From d2fd193bc402dfd40826ad3b9c2d2c518d55309f Mon Sep 17 00:00:00 2001
From: "Edward Z. Yang" <edwardzyang@thewritingpot.com>
Date: Wed, 8 Nov 2006 03:10:43 +0000
Subject: [PATCH] [1.2.0] Implement primitive email regexp to be used for
 mailto. There are many spotty implementation issues, so this code is not
 actually called anywhere else currently.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@517 48356398-32a2-884e-a903-53898d9a118a
---
 library/HTMLPurifier/AttrDef.php              |  5 +--
 library/HTMLPurifier/AttrDef/Email.php        | 17 ++++++++++
 .../AttrDef/Email/SimpleCheck.php             | 23 +++++++++++++
 .../AttrDef/Email/SimpleCheckTest.php         | 16 +++++++++
 tests/HTMLPurifier/AttrDef/EmailHarness.php   | 33 +++++++++++++++++++
 tests/index.php                               |  1 +
 6 files changed, 91 insertions(+), 4 deletions(-)
 create mode 100644 library/HTMLPurifier/AttrDef/Email.php
 create mode 100644 library/HTMLPurifier/AttrDef/Email/SimpleCheck.php
 create mode 100644 tests/HTMLPurifier/AttrDef/Email/SimpleCheckTest.php
 create mode 100644 tests/HTMLPurifier/AttrDef/EmailHarness.php

diff --git a/library/HTMLPurifier/AttrDef.php b/library/HTMLPurifier/AttrDef.php
index 4abe3acc..334a7ace 100644
--- a/library/HTMLPurifier/AttrDef.php
+++ b/library/HTMLPurifier/AttrDef.php
@@ -20,10 +20,7 @@ class HTMLPurifier_AttrDef
     var $minimized = false;
     
     /**
-     * Abstract function defined for functions that validate and clean strings.
-     * 
-     * This function forms the basis for all the subclasses: they must
-     * define this method.
+     * Validates and cleans passed string according to a definition.
      * 
      * @public
      * @param $string String to be validated and cleaned.
diff --git a/library/HTMLPurifier/AttrDef/Email.php b/library/HTMLPurifier/AttrDef/Email.php
new file mode 100644
index 00000000..7a7ad6ab
--- /dev/null
+++ b/library/HTMLPurifier/AttrDef/Email.php
@@ -0,0 +1,17 @@
+<?php
+
+require_once 'HTMLPurifier/AttrDef.php';
+
+class HTMLPurifier_AttrDef_Email extends HTMLPurifier_AttrDef
+{
+    
+    /**
+     * Unpacks a mailbox into its display-name and address
+     */
+    function unpack($string) {
+        // needs to be implemented
+    }
+    
+}
+
+?>
\ No newline at end of file
diff --git a/library/HTMLPurifier/AttrDef/Email/SimpleCheck.php b/library/HTMLPurifier/AttrDef/Email/SimpleCheck.php
new file mode 100644
index 00000000..4b9fdf1a
--- /dev/null
+++ b/library/HTMLPurifier/AttrDef/Email/SimpleCheck.php
@@ -0,0 +1,23 @@
+<?php
+
+require_once 'HTMLPurifier/AttrDef/Email.php';
+
+/**
+ * Primitive email validation class based on the regexp found at 
+ * http://www.regular-expressions.info/email.html
+ */
+class HTMLPurifier_AttrDef_Email_SimpleCheck extends HTMLPurifier_AttrDef_Email
+{
+    
+    function validate($string, $config, &$context) {
+        // no support for named mailboxes i.e. "Bob <bob@example.com>"
+        // that needs more percent encoding to be done
+        if ($string == '') return false;
+        $string = trim($string);
+        $result = preg_match('/^[A-Z0-9._%-]+@[A-Z0-9.-]+\.[A-Z]{2,4}$/i', $string);
+        return $result ? $string : false;
+    }
+    
+}
+
+?>
\ No newline at end of file
diff --git a/tests/HTMLPurifier/AttrDef/Email/SimpleCheckTest.php b/tests/HTMLPurifier/AttrDef/Email/SimpleCheckTest.php
new file mode 100644
index 00000000..70a77f72
--- /dev/null
+++ b/tests/HTMLPurifier/AttrDef/Email/SimpleCheckTest.php
@@ -0,0 +1,16 @@
+<?php
+
+require_once 'HTMLPurifier/AttrDef/Email/SimpleCheck.php';
+require_once 'HTMLPurifier/AttrDef/EmailHarness.php';
+
+class HTMLPurifier_AttrDef_Email_SimpleCheckTest
+    extends HTMLPurifier_AttrDef_EmailHarness
+{
+    
+    function setUp() {
+        $this->def = new HTMLPurifier_AttrDef_Email_SimpleCheck();
+    }
+    
+}
+
+?>
\ No newline at end of file
diff --git a/tests/HTMLPurifier/AttrDef/EmailHarness.php b/tests/HTMLPurifier/AttrDef/EmailHarness.php
new file mode 100644
index 00000000..715a5293
--- /dev/null
+++ b/tests/HTMLPurifier/AttrDef/EmailHarness.php
@@ -0,0 +1,33 @@
+<?php
+
+require_once 'HTMLPurifier/AttrDef/Email.php';
+
+class HTMLPurifier_AttrDef_EmailHarness extends HTMLPurifier_AttrDefHarness
+{
+    
+    /**
+     * Tests common email strings that are obviously pass/fail
+     */
+    function testCore() {
+        $this->assertDef('bob@example.com');
+        $this->assertDef('  bob@example.com  ', 'bob@example.com');
+        $this->assertDef('bob.thebuilder@example.net');
+        $this->assertDef('Bob_the_Builder-the-2nd@example.org');
+        $this->assertDef('Bob%20the%20Builder@white-space.test');
+        
+        // extended format, with real name
+        //$this->assertDef('Bob%20Builder%20%3Cbobby.bob.bob@it.is.example.com%3E');
+        //$this->assertDef('Bob Builder <bobby.bob.bob@it.is.example.com>');
+        
+        // time to fail
+        $this->assertDef('bob', false);
+        $this->assertDef('bob@home@work', false);
+        $this->assertDef('@example.com', false);
+        $this->assertDef('bob@', false);
+        $this->assertDef('', false);
+        
+    }
+    
+}
+
+?>
\ No newline at end of file
diff --git a/tests/index.php b/tests/index.php
index 4ceb8d96..33ad68e6 100644
--- a/tests/index.php
+++ b/tests/index.php
@@ -78,6 +78,7 @@ $test_files[] = 'AttrDef/IPv6Test.php';
 $test_files[] = 'AttrDef/FontTest.php';
 $test_files[] = 'AttrDef/BorderTest.php';
 $test_files[] = 'AttrDef/ListStyleTest.php';
+$test_files[] = 'AttrDef/Email/SimpleCheckTest.php';
 $test_files[] = 'IDAccumulatorTest.php';
 $test_files[] = 'TagTransformTest.php';
 $test_files[] = 'AttrTransform/LangTest.php';