From 9b2b375fcef153ea554925cb2c8133fb64b336e9 Mon Sep 17 00:00:00 2001
From: Brad Jorsch <bjorsch@wikimedia.org>
Date: Fri, 24 Nov 2017 14:22:25 -0500
Subject: [PATCH] ParserOutput: Add 'deduplicateStyles' post-cache
 transformation

This transformation will find <style> tag with a "data-mw-deduplicate"
attribute. For each value of the attribute, the first instance will be
kept as-is, while any subsequent tags with the same value will be
replaced by a <link rel="mw-deduplicated-inline-style"> with its href
referring to the "data-mw-deduplicate" value using a custom scheme.

This also adds an $attribs parameter to Html::inlineStyle() so the
data-mw-deduplicate attribute can be added.

Note this doesn't actually depend on Ib931e25c, but action=mobileview
will break if it starts being used without that patch.

Bug: T160563
Change-Id: I055abdf4d73ec65771eaa4fe0999ec907c831568
Depends-On: Ib931e25ce85072000e62c486bbe5907f03372494
---
 RELEASE-NOTES-1.31                            |  3 ++
 includes/Html.php                             |  7 +++-
 includes/api/ApiParse.php                     |  2 +
 includes/api/i18n/en.json                     |  1 +
 includes/api/i18n/qqq.json                    |  1 +
 includes/parser/ParserOutput.php              | 37 +++++++++++++++++++
 .../includes/parser/ParserOutputTest.php      | 30 +++++++++++++++
 7 files changed, 79 insertions(+), 2 deletions(-)

diff --git a/RELEASE-NOTES-1.31 b/RELEASE-NOTES-1.31
index 88d927ef28..cc08b33e84 100644
--- a/RELEASE-NOTES-1.31
+++ b/RELEASE-NOTES-1.31
@@ -46,6 +46,9 @@ production.
   initial page text for file uploads.
 * (T181651) The info page for File pages now displays the file's base-16 SHA1
   hash value in the table of basic information.
+* Style tags with a 'data-mw-deduplicate' attribute will be deduplicated as a
+  ParserOutput::getText() post-cache transformation. This may be disabled by
+  passing 'deduplicateStyles' => false to that method.
 
 === External library changes in 1.31 ===
 
diff --git a/includes/Html.php b/includes/Html.php
index dfd80a8c43..3bcf13132f 100644
--- a/includes/Html.php
+++ b/includes/Html.php
@@ -589,9 +589,12 @@ class Html {
 	 *
 	 * @param string $contents CSS
 	 * @param string $media A media type string, like 'screen'
+	 * @param array $attribs (since 1.31) Associative array of attributes, e.g., [
+	 *   'href' => 'https://www.mediawiki.org/' ]. See expandAttributes() for
+	 *   further documentation.
 	 * @return string Raw HTML
 	 */
-	public static function inlineStyle( $contents, $media = 'all' ) {
+	public static function inlineStyle( $contents, $media = 'all', $attribs = [] ) {
 		// Don't escape '>' since that is used
 		// as direct child selector.
 		// Remember, in css, there is no "x" for hexadecimal escapes, and
@@ -609,7 +612,7 @@ class Html {
 
 		return self::rawElement( 'style', [
 			'media' => $media,
-		], $contents );
+		] + $attribs, $contents );
 	}
 
