Used ANTLR lexer modes for java regex quotes \Q...\E.

Author: lmasroca

core/src/main/antlr4/org/evomaster/core/parser/RegexJavaLexer.g4

@@ -94,9 +94,6 @@ MINUS                      : '-';
 COMMA                      : ',';
 COLON                      : ':';
 
-Q: 'Q';
-E: 'E';
-
 BaseChar
  // practically all chars but the ones used for control and digits
  : ~[0-9:,^$\\.*+?()[\]{}|-]
@@ -134,3 +131,33 @@ fragment OctalDigit:
 NamedBackReference
  : SLASH 'k<' [a-zA-Z] [a-zA-Z0-9]* '>'
  ;
+
+// Special for Java: \Q...\E literal quoting.
+// Matching '\Q' switches the lexer into QUOTE_MODE (see below).
+QUOTE_OPEN
+ : '\\' 'Q' -> pushMode(QUOTE_MODE)
+ ;
+
+
+
+// --------------------------------------------------------------------------------
+// QUOTE_MODE: everything after \Q and before an optional \E is literal text.
+// By using lexer modes we skip having to explicitly allow all lexer tokens to appear
+// within quotes. This works by having a different set of tokens available when inside
+// a quote block, the default mode and the QUOTE_MODE each keep their own set of tokens.
+// QUOTE_MODE only has the lexer tokens defined below.
+// --------------------------------------------------------------------------------
+mode QUOTE_MODE;
+
+// The only way out of a \Q...\E block. In Java the first literal "\E" sequence ends quoting.
+QUOTE_CLOSE
+ : '\\' 'E' -> popMode
+ ;
+
+// One or more characters that do not form a "\E" sequence, so either:
+// - a backslash immediately followed by something other than 'E',
+// - any single non-backslash character.
+// Note that this cannot consume the sequence QUOTE_CLOSE.
+QUOTE_CONTENT
+ : ( '\\' ~[E] | ~[\\] )+
+ ;

core/src/main/antlr4/org/evomaster/core/parser/RegexJavaParser.g4

@@ -103,19 +103,7 @@ atom
 
 // Special for Java
 quote
- : SLASH Q quoteBlock SLASH E
-;
-
-quoteBlock
-  : quoteChar*
-;
-
-quoteChar
- : classAtom
- | SLASH
- | BRACKET_close
- | Q
- | E
+ : QUOTE_OPEN QUOTE_CONTENT? QUOTE_CLOSE? // both the content and closing the quote is optional in java regex.
 ;
 
 patternCharacter
@@ -125,7 +113,6 @@ patternCharacter
  | COMMA
  | MINUS
  | DecimalDigit
- | E | Q
  // These are also allowed as literals when no matching pair exists
  | BRACE_close
  | BRACKET_close

core/src/main/kotlin/org/evomaster/core/parser/GeneRegexJavaVisitor.kt

@@ -366,8 +366,7 @@ class GeneRegexJavaVisitor(externalRegexFlags: RegexFlags = RegexFlags()) : Rege
 
         if(ctx.quote() != null){
 
-            val block = ctx.quote().quoteBlock().quoteChar().map { it.text }
-                    .joinToString("")
+            val block = ctx.quote().QUOTE_CONTENT()?.text ?: ""
 
             val name = if(block.isBlank()) "blankBlock" else block
 

core/src/test/kotlin/org/evomaster/core/parser/GeneRegexJavaVisitorTest.kt

@@ -56,6 +56,7 @@ class GeneRegexJavaVisitorTest : GeneRegexEcma262VisitorTest() {
         checkSameAsJava("^(\\Q6\\E(n|N)(u|U)(a|A)(q|Q)(w|W)(b|B)\\Q51\\E(y|Y)(w|W)\\Q1\\E(e|E)(r|R)(n|N))$")
         checkSameAsJava("^((z|Z)(l|L)(q|Q)\\Q9\\E(r|R)(e|E)(k|K)(q|Q)(b|B)\\Q6\\E(e|E)(q|Q)(u|U))$")
         checkSameAsJava("^(\\Q81\\E(x|X)(a|A)\\Q3\\E(p|P)(x|X)(d|D))$")
+        checkSameAsJava("a{1}\\Qa{1}\\Qabcd")
     }
 
     @Test