Convert to blocks: upgrade and remove spans if possible

packages/blocks/src/api/raw-handling/anchor-reducer.js

@@ -0,0 +1,25 @@
+export default function anchorReducer( node ) {
+	if ( node.nodeName === 'A' ) {
+		// In jsdom-jscore, 'node.target' can be null.
+		// TODO: Explore fixing this by patching jsdom-jscore.
+		if ( node.target && node.target.toLowerCase() === '_blank' ) {
+			node.rel = 'noreferrer noopener';
+		} else {
+			node.removeAttribute( 'target' );
+			node.removeAttribute( 'rel' );
+		}
+
+		// Saves anchor elements name attribute as id
+		if ( node.name && ! node.id ) {
+			node.id = node.name;
+		}
+
+		// Keeps id only if there is an internal link pointing to it
+		if (
+			node.id &&
+			! node.ownerDocument.querySelector( `[href="#${ node.id }"]` )
+		) {
+			node.removeAttribute( 'id' );
+		}
+	}
+}

packages/blocks/src/api/raw-handling/empty-span-remover.js

@@ -0,0 +1,11 @@
+/**
+ * WordPress dependencies
+ */
+import { unwrap } from '@wordpress/dom';
+
+export default ( node ) => {
+	// Remove spans with no attributes.
+	if ( node.nodeName === 'SPAN' && ! node.attributes.length ) {
+		unwrap( node );
+	}
+};

packages/blocks/src/api/raw-handling/index.js

@@ -14,6 +14,8 @@ import specialCommentConverter from './special-comment-converter';
 import listReducer from './list-reducer';
 import blockquoteNormaliser from './blockquote-normaliser';
 import figureContentReducer from './figure-content-reducer';
+import phrasingContentReducer from './phrasing-content-reducer';
+import emptySpanRemover from './empty-span-remover';
 import shortcodeConverter from './shortcode-converter';
 import { deepFilterHTML, getBlockContentSchema } from './utils';
 
@@ -69,9 +71,11 @@ export function rawHandler( { HTML = '' } ) {
 				specialCommentConverter,
 				// Needed to create media blocks.
 				figureContentReducer,
+				phrasingContentReducer,
 				// Needed to create the quote block, which cannot handle text
 				// without wrapper paragraphs.
 				blockquoteNormaliser( { raw: true } ),
+				emptySpanRemover,
 			];
 
 			piece = deepFilterHTML( piece, filters, blockContentSchema );

packages/blocks/src/api/raw-handling/paste-handler.js

@@ -15,6 +15,7 @@ import specialCommentConverter from './special-comment-converter';
 import commentRemover from './comment-remover';
 import isInlineContent from './is-inline-content';
 import phrasingContentReducer from './phrasing-content-reducer';
+import anchorReducer from './anchor-reducer';
 import headRemover from './head-remover';
 import msListConverter from './ms-list-converter';
 import msListIgnore from './ms-list-ignore';
@@ -48,6 +49,7 @@ function filterInlineHTML( HTML ) {
 		googleDocsUIDRemover,
 		msListIgnore,
 		phrasingContentReducer,
+		anchorReducer,
 		commentRemover,
 	] );
 	HTML = removeInvalidHTML( HTML, getPhrasingContentSchema( 'paste' ), {
@@ -193,6 +195,7 @@ export function pasteHandler( {
 				listReducer,
 				imageCorrector,
 				phrasingContentReducer,
+				anchorReducer,
 				specialCommentConverter,
 				commentRemover,
 				iframeRemover,

packages/blocks/src/api/raw-handling/phrasing-content-reducer.js

@@ -3,6 +3,13 @@
  */
 import { wrap, replaceTag } from '@wordpress/dom';
 
+function removeStyle( node, property ) {
+	node.style[ property ] = '';
+	if ( ! node.style.length ) {
+		node.removeAttribute( 'style' );
+	}
+}
+
 export default function phrasingContentReducer( node, doc ) {
 	// In jsdom-jscore, 'node.style' can be null.
 	// TODO: Explore fixing this by patching jsdom-jscore.
@@ -15,11 +22,17 @@ export default function phrasingContentReducer( node, doc ) {
 			verticalAlign,
 		} = node.style;
 
+		if ( fontWeight === 'normal' || fontWeight === '400' ) {
+			removeStyle( node, 'fontWeight' );
+		}
+
 		if ( fontWeight === 'bold' || fontWeight === '700' ) {
+			removeStyle( node, 'fontWeight' );
 			wrap( doc.createElement( 'strong' ), node );
 		}
 
 		if ( fontStyle === 'italic' ) {
+			removeStyle( node, 'fontStyle' );
 			wrap( doc.createElement( 'em' ), node );
 		}
 
@@ -30,39 +43,20 @@ export default function phrasingContentReducer( node, doc ) {
 			textDecorationLine === 'line-through' ||
 			textDecoration.includes( 'line-through' )
 		) {
+			removeStyle( node, 'textDecoration' );
 			wrap( doc.createElement( 's' ), node );
 		}
 
 		if ( verticalAlign === 'super' ) {
+			removeStyle( node, 'verticalAlign' );
 			wrap( doc.createElement( 'sup' ), node );
 		} else if ( verticalAlign === 'sub' ) {
+			removeStyle( node, 'verticalAlign' );
 			wrap( doc.createElement( 'sub' ), node );
 		}
 	} else if ( node.nodeName === 'B' ) {
 		node = replaceTag( node, 'strong' );
 	} else if ( node.nodeName === 'I' ) {
 		node = replaceTag( node, 'em' );
-	} else if ( node.nodeName === 'A' ) {
-		// In jsdom-jscore, 'node.target' can be null.
-		// TODO: Explore fixing this by patching jsdom-jscore.
-		if ( node.target && node.target.toLowerCase() === '_blank' ) {
-			node.rel = 'noreferrer noopener';
-		} else {
-			node.removeAttribute( 'target' );
-			node.removeAttribute( 'rel' );
-		}
-
-		// Saves anchor elements name attribute as id
-		if ( node.name && ! node.id ) {
-			node.id = node.name;
-		}
-
-		// Keeps id only if there is an internal link pointing to it
-		if (
-			node.id &&
-			! node.ownerDocument.querySelector( `[href="#${ node.id }"]` )
-		) {
-			node.removeAttribute( 'id' );
-		}
 	}
 }

