From 011d4043a31f2142de4e5ddd31402eca3b55caa8 Mon Sep 17 00:00:00 2001
From: Antoine Musso <hashar@users.mediawiki.org>
Date: Fri, 29 Oct 2010 22:03:17 +0000
Subject: [PATCH] Follow up r75627. Implements r75670 in PHP to validate
 emails.

* Server side validation of email according to an HTML5 specifications provided by Simetrical :
http://www.whatwg.org/specs/web-apps/current-work/multipage/states-of-the-type-attribute.html#e-mail-state
* This is NOT a fix of bug 959 (which wants RFC 2822 validation)
* Basic unit tests
---
 includes/User.php                             | 15 ++++-
 .../includes/UserIsValidEmailAddrTest.php     | 63 +++++++++++++++++++
 2 files changed, 76 insertions(+), 2 deletions(-)
 create mode 100644 maintenance/tests/phpunit/includes/UserIsValidEmailAddrTest.php

diff --git a/includes/User.php b/includes/User.php
index 22ea6dcc5c..80ecd460b3 100644
--- a/includes/User.php
+++ b/includes/User.php
@@ -648,8 +648,19 @@ class User {
 		if( !wfRunHooks( 'isValidEmailAddr', array( $addr, &$result ) ) ) {
 			return $result;
 		}
-
-		return strpos( $addr, '@' ) !== false;
+		$rfc5322_atext   = "a-z0-9!#$%&'*+-\/=?^_`{|}—~" ;
+		$rfc1034_ldh_str = "a-z0-9-" ;
+
+		$HTML5_email_regexp = "/
+		^                      # start of string
+		[$rfc5322_atext\\.]+    # user part which is liberal :p
+		@                      # 'apostrophe'
+		[$rfc1034_ldh_str]     # Domain first character
+		[$rfc1034_ldh_str\\.]+  # Second char and following can include dot
+		$                      # End of string
+		/ix" ; // case Insensitive, eXtended 
+
+		return (bool) preg_match( $HTML5_email_regexp, $addr );
 	}
 
 	/**
diff --git a/maintenance/tests/phpunit/includes/UserIsValidEmailAddrTest.php b/maintenance/tests/phpunit/includes/UserIsValidEmailAddrTest.php
new file mode 100644
index 0000000000..4fdf142dda
--- /dev/null
+++ b/maintenance/tests/phpunit/includes/UserIsValidEmailAddrTest.php
@@ -0,0 +1,63 @@
+<?php
+
+class UserIsValidEmailAddrTest extends PHPUnit_Framework_TestCase {
+
+	private function testEmail( $addr, $expected = true, $msg = '') {
+		$this->assertEquals(
+			$expected,
+			User::isValidEmailAddr( $addr ),
+			$msg
+		);
+	}
+	private function valid( $addr, $msg = '' ) {
+		$this->testEmail( $addr, true, $msg );
+	}
+	private function invalid( $addr, $msg = '' ) {
+		$this->testEmail( $addr, false, $msg );
+	}
+
+	function testEmailWellKnownUserAtHostDotTldAreValid() {
+		$this->valid( 'user@example.com' );
+		$this->valid( 'user@example.museum' );
+	}
+	function testEmailWithUpperCaseCharactersAreValid() {
+		$this->valid( 'USER@example.com' );
+		$this->valid( 'user@EXAMPLE.COM' );
+		$this->valid( 'user@Example.com' );
+		$this->valid( 'USER@eXAMPLE.com' );
+	}
+	function testEmailWithAPlusInUserName() {
+		$this->valid( 'user+sub@localdomain' );
+	}
+	function testEmailWithWhiteSpacesBeforeOrAfterAreInvalids() {
+		$this->invalid( " user@host" );
+		$this->invalid( "user@host " );
+		$this->invalid( "\tuser@host" );
+		$this->invalid( "user@host\t" );
+	}
+	function testEmailWithWhiteSpacesAreInvalids() {
+		$this->invalid( "User user@host" );
+		$this->invalid( "first last@mycompany" );
+		$this->invalid( "firstlast@my company" );
+	}
+	function testEmailDomainCanNotBeginWithDot() {
+		$this->invalid( "user@." );
+		$this->invalid( "user@.localdomain" );
+		$this->valid( "user@localdomain." );
+		$this->valid( "user.@localdomain" );
+		$this->valid( ".@localdomain" );
+		$this->valid( ".@a............" );
+	}
+	function testEmailWithFunnyCharacters() {
+		$this->valid( "\$user!ex{this}@123.com" );
+	}
+	function testEmailTopLevelDomainCanBeNumerical() {
+		$this->valid( "user@example.1234" );
+	}
+	function testEmailWithoutAtSignIsInvalid() {
+		$this->invalid( 'useràexample.com' );
+	}
+	function testEmailWithOneCharacterDomainIsInvalid() {
+		$this->invalid( 'user@a' );
+	}
+}
-- 
2.20.1