We must also check the fragment before it is percent decoded as requi…

…red by the html standard. Fixes #1467
lycheeverse · Oct 18, 2024 · d5293d1 · d5293d1
1 parent f5a4f7c
commit d5293d1
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 4 deletions.
diff --git a/fixtures/fragments/file.html b/fixtures/fragments/file.html
@@ -7,6 +7,7 @@
   <body>
     <section id="in-the-beginning" style="height: 100vh;">
       <p id="Upper-ÄÖö">
+        <div id="tangent%3A-kustomize"></div>
         To start
         <a href="file1.md#fragment-1">
           let's run away.
@@ -17,6 +18,7 @@
       <p id="a-word">Word</p>
       <a href="#in-the-beginning">back we go</a><br>
       <a href="#in-THE-begiNNing">back we go upper does not work</a><br>
+      <a href="#tangent%3A-kustomize">id with percent encoding</a><br>
       <a href="#Upper-ÄÖö">back to Upper-ÄÖö</a><br>
       <a href="#Upper-%C3%84%C3%96%C3%B6">back to öüä encoded</a><br>
       <a href="#in-the-end">doesn't exist</a><br>

diff --git a/lychee-lib/src/utils/fragment_checker.rs b/lychee-lib/src/utils/fragment_checker.rs
@@ -47,7 +47,7 @@ impl FragmentChecker {
         let Some(fragment) = url.fragment() else {
             return Ok(true);
         };
-        let mut fragment = percent_decode_str(fragment).decode_utf8()?;
+        let mut fragment_decoded = percent_decode_str(fragment).decode_utf8()?;
         let url_without_frag = Self::remove_fragment(url.clone());
 
         let file_type = FileType::from(path);
@@ -57,15 +57,19 @@ impl FragmentChecker {
             FileType::Plaintext => return Ok(true),
         };
         if file_type == FileType::Markdown {
-            fragment = fragment.to_lowercase().into();
+            fragment_decoded = fragment_decoded.to_lowercase().into();
         }
         match self.cache.lock().await.entry(url_without_frag) {
             Entry::Vacant(entry) => {
                 let content = fs::read_to_string(path).await?;
                 let file_frags = extractor(&content);
-                Ok(entry.insert(file_frags).contains(&fragment as &str))
+                let contains_fragment =
+                    file_frags.contains(fragment) || file_frags.contains(&fragment_decoded as &str);
+                entry.insert(file_frags);
+                Ok(contains_fragment)
             }
-            Entry::Occupied(entry) => Ok(entry.get().contains(&fragment as &str)),
+            Entry::Occupied(entry) => Ok(entry.get().contains(&fragment as &str)
+                || entry.get().contains(&fragment_decoded as &str)),
         }
     }