Fix shortcode/continue-reading parsing with inline HTML

components/content/src/page.rs

@@ -31,10 +31,6 @@ static RFC3339_DATE: Lazy<Regex> = Lazy::new(|| {
     ).unwrap()
 });
 
-static FOOTNOTES_RE: Lazy<Regex> = Lazy::new(|| {
-    Regex::new(r#"<sup class="footnote-reference"><a href=\s*.*?>\s*.*?</a></sup>"#).unwrap()
-});
-
 #[derive(Clone, Debug, Default, PartialEq, Eq)]
 pub struct Page {
     /// All info about the actual file
@@ -232,10 +228,7 @@ impl Page {
         let res = render_content(&self.raw_content, &context)
             .with_context(|| format!("Failed to render content of {}", self.file.path.display()))?;
 
-        self.summary = res
-            .summary_len
-            .map(|l| &res.body[0..l])
-            .map(|s| FOOTNOTES_RE.replace_all(s, "").into_owned());
+        self.summary = res.summary;
         self.content = res.body;
         self.toc = res.toc;
         self.external_links = res.external_links;

components/markdown/src/markdown.rs

@@ -24,6 +24,7 @@ use crate::codeblock::{CodeBlock, FenceSettings};
 use crate::shortcode::{Shortcode, SHORTCODE_PLACEHOLDER};
 
 const CONTINUE_READING: &str = "<span id=\"continue-reading\"></span>";
+const SUMMARY_CUTOFF: &str = "<span class=\"summary-cutoff\"></span>";
 const ANCHOR_LINK_TEMPLATE: &str = "anchor-link.html";
 static EMOJI_REPLACER: Lazy<EmojiReplacer> = Lazy::new(EmojiReplacer::new);
 
@@ -36,6 +37,10 @@ static MORE_DIVIDER_RE: Lazy<Regex> = Lazy::new(|| {
         .unwrap()
 });
 
+static FOOTNOTES_RE: Lazy<Regex> = Lazy::new(|| {
+    Regex::new(r#"<sup class="footnote-reference"><a href=\s*.*?>\s*.*?</a></sup>"#).unwrap()
+});
+
 /// Although there exists [a list of registered URI schemes][uri-schemes], a link may use arbitrary,
 /// private schemes. This regex checks if the given string starts with something that just looks
 /// like a scheme, i.e., a case-insensitive identifier followed by a colon.
@@ -78,7 +83,7 @@ fn is_colocated_asset_link(link: &str) -> bool {
 #[derive(Debug)]
 pub struct Rendered {
     pub body: String,
-    pub summary_len: Option<usize>,
+    pub summary: Option<String>,
     pub toc: Vec<Heading>,
     /// Links to site-local pages: relative path plus optional anchor target.
     pub internal_links: Vec<(String, Option<String>)>,
@@ -405,6 +410,7 @@ pub fn markdown_to_html(
         .map(|x| x.as_object().unwrap().get("relative_path").unwrap().as_str().unwrap());
     // the rendered html
     let mut html = String::with_capacity(content.len());
+    let mut summary = None;
     // Set while parsing
     let mut error = None;
 
@@ -679,17 +685,15 @@ pub fn markdown_to_html(
                         event
                     });
                 }
-                Event::Html(text) => {
-                    if !has_summary && MORE_DIVIDER_RE.is_match(&text) {
-                        has_summary = true;
-                        events.push(Event::Html(CONTINUE_READING.into()));
-                        continue;
-                    }
-                    if !contains_shortcode(text.as_ref()) {
-                        events.push(Event::Html(text));
-                        continue;
-                    }
-
+                Event::Html(text) | Event::InlineHtml(text)
+                    if !has_summary && MORE_DIVIDER_RE.is_match(text.as_ref()) =>
+                {
+                    has_summary = true;
+                    events.push(Event::Html(CONTINUE_READING.into()));
+                }
+                Event::Html(text) | Event::InlineHtml(text)
+                    if contains_shortcode(text.as_ref()) =>
+                {
                     render_shortcodes!(false, text, range);
                 }
                 _ => events.push(event),
@@ -781,14 +785,57 @@ pub fn markdown_to_html(
             convert_footnotes_to_github_style(&mut events);
         }
 
