Merge pull request #682 from kivikakk/push-txtvvkrsnxqk

don't append a newline at EOF; handle it in the parser.

Author: kivikakk

fuzz/fuzz_targets/all_options.rs

@@ -54,6 +54,8 @@ struct FuzzExtensionOptions {
     subscript: bool,
     subtext: bool,
     cjk_friendly_emphasis: bool,
+    highlight: bool,
+    // non-bool below
     header_ids: bool,
     front_matter_delimiter: bool,
     image_url_rewriter: bool,
@@ -88,6 +90,8 @@ impl FuzzExtensionOptions {
             subscript: self.subscript,
             subtext: self.subtext,
             cjk_friendly_emphasis: self.cjk_friendly_emphasis,
+            highlight: self.highlight,
+            // non-bool below
             header_ids: if self.header_ids {
                 Some("user-content-".into())
             } else {

src/arena_tree.rs

@@ -205,6 +205,13 @@ impl<'a, T> Node<'a, T> {
         self.last_child.set(Some(new_child));
     }
 
+    /// Append multiple new children to this node, after existing children.
+    pub fn extend(&'a self, new_children: impl IntoIterator<Item = &'a Node<'a, T>>) {
+        for child in new_children.into_iter() {
+            self.append(child);
+        }
+    }
+
     /// Prepend a new child to this node, before existing children.
     pub fn prepend(&'a self, new_child: &'a Node<'a, T>) {
         new_child.detach();

src/html.rs

@@ -1622,9 +1622,10 @@ fn tagfilter(literal: &str) -> bool {
     for t in TAGFILTER_BLACKLIST.iter() {
         if lc.starts_with(t) {
             let j = i + t.len();
-            return isspace(bytes[j])
-                || bytes[j] == b'>'
-                || (bytes[j] == b'/' && bytes.len() >= j + 2 && bytes[j + 1] == b'>');
+            let Some(&b) = bytes.get(j) else {
+                return false;
+            };
+            return isspace(b) || b == b'>' || (b == b'/' && bytes.get(j + 1) == Some(&b'>'));
         }
     }
 

src/nodes.rs

@@ -173,7 +173,7 @@ pub enum NodeValue {
     Image(Box<NodeLink>),
 
     /// **Inline**.  A footnote reference.
-    FootnoteReference(NodeFootnoteReference),
+    FootnoteReference(Box<NodeFootnoteReference>),
 
     #[cfg(feature = "shortcodes")]
     /// **Inline**. An Emoji character generated from a shortcode. Enable with feature "shortcodes".
@@ -459,6 +459,9 @@ pub struct NodeFootnoteReference {
     /// The name of the footnote.
     pub name: String,
 
+    /// The original text elements of the footnote, including their source position spans.
+    pub texts: Vec<(String, usize)>,
+
     /// The index of reference to the same footnote
     pub ref_num: u32,
 

src/parser/inlines.rs

@@ -994,11 +994,12 @@ impl<'a, 'r, 'o, 'd, 'c, 'p> Subject<'a, 'r, 'o, 'd, 'c, 'p> {
 
         // Create the footnote reference node
         let ref_node = self.make_inline(
-            NodeValue::FootnoteReference(NodeFootnoteReference {
+            NodeValue::FootnoteReference(Box::new(NodeFootnoteReference {
                 name: name.clone(),
+                texts: vec![], // Unused.
                 ref_num: 0,
                 ix: 0,
-            }),
+            })),
             startpos,
             endpos,
         );
@@ -1734,13 +1735,33 @@ impl<'a, 'r, 'o, 'd, 'c, 'p> Subject<'a, 'r, 'o, 'd, 'c, 'p> {
             // do anything fancy here at all.
             let mut sussy = false;
 
+            let mut texts = vec![];
+
             for sibling in sibling_iterator {
-                match sibling.data().value {
+                let sibling_ast = sibling.data();
+                if sibling_ast.sourcepos.start.line != sibling_ast.sourcepos.end.line
+                    || sibling_ast.sourcepos.end.column < sibling_ast.sourcepos.start.column
+                {
+                    sussy = true;
+                    break;
+                }
+
+                match sibling_ast.value {
                     NodeValue::Text(ref literal) => {
                         text.push_str(literal);
+                        texts.push((
+                            literal.to_string(),
+                            sibling_ast.sourcepos.end.column - sibling_ast.sourcepos.start.column
+                                + 1,
+                        ));
                     }
                     NodeValue::HtmlInline(ref literal) => {
                         text.push_str(literal);
+                        texts.push((
+                            literal.to_string(),
+                            sibling_ast.sourcepos.end.column - sibling_ast.sourcepos.start.column
+                                + 1,
+                        ));
                     }
                     _ => {
                         sussy = true;
@@ -1751,11 +1772,12 @@ impl<'a, 'r, 'o, 'd, 'c, 'p> Subject<'a, 'r, 'o, 'd, 'c, 'p> {
 
             if !sussy && text.len() > 1 {
                 let inl = self.make_inline(
-                    NodeValue::FootnoteReference(NodeFootnoteReference {
+                    NodeValue::FootnoteReference(Box::new(NodeFootnoteReference {
                         name: text[1..].to_string(),
+                        texts,
                         ref_num: 0,
                         ix: 0,
-                    }),
+                    })),
                     // Overridden immediately below.
                     self.scanner.pos,
                     self.scanner.pos,
@@ -2341,7 +2363,7 @@ pub(crate) fn manual_scan_link_url_2(input: &str) -> Option<(&str, usize)> {
         }
     }
 
-    if i >= len || nb_p != 0 {
+    if len == 0 || nb_p != 0 {
         None
     } else {
         Some((&input[..i], i))