From 17b7e7de58d0923fc6419b62f836566c4c6e1888 Mon Sep 17 00:00:00 2001
From: Jaapio <jaap@ijaap.nl>
Date: Thu, 29 Aug 2024 21:32:22 +0200
Subject: [PATCH] Fix invalid url parsing

I did not found any specification where an URL can contain parenthesis
in the url, nor we do have a test that fails on this case. If we need
it anyway we can introduce it later on. But for now this fixes a bug
where the fetch of a url is way to gredy.

(cherry picked from commit 075539c228d3621bd697b1366e56a47d325dc45c)
---
 .../RestructuredText/Parser/InlineLexer.php   |  2 +-
 .../tests/unit/Parser/InlineLexerTest.php     | 34 +++++++++++++++++++
 2 files changed, 35 insertions(+), 1 deletion(-)

diff --git a/packages/guides-restructured-text/src/RestructuredText/Parser/InlineLexer.php b/packages/guides-restructured-text/src/RestructuredText/Parser/InlineLexer.php
index bebdfa445..94aaf19ba 100644
--- a/packages/guides-restructured-text/src/RestructuredText/Parser/InlineLexer.php
+++ b/packages/guides-restructured-text/src/RestructuredText/Parser/InlineLexer.php
@@ -74,7 +74,7 @@ protected function getCatchablePatterns(): array
             '|',
             '\\*\\*',
             '\\*',
-            '\b(?<!:)[a-z0-9\\.\-+]{2,}:\\/\\/[-a-zA-Z0-9()@:%_\\+.~#?&\\/=]*[-a-zA-Z0-9()@%_\\+~#&\\/=]', // standalone hyperlinks
+            '\b(?<!:)[a-z0-9\\.\-+]{2,}:\\/\\/[-a-zA-Z0-9@:%_\\+.~#?&\\/=]*[-a-zA-Z0-9@%_\\+~#&\\/=]', // standalone hyperlinks
         ];
     }
 
diff --git a/packages/guides-restructured-text/tests/unit/Parser/InlineLexerTest.php b/packages/guides-restructured-text/tests/unit/Parser/InlineLexerTest.php
index f4f49dfb3..705f9fe15 100644
--- a/packages/guides-restructured-text/tests/unit/Parser/InlineLexerTest.php
+++ b/packages/guides-restructured-text/tests/unit/Parser/InlineLexerTest.php
@@ -17,6 +17,7 @@
 use PHPUnit\Framework\TestCase;
 
 use function PHPUnit\Framework\assertEquals;
+use function trim;
 
 final class InlineLexerTest extends TestCase
 {
@@ -83,4 +84,37 @@ public static function inlineLexerProvider(): array
             ],
         ];
     }
+
+    #[DataProvider('hyperlinkProvider')]
+    public function testHyperlinkEndsBeforeParenthesis(string $url): void
+    {
+        $input = '(text in parenthesis ' . $url . ').';
+        $lexer = new InlineLexer();
+
+        $lexer->setInput($input);
+        $lexer->moveNext();
+
+        for ($i = 0; $i < 21; $i++) {
+            $lexer->moveNext();
+            assertEquals(
+                trim($input[$i]) === '' ? InlineLexer::WHITESPACE : InlineLexer::WORD,
+                $lexer->token?->type,
+            );
+            assertEquals($input[$i], $lexer->token?->value);
+        }
+
+        $lexer->moveNext();
+        assertEquals(InlineLexer::HYPERLINK, $lexer->token?->type);
+        assertEquals($url, $lexer->token?->value);
+    }
+
+    /** @return array<string, array<string>> */
+    public static function hyperlinkProvider(): array
+    {
+        return [
+            'Url with parenthesis' => ['https://www.test.com'],
+            'Url with parenthesis and query' => ['https://www.test.com?query=1'],
+            'Url with parenthesis and query and fragment' => ['https://www.test.com?query=1#fragment'],
+        ];
+    }
 }