Match <!--more--> in addition to <!-- more --> (getzola#2397)

* Match <!-- more --> without spaces

* Add tests for new <!-- more --> handling, with a note on pulldown-cmark bug

Author: clarfonthey

components/markdown/src/markdown.rs

@@ -10,7 +10,7 @@ use utils::net::is_external_link;
 use crate::context::RenderContext;
 use errors::{Context, Error, Result};
 use libs::pulldown_cmark::escape::escape_html;
-use libs::regex::Regex;
+use libs::regex::{Regex, RegexBuilder};
 use utils::site::resolve_internal_link;
 use utils::slugs::slugify_anchors;
 use utils::table_of_contents::{make_table_of_contents, Heading};
@@ -24,6 +24,15 @@ const CONTINUE_READING: &str = "<span id=\"continue-reading\"></span>";
 const ANCHOR_LINK_TEMPLATE: &str = "anchor-link.html";
 static EMOJI_REPLACER: Lazy<EmojiReplacer> = Lazy::new(EmojiReplacer::new);
 
+/// Set as a regex to help match some extra cases. This way, spaces and case don't matter.
+static MORE_DIVIDER_RE: Lazy<Regex> = Lazy::new(|| {
+    RegexBuilder::new(r#"<!--\s*more\s*-->"#)
+        .case_insensitive(true)
+        .dot_matches_new_line(true)
+        .build()
+        .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.
@@ -485,7 +494,7 @@ pub fn markdown_to_html(
                     });
                 }
                 Event::Html(text) => {
-                    if text.contains("<!-- more -->") {
+                    if !has_summary && MORE_DIVIDER_RE.is_match(&text) {
                         has_summary = true;
                         events.push(Event::Html(CONTINUE_READING.into()));
                         continue;
@@ -600,6 +609,8 @@ pub fn markdown_to_html(
 
 #[cfg(test)]
 mod tests {
+    use config::Config;
+
     use super::*;
     #[test]
 
@@ -644,4 +655,31 @@ mod tests {
             assert!(!is_colocated_asset_link(link));
         }
     }
+
+    #[test]
+    // Tests for summary being split out
+    fn test_summary_split() {
+        let top = "Here's a compelling summary.";
+        let top_rendered = format!("<p>{top}</p>");
+        let bottom = "Here's the compelling conclusion.";
+        let bottom_rendered = format!("<p>{bottom}</p>");
+        // FIXME: would add a test that includes newlines, but due to the way pulldown-cmark parses HTML nodes, these are passed as separate HTML events. see: https://github.com/raphlinus/pulldown-cmark/issues/803
+        let mores =
+            ["<!-- more -->", "<!--more-->", "<!-- MORE -->", "<!--MORE-->", "<!--\t MoRe \t-->"];
+        let config = Config::default();
+        let context = RenderContext::from_config(&config);
+        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();
+            let continue_reading = &body[..CONTINUE_READING.len()];
+            let body = &body[CONTINUE_READING.len()..].trim();
+            assert_eq!(summary, &top_rendered);
+            assert_eq!(continue_reading, CONTINUE_READING);
+            assert_eq!(body, &bottom_rendered);
+        }
+    }
 }