Merge branch 'next' into ExtractComponentsFromQMNsSecondTry

Author: MDoerner

Rubberduck.Parsing/Grammar/VBALexer.g4

@@ -295,8 +295,7 @@ SINGLEQUOTE : '\'';
 UNDERSCORE : '_';
 WS : [ \t];
 GUIDLITERAL : '{' [0-9A-F]+ '-' [0-9A-F]+ '-' [0-9A-F]+ '-' [0-9A-F]+ '-' [0-9A-F]+ '}';
-LETTER_RANGE : LETTER WS? MINUS WS? LETTER;
-IDENTIFIER :  ~[[\]()\r\n\t.,'"|!@#$%^&*\-+:=; 0-9-/\\] ~[[\]()\r\n\t.,'"|!@#$%^&*\-+:=; ]*;
+IDENTIFIER :  ~[[\]()\r\n\t.,'"|!@#$%^&*\-+:=; 0-9-/\\-] ~[[\]()\r\n\t.,'"|!@#$%^&*\-+:=; -]*;
 LINE_CONTINUATION : [ \t]* UNDERSCORE [ \t]* '\r'? '\n';
 fragment LETTER : [a-zA-Z_äöüÄÖÜ];
 fragment DIGIT : [0-9];

Rubberduck.Parsing/Grammar/VBAParser.g4

@@ -21,6 +21,8 @@ parser grammar VBAParser;
 
 options { tokenVocab = VBALexer; }
 
+@header { using System.Text.RegularExpressions; }
+
 startRule : module EOF;
 
 module :
@@ -289,16 +291,17 @@ defType :
 // singleLetter must appear at the end to prevent premature bailout
 letterSpec : universalLetterRange | letterRange | singleLetter;
 
-// we're purpusefully not actually making sure this is an ASCII letter
-// that'd be too much of a hassle with the lexer to get working
-singleLetter : {_input.Lt(1).Text.Length == 1}? IDENTIFIER;
+singleLetter : {_input.Lt(1).Text.Length == 1 && Regex.Match(_input.Lt(1).Text, @"[a-zA-Z]").Success}? IDENTIFIER;
+
 // We make a separate universalLetterRange rule because it is treated specially in VBA. This makes it easy for users of the parser
 // to identify this case. Quoting MS VBAL:
 // "A <universal-letter-range> defines a single implicit declared type for every <IDENTIFIER> within 
 // a module, even those with a first character that would otherwise fall outside this range if it was 
 // interpreted as a <letter-range> from A-Z.""
-universalLetterRange : {_input.Lt(1).Text.StartsWith("A") && _input.Lt(1).Text.EndsWith("Z")}? LETTER_RANGE;
-letterRange : LETTER_RANGE;
+universalLetterRange : {_input.Lt(1).Text.Equals("A") && _input.Lt(3).Text.Equals("Z")}? IDENTIFIER MINUS IDENTIFIER;
+ 
+letterRange : singleLetter MINUS singleLetter;
+
 
 doLoopStmt :
     DO endOfStatement 

RubberduckTests/Grammar/VBAParserTests.cs

@@ -4,7 +4,6 @@
 using Antlr4.Runtime.Tree.Xpath;
 using NUnit.Framework;
 using Rubberduck.Parsing.Grammar;
-using Rubberduck.Parsing.Symbols.ParsingExceptions;
 using System;
 using System.Collections.Generic;
 using System.Diagnostics;
@@ -392,7 +391,7 @@ public void TestDefDirectiveLetterRange()
         [Test]
         public void TestDefDirectiveUniversalLetterRange()
         {
-            string code = @"DefBool A - Z";
+            string code = @"DefBool A-Z";
             var parseResult = Parse(code);
             AssertTree(parseResult.Item1, parseResult.Item2, "//universalLetterRange");
         }
@@ -2136,6 +2135,26 @@ End Sub
             AssertTree(parseResult.Item1, parseResult.Item2, "//attributeStmt", matches => matches.Count == 1);
         }
 
+        [Category("Parser")]
+        [Test]
+        public void SubtractionExpressionsAreNoLetterRanges()
+        {
+            const string code = @"
+Public Sub Foo()
+    Dim a As Long
+    Dim b As Long
+    Dim z As Long
+    a = 1
+    b = 2
+    z = a-b
+    b = a-z
+End Sub
+";
+            var parseResult = Parse(code);
+            AssertTree(parseResult.Item1, parseResult.Item2, "//letterRange", matches => matches.Count == 0);
+            AssertTree(parseResult.Item1, parseResult.Item2, "//universalLetterRange", matches => matches.Count == 0);
+        }
+
         private Tuple<VBAParser, ParserRuleContext> Parse(string code, PredictionMode predictionMode = null)
         {
             var stream = new AntlrInputStream(code);
@@ -2157,6 +2176,7 @@ private Tuple<VBAParser, ParserRuleContext> Parse(string code, PredictionMode pr
                 {
                     // If SLL fails we want to get notified ASAP so we can fix it, that's why we don't retry using LL.
                     // If LL mode fails, we're done.
+
                     throw;
                 }