Match html5 unquoted attribute parsing

 * Brings us closer to the html5 attribute parsing algorithm described
   in http://www.w3.org/TR/html5/syntax.html#before-attribute-value-state

 * There's a similar patch for the Parsoid in,
   I2160a23b2a3c914eb369347bbf5d58328440041d

 * The spec says <div class=  style="123">hi</div> should parse as
   <div class="style=\"123\"">hi</div>, which it now does, whereas it
   used to yield <div class="" style="123">hi</div>.

 * Merge with caution. This is going to break pages like,
   frwikisource/La_Mirlitantouille_(Lenotre)?oldid=4669681

Bug: T108134
Change-Id: Ic2fc1b573a55a847e6c05707678b58c1189ecc52

diff --git a/includes/Sanitizer.php b/includes/Sanitizer.php
index b84adc3..d52bc07 100644 (file)
--- a/includes/Sanitizer.php
+++ b/includes/Sanitizer.php
@@ -332,7 +332,8 @@ class Sanitizer {
 
        /**
         * Regular expression to match HTML/XML attribute pairs within a tag.
-        * Allows some... latitude.
+        * Allows some... latitude. Based on,
+        * http://www.w3.org/TR/html5/syntax.html#before-attribute-value-state
         * Used in Sanitizer::fixTagAttributes and Sanitizer::decodeTagAttributes
         * @return string
         */
@@ -340,15 +341,15 @@ class Sanitizer {
                if ( self::$attribsRegex === null ) {
                        $attribFirst = '[:A-Z_a-z0-9]';
                        $attrib = '[:A-Z_a-z-.0-9]';
-                       $space = '[\x09\x0a\x0d\x20]';
+                       $space = '[\x09\x0a\x0c\x0d\x20]';
                        self::$attribsRegex =
                                "/(?:^|$space)({$attribFirst}{$attrib}*)
                                  ($space*=$space*
                                        (?:
                                         # The attribute value: quoted or alone
-                                         \"([^<\"]*)(?:\"|\$)
-                                        | '([^<']*)(?:'|\$)
-                                        |  ([a-zA-Z0-9!#$%&()*,\\-.\\/:;<>?@[\\]^_`{|}~]+)
+                                         \"([^\"]*)(?:\"|\$)
+                                        | '([^']*)(?:'|\$)
+                                        |  (((?!$space|>).)*)
                                        )
                                )?(?=$space|\$)/sx";
                }

diff --git a/tests/parser/parserTests.txt b/tests/parser/parserTests.txt
index d4e7119..9ad7bd5 100644 (file)
--- a/tests/parser/parserTests.txt
+++ b/tests/parser/parserTests.txt
@@ -6552,14 +6552,18 @@ Table with empty line following the start tag
 
 !! end
 
-# FIXME: Preserve the attribute properly (with an empty string as value) in
-# the PHP parser. Parsoid implements the behavior below.
 !! test
 Table attributes with empty value
 !! wikitext
 {|
 | style=| hello
 |}
+!! html/php
+<table>
+<tr>
+<td style=""> hello
+</td></tr></table>
+
 !! html/parsoid
 <table>
 <tbody>
@@ -15451,12 +15455,8 @@ div with illegal double attributes
 
 !!end
 
-# FIXME: produce empty string instead of "class" in the PHP parser, following
-# the HTML5 spec.
 !! test
 div with empty attribute value, space before equals
-!! options
-parsoid
 !! wikitext
 <div class =>HTML rocks</div>
 !! html
@@ -15464,26 +15464,36 @@ parsoid
 
 !! end
 
+# FIXME: Parsoid doesn't match the html5 spec
 !! test
 div with multiple empty attribute values
 !! options
-parsoid
+parsoid=wt2html,html2html
 !! wikitext
 <div id= title=>HTML rocks</div>
-!! html
-<div id="" title="">HTML rocks</div>
+!! html/php
+<div id="title.3D">HTML rocks</div>
 
+!! html/parsoid
+<div id="" title="">HTML rocks</div>
 !! end
 
+# FIXME: Parsoid doesn't match the html5 spec
 !! test
 table with multiple empty attribute values
 !! options
-parsoid
+parsoid=wt2html,html2html
 !! wikitext
 {| title= id=
 | hi
 |}
-!! html
+!! html/php
+<table title="id=">
+<tr>
+<td> hi
+</td></tr></table>
+
+!! html/parsoid
 <table title="" id="">
 <tbody><tr><td> hi</td></tr>
 </tbody></table>
@@ -15500,13 +15510,6 @@ div with braces in attribute value
 <div title="{}">Foo</div>
 !! end
 
-# This it very inconsistent in the PHP parser: it returns 
-# class="class" if there is a space between the name and the equal sign (see
-# 'div with empty attribute value, space before equals'), but strips the
-# attribute completely if the space is missing. We hope that not much content
-# depends on this, so are implementing the behavior below in Parsoid for
-# consistencies' sake.
-# FIXME: fix this behavior in the PHP parser?
 !! test
 div with empty attribute value, no space before equals
 !! options
@@ -15514,7 +15517,7 @@ parsoid=wt2html,html2html
 !! wikitext
 <div class=>HTML rocks</div>
 !! html/php
-<div>HTML rocks</div>
+<div class="">HTML rocks</div>
 
 !! html/parsoid
 <div class="">HTML rocks</div>
@@ -15836,7 +15839,7 @@ Attribute test: equals, then nothing
 !! wikitext
 <font color=>foo</font>
 !! html
-<p><font>foo</font>
+<p><font color="">foo</font>
 </p>
 !! end
 
@@ -23909,14 +23912,15 @@ HTML tag with 'unnecessary' entity encoding in attributes
 
 !! test
 HTML tag with broken attribute value quoting
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 <span title="Hello world>Foo</span>
 !! html/php
 <p><span title="Hello world">Foo</span>
 </p>
 !! html/parsoid
-<p><span title="Hello world">Foo</span>
-</p>
+<p><span title="Hello world">Foo</span></p>
 !! end
 
 !! test
@@ -23934,6 +23938,8 @@ parsoid=wt2html,html2html
 
 !! test
 Table with broken attribute value quoting
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 {|
 | title="Hello world|Foo
@@ -23954,6 +23960,8 @@ Table with broken attribute value quoting
 
 !! test
 Table with broken attribute value quoting on consecutive lines
+!! options
+parsoid=wt2html,html2html
 !! wikitext
 {|
 | title="Hello world|Foo