.)|(\\p{Cc}+)", Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE
         );
         Matcher matcher = pattern.matcher(input);
 

src/main/java/emailvalidator4j/lexer/Tokens.java

@@ -35,6 +35,9 @@ public enum Tokens implements TokenInterface {
     public static final String INVALID = "INVALID";
 
     private static final HashMap<String, TokenInterface> tokensMap = new HashMap<String, TokenInterface>();
+    private static final Pattern invalidUTF8 = Pattern.compile("\\p{Cc}+",
+            Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE);
+
     private final String name;
     private final String text;
 
@@ -64,10 +67,7 @@ public static TokenInterface get(String value) {
     }
 
     private static boolean isUTF8Invalid(String match) {
-        Pattern pattern = Pattern.compile("\\p{Cc}+", Pattern.UNICODE_CASE | Pattern.CASE_INSENSITIVE);
-        Matcher matcher = pattern.matcher(match);
-
-        return matcher.find();
+        return invalidUTF8.matcher(match).find();
     }
 
     public boolean equals(TokenInterface that) {