From e4346b778d2fcad3ec573bcdf0b9c1c717657caa Mon Sep 17 00:00:00 2001
From: Bowei Zhang <zbw0517@gmail.com>
Date: Mon, 11 Nov 2024 14:15:30 -0800
Subject: [PATCH] fix: better handling of hyperlinks with parentheses

---
 config/src/config.rs     |  8 ++++---
 termwiz/src/hyperlink.rs | 50 ++++++++++++++++++++++++++++++++++++++++
 2 files changed, 55 insertions(+), 3 deletions(-)

diff --git a/config/src/config.rs b/config/src/config.rs
index 113cdbfba4c..644c2def21f 100644
--- a/config/src/config.rs
+++ b/config/src/config.rs
@@ -1666,9 +1666,11 @@ pub fn default_hyperlink_rules() -> Vec<hyperlink::Rule> {
         hyperlink::Rule::with_highlight(r"\((\w+://\S+)\)", "$1", 1).unwrap(),
         hyperlink::Rule::with_highlight(r"\[(\w+://\S+)\]", "$1", 1).unwrap(),
         hyperlink::Rule::with_highlight(r"<(\w+://\S+)>", "$1", 1).unwrap(),
-        // Then handle URLs not wrapped in brackets
-        // and include terminating ), / or - characters, if any
-        hyperlink::Rule::new(r"\b\w+://\S+[)/a-zA-Z0-9-]+", "$0").unwrap(),
+        // Then handle URLs not wrapped in brackets that
+        // 1) have a balanced ending parenthesis or
+        hyperlink::Rule::new(hyperlink::CLOSING_PARENTHESIS_HYPERLINK_PATTERN, "$0").unwrap(),
+        // 2) include terminating _, / or - characters, if any
+        hyperlink::Rule::new(hyperlink::GENERIC_HYPERLINK_PATTERN, "$0").unwrap(),
         // implicit mailto link
         hyperlink::Rule::new(r"\b\w+@[\w-]+(\.[\w-]+)+\b", "mailto:$0").unwrap(),
     ]
diff --git a/termwiz/src/hyperlink.rs b/termwiz/src/hyperlink.rs
index 159298f7cde..24749ae77e2 100644
--- a/termwiz/src/hyperlink.rs
+++ b/termwiz/src/hyperlink.rs
@@ -265,6 +265,9 @@ impl<'t> Match<'t> {
         result
     }
 }
+pub const CLOSING_PARENTHESIS_HYPERLINK_PATTERN: &str =
+    r"\b\w+://[^\s()]*\(\S*\)(?=\s|$|[^_-/a-zA-Z0-9])";
+pub const GENERIC_HYPERLINK_PATTERN: &str = r"\b\w+://\S+[_-/a-zA-Z0-9]";
 
 impl Rule {
     /// Construct a new rule.  It may fail if the regex is invalid.
@@ -347,4 +350,51 @@ mod test {
             ]
         );
     }
+
+    #[test]
+    fn parse_with_parentheses() {
+        fn assert_helper(test_uri: &str, expected_uri: &str, msg: &str) {
+            let rules = vec![
+                Rule::new(CLOSING_PARENTHESIS_HYPERLINK_PATTERN, "$0").unwrap(),
+                Rule::new(GENERIC_HYPERLINK_PATTERN, "$0").unwrap(),
+            ];
+
+            assert_eq!(
+                Rule::match_hyperlinks(test_uri, &rules)[0].link.uri,
+                expected_uri,
+                "{}",
+                msg,
+            );
+        }
+
+        assert_helper(
+            "   http://example.com)",
+            "http://example.com",
+            "Unblanced terminating parenthesis should not be captured.",
+        );
+
+        assert_helper(
+            "http://example.com/(complete_parentheses)",
+            "http://example.com/(complete_parentheses)",
+            "Balanced terminating parenthesis should be captureed.",
+        );
+
+        assert_helper(
+            "http://example.com/(complete_parentheses)>",
+            "http://example.com/(complete_parentheses)",
+            "Non-URL characters after a balanced terminating parenthesis should be dropped.",
+        );
+
+        assert_helper(
+            "http://example.com/(complete_parentheses))",
+            "http://example.com/(complete_parentheses))",
+            "Non-terminating parentheses should not impact matching the entire URL - Terminated with )",
+        );
+
+        assert_helper(
+            "http://example.com/(complete_parentheses)-((-)-()-_-",
+            "http://example.com/(complete_parentheses)-((-)-()-_-",
+            "Non-terminating parentheses should not impact matching the entire URL - Terminated with a valid character",
+        );
+    }
 }