Fix lookahead/lookbehind breaking capture group count (#2)

c272 · Jul 9, 2021 · 1dbf21c · 1dbf21c
1 parent 44e6fd5
commit 1dbf21c
Show file tree

Hide file tree

Showing 9 changed files with 27 additions and 9 deletions.
diff --git a/.gitignore b/.gitignore
@@ -265,4 +265,5 @@ paket-files/
 
 # Python Tools for Visual Studio (PTVS)
 __pycache__/
-*.pyc
+*.pyc
+/iroBaseVisitor.cs
diff --git a/Compiler/Compiler.cs b/Compiler/Compiler.cs
@@ -20,8 +20,11 @@ public static bool GroupsMatch(string data, int expectedGroups)
             //Get groups out, ignoring escape chars.
             int realGroups = 0;
             bool ignoreNext = false, inCharSet = false;
-            foreach (var c in data)
+            for (int i=0; i<data.Length; i++)
             {
+                //Get current character.
+                char c = data[i];
+
                 //Ignore escape chars.
                 if (c == '\\') { ignoreNext = true; continue; }
                 if (ignoreNext) { ignoreNext = false; continue; }
@@ -31,7 +34,17 @@ public static bool GroupsMatch(string data, int expectedGroups)
                 if (c == ']') { inCharSet = false; }
                 if (inCharSet) { continue; }
 
-                if (c == '(') { realGroups++; }
+                //If there's an open bracket, this might be a group.
+                if (c == '(') 
+                { 
+                    //Is this a lookahead/lookbehind?
+                    if (i<data.Length-1 && data[i+1] == '?')
+                    {
+                        //Yes, ignore.
+                        continue;
+                    }
+                    realGroups++; 
+                }
             }
 
             //Return equality.

diff --git a/obj/Debug/.NETFramework,Version=v4.7.1.AssemblyAttributes.cs b/obj/Debug/.NETFramework,Version=v4.7.1.AssemblyAttributes.cs
@@ -0,0 +1,4 @@
+// <autogenerated />
+using System;
+using System.Reflection;
+[assembly: global::System.Runtime.Versioning.TargetFrameworkAttribute(".NETFramework,Version=v4.7.1", FrameworkDisplayName = ".NET Framework 4.7.1")]
diff --git a/obj/Debug/iroBaseListener.cs b/obj/Debug/iroBaseListener.cs
@@ -8,7 +8,7 @@
 // </auto-generated>
 //------------------------------------------------------------------------------
 
-// Generated from C:\Users\Larry\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
+// Generated from C:\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
 
 // Unreachable code detected
 #pragma warning disable 0162

diff --git a/obj/Debug/iroBaseVisitor.cs b/obj/Debug/iroBaseVisitor.cs
@@ -8,7 +8,7 @@
 // </auto-generated>
 //------------------------------------------------------------------------------
 
-// Generated from C:\Users\Larry\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
+// Generated from C:\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
 
 // Unreachable code detected
 #pragma warning disable 0162

diff --git a/obj/Debug/iroLexer.cs b/obj/Debug/iroLexer.cs
@@ -8,7 +8,7 @@
 // </auto-generated>
 //------------------------------------------------------------------------------
 
-// Generated from C:\Users\Larry\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
+// Generated from C:\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
 
 // Unreachable code detected
 #pragma warning disable 0162

diff --git a/obj/Debug/iroListener.cs b/obj/Debug/iroListener.cs
@@ -8,7 +8,7 @@
 // </auto-generated>
 //------------------------------------------------------------------------------
 
-// Generated from C:\Users\Larry\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
+// Generated from C:\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
 
 // Unreachable code detected
 #pragma warning disable 0162

diff --git a/obj/Debug/iroParser.cs b/obj/Debug/iroParser.cs
@@ -8,7 +8,7 @@
 // </auto-generated>
 //------------------------------------------------------------------------------
 
-// Generated from C:\Users\Larry\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
+// Generated from C:\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
 
 // Unreachable code detected
 #pragma warning disable 0162

diff --git a/obj/Debug/iroVisitor.cs b/obj/Debug/iroVisitor.cs
@@ -8,7 +8,7 @@
 // </auto-generated>
 //------------------------------------------------------------------------------
 
-// Generated from C:\Users\Larry\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
+// Generated from C:\Files\Programming\iro4cli\Grammar\iro.g4 by ANTLR 4.6.6
 
 // Unreachable code detected
 #pragma warning disable 0162