fix(html): parse inline SVG with XML declarations (#3005)

crates/mdbook-html/src/html/tests.rs

@@ -37,6 +37,28 @@ if (3 < 5 > 10)
     assert_eq!(output, script);
 }
 
+#[test]
+fn parse_html_svg_with_xml_decl() {
+    let html = r#"<svg xmlns="http://www.w3.org/2000/svg"><?xml version="1.0"?><rect/></svg>"#;
+    let ts = parse_html(html);
+    for t in &ts {
+        if let Token::ParseError(e) = t {
+            panic!("unexpected parse error: {e:?}");
+        }
+    }
+}
+
+#[test]
+fn parse_html_pre_with_svg_xml_decl() {
+    let html = r#"<pre><?xml version="1.0" encoding="utf-8"?><svg xmlns="http://www.w3.org/2000/svg"><rect/></svg></pre>"#;
+    let ts = parse_html(html);
+    for t in &ts {
+        if let Token::ParseError(e) = t {
+            panic!("unexpected parse error: {e:?}");
+        }
+    }
+}
+
 // What happens if a script doesn't end.
 #[test]
 fn parse_html_script_unclosed() {

crates/mdbook-html/src/html/tokenizer.rs

@@ -9,6 +9,7 @@ use html5ever::tokenizer::states::RawKind;
 use html5ever::tokenizer::{
     BufferQueue, TagKind, Token, TokenSink, TokenSinkResult, Tokenizer, TokenizerOpts,
 };
+use std::borrow::Cow;
 use std::cell::RefCell;
 
 /// Collector for HTML tokens.
@@ -45,6 +46,15 @@ impl TokenSink for TokenCollector {
                         TagKind::EndTag => {}
                     }
                 }
+                if tag_name == b"svg" {
+                    match tag.kind {
+                        TagKind::StartTag => {
+                            self.tokens.borrow_mut().push(token);
+                            return TokenSinkResult::RawData(RawKind::Rawtext);
+                        }
+                        TagKind::EndTag => {}
+                    }
+                }
                 self.tokens.borrow_mut().push(token);
             }
             Token::CommentToken(_) => {
@@ -63,8 +73,32 @@ impl TokenSink for TokenCollector {
     }
 }
 
+/// Strips XML processing instructions (e.g. `<?xml ...?>`) that are invalid in HTML
+/// but commonly appear in inline SVG emitted by preprocessors.
+fn strip_xml_processing_instructions(html: &str) -> Cow<'_, str> {
+    let mut out = String::new();
+    let mut rest = html;
+    let mut changed = false;
+    while let Some(start) = rest.find("<?") {
+        changed = true;
+        out.push_str(&rest[..start]);
+        let after = &rest[start + 2..];
+        let Some(end) = after.find("?>") else {
+            out.push_str(&rest[start..]);
+            return Cow::Owned(out);
+        };
+        rest = &after[end + 2..];
+    }
+    if !changed {
+        return Cow::Borrowed(html);
+    }
+    out.push_str(rest);
+    Cow::Owned(out)
+}
+
 /// Parse HTML into tokens.
 pub(crate) fn parse_html(html: &str) -> Vec<Token> {
+    let html = strip_xml_processing_instructions(html);
     let tendril: ByteTendril = html.as_bytes().into();
     let mut queue = BufferQueue::default();
     queue.push_back(tendril.try_reinterpret().unwrap());

crates/mdbook-html/src/html/tree.rs

@@ -662,7 +662,7 @@ where
     /// Adds an open HTML tag.
     fn start_html_tag(&mut self, tag: html5ever::tokenizer::Tag, is_raw: &mut bool) {
         let is_closed = is_void_element(&tag.name) || tag.self_closing;
-        *is_raw = matches!(&*tag.name, "script" | "style");
+        *is_raw = matches!(&*tag.name, "script" | "style" | "svg");
         let name = QualName::new(None, html5ever::ns!(html), tag.name);
         let attrs = tag
             .attrs