From 34047f8da8e191324121023c7cb040a120e2a523 Mon Sep 17 00:00:00 2001
From: Kafji <k+github@kafji.net>
Date: Mon, 10 Mar 2025 10:24:48 +0700
Subject: [PATCH] Allow domain-like as URL location input (#35756)

* Allow domain-like as URL location input

Before this patch, domain with subdomain (e.g. book.servo.org) won't be
treated as URL location.

This patch retifies that by adding Firefox's location bar behavior:

  - book.servo.org is URL location
  - book.servo.org. is URL location
  - .book.servo.org is not URL location

Fixes #35754.

Signed-off-by: Kafji <k@kafji.net>

* Chain location input interpretation attempts

Signed-off-by: Kafji <k@kafji.net>

---------

Signed-off-by: Kafji <k@kafji.net>
---
 ports/servoshell/parser.rs | 40 ++++++++++++++++++------
 ports/servoshell/test.rs   | 63 ++++++++++++++++++++++++++++++++++++++
 2 files changed, 94 insertions(+), 9 deletions(-)

diff --git a/ports/servoshell/parser.rs b/ports/servoshell/parser.rs
index d52fc64bead..f28c6698463 100644
--- a/ports/servoshell/parser.rs
+++ b/ports/servoshell/parser.rs
@@ -63,13 +63,35 @@ pub fn get_default_url(
 /// interpret the string as a search term.
 pub(crate) fn location_bar_input_to_url(request: &str, searchpage: &str) -> Option<ServoUrl> {
     let request = request.trim();
-    ServoUrl::parse(request).ok().or_else(|| {
-        if request.starts_with('/') {
-            ServoUrl::parse(&format!("file://{}", request)).ok()
-        } else if request.contains('/') || is_reg_domain(request) {
-            ServoUrl::parse(&format!("https://{}", request)).ok()
-        } else {
-            ServoUrl::parse(&searchpage.replace("%s", request)).ok()
-        }
-    })
+    ServoUrl::parse(request)
+        .ok()
+        .or_else(|| try_as_file(request))
+        .or_else(|| try_as_domain(request))
+        .or_else(|| try_as_search_page(request, searchpage))
+}
+
+fn try_as_file(request: &str) -> Option<ServoUrl> {
+    if request.starts_with('/') {
+        return ServoUrl::parse(&format!("file://{}", request)).ok();
+    }
+    None
+}
+
+fn try_as_domain(request: &str) -> Option<ServoUrl> {
+    fn is_domain_like(s: &str) -> bool {
+        !s.starts_with('/') && s.contains('/') ||
+            (!s.contains(' ') && !s.starts_with('.') && s.split('.').count() > 1)
+    }
+
+    if !request.contains(' ') && is_reg_domain(request) || is_domain_like(request) {
+        return ServoUrl::parse(&format!("https://{}", request)).ok();
+    }
+    None
+}
+
+fn try_as_search_page(request: &str, searchpage: &str) -> Option<ServoUrl> {
+    if request.is_empty() {
+        return None;
+    }
+    ServoUrl::parse(&searchpage.replace("%s", request)).ok()
 }
diff --git a/ports/servoshell/test.rs b/ports/servoshell/test.rs
index 0ff470f8b90..0ffb149905d 100644
--- a/ports/servoshell/test.rs
+++ b/ports/servoshell/test.rs
@@ -220,3 +220,66 @@ fn test_cmd_and_location_bar_url() {
         "file:///dev/null",
     );
 }
+
+/// Like [test_url] but will produce test for Windows or non Windows using `#[cfg(target_os)]` internally.
+fn test_url_any_os(
+    input: &str,
+    location: &str,
+    #[allow(unused)] if_exists: &str,
+    #[allow(unused)] if_exists_windows: &str,
+    otherwise: &str,
+) {
+    #[cfg(not(target_os = "windows"))]
+    test_url(input, location, if_exists, otherwise);
+
+    #[cfg(target_os = "windows")]
+    test_url(input, location, if_exists_windows, otherwise);
+}
+
+// https://github.com/servo/servo/issues/35754
+#[test]
+fn test_issue_35754() {
+    test_url_any_os(
+        "leah.chromebooks.lol",
+        "https://leah.chromebooks.lol/",
+        "file:///fake/cwd/leah.chromebooks.lol",
+        "file:///C:/fake/cwd/leah.chromebooks.lol",
+        "https://leah.chromebooks.lol/",
+    );
+
+    // ends with dot
+    test_url_any_os(
+        "leah.chromebooks.lol.",
+        "https://leah.chromebooks.lol./",
+        "file:///fake/cwd/leah.chromebooks.lol.",
+        "file:///C:/fake/cwd/leah.chromebooks.lol.",
+        "https://leah.chromebooks.lol./",
+    );
+
+    // starts with dot
+    test_url_any_os(
+        ".leah.chromebooks.lol",
+        "https://duckduckgo.com/html/?q=.leah.chromebooks.lol",
+        "file:///fake/cwd/.leah.chromebooks.lol",
+        "file:///C:/fake/cwd/.leah.chromebooks.lol",
+        "https://duckduckgo.com/html/?q=.leah.chromebooks.lol",
+    );
+
+    // contains spaces
+    test_url_any_os(
+        "3.5 kg in lb",
+        "https://duckduckgo.com/html/?q=3.5%20kg%20in%20lb",
+        "file:///fake/cwd/3.5%20kg%20in%20lb",
+        "file:///C:/fake/cwd/3.5%20kg%20in%20lb",
+        "https://duckduckgo.com/html/?q=3.5%20kg%20in%20lb",
+    );
+
+    // user-local domain
+    test_url_any_os(
+        "foo/bar",
+        "https://foo/bar",
+        "file:///fake/cwd/foo/bar",
+        "file:///C:/fake/cwd/foo/bar",
+        "https://foo/bar",
+    );
+}