1 files changed, 46 insertions, 39 deletions

diff --git a/compiler/rustc_lexer/src/lib.rs b/compiler/rustc_lexer/src/lib.rs
index e80196ed567..f6790f7ed1e 100644
--- a/compiler/rustc_lexer/src/lib.rs
+++ b/compiler/rustc_lexer/src/lib.rs
@@ -331,24 +331,37 @@ pub fn is_whitespace(c: char) -> bool {
 
     matches!(
         c,
-        // Usual ASCII suspects
-        '\u{0009}'   // \t
-        | '\u{000A}' // \n
+        // End-of-line characters
+        | '\u{000A}' // line feed (\n)
         | '\u{000B}' // vertical tab
         | '\u{000C}' // form feed
-        | '\u{000D}' // \r
-        | '\u{0020}' // space
-
-        // NEXT LINE from latin1
-        | '\u{0085}'
+        | '\u{000D}' // carriage return (\r)
+        | '\u{0085}' // next line (from latin1)
+        | '\u{2028}' // LINE SEPARATOR
+        | '\u{2029}' // PARAGRAPH SEPARATOR
 
-        // Bidi markers
+        // `Default_Ignorable_Code_Point` characters
         | '\u{200E}' // LEFT-TO-RIGHT MARK
         | '\u{200F}' // RIGHT-TO-LEFT MARK
 
-        // Dedicated whitespace characters from Unicode
-        | '\u{2028}' // LINE SEPARATOR
-        | '\u{2029}' // PARAGRAPH SEPARATOR
+        // Horizontal space characters
+        | '\u{0009}'   // tab (\t)
+        | '\u{0020}' // space
+    )
+}
+
+/// True if `c` is considered horizontal whitespace according to Rust language definition.
+pub fn is_horizontal_whitespace(c: char) -> bool {
+    // This is Pattern_White_Space.
+    //
+    // Note that this set is stable (ie, it doesn't change with different
+    // Unicode versions), so it's ok to just hard-code the values.
+
+    matches!(
+        c,
+        // Horizontal space characters
+        '\u{0009}'   // tab (\t)
+        | '\u{0020}' // space
     )
 }
 
@@ -538,40 +551,32 @@ impl Cursor<'_> {
         debug_assert!(length_opening >= 3);
 
         // whitespace between the opening and the infostring.
-        self.eat_while(|ch| ch != '\n' && is_whitespace(ch));
+        self.eat_while(|ch| ch != '\n' && is_horizontal_whitespace(ch));
 
-        // copied from `eat_identifier`, but allows `.` in infostring to allow something like
+        // copied from `eat_identifier`, but allows `-` and `.` in infostring to allow something like
         // `---Cargo.toml` as a valid opener
         if is_id_start(self.first()) {
             self.bump();
-            self.eat_while(|c| is_id_continue(c) || c == '.');
+            self.eat_while(|c| is_id_continue(c) || c == '-' || c == '.');
         }
 
-        self.eat_while(|ch| ch != '\n' && is_whitespace(ch));
+        self.eat_while(|ch| ch != '\n' && is_horizontal_whitespace(ch));
         let invalid_infostring = self.first() != '\n';
 
-        let mut s = self.as_str();
         let mut found = false;
-        let mut size = 0;
-        while let Some(closing) = s.find(&"-".repeat(length_opening as usize)) {
-            let preceding_chars_start = s[..closing].rfind("\n").map_or(0, |i| i + 1);
-            if s[preceding_chars_start..closing].chars().all(is_whitespace) {
-                // candidate found
-                self.bump_bytes(size + closing);
-                // in case like
-                // ---cargo
-                // --- blahblah
-                // or
-                // ---cargo
-                // ----
-                // combine those stuff into this frontmatter token such that it gets detected later.
-                self.eat_until(b'\n');
-                found = true;
-                break;
-            } else {
-                s = &s[closing + length_opening as usize..];
-                size += closing + length_opening as usize;
-            }
+        let nl_fence_pattern = format!("\n{:-<1$}", "", length_opening as usize);
+        if let Some(closing) = self.as_str().find(&nl_fence_pattern) {
+            // candidate found
+            self.bump_bytes(closing + nl_fence_pattern.len());
+            // in case like
+            // ---cargo
+            // --- blahblah
+            // or
+            // ---cargo
+            // ----
+            // combine those stuff into this frontmatter token such that it gets detected later.
+            self.eat_until(b'\n');
+            found = true;
         }
 
         if !found {
@@ -594,14 +599,16 @@ impl Cursor<'_> {
             if potential_closing.is_none() {
                 // a less fortunate recovery if all else fails which finds any dashes preceded by whitespace
                 // on a standalone line. Might be wrong.
+                let mut base_index = 0;
                 while let Some(closing) = rest.find("---") {
                     let preceding_chars_start = rest[..closing].rfind("\n").map_or(0, |i| i + 1);
-                    if rest[preceding_chars_start..closing].chars().all(is_whitespace) {
+                    if rest[preceding_chars_start..closing].chars().all(is_horizontal_whitespace) {
                         // candidate found
-                        potential_closing = Some(closing);
+                        potential_closing = Some(closing + base_index);
                         break;
                     } else {
                         rest = &rest[closing + 3..];
+                        base_index += closing + 3;
                     }
                 }
             }