 	/**
diff --git a/includes/api/ApiParse.php b/includes/api/ApiParse.php
index 2839ab9453..3326fab5e4 100644
--- a/includes/api/ApiParse.php
+++ b/includes/api/ApiParse.php
@@ -345,6 +345,7 @@ class ApiParse extends ApiBase {
 				'allowTOC' => !$params['disabletoc'],
 				'enableSectionEditLinks' => !$params['disableeditsection'],
 				'unwrap' => $params['wrapoutputclass'] === '',
+				'deduplicateStyles' => !$params['disablestylededuplication'],
 			] );
 			$result_array[ApiResult::META_BC_SUBELEMENTS][] = 'text';
 		}
@@ -877,6 +878,7 @@ class ApiParse extends ApiBase {
 			'disablelimitreport' => false,
 			'disableeditsection' => false,
 			'disabletidy' => false,
+			'disablestylededuplication' => false,
 			'generatexml' => [
 				ApiBase::PARAM_DFLT => false,
 				ApiBase::PARAM_HELP_MSG => [
diff --git a/includes/api/i18n/en.json b/includes/api/i18n/en.json
index 729c4c75f5..168901971b 100644
--- a/includes/api/i18n/en.json
+++ b/includes/api/i18n/en.json
@@ -391,6 +391,7 @@
 	"apihelp-parse-param-disablepp": "Use <var>$1disablelimitreport</var> instead.",
 	"apihelp-parse-param-disableeditsection": "Omit edit section links from the parser output.",
 	"apihelp-parse-param-disabletidy": "Do not run HTML cleanup (e.g. tidy) on the parser output.",
+	"apihelp-parse-param-disablestylededuplication": "Do not deduplicate inline stylesheets in the parser output.",
 	"apihelp-parse-param-generatexml": "Generate XML parse tree (requires content model <code>$1</code>; replaced by <kbd>$2prop=parsetree</kbd>).",
 	"apihelp-parse-param-preview": "Parse in preview mode.",
 	"apihelp-parse-param-sectionpreview": "Parse in section preview mode (enables preview mode too).",
diff --git a/includes/api/i18n/qqq.json b/includes/api/i18n/qqq.json
index 1e4bfc8d29..e76988058b 100644
--- a/includes/api/i18n/qqq.json
+++ b/includes/api/i18n/qqq.json
@@ -371,6 +371,7 @@
 	"apihelp-parse-param-disablepp": "{{doc-apihelp-param|parse|disablepp}}",
 	"apihelp-parse-param-disableeditsection": "{{doc-apihelp-param|parse|disableeditsection}}",
 	"apihelp-parse-param-disabletidy": "{{doc-apihelp-param|parse|disabletidy}}",
+	"apihelp-parse-param-disablestylededuplication": "{{doc-apihelp-param|parse|disablestylededuplication}}",
 	"apihelp-parse-param-generatexml": "{{doc-apihelp-param|parse|generatexml|params=* $1 - Value of the constant CONTENT_MODEL_WIKITEXT|paramstart=2}}",
 	"apihelp-parse-param-preview": "{{doc-apihelp-param|parse|preview}}",
 	"apihelp-parse-param-sectionpreview": "{{doc-apihelp-param|parse|sectionpreview}}",
diff --git a/includes/parser/ParserOutput.php b/includes/parser/ParserOutput.php
index 346a151bee..19375e0c21 100644
--- a/includes/parser/ParserOutput.php
+++ b/includes/parser/ParserOutput.php
@@ -21,6 +21,7 @@
  * @file
  * @ingroup Parser
  */
+
 class ParserOutput extends CacheTime {
 	/**
 	 * Feature flags to indicate to extensions that MediaWiki core supports and
@@ -270,6 +271,12 @@ class ParserOutput extends CacheTime {
 	 *     section edit link tokens are present in the HTML. Default is true,
 	 *     but might be statefully overridden.
 	 *  - unwrap: (bool) Remove a wrapping mw-parser-output div. Default is false.
+	 *  - deduplicateStyles: (bool) When true, which is the default, `<style>`
+	 *    tags with the `data-mw-deduplicate` attribute set are deduplicated by
+	 *    value of the attribute: all but the first will be replaced by `<link
+	 *    rel="mw-deduplicated-inline-style" href="mw-data:..."/>` tags, where
+	 *    the scheme-specific-part of the href is the (percent-encoded) value
+	 *    of the `data-mw-deduplicate` attribute.
 	 * @return string HTML
 	 */
 	public function getText( $options = [] ) {
@@ -290,6 +297,7 @@ class ParserOutput extends CacheTime {
 			'allowTOC' => !empty( $this->mTOCEnabled ),
 			'enableSectionEditLinks' => $this->mEditSectionTokens,
 			'unwrap' => false,
+			'deduplicateStyles' => true,
 		];
 		$text = $this->mText;
 
