Merge pull request #15363 from ckeditor/ck/15333-metadata-retained-wh…

…en-copying-from-word-in-windows Fix (paste-from-office): Content from Word documents should be pasted correctly (without HTML styles tag content) on Windows systems. Closes #15333. Closes #9002.
ckeditor · Nov 23, 2023 · 18a88ed · 18a88ed
2 parents f73874c + f9568e7
commit 18a88ed
Show file tree

Hide file tree

Showing 8 changed files with 1,168 additions and 1 deletion.
diff --git a/packages/ckeditor5-paste-from-office/src/filters/parse.ts b/packages/ckeditor5-paste-from-office/src/filters/parse.ts
@@ -29,6 +29,10 @@ export function parseHtml( htmlString: string, stylesProcessor: StylesProcessor
 	// Remove Word specific "if comments" so content inside is not omitted by the parser.
 	htmlString = htmlString.replace( /<!--\[if gte vml 1]>/g, '' );
 
+	// Clean the <head> section of MS Windows specific tags. See https://github.com/ckeditor/ckeditor5/issues/15333.
+	// The regular expression matches the <o:SmartTagType> tag with optional attributes (with or without values).
+	htmlString = htmlString.replace( /<o:SmartTagType(?:\s+[^\s>=]+(?:="[^"]*")?)*\s*\/?>/gi, '' );
+
 	const normalizedHtml = normalizeSpacing( cleanContentAfterBody( htmlString ) );
 
 	// Parse htmlString as native Document object.

diff --git a/packages/ckeditor5-paste-from-office/tests/_data/other/index.js b/packages/ckeditor5-paste-from-office/tests/_data/other/index.js
@@ -0,0 +1,16 @@
+/**
+ * @license Copyright (c) 2003-2023, CKSource Holding sp. z o.o. All rights reserved.
+ * For licensing, see LICENSE.md or https://ckeditor.com/legal/ckeditor-oss-license
+ */
+
+import smartTags from './smart-tags/input.html';
+import smartTagsModel from './smart-tags/model.html';
+
+export const fixtures = {
+	input: {
+		smartTags
+	},
+	model: {
+		smartTags: smartTagsModel
+	}
+};
diff --git a/packages/ckeditor5-paste-from-office/tests/_data/other/smart-tags/15333.docx b/packages/ckeditor5-paste-from-office/tests/_data/other/smart-tags/15333.docx
diff --git a/packages/ckeditor5-paste-from-office/tests/_data/other/smart-tags/input.html b/packages/ckeditor5-paste-from-office/tests/_data/other/smart-tags/input.html
diff --git a/packages/ckeditor5-paste-from-office/tests/_data/other/smart-tags/model.html b/packages/ckeditor5-paste-from-office/tests/_data/other/smart-tags/model.html
@@ -0,0 +1,42 @@
+<paragraph><$text bold="true" fontColor="red">A/K: </$text></paragraph>
+<paragraph><$text bold="true">T ID: 1</$text></paragraph>
+<paragraph></paragraph>
+<paragraph><$text bold="true" fontColor="black">ID 1 –adXY</$text></paragraph>
+<paragraph>M for F Der Transactions with XXXX SYXE dated 06th March 2013</paragraph>
+<paragraph></paragraph>
+<paragraph>Type of Entity:  XXXX company</paragraph>
+<paragraph>Jurisdiction:  Germany</paragraph>
+<paragraph><$text bold="true">C: </$text>Addendum covering </paragraph>
+<paragraph>One-Way:  Two Way </paragraph>
+<paragraph>Go:  German </paragraph>
+<paragraph>Sub:  N/A</paragraph>
+<paragraph>Ind:  zero </paragraph>
+<paragraph>Thresholds:  R: zero</paragraph>
+<paragraph> Co: zero</paragraph>
+<paragraph>Minimum Transfer Amount:  R: EUR1million</paragraph>
+<paragraph> Counterparty: EUR1million</paragraph>
+<paragraph>Rounding:  R: EUR1million</paragraph>
+<paragraph> Co: EUR1million</paragraph>
+<paragraph></paragraph>
+<paragraph>M re to EUR 2,000 (and rounding to EUR 1,000) if either party's rating falls below C/C2</paragraph>
+<paragraph><$text bold="true">El 2-day C:  </$text>Yes</paragraph>
+<paragraph></paragraph>
+<paragraph><$text bold="true" fontColor="black">ID 1 –addfSE</$text></paragraph>
+<paragraph>M A for F D Transactions with XXXX SETXY dated 06th March 2013</paragraph>
+<paragraph></paragraph>
+<paragraph>Type of Entity:  XXX company</paragraph>
+<paragraph>Jurisdiction:  Germany</paragraph>
+<paragraph><$text bold="true" fontColor="black">ID 1 – fSETXX</$text></paragraph>
+<paragraph>Master Agreement for Financial Derivatives Transactions with XXXX STXY dated 06th March 2013</paragraph>
+<paragraph></paragraph>
+<paragraph>Type of Entity:  XXX company</paragraph>
+<paragraph>Jurisdiction:  Germany</paragraph>
+<paragraph></paragraph>
+<paragraph></paragraph>
+<paragraph></paragraph>
+<paragraph><$text bold="true" fontColor="black">GID XXXXX –adE</$text></paragraph>
+<paragraph><$text fontColor="black">Master Agreement for Financial Derivatives Transactions with XYZ COMPANY SE dated 06th March 2013</$text></paragraph>
+<paragraph></paragraph>
+<paragraph><$text fontColor="black">Type of Entity:  XXXYYY company</$text></paragraph>
+<paragraph><$text fontColor="black">Jurisdiction:  Germany</$text></paragraph>
+<paragraph></paragraph>
diff --git a/packages/ckeditor5-paste-from-office/tests/_utils/fixtures.js b/packages/ckeditor5-paste-from-office/tests/_utils/fixtures.js
@@ -16,6 +16,7 @@ import { fixtures as table } from '../_data/table/index.js';
 import { fixtures as pageBreak } from '../_data/page-break/index.js';
 import { fixtures as fontWithoutTableProperties } from '../_data/font-without-table-properties/index';
 import { fixtures as googleDocsBrParagraphs } from '../_data/paste-from-google-docs/br-paragraph/index';
+import { fixtures as smartTags } from '../_data/other/index';
 
