From 670fc5cdd2a2446fc51634daf6eddedfc73074a5 Mon Sep 17 00:00:00 2001
From: Mateusz Turcza <xemlock@gmail.com>
Date: Tue, 14 Aug 2018 23:24:31 +0200
Subject: [PATCH] Add <ruby> element (#25)

---
 library/HTMLPurifier/HTML5Config.php          |   2 +-
 library/HTMLPurifier/HTML5Definition.php      |   3 +
 .../HTMLPurifier/HTMLModule/HTML5/Ruby.php    |  36 +++++
 .../HTMLModule/HTML5/RubyTest.php             | 147 ++++++++++++++++++
 4 files changed, 187 insertions(+), 1 deletion(-)
 create mode 100644 library/HTMLPurifier/HTMLModule/HTML5/Ruby.php
 create mode 100644 tests/HTMLPurifier/HTMLModule/HTML5/RubyTest.php

diff --git a/library/HTMLPurifier/HTML5Config.php b/library/HTMLPurifier/HTML5Config.php
index a88e4f4..42cfc4d 100644
--- a/library/HTMLPurifier/HTML5Config.php
+++ b/library/HTMLPurifier/HTML5Config.php
@@ -2,7 +2,7 @@
 
 class HTMLPurifier_HTML5Config extends HTMLPurifier_Config
 {
-    const REVISION = 2018061701;
+    const REVISION = 2018081401;
 
     /**
      * @param  string|array|HTMLPurifier_Config $config
diff --git a/library/HTMLPurifier/HTML5Definition.php b/library/HTMLPurifier/HTML5Definition.php
index db98066..644a3a2 100644
--- a/library/HTMLPurifier/HTML5Definition.php
+++ b/library/HTMLPurifier/HTML5Definition.php
@@ -17,6 +17,9 @@ public static function setup(HTMLPurifier_HTMLDefinition $def)
         // add support for Floating point number attributes
         $def->manager->attrTypes->set('Float', new HTMLPurifier_AttrDef_Float());
 
