Merge pull request #668 from notriddle/notriddle/lowercase-html-decla…

…rations Allow lowercase letters in HTML declarations
pulldown-cmark · Jun 21, 2023 · ab7c7bf · ab7c7bf
2 parents 34c2bb5 + d4d9b78
commit ab7c7bf
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 10 deletions.
diff --git a/specs/regression.txt b/specs/regression.txt
@@ -1054,3 +1054,12 @@ ISSUE 656
 →|-----|------|
 | baz | alef |</p>
 ````````````````````````````````
+
+
+ISSUE 659
+
+```````````````````````````````` example
+<!doctype html>
+.
+<!doctype html>
+````````````````````````````````
diff --git a/src/firstpass.rs b/src/firstpass.rs
@@ -1466,8 +1466,7 @@ fn get_html_end_tag(text_bytes: &[u8]) -> Option<&'static str> {
 
     if text_bytes.len() > 1
         && text_bytes[0] == b'!'
-        && text_bytes[1] >= b'A'
-        && text_bytes[1] <= b'Z'
+        && text_bytes[1].is_ascii_alphabetic()
     {
         Some(">")
     } else {

diff --git a/src/scanners.rs b/src/scanners.rs
@@ -1312,6 +1312,8 @@ pub(crate) fn scan_inline_html_comment(
     let c = *bytes.get(ix)?;
     ix += 1;
     match c {
+        // An HTML comment consists of `<!--` + text + `-->`, where text does not start with `>`
+        // or `->`, does not end with -, and does not contain --.
         b'-' => {
             let dashes = scan_ch_repeat(&bytes[ix..], b'-');
             if dashes < 1 {
@@ -1336,6 +1338,8 @@ pub(crate) fn scan_inline_html_comment(
             }
             None
         }
+        // A CDATA section consists of the string `<![CDATA[`, a string of characters not
+        // including the string `]]>`, and the string `]]>`.
         b'[' if bytes[ix..].starts_with(b"CDATA[") && ix > scan_guard.cdata => {
             ix += b"CDATA[".len();
             ix = memchr(b']', &bytes[ix..]).map_or(bytes.len(), |x| ix + x);
@@ -1349,14 +1353,9 @@ pub(crate) fn scan_inline_html_comment(
                 Some(ix + 1)
             }
         }
-        b'A'..=b'Z' if ix > scan_guard.declaration => {
-            // Scan declaration.
-            ix += scan_while(&bytes[ix..], |c| c >= b'A' && c <= b'Z');
-            let whitespace = scan_while(&bytes[ix..], is_ascii_whitespace);
-            if whitespace == 0 {
-                return None;
-            }
-            ix += whitespace;
+        // A declaration consists of the string `<!`, an ASCII letter, zero or more characters not
+        // including the character >, and the character >.
+        _ if c.is_ascii_alphabetic() && ix > scan_guard.declaration => {
             ix = memchr(b'>', &bytes[ix..]).map_or(bytes.len(), |x| ix + x);
             if scan_ch(&bytes[ix..], b'>') == 0 {
                 scan_guard.declaration = ix;

diff --git a/tests/suite/regression.rs b/tests/suite/regression.rs
@@ -1241,3 +1241,13 @@ fn regression_test_83() {
 
     test_markdown_html(original, expected, false, false, false);
 }
+
+#[test]
+fn regression_test_84() {
+    let original = r##"<!doctype html>
+"##;
+    let expected = r##"<!doctype html>
+"##;
+
+    test_markdown_html(original, expected, false, false, false);
+}