fix: do not fail on empty # and #top fragments

The empty "#" and "#top" fragments are always valid without related HTML element. Browser will scroll to the top of the page. Hence lychee must not fail on those.

Credits go to @thiru-appitap for initial attempt and helping to find missing parts of the implementation.

Solves: lycheeverse#1599

Signed-off-by: MichaIng <[email protected]>

Author: MichaIng

fixtures/fragments/file.html

@@ -22,6 +22,8 @@
       <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>
+      <a href="#">To the top</a><br>
+      <a href="#top">To the top alt</a><br>
     </section>
   </body>
 </html>

fixtures/fragments/file1.md

@@ -54,4 +54,12 @@ Therefore we put the test into a code block for now to prevent false positives.
 [Link to umlauts wrong case](#fünf-sÜße-Äpfel)
 [Link to umlauts with percent encoding](#f%C3%BCnf-s%C3%BC%C3%9Fe-%C3%A4pfel)
 
+# To top fragments
+
+The empty "#" and "#top" fragments are always valid
+without related HTML element. Browser will scroll to the top of the page.
+
+[Link to top of file2](file2.md#)
+[Alternative link to top of file2](file2.md#top)
+
 ##### Lets wear a hat: être

lychee-bin/tests/cli.rs

@@ -1834,8 +1834,10 @@ mod cli {
             .stderr(contains(
                 "fixtures/fragments/file1.md#kebab-case-fragment-1",
             ))
-            .stdout(contains("21 Total"))
-            .stdout(contains("17 OK"))
+            .stderr(contains("fixtures/fragments/file.html#top"))
+            .stderr(contains("fixtures/fragments/file2.md#top"))
+            .stdout(contains("25 Total"))
+            .stdout(contains("21 OK"))
             // 4 failures because of missing fragments
             .stdout(contains("4 Errors"));
     }

lychee-lib/src/utils/fragment_checker.rs

@@ -59,19 +59,20 @@ impl FragmentChecker {
         if file_type == FileType::Markdown {
             fragment_decoded = fragment_decoded.to_lowercase().into();
         }
+        // Empty # and #top fragments are always valid, triggering the browser to scroll to top.
+        let is_emtpty_or_top_fragment = fragment.is_empty() || fragment.eq_ignore_ascii_case("top");
         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);
                 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)
-                    || entry.get().contains(&fragment_decoded as &str))
+                Ok(is_emtpty_or_top_fragment || contains_fragment)
             }
+            Entry::Occupied(entry) => Ok(is_emtpty_or_top_fragment
+                || entry.get().contains(fragment)
+                || entry.get().contains(&fragment_decoded as &str)),
         }
     }