+        // add support for Ruby markup
+        $def->manager->addModule('HTML5_Ruby');
+
         // http://developers.whatwg.org/sections.html
         $def->addElement('section', 'Block', 'Flow', 'Common');
         $def->addElement('nav', 'Block', 'Flow', 'Common');
diff --git a/library/HTMLPurifier/HTMLModule/HTML5/Ruby.php b/library/HTMLPurifier/HTMLModule/HTML5/Ruby.php
new file mode 100644
index 0000000..45fdd82
--- /dev/null
+++ b/library/HTMLPurifier/HTMLModule/HTML5/Ruby.php
@@ -0,0 +1,36 @@
+<?php
+
+/**
+ * HTML 5.2 Ruby markup
+ * https://html.spec.whatwg.org/multipage/text-level-semantics.html#the-ruby-element
+ *
+ * Note: {@link HTMLPurifier_HTMLModule_Ruby} implementation is based on
+ * XHTML 1.1 Ruby Annotation module which differs from HTML5 spec.
+ */
+class HTMLPurifier_HTMLModule_HTML5_Ruby extends HTMLPurifier_HTMLModule
+{
+    /**
+     * @type string
+     */
+    public $name = 'HTML5_Ruby';
+
+    /**
+     * @param HTMLPurifier_Config $config
+     */
+    public function setup($config)
+    {
+        $this->addElement(
+            'ruby',
+            'Inline',
+            'Custom: ((rb | Inline | #PCDATA)*, (rt | (rp, rt, rp) | rtc))+',
+            'Common'
+        );
+        $this->addElement('rtc', false, 'Custom: (rt | rp | Inline | #PCDATA)*', 'Common');
+        $this->addElement('rb', false, 'Custom: (Inline | #PCDATA)*', 'Common');
+        $this->addElement('rt', false, 'Custom: (Inline | #PCDATA)*', 'Common');
+        $this->addElement('rp', false, 'Custom: (Inline | #PCDATA)*', 'Common');
+
+        // <ruby> elements can be nested as children of <rtc>, <rb>, <rt> and <rp>
+        // https://www.w3.org/TR/2014/NOTE-html-ruby-extensions-20140204/#changes-compared-to-the-current-ruby-model
+    }
+}
diff --git a/tests/HTMLPurifier/HTMLModule/HTML5/RubyTest.php b/tests/HTMLPurifier/HTMLModule/HTML5/RubyTest.php
new file mode 100644
index 0000000..4530a7e
--- /dev/null
+++ b/tests/HTMLPurifier/HTMLModule/HTML5/RubyTest.php
@@ -0,0 +1,147 @@
+<?php
+
+class HTMLPurifier_HTMLModule_HTML5_RubyTest extends PHPUnit_Framework_TestCase
+{
+    public function getPurifier($config = null)
+    {
+        $config = HTMLPurifier_HTML5Config::create($config);
+        $config->set('Cache.DefinitionImpl', null);
+        $purifier = new HTMLPurifier($config);
+        return $purifier;
+    }
+
+    public function rubyInput()
+    {
+        return array(
+            array(
+                '<ruby>明日<rp>(</rp><rt>あした</rt><rp>)</rp></ruby>',
+            ),
+            array(
+                '<ruby>明日<rt>あした</rt></ruby>',
+            ),
+            array(
+                '<ruby>明日<rt></rt></ruby>',
+            ),
+            array(
+                '<ruby>明日<rp></rp><rt></rt><rp></rp></ruby>',
+            ),
+            array(
+                '<ruby>Foo<rp><rt><rp></ruby>',
+                '<ruby>Foo<rp></rp><rt></rt><rp></rp></ruby>',
+            ),
+            array(
+                // Element ruby is missing a required instance of child element rp.
+                '<ruby>Foo<rp><rt></ruby>',
+                '',
+            ),
+            array(
+                // Element ruby is missing a required instance of one or more of
+                // the following child elements: rp, rt, rtc.
+                '<ruby>Foo<rp></rp></ruby>',
+                '',
+            ),
+            array(
+                // Element ruby is missing a required instance of one or more of
+                // the following child elements: rp, rt, rtc.
+                '<ruby>Foo</ruby>',
+                '',
+            ),
+            array(
+                '<ruby><rb>Bar</rb><rt>Baz</rt></ruby>',
+            ),
+            array(
+                // Text before <rb> element
+                '<ruby>Foo<rb>Bar</rb><rt>Baz</rt></ruby>',
+            ),
+            array(
+                '<ruby><rb>Bar</rb><rp>(</rp><rt>Baz</rt><rp>)</rp></ruby>',
+            ),
+            array(
+                // Multiple sequences of <rb><rt> elements
+                '<ruby><rb>Bar</rb><rt>Baz</rt><rb>Qux</rb><rt>Quux</rt></ruby>',
+            ),
+            array(
+                // Text after <rb> element
+                '<ruby><rb>Bar</rb>Baz<rt>Qux</rt></ruby>',
+            ),
+            array(
+                // Multiple consecutive <rb> elements
+                '<ruby><rb>Bar</rb><rb>Baz</rb><rb>Qux</rb>Quux<rt>Quuz</rt></ruby>',
+            ),
+            array(
+                // Element ruby is missing a required instance of one or more of
+                // the following child elements: rp, rt, rtc.
+                '<ruby>Foo<rb>Bar</rb><rt>Baz</rt><rb>Qux</rb></ruby>',
+                '',
+            ),
+            array(
+                // Inline element in <ruby>
+                '<ruby><span>Foo</span><rt></rt></ruby>',
+            ),
+            array(
+                // Block-level elements are moved outside <ruby>
+                '<ruby><div>Foo</div><rt></rt></ruby>',
+                '<div>Foo</div>',
+            ),
+            array(
+                '<ruby><rb>Foo</rb><rt>Bar</rt></ruby>'
+            ),
+            array(
+                '<ruby>Foo<rb>Bar</rb><rb>Baz</rb><rb>Qux</rb><rt></rt><rb>Quux</rb><rt></rt></ruby>',
+            ),
+            array(
+                // Empty <rtc> element
+                '<ruby><rtc></rtc></ruby>',
+            ),
+            array(
+                // Properly auto-close <rtc> element
+                '<ruby><rtc>Foo<rtc>Bar</ruby>',
+                '<ruby><rtc>Foo</rtc><rtc>Bar</rtc></ruby>',
+            ),
+            array(
+                // Multiple <rtc> elements
+                '<ruby><rtc>Foo<rp></rp>Bar<rt></rt></rtc><rtc>Baz</rtc></ruby>',
+            ),
+            array(
+                // Block-level elements are moved outside <rtc>
+                '<ruby><rtc><div>Foo</div></rtc></ruby>',
+                '<ruby><rtc></rtc></ruby><div>Foo</div>',
+            ),
+            array(
+                // <rt> element can only be child of <ruby> and <rtc> elements
+                '<rt>Foo</rt>',
+                'Foo',
+            ),
+            array(
+                // <rp> element can only be child of <ruby> and <rtc> elements
+                '<rp>Foo</rp>',
+                'Foo',
+            ),
+            array(
+                // <rb> element can only be child of <ruby> element
+                '<rb>Foo</rb>',
+                'Foo',
+            ),
+            array(
+                // <rtc> element can only be child of <ruby> element
+                '<rtc>Foo</rtc>',
+                'Foo',
+            ),
+            array(
+                // Nested <ruby> elements
+                '<ruby><rtc><ruby>Foo<rt></rt></ruby><rt><ruby>Bar<rt></rt></ruby></rt></rtc></ruby>',
+            ),
+        );
+    }
+
+    /**
+     * @param string $input
+     * @param string $expectedOutput OPTIONAL
+     * @dataProvider rubyInput
+     */
+    public function testRuby($input, $expectedOutput = null)
+    {
+        $output = $this->getPurifier()->purify($input);
+        $this->assertEquals($expectedOutput !== null ? $expectedOutput : $input, $output);
+    }
+}