@@ -350,6 +358,35 @@ class ParserOutput extends CacheTime {
 			);
 		}
 
+		if ( $options['deduplicateStyles'] ) {
+			$seen = [];
+			$text = preg_replace_callback(
+				'#<style\s+([^>]*data-mw-deduplicate\s*=[^>]*)>.*?</style>#s',
+				function ( $m ) use ( &$seen ) {
+					$attr = Sanitizer::decodeTagAttributes( $m[1] );
+					if ( !isset( $attr['data-mw-deduplicate'] ) ) {
+						return $m[0];
+					}
+
+					$key = $attr['data-mw-deduplicate'];
+					if ( !isset( $seen[$key] ) ) {
+						$seen[$key] = true;
+						return $m[0];
+					}
+
+					// We were going to use an empty <style> here, but there
+					// was concern that would be too much overhead for browsers.
+					// So let's hope a <link> with a non-standard rel and href isn't
+					// going to be misinterpreted or mangled by any subsequent processing.
+					return Html::element( 'link', [
+						'rel' => 'mw-deduplicated-inline-style',
+						'href' => "mw-data:" . wfUrlencode( $key ),
+					] );
+				},
+				$text
+			);
+		}
+
 		return $text;
 	}
 
diff --git a/tests/phpunit/includes/parser/ParserOutputTest.php b/tests/phpunit/includes/parser/ParserOutputTest.php
index 1f3ee67433..44c1773adb 100644
--- a/tests/phpunit/includes/parser/ParserOutputTest.php
+++ b/tests/phpunit/includes/parser/ParserOutputTest.php
@@ -153,6 +153,19 @@ class ParserOutputTest extends MediaWikiTestCase {
 <h2><span class="mw-headline" id="Section_3">Section 3</span><mw:editsection page="Test Page" section="4">Section 3</mw:editsection></h2>
 <p>Three
 </p></div>
+EOF;
+
+		$dedupText = <<<EOF
+<p>This is a test document.</p>
+<style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
+<style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
+<style data-mw-deduplicate="duplicate2">.Duplicate2 {}</style>
+<style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
+<style data-mw-deduplicate="duplicate2">.Duplicate2 {}</style>
+<style data-mw-not-deduplicate="duplicate1">.Duplicate1 {}</style>
+<style data-mw-deduplicate="duplicate1">.Same-attribute-different-content {}</style>
+<style data-mw-deduplicate="duplicate3">.Duplicate1 {}</style>
+<style>.Duplicate1 {}</style>
 EOF;
 
 		return [
@@ -354,6 +367,23 @@ EOF
 <!-- Saved in parser cache... -->', '<p>Test document.</p>
 <!-- Saved in parser cache... -->'
 			],
+			'Style deduplication' => [
+				[], [], $dedupText, <<<EOF
+<p>This is a test document.</p>
+<style data-mw-deduplicate="duplicate1">.Duplicate1 {}</style>
+<link rel="mw-deduplicated-inline-style" href="mw-data:duplicate1"/>
+<style data-mw-deduplicate="duplicate2">.Duplicate2 {}</style>
+<link rel="mw-deduplicated-inline-style" href="mw-data:duplicate1"/>
+<link rel="mw-deduplicated-inline-style" href="mw-data:duplicate2"/>
+<style data-mw-not-deduplicate="duplicate1">.Duplicate1 {}</style>
+<link rel="mw-deduplicated-inline-style" href="mw-data:duplicate1"/>
+<style data-mw-deduplicate="duplicate3">.Duplicate1 {}</style>
+<style>.Duplicate1 {}</style>
+EOF
+			],
+			'Style deduplication disabled' => [
+				[ 'deduplicateStyles' => false ], [], $dedupText, $dedupText
+			],
 		];
 		// phpcs:enable
 	}
-- 
2.20.1