[assembler] Support super/subscript arrow.

Author: youngman-g

assembler/src/asmlib/lexer.rs

@@ -98,7 +98,7 @@ pub(crate) enum Token {
     /// section 6-3.2, "RULES FOR SYMEX FORMATION".
     Hold,
     NotHold, // handled specially, there is no glyph for this.
-    Arrow,
+    Arrow(Script),
     Hand(Script),
     Hash(Script),
     Equals(Script),
@@ -177,7 +177,7 @@ impl Display for Token {
             Token::RightParen(script) => write_elevated(script, ")"),
             Token::Hold => f.write_char('h'),
             Token::NotHold => f.write_char('ℏ'),
-            Token::Arrow => f.write_str("->"),
+            Token::Arrow(script) => write_elevated(script, "->"),
             Token::Hand(script) => write_elevated(script, "☛"),
             Token::Asterisk => f.write_char('*'),
             Token::Dot(script) => write_elevated(script, DOT_STR),
@@ -495,10 +495,7 @@ mod lexer_impl_new {
             GlyphShape::Solidus => Some(Token::Solidus(script)),
             GlyphShape::Times => Some(Token::Times(script)),
             GlyphShape::Hash => Some(Token::Hash(script)),
-            GlyphShape::Arrow => Some(match script {
-                Script::Super | Script::Sub => unimplemented!(),
-                Script::Normal => Token::Arrow,
-            }),
+            GlyphShape::Arrow => Some(Token::Arrow(script)),
             GlyphShape::LessThan => only_normal(Token::LessThan),
             GlyphShape::GreaterThan => only_normal(Token::GreaterThan),
             GlyphShape::Overbar | GlyphShape::Square | GlyphShape::n => make_symex(),
@@ -576,7 +573,7 @@ mod lexer_impl_new {
         let merged_span = current_span.start..incoming_span.end;
         match current {
             Token::Minus(Script::Normal) if incoming == Token::GreaterThan => {
-                TokenMergeResult::Merged(Token::Arrow, merged_span)
+                TokenMergeResult::Merged(Token::Arrow(Script::Normal), merged_span)
             }
             Token::SymexSyllable(existing_script, mut existing_name) => match incoming {
                 Token::Hold if existing_script == Script::Normal => {

assembler/src/asmlib/lexer/tests.rs

@@ -138,20 +138,31 @@ fn test_hand_hand_normal() {
 
 #[test]
 fn test_arrow_plain() {
-    assert_eq!(scan_tokens_only("->"), Ok(vec![Token::Arrow]));
+    assert_eq!(
+        scan_tokens_only("->"),
+        Ok(vec![Token::Arrow(Script::Normal)])
+    );
+    // Since the user has @arr@, @sup_arr@, etc. we don't actually
+    // require @sub_minus@@sub_greater@ to be understood as an arrow
+    // (though it likely is).
 }
 
 #[test]
 fn test_double_arrow_plain() {
+    use Script::Normal;
     assert_eq!(
         scan_tokens_only("->->"),
-        Ok(vec![Token::Arrow, Token::Arrow])
+        Ok(vec![Token::Arrow(Normal), Token::Arrow(Normal)])
     );
 }
 
 #[test]
 fn test_arrow_as_glyph() {
-    assert_eq!(scan_tokens_only("@arr@"), Ok(vec![Token::Arrow]));
+    use Script::*;
+    assert_eq!(scan_tokens_only("@arr@"), Ok(vec![Token::Arrow(Normal)]));
+
+    assert_eq!(scan_tokens_only("@sup_arr@"), Ok(vec![Token::Arrow(Super)]));
+    assert_eq!(scan_tokens_only("@sub_arr@"), Ok(vec![Token::Arrow(Sub)]));
 }
 
 #[test]
@@ -163,7 +174,7 @@ fn test_upper_lexer_span() {
         scan_slices("{->}"),
         Ok(vec![
             (Token::LeftBrace, "{"),
-            (Token::Arrow, "->"),
+            (Token::Arrow(Script::Normal), "->"),
             (Token::RightBrace, "}"),
         ])
     );
@@ -1084,9 +1095,10 @@ fn test_superscript_minus_unicode() {
 
 #[test]
 fn test_annotations_are_ignored() {
+    use Script::Normal;
     assert_eq!(
         scan_tokens_only("->[THIS IS AN ANNOTATION]->"),
-        Ok(vec![Token::Arrow, Token::Arrow,])
+        Ok(vec![Token::Arrow(Normal), Token::Arrow(Normal),])
     );
 }
 

assembler/src/asmlib/parser.rs

@@ -188,7 +188,7 @@ where
             name,
             span: extra.span(),
         })
-        .then_ignore(just(Tok::Arrow))
+        .then_ignore(just(Tok::Arrow(Script::Normal)))
         .labelled("tag definition")
 }
 
@@ -313,7 +313,7 @@ where
         just(Tok::Hand(Script::Normal)).to('☛'),
         just(Tok::Dot(Script::Normal)).to(lexer::DOT_CHAR),
         just(Tok::Equals(Script::Normal)).to('='),
-        just(Tok::Arrow).to('→'),
+        just(Tok::Arrow(Script::Normal)).to('→'),
         just(Tok::Pipe(Script::Normal)).to('|'),
         just(Tok::ProperSuperset).to('⊃'),
         just(Tok::IdenticalTo).to('≡'),