 // Generic fixtures.
 export const generic = {
@@ -29,7 +30,8 @@ export const generic = {
 	'google-docs-br-paragraphs': googleDocsBrParagraphs,
 	table,
 	'page-break': pageBreak,
-	'font-without-table-properties': fontWithoutTableProperties
+	'font-without-table-properties': fontWithoutTableProperties,
+	'smart-tags': smartTags
 };
 
 // Browser specific fixtures.

diff --git a/packages/ckeditor5-paste-from-office/tests/data/integration.js b/packages/ckeditor5-paste-from-office/tests/data/integration.js
@@ -130,6 +130,13 @@ describe( 'PasteFromOffice - integration', () => {
 		}
 	} );
 
+	generateIntegrationTests( {
+		input: 'smart-tags',
+		editorConfig: {
+			plugins: [ Clipboard, Paragraph, Bold, PasteFromOffice, FontColor ]
+		}
+	} );
+
 	function generateIntegrationTests( config ) {
 		const commonIntegrationConfig = {
 			type: 'integration',

diff --git a/packages/ckeditor5-paste-from-office/tests/filters/parse.js b/packages/ckeditor5-paste-from-office/tests/filters/parse.js
@@ -172,6 +172,79 @@ describe( 'PasteFromOffice - filters', () => {
 
 				expect( body.getChild( 0 ).name ).to.equal( 'p' );
 			} );
+
+			// See https://github.com/ckeditor/ckeditor5/issues/15333.
+			describe( 'should remove MS Windows specific tags to prevent incorrect parsing of HTML', () => {
+				it( 'should remove <o:SmartTagType> empty tag (with or without `/` at the end)', () => {
+					const html =
+						'<html>' +
+							'<head>' +
+								'<o:SmartTagType/>' +
+								'<o:SmartTagType>' +
+							'</head>' +
+							'<body>' +
+								'<p>foo</p>' +
+							'</body>' +
+						'</html>';
+					const { body, bodyString } = parseHtml( html );
+
+					expect( body ).to.instanceof( DocumentFragment );
+					expect( body.childCount ).to.equal( 1 );
+					expect( bodyString ).to.equal( '<p>foo</p>' );
+				} );
+
+				it( 'should remove <o:SmartTagType> empty tag with white space before the ending', () => {
+					const html =
+						'<html>' +
+							'<head>' +
+								'<o:SmartTagType />' +
+								'<o:SmartTagType >' +
+							'</head>' +
+							'<body>' +
+								'<p>foo</p>' +
+							'</body>' +
+						'</html>';
+					const { body, bodyString } = parseHtml( html );
+
+					expect( body ).to.instanceof( DocumentFragment );
+					expect( body.childCount ).to.equal( 1 );
+					expect( bodyString ).to.equal( '<p>foo</p>' );
+				} );
+
+				it( 'should remove <o:SmartTagType> tag with attributes (with and without values)', () => {
+					const html =
+						'<html>' +
+							'<head>' +
+								'<o:SmartTagType namespaceuri="foo:bar:smarttags" baz />' +
+							'</head>' +
+							'<body>' +
+								'<p>foo</p>' +
+							'</body>' +
+						'</html>';
+					const { body, bodyString } = parseHtml( html );
+
+					expect( body ).to.instanceof( DocumentFragment );
+					expect( body.childCount ).to.equal( 1 );
+					expect( bodyString ).to.equal( '<p>foo</p>' );
+				} );
+
+				it( 'should remove <o:SmartTagType> tag with attributes containing `>`', () => {
+					const html =
+						'<html>' +
+							'<head>' +
+								'<o:SmartTagType namespaceuri="foo>bar>smarttags" />' +
+							'</head>' +
+							'<body>' +
+								'<p>foo</p>' +
+							'</body>' +
+						'</html>';
+					const { body, bodyString } = parseHtml( html );
+
+					expect( body ).to.instanceof( DocumentFragment );
+					expect( body.childCount ).to.equal( 1 );
+					expect( bodyString ).to.equal( '<p>foo</p>' );
+				} );
+			} );
 		} );
 	} );
 } );