packages/blocks/src/api/raw-handling/test/anchor-reducer.js

@@ -0,0 +1,35 @@
+/**
+ * Internal dependencies
+ */
+import anchorReducer from '../anchor-reducer';
+import { deepFilterHTML } from '../utils';
+
+describe( 'anchorReducer', () => {
+	it( 'should normalise the rel attribute', () => {
+		const input =
+			'<a href="https://wordpress.org" target="_blank">WordPress</a>';
+		const output =
+			'<a href="https://wordpress.org" target="_blank" rel="noreferrer noopener">WordPress</a>';
+		expect( deepFilterHTML( input, [ anchorReducer ], {} ) ).toEqual(
+			output
+		);
+	} );
+
+	it( 'should only allow target="_blank"', () => {
+		const input =
+			'<a href="https://wordpress.org" target="_self">WordPress</a>';
+		const output = '<a href="https://wordpress.org">WordPress</a>';
+		expect( deepFilterHTML( input, [ anchorReducer ], {} ) ).toEqual(
+			output
+		);
+	} );
+
+	it( 'should remove the rel attribute when target is not set', () => {
+		const input =
+			'<a href="https://wordpress.org" rel="noopener">WordPress</a>';
+		const output = '<a href="https://wordpress.org">WordPress</a>';
+		expect( deepFilterHTML( input, [ anchorReducer ], {} ) ).toEqual(
+			output
+		);
+	} );
+} );

packages/blocks/src/api/raw-handling/test/phrasing-content-reducer.js

@@ -12,9 +12,7 @@ describe( 'phrasingContentReducer', () => {
 				[ phrasingContentReducer ],
 				{}
 			)
-		).toEqual(
-			'<strong><span style="font-weight:bold">test</span></strong>'
-		);
+		).toEqual( '<strong><span>test</span></strong>' );
 	} );
 
 	it( 'should transform numeric font weight', () => {
@@ -24,9 +22,7 @@ describe( 'phrasingContentReducer', () => {
 				[ phrasingContentReducer ],
 				{}
 			)
-		).toEqual(
-			'<strong><span style="font-weight:700">test</span></strong>'
-		);
+		).toEqual( '<strong><span>test</span></strong>' );
 	} );
 
 	it( 'should transform font style', () => {
@@ -36,7 +32,7 @@ describe( 'phrasingContentReducer', () => {
 				[ phrasingContentReducer ],
 				{}
 			)
-		).toEqual( '<em><span style="font-style:italic">test</span></em>' );
+		).toEqual( '<em><span>test</span></em>' );
 	} );
 
 	it( 'should transform nested formatting', () => {
@@ -46,36 +42,6 @@ describe( 'phrasingContentReducer', () => {
 				[ phrasingContentReducer ],
 				{}
 			)
-		).toEqual(
-			'<strong><em><span style="font-style:italic;font-weight:bold">test</span></em></strong>'
-		);
-	} );
-
-	it( 'should normalise the rel attribute', () => {
-		const input =
-			'<a href="https://wordpress.org" target="_blank">WordPress</a>';
-		const output =
-			'<a href="https://wordpress.org" target="_blank" rel="noreferrer noopener">WordPress</a>';
-		expect(
-			deepFilterHTML( input, [ phrasingContentReducer ], {} )
-		).toEqual( output );
-	} );
-
-	it( 'should only allow target="_blank"', () => {
-		const input =
-			'<a href="https://wordpress.org" target="_self">WordPress</a>';
-		const output = '<a href="https://wordpress.org">WordPress</a>';
-		expect(
-			deepFilterHTML( input, [ phrasingContentReducer ], {} )
-		).toEqual( output );
-	} );
-
-	it( 'should remove the rel attribute when target is not set', () => {
-		const input =
-			'<a href="https://wordpress.org" rel="noopener">WordPress</a>';
-		const output = '<a href="https://wordpress.org">WordPress</a>';
-		expect(
-			deepFilterHTML( input, [ phrasingContentReducer ], {} )
-		).toEqual( output );
+		).toEqual( '<strong><em><span>test</span></em></strong>' );
 	} );
 } );

test/integration/__snapshots__/blocks-raw-handling.test.js.snap

@@ -218,3 +218,9 @@ exports[`rawHandler should preserve all paragraphs 1`] = `
 <p></p>
 <!-- /wp:paragraph -->"
 `;
+
+exports[`rawHandler should remove convertable spans 1`] = `
+"<!-- wp:paragraph -->
+<p>a</p>
+<!-- /wp:paragraph -->"
+`;

test/integration/blocks-raw-handling.test.js

@@ -569,4 +569,9 @@ describe( 'rawHandler', () => {
 <p style="border: 1px solid tomato;"></p>`;
 		expect( serialize( rawHandler( { HTML } ) ) ).toMatchSnapshot();
 	} );
+
+	it( 'should remove convertable spans', () => {
+		const HTML = `<span style="font-weight: 400">a</span>`;
+		expect( serialize( rawHandler( { HTML } ) ) ).toMatchSnapshot();
+	} );
 } );