Improved parsing of URLs with symbols (#432)

* Improved parsing of URLs with symbols * Remove NUM from LocalPartAccepting token group Interferes with domain tokens when parsing numbers such as www.drive1.com * Complex email HTML test
nfrasser · Mar 25, 2023 · afc51f1 · afc51f1
1 parent faab3df
commit afc51f1
Show file tree

Hide file tree

Showing 5 changed files with 59 additions and 8 deletions.
diff --git a/packages/linkifyjs/src/parser.js b/packages/linkifyjs/src/parser.js
@@ -81,7 +81,6 @@ export function init({ groups }) {
 		tk.DOLLAR,
 		tk.EQUALS,
 		tk.HYPHEN,
-		tk.NUM,
 		tk.OPENBRACE,
 		tk.PERCENT,
 		tk.PIPE,
@@ -184,11 +183,11 @@ export function init({ groups }) {
 	tt(DomainDotTldColonPort, tk.SLASH, Url);
 
 	// Note that domains that begin with schemes are treated slighly differently
-	const UriPrefix = tt(Scheme, tk.COLON); // e.g., 'mailto:' or 'http://'
+	const SchemeColon = tt(Scheme, tk.COLON); // e.g., 'mailto:'
 	const SlashSchemeColon = tt(SlashScheme, tk.COLON); // e.g., 'http:'
 	const SlashSchemeColonSlash = tt(SlashSchemeColon, tk.SLASH); // e.g., 'http:/'
 
-	tt(SlashSchemeColonSlash, tk.SLASH, UriPrefix);
+	const UriPrefix = tt(SlashSchemeColonSlash, tk.SLASH); // e.g., 'http://'
 
 	// Scheme states can transition to domain states
 	ta(Scheme, groups.domain, Domain);
@@ -199,7 +198,10 @@ export function init({ groups }) {
 	tt(SlashScheme, tk.HYPHEN, DomainHyphen);
 
 	// Force URL with scheme prefix followed by anything sane
+	ta(SchemeColon, groups.domain, Url);
+	tt(SchemeColon, tk.SLASH, Url);
 	ta(UriPrefix, groups.domain, Url);
+	ta(UriPrefix, qsAccepting, Url);
 	tt(UriPrefix, tk.SLASH, Url);
 
 	// URL, followed by an opening bracket

diff --git a/test/spec/html/email.html b/test/spec/html/email.html
diff --git a/test/spec/html/options.js b/test/spec/html/options.js
@@ -17,6 +17,7 @@ export default {
 		.split('\n'),
 
 	extra: fs.readFileSync(__dirname + '/extra.html', 'utf8').trim(), // for jQuery plugin tests
+	email: fs.readFileSync(__dirname + '/email.html', 'utf8').trim(), // for linkify-html performance tests
 	altOptions: {
 		className: 'linkified',
 		rel: 'nofollow',

diff --git a/test/spec/linkify-html.test.js b/test/spec/linkify-html.test.js
@@ -199,4 +199,8 @@ describe('linkify-html', () => {
 		const input = '這禮拜是我們新的循環 (3/23-4/19), 我將於這週日給 Jeffrey 補課，並且我們會在這期間選另外一個可以上課的日期。';
 		expect(linkifyHtml(input)).to.be.ok;
 	});
+
+	it('Handles complex email page', () => {
+		expect(linkifyHtml(htmlOptions.email)).to.be.ok;
+	});
 });
diff --git a/test/spec/linkifyjs/parser.test.js b/test/spec/linkifyjs/parser.test.js
@@ -220,6 +220,10 @@ const tests = [
 		'~@example.org',
 		[Email],
 		['~@example.org']
+	], [
+		'~emersion/soju-dev@lists.sr.ht',
+		[Email],
+		['~emersion/soju-dev@lists.sr.ht']
 	], [
 		'test@example2.com',
 		[Email],
@@ -228,10 +232,6 @@ const tests = [
 		'noreply@500px.so',
 		[Email],
 		['noreply@500px.so']
-	], [
-		'~emersion/soju-dev@lists.sr.ht',
-		[Email],
-		['~emersion/soju-dev@lists.sr.ht']
 	], [
 		'http@example.com',
 		[Email],
@@ -264,7 +264,15 @@ const tests = [
 		'Hello\nWorld',
 		[Text, Nl, Text],
 		['Hello', '\n', 'World'],
-	]
+	], [
+		'And http://↑↑↓↓←→←→ba.tk/ is also a URL',
+		[Text, Url, Text],
+		['And ', 'http://↑↑↓↓←→←→ba.tk/', ' is also a URL']
+	], [
+		'This Url www.drive1.com with www and digits also www.500px.com',
+		[Text, Url, Text, Url],
+		['This Url ', 'www.drive1.com', ' with www and digits also ', 'www.500px.com']
+	],
 ];