Fix strikethrough, attention interplay

Closes GH-84. Closes GH-88. Co-authored-by: awilder <rawilder0506@gmail.com>
wooorm · Dec 5, 2023 · f2c94af · f2c94af
1 parent 137a299
commit f2c94af
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 11 deletions.
diff --git a/src/construct/attention.rs b/src/construct/attention.rs
@@ -239,18 +239,22 @@ fn get_sequences(tokenizer: &mut Tokenizer) -> Vec<Sequence> {
                 let exit = &tokenizer.events[end];
 
                 let marker = tokenizer.parse_state.bytes[enter.point.index];
-                let before = classify_opt(char_before_index(
-                    tokenizer.parse_state.bytes,
-                    enter.point.index,
-                ));
-                let after = classify_opt(char_after_index(
-                    tokenizer.parse_state.bytes,
-                    exit.point.index,
-                ));
+                let before_char = char_before_index(tokenizer.parse_state.bytes, enter.point.index);
+                let before = classify_opt(before_char);
+                let after_char = char_after_index(tokenizer.parse_state.bytes, exit.point.index);
+                let after = classify_opt(after_char);
                 let open = after == CharacterKind::Other
-                    || (after == CharacterKind::Punctuation && before != CharacterKind::Other);
+                    || (after == CharacterKind::Punctuation && before != CharacterKind::Other)
+                    // For regular attention markers (not strikethrough), the
+                    // other attention markers can be used around them
+                    || (marker != b'~' && matches!(after_char, Some('*' | '_')))
+                    || (marker != b'~' && tokenizer.parse_state.options.constructs.gfm_strikethrough && matches!(after_char, Some('~')));
                 let close = before == CharacterKind::Other
-                    || (before == CharacterKind::Punctuation && after != CharacterKind::Other);
+                    || (before == CharacterKind::Punctuation && after != CharacterKind::Other)
+                    || (marker != b'~' && matches!(before_char, Some('*' | '_')))
+                    || (marker != b'~'
+                        && tokenizer.parse_state.options.constructs.gfm_strikethrough
+                        && matches!(before_char, Some('~')));
 
                 sequences.push(Sequence {
                     index,

diff --git a/src/util/char.rs b/src/util/char.rs
@@ -93,7 +93,7 @@ pub fn classify(char: char) -> Kind {
         Kind::Whitespace
     }
     // Unicode punctuation.
-    else if PUNCTUATION.contains(&char) {
+    else if char.is_ascii_punctuation() || PUNCTUATION.contains(&char) {
         Kind::Punctuation
     }
     // Everything else.

diff --git a/tests/gfm_strikethrough.rs b/tests/gfm_strikethrough.rs
@@ -363,6 +363,18 @@ u ~**xxx**~ zzz
         "should handle interplay like GitHub"
     );
 
+    assert_eq!(
+        to_html_with_options("a*~b~*c\n\na*.b.*c", &Options::gfm())?,
+        "<p>a<em><del>b</del></em>c</p>\n<p>a*.b.*c</p>",
+        "should handle interplay w/ other attention markers (GFM)"
+    );
+
+    assert_eq!(
+        to_html("a*~b~*c\n\na*.b.*c"),
+        "<p>a*~b~*c</p>\n<p>a*.b.*c</p>",
+        "should handle interplay w/ other attention markers (CM reference)"
+    );
+
     assert_eq!(
         to_html_with_options(
             "a ~b~ ~~c~~ d",