From a816baedd98c6d9d0576fb173538d65fd9c41e80 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Bartosz=20Dziewo=C5=84ski?= <matma.rex@gmail.com>
Date: Fri, 7 Feb 2020 01:15:49 +0100
Subject: [PATCH] Fix handling of pasted internal red links

Bug: T239550
Change-Id: Iacfba4b46bea8294f12a0c010344fda317f75df6
---
 .../ve.dm.MWInternalLinkAnnotation.js         | 22 +++++++++----
 .../dm/ve.dm.MWInternalLinkAnnotation.test.js | 32 +++++++++++++++++--
 2 files changed, 46 insertions(+), 8 deletions(-)

diff --git a/modules/ve-mw/dm/annotations/ve.dm.MWInternalLinkAnnotation.js b/modules/ve-mw/dm/annotations/ve.dm.MWInternalLinkAnnotation.js
index 3afb2d2cc3..ed77449793 100644
--- a/modules/ve-mw/dm/annotations/ve.dm.MWInternalLinkAnnotation.js
+++ b/modules/ve-mw/dm/annotations/ve.dm.MWInternalLinkAnnotation.js
@@ -128,28 +128,38 @@ ve.dm.MWInternalLinkAnnotation.static.newFromTitle = function ( title, rawTitle
  *    True if the href pointed to the local wiki, false if href is external
  */
 ve.dm.MWInternalLinkAnnotation.static.getTargetDataFromHref = function ( href, doc ) {
-	var relativeBase, relativeBaseRegex, relativeHref, isInternal, matches, data;
+	var relativeBase, relativeBaseRegex, relativeHref, isInternal, matches, data, uri;
 
 	function regexEscape( str ) {
 		return str.replace( /([.?*+^$[\]\\(){}|-])/g, '\\$1' );
 	}
 
-	// Protocol relative base
-	relativeBase = ve.resolveUrl( mw.config.get( 'wgArticlePath' ), doc ).replace( /^https?:/i, '' );
-	relativeBaseRegex = new RegExp( regexEscape( relativeBase ).replace( regexEscape( '$1' ), '(.*)' ) );
 	// Protocol relative href
 	relativeHref = href.replace( /^https?:/i, '' );
 	// Paths without a host portion are assumed to be internal
 	isInternal = !/^\/\//.test( relativeHref );
-	// Check if this matches the server's article path
-	matches = relativeHref.match( relativeBaseRegex );
 
+	// Check if this matches the server's article path
+	// Protocol relative base
+	relativeBase = ve.resolveUrl( mw.config.get( 'wgArticlePath' ), doc ).replace( /^https?:/i, '' );
+	relativeBaseRegex = new RegExp( regexEscape( relativeBase ).replace( regexEscape( '$1' ), '(.*)' ) );
+	matches = relativeHref.match( relativeBaseRegex );
 	if ( matches && matches[ 1 ].split( '#' )[ 0 ].indexOf( '?' ) === -1 ) {
 		// Take the relative path
 		href = matches[ 1 ];
 		isInternal = true;
 	}
 
+	// Check if this matches the server's script path (as used by red links)
+	relativeBase = ve.resolveUrl( mw.config.get( 'wgScript' ), doc ).replace( /^https?:/i, '' );
+	if ( relativeHref.indexOf( relativeBase ) === 0 ) {
+		uri = new mw.Uri( relativeHref );
+		if ( uri.query.title ) {
+			href = uri.query.title;
+			isInternal = true;
+		}
+	}
+
 	// This href doesn't necessarily come from Parsoid (and it might not have the "./" prefix), but
 	// this method will work fine.
 	data = ve.parseParsoidResourceName( href );
diff --git a/modules/ve-mw/tests/dm/ve.dm.MWInternalLinkAnnotation.test.js b/modules/ve-mw/tests/dm/ve.dm.MWInternalLinkAnnotation.test.js
index 0bcc8409d1..abd009405e 100644
--- a/modules/ve-mw/tests/dm/ve.dm.MWInternalLinkAnnotation.test.js
+++ b/modules/ve-mw/tests/dm/ve.dm.MWInternalLinkAnnotation.test.js
@@ -9,12 +9,27 @@ QUnit.module( 've.dm.MWInternalLinkAnnotation' );
 QUnit.test( 'toDataElement', function ( assert ) {
 	var i, l,
 		doc = ve.dm.example.createExampleDocument(),
-		internalLink = function ( pageTitle ) {
+		externalLink = function ( href ) {
 			var link = document.createElement( 'a' );
-			link.setAttribute( 'href', location.origin + mw.Title.newFromText( pageTitle ).getUrl() );
+			link.setAttribute( 'href', href );
+			return link;
+		},
+		internalLink = function ( pageTitle, params ) {
+			var link = document.createElement( 'a' );
+			link.setAttribute( 'href', location.origin + mw.Title.newFromText( pageTitle ).getUrl( params ) );
 			return link;
 		},
 		cases = [
+			{
+				msg: 'Not an internal link',
+				element: externalLink( 'http://example.com/' ),
+				expected: {
+					type: 'link/mwExternal',
+					attributes: {
+						href: 'http://example.com/'
+					}
+				}
+			},
 			{
 				msg: 'Simple',
 				element: internalLink( 'Foo' ),
@@ -28,6 +43,19 @@ QUnit.test( 'toDataElement', function ( assert ) {
 					}
 				}
 			},
+			{
+				msg: 'Red link',
+				element: internalLink( 'Foo', { action: 'edit', redlink: '1' } ),
+				expected: {
+					type: 'link/mwInternal',
+					attributes: {
+						lookupTitle: 'Foo',
+						normalizedTitle: 'Foo',
+						origTitle: 'Foo',
+						title: 'Foo'
+					}
+				}
+			},
 			{
 				// Because percent-encoded URLs aren't valid titles, but what they decode to might be
 				msg: 'Percent encoded characters',