From 78e5c8d7af310c0f81da84f232c5a9c0eaf019cd Mon Sep 17 00:00:00 2001
From: "James D. Forrester" <jforrester@wikimedia.org>
Date: Wed, 4 Apr 2018 10:18:56 -0700
Subject: [PATCH] Default installations to using RemexHtml for tidying
MIME-Version: 1.0
Content-Type: text/plain; charset=utf8
Content-Transfer-Encoding: 8bit

This combines two changes – defaulting tidying to on, previously off, and
defaulting the tidying library to RemexHtml, previously the tidy binary.
Config options are going to be a bit of a mess until we drop support for
the old tidy binary config route.

Bug: T185753
Depends-On: I0a8973f508fbf65160177b003260831639828eeb
Change-Id: I6879a77a78d780c7c056d807dde20682c6097d1b
---
 RELEASE-NOTES-1.31                         | 4 ++++
 includes/DefaultSettings.php               | 7 ++++---
 tests/phpunit/includes/parser/TidyTest.php | 4 ++--
 3 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/RELEASE-NOTES-1.31 b/RELEASE-NOTES-1.31
index f40a422adb..2fde0bd505 100644
--- a/RELEASE-NOTES-1.31
+++ b/RELEASE-NOTES-1.31
@@ -31,6 +31,10 @@ production.
 * (T188472) The 'comma' value for $wgArticleCountMethod is no longer supported for
   performance reasons, and installations with this setting will now work as if it
   was configured with 'any'.
+* (T185753) MediaWiki now defaults to using RemexHtml to tidy up user input, rather than
+  being off by default. If you wish to disable HTML tidying entirely, set $wgTidyConfig
+  to null; if you wish to use the old, deprecated Tidy external binary, both
+  set $wgTidyConfig to null and also set $wgUseTidy to true.
 * $wgLogAutopatrol now defaults to false instead of true.
 * $wgValidateAllHtml was removed and will be ignored.
 
diff --git a/includes/DefaultSettings.php b/includes/DefaultSettings.php
index 22f587ee15..a0f849e43a 100644
--- a/includes/DefaultSettings.php
+++ b/includes/DefaultSettings.php
@@ -4264,8 +4264,9 @@ $wgAllowImageTag = false;
 
 /**
  * Configuration for HTML postprocessing tool. Set this to a configuration
- * array to enable an external tool. Dave Raggett's "HTML Tidy" is typically
- * used. See https://www.w3.org/People/Raggett/tidy/
+ * array to enable an external tool. By default, we now use the RemexHtml
+ * library; historically, Dave Raggett's "HTML Tidy" was typically used.
+ * See https://www.w3.org/People/Raggett/tidy/
  *
  * If this is null and $wgUseTidy is true, the deprecated configuration
  * parameters will be used instead.
@@ -4286,7 +4287,7 @@ $wgAllowImageTag = false;
  *  - tidyBin: For RaggettExternal, the path to the tidy binary.
  *  - tidyCommandLine: For RaggettExternal, additional command line options.
  */
-$wgTidyConfig = null;
+$wgTidyConfig = [ 'driver' => 'RemexHtml' ];
 
 /**
  * Set this to true to use the deprecated tidy configuration parameters.
diff --git a/tests/phpunit/includes/parser/TidyTest.php b/tests/phpunit/includes/parser/TidyTest.php
index 62b84aa129..be5125c7e3 100644
--- a/tests/phpunit/includes/parser/TidyTest.php
+++ b/tests/phpunit/includes/parser/TidyTest.php
@@ -55,8 +55,8 @@ MathML;
 				'<editsection> should survive tidy'
 			],
 			[ '<mw:toc>foo</mw:toc>', '<mw:toc>foo</mw:toc>', '<mw:toc> should survive tidy' ],
-			[ "<link foo=\"bar\" />\nfoo", '<link foo="bar"/>foo', '<link> should survive tidy' ],
-			[ "<meta foo=\"bar\" />\nfoo", '<meta foo="bar"/>foo', '<meta> should survive tidy' ],
+			[ "<link foo=\"bar\" />foo", '<link foo="bar"/>foo', '<link> should survive tidy' ],
+			[ "<meta foo=\"bar\" />foo", '<meta foo="bar"/>foo', '<meta> should survive tidy' ],
 			[ $testMathML, $testMathML, '<math> should survive tidy' ],
 		];
 	}
-- 
2.20.1