From c418ae577c4d99202c2b630b06ee5eca76a523b0 Mon Sep 17 00:00:00 2001
From: Erlend Lind Madsen <erlendf80@gmail.com>
Date: Fri, 2 Feb 2024 04:55:07 +0100
Subject: [PATCH] url: add missing rarer (all) characters allowed in URL

---
 src/config/url.zig | 12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/src/config/url.zig b/src/config/url.zig
index e3681c11a..f8969872f 100644
--- a/src/config/url.zig
+++ b/src/config/url.zig
@@ -22,7 +22,7 @@ const oni = @import("oniguruma");
 ///
 /// There are many complicated cases where these heuristics break down, but
 /// handling them well requires a non-regex approach.
-pub const regex = "(?:" ++ url_scheme ++ ")(?:[\\w./+:@%?=&-]+(?:\\(\\w*\\))?)+(?<!\\.)";
+pub const regex = "(?:" ++ url_scheme ++ ")(?:[\\w\\-.~:/?#\\[\\]@!$&*+,;=%]+(?:\\(\\w*\\))?)+(?<!\\.)";
 const url_scheme = "https?://|mailto:|ftp://|file:|ssh:|git://|ssh://|tel:|magnet:|ipfs://|ipns://|gemini://|gopher://|news:";
 
 test "url regex" {
@@ -97,6 +97,16 @@ test "url regex" {
             .input = "url with dashes [mode 2027](https://github.com/contour-terminal/terminal-unicode-core) for better unicode support",
             .expect = "https://github.com/contour-terminal/terminal-unicode-core",
         },
+        // weird characters in URL
+        .{
+            .input = "weird characters https://example.com/~user/?query=1&other=2#hash and more",
+            .expect = "https://example.com/~user/?query=1&other=2#hash",
+        },
+        // square brackets in URL
+        .{
+            .input = "square brackets https://example.com/[foo] and more",
+            .expect = "https://example.com/[foo]",
+        },
         // remaining URL schemes tests
         .{
             .input = "match ftp://example.com ftp links",