From c52b15d61773c95dd29d3b7ba1e18ca627c347be Mon Sep 17 00:00:00 2001
From: Sean Kim <40130483+theskim@users.noreply.github.com>
Date: Wed, 27 Nov 2024 10:42:24 -0500
Subject: [PATCH] Normalize URL paths: convert /.//p, /..//p, and //p to p

---
 url/src/lib.rs                  | 53 +++++++++++++++++++++++++++++++++
 url/tests/expected_failures.txt |  4 ---
 2 files changed, 53 insertions(+), 4 deletions(-)

diff --git a/url/src/lib.rs b/url/src/lib.rs
index 3eacded1..a74062ff 100644
--- a/url/src/lib.rs
+++ b/url/src/lib.rs
@@ -1757,6 +1757,39 @@ impl Url {
         let old_after_path_pos = to_u32(self.serialization.len()).unwrap();
         let cannot_be_a_base = self.cannot_be_a_base();
         let scheme_type = SchemeType::from(self.scheme());
+        let mut path_empty = false;
+
+        // Check ':' and then see if the next character is '/'
+        let mut has_host = if let Some(index) = self.serialization.find(":") {
+            if self.serialization.len() > index + 1
+                && self.serialization.as_bytes().get(index + 1) == Some(&b'/')
+            {
+                let rest = &self.serialization[(index + ":/".len())..];
+                let host_part = rest.split('/').next().unwrap_or("");
+                path_empty = rest.is_empty();
+                !host_part.is_empty() && !host_part.contains('@')
+            } else {
+                false
+            }
+        } else {
+            false
+        };
+
+        // Ensure the path length is greater than 1 to account
+        // for cases where "/." is already appended from serialization
+        // If we set path, then we already checked the other two conditions:
+        // https://url.spec.whatwg.org/#url-serializing
+        // 1. The host is null
+        // 2. the first segment of the URL's path is an empty string
+        if path.len() > 1 {
+            if let Some(index) = self.serialization.find(":") {
+                let removal_start = index + ":".len();
+                if self.serialization[removal_start..].starts_with("/.") {
+                    self.path_start -= "/.".len() as u32;
+                }
+            }
+        }
+
         self.serialization.truncate(self.path_start as usize);
         self.mutate(|parser| {
             if cannot_be_a_base {
@@ -1774,6 +1807,26 @@ impl Url {
                 );
             }
         });
+
+        // For cases where normalization is applied across both the serialization and the path.
+        // Append "/." immediately after the scheme (up to ":")
+        // This is done if three conditions are met.
+        // https://url.spec.whatwg.org/#url-serializing
+        // 1. The host is null
+        // 2. The url's path length is greater than 1
+        // 3. the first segment of the URL's path is an empty string
+        if !has_host && path.len() > 1 && path_empty {
+            if let Some(index) = self.serialization.find(":") {
+                if self.serialization.len() > index + 2
+                    && self.serialization.as_bytes().get(index + 1) == Some(&b'/')
+                    && self.serialization.as_bytes().get(index + 2) == Some(&b'/')
+                {
+                    self.serialization.insert_str(index + ":".len(), "/.");
+                    self.path_start += "/.".len() as u32;
+                }
+            }
+        }
+
         self.restore_after_path(old_after_path_pos, &after_path);
     }
 
diff --git a/url/tests/expected_failures.txt b/url/tests/expected_failures.txt
index 238f7a66..9bf60b34 100644
--- a/url/tests/expected_failures.txt
+++ b/url/tests/expected_failures.txt
@@ -41,7 +41,3 @@
 <file://monkey/> set pathname to <\\\\>
 <file:///unicorn> set pathname to <//\\/>
 <file:///unicorn> set pathname to <//monkey/..//>
-<non-spec:/> set pathname to </.//p>
-<non-spec:/> set pathname to </..//p>
-<non-spec:/> set pathname to <//p>
-<non-spec:/.//> set pathname to <p>