Improve code with tests and adding check for TLDs

packages/block-editor/src/components/link-control/is-url-like.js

@@ -24,11 +24,38 @@ export default function isURLLike( val ) {
 	const protocol = getProtocol( val );
 	const protocolIsValid = isValidProtocol( protocol );
 
+	const mayBeTLD = hasPossibleTLD( val );
+
 	const isWWW = val?.startsWith( 'www.' );
 	const isHTTPProtocol = /^(http|https)/.test( val ) && protocolIsValid;
 	const isMailTo = val?.startsWith( 'mailto:' ) && protocolIsValid;
 	const isTel = val?.startsWith( 'tel:' ) && protocolIsValid;
 	const isInternal = val?.startsWith( '#' ) && isValidFragment( val );
 
-	return isHTTPProtocol || isWWW || isMailTo || isTel || isInternal;
+	return (
+		isHTTPProtocol || isWWW || isMailTo || isTel || isInternal || mayBeTLD
+	);
+}
+
+/**
+ * Checks if a given URL has a valid Top-Level Domain (TLD).
+ *
+ * @param {string} url       - The URL to check.
+ * @param {number} maxLength - The maximum length of the TLD.
+ * @return {boolean} Returns true if the URL has a valid TLD, false otherwise.
+ */
+function hasPossibleTLD( url, maxLength = 6 ) {
+	// Clean the URL by removing anything after the first occurrence of "?" or "#".
+	const cleanedURL = url.split( /[?#]/ )[ 0 ];
+
+	// Regular expression explanation:
+	// - (?<=\S)                  : Positive lookbehind assertion to ensure there is at least one non-whitespace character before the TLD
+	// - \.                       : Matches a literal dot (.)
+	// - [a-zA-Z_]{2,maxLength}   : Matches 2 to maxLength letters or underscores, representing the TLD
+	// - (?:\/|$)                 : Non-capturing group that matches either a forward slash (/) or the end of the string
+	const regex = new RegExp(
+		`(?<=\\S)\\.(?:[a-zA-Z_]{2,${ maxLength }})(?:\\/|$)`
+	);
+
+	return regex.test( cleanedURL );
 }

packages/block-editor/src/components/link-control/test/is-url-like.js

@@ -40,4 +40,27 @@ describe( 'isURLLike', () => {
 	it( 'returns true for internal anchor ("hash") links.', () => {
 		expect( isURLLike( '#someinternallink' ) ).toBe( true );
 	} );
+
+	// use .each to test multiple cases
+	it.each( [
+		[ true, 'http://example.com' ],
+		[ true, 'https://test.co.uk?query=param' ],
+		[ true, 'ftp://openai.ai?param=value#section' ],
+		[ true, 'example.com' ],
+		[ true, 'http://example.com?query=param#section' ],
+		[ true, 'https://test.co.uk/some/path' ],
+		[ true, 'ftp://openai.ai/some/path' ],
+		[ true, 'example.org/some/path' ],
+		[ true, 'example_test.tld' ],
+		[ true, 'example_test.com' ],
+		[ false, 'example' ],
+		[ false, '.com' ],
+		[ true, '_test.com' ],
+		[ true, 'http://example_test.com' ],
+	] )(
+		'returns %s when testing against string "%s" for a valid TLD',
+		( expected, testString ) => {
+			expect( isURLLike( testString ) ).toBe( expected );
+		}
+	);
 } );