Ignore textContent links in html nodes (#1528)

This fixes issue #1462 by removing plaintext URI parsing
in html5ever and pruning attribute-less URIs in html5gum

lychee-lib/src/collector.rs

@@ -148,6 +148,14 @@ mod tests {
         responses.map(|r| r.unwrap().uri).collect().await
     }
 
+    // Helper function for collecting verbatim links
+    async fn collect_verbatim(inputs: Vec<Input>, base: Option<Base>) -> HashSet<Uri> {
+        let responses = Collector::new(base)
+            .include_verbatim(true)
+            .collect_links(inputs);
+        responses.map(|r| r.unwrap().uri).collect().await
+    }
+
     const TEST_STRING: &str = "http://test-string.com";
     const TEST_URL: &str = "https://test-url.org";
     const TEST_FILE: &str = "https://test-file.io";
@@ -233,7 +241,7 @@ mod tests {
             },
         ];
 
-        let links = collect(inputs, None).await;
+        let links = collect_verbatim(inputs, None).await;
 
         let expected_links = HashSet::from_iter([
             website(TEST_STRING),

lychee-lib/src/extract/html/html5ever.rs

@@ -28,9 +28,11 @@ impl TokenSink for LinkExtractor {
                 if self.current_verbatim_element_name.borrow().is_some() {
                     return TokenSinkResult::Continue;
                 }
-                self.links
-                    .borrow_mut()
-                    .extend(extract_raw_uri_from_plaintext(&raw));
+                if self.include_verbatim {
+                    self.links
+                        .borrow_mut()
+                        .extend(extract_raw_uri_from_plaintext(&raw));
+                }
             }
             Token::TagToken(tag) => {
                 let Tag {
@@ -414,6 +416,21 @@ mod tests {
         assert!(uris.is_empty());
     }
 
+    #[test]
+    fn test_ignore_text_content_links() {
+        let input = r#"
+            <a href="https://example.com">https://ignoreme.com</a>
+        "#;
+        let expected = vec![RawUri {
+            text: "https://example.com".to_string(),
+            element: Some("a".to_string()),
+            attribute: Some("href".to_string()),
+        }];
+
+        let uris = extract_html(input, false);
+        assert_eq!(uris, expected);
+    }
+
     #[test]
     fn test_skip_dns_prefetch() {
         let input = r#"

lychee-lib/src/extract/html/html5gum.rs

@@ -4,6 +4,15 @@ use std::collections::{HashMap, HashSet};
 use super::{is_email_link, is_verbatim_elem, srcset};
 use crate::{extract::plaintext::extract_raw_uri_from_plaintext, types::uri::raw::RawUri};
 
+#[derive(Clone, Default, Debug)]
+struct Element {
+    /// Current element name being processed.
+    /// This is called a tag in html5gum.
+    name: String,
+    /// Whether the current element is a closing tag.
+    is_closing: bool,
+}
+
 /// Extract links from HTML documents.
 ///
 /// This is the main driver for the html5gum tokenizer.
@@ -16,7 +25,7 @@ use crate::{extract::plaintext::extract_raw_uri_from_plaintext, types::uri::raw:
 ///
 /// The `links` vector contains all links extracted from the HTML document and
 /// the `fragments` set contains all fragments extracted from the HTML document.
-#[derive(Clone, Default)]
+#[derive(Clone, Default, Debug)]
 struct LinkExtractor {
     /// Links extracted from the HTML document.
     links: Vec<RawUri>,
@@ -39,15 +48,6 @@ struct LinkExtractor {
     verbatim_stack: Vec<String>,
 }
 
-#[derive(Clone, Default)]
-struct Element {
-    /// Current element name being processed.
-    /// This is called a tag in html5gum.
-    name: String,
-    /// Whether the current element is a closing tag.
-    is_closing: bool,
-}
-
 impl LinkExtractor {
     /// Create a new `LinkExtractor`.
     ///
@@ -326,7 +326,11 @@ pub(crate) fn extract_html(buf: &str, include_verbatim: bool) -> Vec<RawUri> {
     let mut extractor = LinkExtractor::new(include_verbatim);
     let mut tokenizer = Tokenizer::new_with_emitter(buf, &mut extractor).infallible();
     assert!(tokenizer.next().is_none());
-    extractor.links
+    extractor
+        .links
+        .into_iter()
+        .filter(|link| link.attribute.is_some() || include_verbatim)
+        .collect()
 }
 
 /// Extract fragments from id attributes within a HTML string.
@@ -609,6 +613,21 @@ mod tests {
         assert!(uris.is_empty());
     }
 
+    #[test]
+    fn test_ignore_text_content_links() {
+        let input = r#"
+            <a href="https://example.com">https://ignoreme.com</a>
+        "#;
+        let expected = vec![RawUri {
+            text: "https://example.com".to_string(),
+            element: Some("a".to_string()),
+            attribute: Some("href".to_string()),
+        }];
+
+        let uris = extract_html(input, false);
+        assert_eq!(uris, expected);
+    }
+
     #[test]
     fn test_skip_dns_prefetch() {
         let input = r#"