complete the relationship as you type

src/blogroll/__tests__/deprecated.test.js

@@ -9,7 +9,9 @@ const legacy = { links: [ { url: 'https://a.example/', name: 'A' } ] };
 
 describe( 'hasLinkBlocks', () => {
 	it( 'counts link blocks only', () => {
-		expect( hasLinkBlocks( [ link( 'https://a.example/' ) ] ) ).toBe( true );
+		expect( hasLinkBlocks( [ link( 'https://a.example/' ) ] ) ).toBe(
+			true
+		);
 		expect( hasLinkBlocks( [ paragraph ] ) ).toBe( false );
 		expect( hasLinkBlocks( [] ) ).toBe( false );
 		expect( hasLinkBlocks( undefined ) ).toBe( false );

src/blogroll/components/xfn-control.js

@@ -15,6 +15,10 @@ import { ALL_TOKENS, sanitizeXfn } from './xfn';
  * A token field over the XFN vocabulary; the group rules are applied
  * on every change, so conflicting tokens replace each other.
  *
+ * Typing completes: Enter takes the first match, so "fr" becomes
+ * "friend". A word that is no relationship stays in the field instead of
+ * being swallowed, the vocabulary is the one of the XFN spec.
+ *
  * @param {Object}   props          Component props.
  * @param {string[]} props.value    Selected tokens.
  * @param {Function} props.onChange Change handler.
@@ -25,6 +29,10 @@ export default function XfnControl( { value = [], onChange } ) {
 			__next40pxDefaultSize
 			__nextHasNoMarginBottom
 			__experimentalExpandOnFocus
+			__experimentalAutoSelectFirstMatch
+			__experimentalValidateInput={ ( token ) =>
+				ALL_TOKENS.includes( token )
+			}
 			label={ __( 'Relationship', 'blockroll' ) }
 			value={ value }
 			suggestions={ ALL_TOKENS }