-        cmark::html::push_html(&mut html, events.into_iter());
+        let continue_reading = events
+            .iter()
+            .position(|e| matches!(e, Event::Html(CowStr::Borrowed(CONTINUE_READING))))
+            .unwrap_or(events.len());
+
+        // determine closing tags missing from summary
+        let mut tags = Vec::new();
+        for event in &events[..continue_reading] {
+            match event {
+                Event::Start(Tag::HtmlBlock) | Event::End(TagEnd::HtmlBlock) => (),
+                Event::Start(tag) => tags.push(tag.to_end()),
+                Event::End(tag) => {
+                    tags.truncate(tags.iter().rposition(|t| *t == *tag).unwrap_or(0));
+                }
+                _ => (),
+            }
+        }
+
+        let mut events = events.into_iter();
+
+        // emit everything up to summary
+        cmark::html::push_html(&mut html, events.by_ref().take(continue_reading));
+
+        if has_summary {
+            // remove footnotes
+            let mut summary_html = FOOTNOTES_RE.replace_all(&html, "").into_owned();
+
+            // truncate trailing whitespace
+            summary_html.truncate(summary_html.trim_end().len());
+
+            // add cutoff placeholder
+            if !tags.is_empty() {
+                dbg!(&tags);
+                summary_html.push_str(SUMMARY_CUTOFF);
+            }
+
+            // close remaining tags
+            cmark::html::push_html(&mut summary_html, tags.into_iter().rev().map(Event::End));
+
+            summary = Some(summary_html)
+        }
+
+        // emit everything after summary
+        cmark::html::push_html(&mut html, events);
     }
 
     if let Some(e) = error {
         Err(e)
     } else {
         Ok(Rendered {
-            summary_len: if has_summary { html.find(CONTINUE_READING) } else { None },
+            summary,
             body: html,
             toc: make_table_of_contents(headings),
             internal_links,
@@ -861,10 +908,10 @@ mod tests {
         for more in mores {
             let content = format!("{top}\n\n{more}\n\n{bottom}");
             let rendered = markdown_to_html(&content, &context, vec![]).unwrap();
-            assert!(rendered.summary_len.is_some(), "no summary when splitting on {more}");
-            let summary_len = rendered.summary_len.unwrap();
-            let summary = &rendered.body[..summary_len].trim();
-            let body = &rendered.body[summary_len..].trim();
+            assert!(rendered.summary.is_some(), "no summary when splitting on {more}");
+            let summary = rendered.summary.unwrap();
+            let summary = summary.trim();
+            let body = rendered.body[summary.len()..].trim();
             let continue_reading = &body[..CONTINUE_READING.len()];
             let body = &body[CONTINUE_READING.len()..].trim();
             assert_eq!(summary, &top_rendered);

components/markdown/tests/shortcodes.rs

@@ -311,3 +311,15 @@ fn can_use_shortcodes_in_quotes() {
     .body;
     insta::assert_snapshot!(body);
 }
+
+#[test]
+fn can_render_with_inline_html() {
+    let body = common::render(
+        r#"
+Here is <span>{{ ex1(page="") }}</span> example.
+    "#,
+    )
+    .unwrap()
+    .body;
+    insta::assert_snapshot!(body);
+}

components/markdown/tests/snapshots/shortcodes__can_render_with_inline_html.snap

@@ -0,0 +1,5 @@
+---
+source: components/markdown/tests/shortcodes.rs
+expression: body
+---
+<p>Here is <span>1</span> example.</p>

components/markdown/tests/snapshots/summary__foontnotes_summary.snap

@@ -0,0 +1,5 @@
+---
+source: components/markdown/tests/summary.rs
+expression: body
+---
+<p>Hello world.</p>

components/markdown/tests/snapshots/summary__truncated_summary.snap

@@ -0,0 +1,8 @@
+---
+source: components/markdown/tests/summary.rs
+expression: body
+---
+<p>Things to do:</p>
+<ul>
+<li>Program<span class="summary-cutoff"></span></li>
+</ul>

components/markdown/tests/summary.rs

@@ -1,10 +1,7 @@
 mod common;
 
 fn get_summary(content: &str) -> String {
-    let rendered = common::render(content).unwrap();
-    assert!(rendered.summary_len.is_some());
-    let summary_len = rendered.summary_len.unwrap();
-    rendered.body[..summary_len].to_owned()
+    common::render(content).expect("couldn't render").summary.expect("had no summary")
 }
 
 #[test]
@@ -45,3 +42,32 @@ And some content after
     );
     insta::assert_snapshot!(body);
 }
+
+#[test]
+fn truncated_summary() {
+    let body = get_summary(
+        r#"
+Things to do:
+* Program <!-- more --> something
+* Eat
+* Sleep
+    "#,
+    );
+    insta::assert_snapshot!(body);
+}
+
+#[test]
+fn foontnotes_summary() {
+    let body = get_summary(
+        r#"
+Hello world[^1].
+
+<!-- more -->
+
+Good bye.
+
+[^1]: "World" is a placeholder.
+    "#,
+    );
+    insta::assert_snapshot!(body);
+}