From c5991fc814a8151cbc8136ed0a70ea25350e20d7 Mon Sep 17 00:00:00 2001
From: Tanishq <tanishq.cse@gmail.com>
Date: Wed, 12 Jun 2024 22:08:51 +0530
Subject: [PATCH] fix ignore links to #top when checking anchors (#2519)

* fix ignore links to #top when checking anchors

* move logic to check internal links

---------

Co-authored-by: Tanishq <tanishq@levels.fyi>
---
 components/site/src/link_checking.rs |  2 ++
 components/utils/src/anchors.rs      | 16 +++++++++++++++-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/components/site/src/link_checking.rs b/components/site/src/link_checking.rs
index c7e4678195..aff8879b73 100644
--- a/components/site/src/link_checking.rs
+++ b/components/site/src/link_checking.rs
@@ -10,6 +10,7 @@ use crate::Site;
 use errors::{bail, Result};
 use libs::rayon;
 use libs::url::Url;
+use utils::anchors::is_special_anchor;
 
 /// Check whether all internal links pointing to explicit anchor fragments are valid.
 ///
@@ -40,6 +41,7 @@ pub fn check_internal_links_with_anchors(site: &Site) -> Vec<String> {
             (md_path, Some(anchor)) => Some((page_path, md_path, anchor)),
             _ => None,
         })
+        .filter(|(_, _, anchor)| !is_special_anchor(anchor))
         .inspect(|_| anchors_total = anchors_total.saturating_add(1));
 
     // Check for targets existence (including anchors), then keep only the faulty
diff --git a/components/utils/src/anchors.rs b/components/utils/src/anchors.rs
index a5706e79d9..5c1bd2d16c 100644
--- a/components/utils/src/anchors.rs
+++ b/components/utils/src/anchors.rs
@@ -10,9 +10,15 @@ fn anchor_id_checks(anchor: &str) -> Regex {
     Regex::new(&format!(r#"\s(?i)(id|name) *= *("|')*{}("|'| |>)+"#, escape(anchor))).unwrap()
 }
 
+/// Checks if anchor has a special meaning in HTML
+/// https://html.spec.whatwg.org/#select-the-indicated-part
+pub fn is_special_anchor(anchor: &str) -> bool {
+    anchor.is_empty() || anchor.eq_ignore_ascii_case("top")
+}
+
 #[cfg(test)]
 mod tests {
-    use super::anchor_id_checks;
+    use super::{anchor_id_checks, is_special_anchor};
 
     fn check(anchor: &str, content: &str) -> bool {
         anchor_id_checks(anchor).is_match(content)
@@ -52,4 +58,12 @@ id="fred">"#));
         // Non matchers
         assert!(!m(r#"<a notid="fred">"#));
     }
+
+    #[test]
+    fn test_is_special_anchor() {
+        assert!(is_special_anchor(""));
+        assert!(is_special_anchor("top"));
+        assert!(is_special_anchor("Top"));
+        assert!(!is_special_anchor("anchor"));
+    }
 }