Preserve preformatted boundaries and empty code separation

moli-html2md/src/machine.rs

@@ -384,10 +384,10 @@ impl<'a, D: Dom + ?Sized> Machine<'a, D> {
             NodeKind::Text(text) => self.raw.push_str(text),
             // Inline code normalizes line endings to spaces. Keep explicit
             // breaks without converting descendant formatting into Markdown.
-            NodeKind::Element("br") if inline => self.raw.push('\n'),
+            NodeKind::Element("br") => self.raw.push('\n'),
             NodeKind::Document | NodeKind::Element(_) if depth + 1 < self.options.max_depth => {
                 if !inline
-                    && matches!(self.dom.node_kind(node), NodeKind::Element(tag) if is_block(tag))
+                    && matches!(self.dom.node_kind(node), NodeKind::Element(tag) if is_structural_block(tag))
                 {
                     if !self.raw.is_empty() && !self.raw.ends_with('\n') {
                         self.raw.push('\n');
@@ -449,17 +449,20 @@ fn class_language(class: Option<&str>) -> Option<&str> {
 }
 
 fn is_block(tag: &str) -> bool {
+    is_structural_block(tag)
+        || matches!(
+            tag,
+            "audio" | "canvas" | "frameset" | "isindex" | "noframes" | "output"
+        )
+}
+
+fn is_structural_block(tag: &str) -> bool {
     matches!(
         tag,
         "address"
-            | "audio"
-            | "canvas"
-            | "frameset"
-            | "isindex"
-            | "noframes"
-            | "output"
             | "article"
             | "aside"
+            | "blockquote"
             | "body"
             | "caption"
             | "center"
@@ -475,14 +478,23 @@ fn is_block(tag: &str) -> bool {
             | "figure"
             | "footer"
             | "form"
+            | "h1"
+            | "h2"
+            | "h3"
+            | "h4"
+            | "h5"
+            | "h6"
             | "header"
             | "hgroup"
             | "html"
             | "legend"
+            | "li"
             | "main"
             | "menu"
             | "nav"
+            | "ol"
             | "p"
+            | "pre"
             | "section"
             | "summary"
             | "table"
@@ -492,5 +504,6 @@ fn is_block(tag: &str) -> bool {
             | "th"
             | "thead"
             | "tr"
+            | "ul"
     )
 }

moli-html2md/src/writer.rs

@@ -171,6 +171,11 @@ impl<'a> Writer<'a> {
     }
 
     pub(crate) fn code_with_edges(&mut self, text: &str, preformatted: bool) {
+        // An empty element has no visible edge. Keep the pending code until
+        // the next visible text or element decides whether it needs a gap.
+        if text.is_empty() {
+            return;
+        }
         // Each call represents a separate HTML code element. Keep adjacent
         // elements distinct even when the source has no whitespace between them.
         if self.code.is_some() {

moli-html2md/tests/inline_boundaries.rs

@@ -106,6 +106,32 @@ fn pre_keeps_raw_text_and_line_endings() {
     );
 }
 
+#[test]
+fn preformatted_examples_keep_inline_values_and_separate_display_lines() {
+    for (source, expected) in [
+        (
+            "<pre>const answer = <output>42</output>;</pre>",
+            "const answer = 42;",
+        ),
+        (
+            "<pre><h3>TypeScript</h3><code>function identity() {}</code></pre>",
+            "TypeScript\nfunction identity() {}",
+        ),
+        (
+            "<pre><ol><li>first()</li><li>second()</li></ol></pre>",
+            "first()\nsecond()",
+        ),
+        ("<pre>first<br>second</pre>", "first\nsecond"),
+    ] {
+        let result = markdown(source);
+        assert_eq!(
+            rendered_html(&result),
+            format!("<pre><code>{expected}\n</code></pre>\n"),
+            "{source}: {result}"
+        );
+    }
+}
+
 #[test]
 fn nested_emphasis_closes_with_markdown_before_a_word() {
     for (html, expected, expected_html) in [

moli-html2md/tests/regressions.rs

@@ -158,6 +158,21 @@ fn adjacent_code_elements_keep_distinct_values() {
     );
 }
 
+#[test]
+fn empty_code_between_values_does_not_join_or_invent_code_text() {
+    for empty in ["<code></code>", "<code><!-- source note --></code>"] {
+        let source = format!("<code>name</code>{empty}<code>string</code>");
+        for preformatted in [false, true] {
+            let result = markdown(&source, preformatted);
+            assert_eq!(
+                rendered_html(&result),
+                "<p><code>name</code> <code>string</code></p>\n",
+                "{source}, preformatted={preformatted}: {result}"
+            );
+        }
+    }
+}
+
 #[test]
 fn preformatted_block_children_keep_text_boundaries() {
     let html = "<pre><div>ts</div><div><code>function identity() {}</code></div></pre>";