From 8fcd557e74c55774ac7370320d1348ddb975e9f6 Mon Sep 17 00:00:00 2001
From: Alain O'Dea <AlainODea@users.noreply.github.com>
Date: Tue, 15 Jan 2019 17:46:19 -0330
Subject: [PATCH] fix Accepts invalid domains (#7)

- Broaden GENERIC lexer regex to include non-ASCII for IDNs

 - Consider consecutive GENERIC tokens invalid in domain-part

 - Consider a leading hyphen in any part of a domain name an error
---
 src/main/java/emailvalidator4j/lexer/EmailLexer.java     | 2 +-
 src/main/java/emailvalidator4j/parser/DomainPart.java    | 9 +++++++++
 .../parser/exception/ConsecutiveGeneric.java             | 7 +++++++
 .../java/emailvalidator4j/parser/DomainPartTest.java     | 4 ++++
 4 files changed, 21 insertions(+), 1 deletion(-)
 mode change 100644 => 100755 src/main/java/emailvalidator4j/parser/DomainPart.java
 create mode 100644 src/main/java/emailvalidator4j/parser/exception/ConsecutiveGeneric.java

diff --git a/src/main/java/emailvalidator4j/lexer/EmailLexer.java b/src/main/java/emailvalidator4j/lexer/EmailLexer.java
index 7356eb9..b958a8f 100644
--- a/src/main/java/emailvalidator4j/lexer/EmailLexer.java
+++ b/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+?)|(.)|(\\p{Cc}+)", Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE
+                "(([a-zA-Z_]|[^\\u0000-\\u007F])+[46]?)|([0-9]+)|(\r\n)|(::)|(\\s+?)|(.)|(\\p{Cc}+)", Pattern.CASE_INSENSITIVE | Pattern.UNICODE_CASE
         );
         Matcher matcher = pattern.matcher(input);
 
diff --git a/src/main/java/emailvalidator4j/parser/DomainPart.java b/src/main/java/emailvalidator4j/parser/DomainPart.java
old mode 100644
new mode 100755
index 1eae87a..ac7e3d0
--- a/src/main/java/emailvalidator4j/parser/DomainPart.java
+++ b/src/main/java/emailvalidator4j/parser/DomainPart.java
@@ -55,6 +55,10 @@ private void doParseDomainPart() throws InvalidEmail {
         int domainPartOpenedParenthesis = 0;
         boolean openBrackets = false;
         do {
+            if (this.lexer.getCurrent().equals(Tokens.HYPHEN)) {
+                throw new DomainHyphen("Found -  in domain part");
+            }
+
             if (this.lexer.getCurrent().equals(Tokens.SEMICOLON)) {
                 throw new ExpectedATEXT("Expected ATEXT");
             }
@@ -149,6 +153,11 @@ private void checkExceptions() throws InvalidEmail {
         if (this.lexer.getCurrent().equals(Tokens.BACKSLASH) && this.lexer.isNextToken(Tokens.get("GENERIC"))) {
             throw new ExpectedATEXT("Found BACKSLASH");
         }
+
+        if (this.lexer.getCurrent().equals(Tokens.get("GENERIC")) && this.lexer.isNextToken(Tokens.get("GENERIC"))) {
+            this.lexer.next();
+            throw new ConsecutiveGeneric("Found " + this.lexer.getCurrent().getText());
+        }
     }
 
     private void checkIPv6Tag(String literal) {
diff --git a/src/main/java/emailvalidator4j/parser/exception/ConsecutiveGeneric.java b/src/main/java/emailvalidator4j/parser/exception/ConsecutiveGeneric.java
new file mode 100644
index 0000000..b1d3a1d
--- /dev/null
+++ b/src/main/java/emailvalidator4j/parser/exception/ConsecutiveGeneric.java
@@ -0,0 +1,7 @@
+package emailvalidator4j.parser.exception;
+
+public class ConsecutiveGeneric extends InvalidEmail {
+    public ConsecutiveGeneric(String message) {
+        super(message);
+    }
+}
diff --git a/src/test/java/emailvalidator4j/parser/DomainPartTest.java b/src/test/java/emailvalidator4j/parser/DomainPartTest.java
index 88e594f..0ead3e0 100644
--- a/src/test/java/emailvalidator4j/parser/DomainPartTest.java
+++ b/src/test/java/emailvalidator4j/parser/DomainPartTest.java
@@ -47,10 +47,14 @@ public static Object[][] invalidDomainParts() {
                 {ConsecutiveAT.class, "@@start"},
                 {ExpectedATEXT.class, "@at[start"},
                 {DomainHyphen.class, "@atstart-.com"},
+                {DomainHyphen.class, "@bb.-cc"},
+                {DomainHyphen.class, "@bb.-cc-"},
+                {DomainHyphen.class, "@bb.cc-"},
                 {DomainNotAllowedCharacter.class, "@atst\\art.com"},
                 {DomainNotAllowedCharacter.class, "@example\\"},
                 {DomainNotAllowedCharacter.class, "@exa\\mple"},
                 {UnclosedDomainLiteral.class, "@example]"},
+                {ConsecutiveGeneric.class, "@example'"},
         };
     }