Fix parsing ambiguity around adjacent > and = tokens

Author: cirras

CHANGELOG.md

@@ -7,6 +7,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+### Fixed
+
+- Parsing errors where adjacent `>` and `=` tokens were wrongly interpreted as the `>=` operator.
+
 ## [1.13.0] - 2025-02-05
 
 ### Added

delphi-frontend/src/main/antlr3/au/com/integradev/delphi/antlr/Delphi.g

@@ -71,6 +71,8 @@ tokens {
   TkArgument;
   TkAnonymousMethod;
   TkAnonymousMethodHeading;
+  TkLessThanEqual;
+  TkGreaterThanEqual;
 }
 
 @header
@@ -328,13 +330,19 @@ import org.apache.commons.lang3.StringUtils;
     return t;
   }
 
-  private Token combineLastNTokens(int count) {
+  private Token combineLastNTokens(int type, int count) {
     CommonToken firstToken = (CommonToken) input.LT(-count);
     CommonToken lastToken = (CommonToken) input.LT(-1);
+    lastToken.setType(type);
     lastToken.setStartIndex(firstToken.getStartIndex());
     return lastToken;
   }
 
+  private BinaryExpressionNodeImpl createBinaryExpression(Object operator) {
+    Token token = adaptor.getToken(operator);
+    return new BinaryExpressionNodeImpl(token);
+  }
+
   @Override
   public void reportError(RecognitionException e) {
     String hdr = this.getErrorHeader(e);
@@ -965,12 +973,21 @@ unaryOperator                : NOT<UnaryExpressionNodeImpl>
 relationalOperator           : '='<BinaryExpressionNodeImpl>
                              | '>'<BinaryExpressionNodeImpl>
                              | '<'<BinaryExpressionNodeImpl>
-                             | '<='<BinaryExpressionNodeImpl>
-                             | '>='<BinaryExpressionNodeImpl>
+                             | op=lessThanEqualOperator -> {createBinaryExpression(op.getTree())}
+                             | op=greaterThanEqualOperator -> {createBinaryExpression(op.getTree())}
                              | '<>'<BinaryExpressionNodeImpl>
                              | IN<BinaryExpressionNodeImpl>
                              | IS<BinaryExpressionNodeImpl>
                              ;
+// We're only doing this for symmetry with greaterThanEqualOperator. (see comment below)
+lessThanEqualOperator        : '<' '=' -> ^({combineLastNTokens(TkLessThanEqual, 2)})
+                             ;
+// We construct the "greater than equal" tokens while parsing binary expressions to preserve the
+// individual '>' and '=' tokens in other cases like `const Foo: TArray<Byte>=[1, 2, 3];`, which
+// we otherwise couldn't parse since the `>=` token would consume the closing angle bracket of the
+// generic type arguments and the const assignment operator.
+greaterThanEqualOperator     : '>' '=' -> ^({combineLastNTokens(TkGreaterThanEqual, 2)})
+                             ;
 constExpression              : expression
                              | recordExpression
                              | arrayExpression
@@ -1372,8 +1389,6 @@ COLON                : ':'  ;
 EQUAL                : '='  ;
 NOT_EQUAL            : '<>' ;
 LESS_THAN            : '<'  ;
-LESS_THAN_EQUAL      : '<=' ;
-GREATER_THAN_EQUAL   : '>=' ;
 GREATER_THAN         : '>'  ;
 SQUARE_BRACKET_LEFT  : '['  ;
 SQUARE_BRACKET_RIGHT : ']'  ;

delphi-frontend/src/test/java/au/com/integradev/delphi/antlr/GrammarTest.java

@@ -353,4 +353,9 @@ void testConditionalAsm() {
   void testSemicolonSeparatedGenericArguments() {
     assertParsed("SemicolonSeparatedGenericArguments.pas");
   }
+
+  @Test
+  void testGreaterThanEqualAmbiguity() {
+    assertParsed("GreaterThanEqualAmbiguity.pas");
+  }
 }

delphi-frontend/src/test/resources/au/com/integradev/delphi/grammar/GreaterThanEqualAmbiguity.pas

@@ -0,0 +1,30 @@
+unit GreaterThanEqualAmbiguity;
+
+interface
+
+type
+  TFoo = class(TObject)
+    procedure Bar(Baz: IBaz<string>=nil);
+  end;
+
+implementation
+
+procedure TFoo.Bar(Baz: IBaz<string>=nil);
+begin
+  // do nothing
+end;
+
+function Flarp: TArray<Byte>;
+const
+  Bytes: TArray<Byte>=[1, 2, 3];
+begin
+  Result := Bytes;
+end;
+
+function FlimFlam: TArray<Byte>;
+begin
+  const Bytes: TArray<Byte>=[1, 2, 3];
+  Result := Bytes;
+end;
+
+end.