works, we just back off a little early

Author: milseman

Sources/_MatchingEngine/Regex/DSLTree.swift

@@ -21,6 +21,8 @@ extension DSLTree {
       AST.Quantification.Kind,
       Node)
 
+    case customCharacterClass(CustomCharacterClass)
+
     /// Comments, non-semantic whitespace, etc
     // TODO: Do we want this? Could be interesting
     case trivia(String)
@@ -34,11 +36,16 @@ extension DSLTree {
     /// An embedded literal
     case regexLiteral(AST)
 
+    // MARK: - Tree conversions
+
     /// The target of AST conversion.
     ///
     /// Keeps original AST around for rich syntatic and source information
     case convertedRegexLiteral(Node, AST)
 
+    // Fall-back for when conversion fails
+    case unconvertedRegexLiteral(AST)
+
     // MARK: - Extensibility points
 
     /// A capturing group (TODO: is it?) with a transformation function
@@ -51,7 +58,19 @@ extension DSLTree {
 
     case consumerValidator(_ConsumerValidatorInterface)
 
-    case customCharacterClass(_CharacterPredicateInterface)
+    // TODO: Would this just boil down to a consumer?
+    case characterPredicate(_CharacterPredicateInterface)
+  }
+}
+
+extension DSLTree {
+  enum CustomCharacterClass {
+    // TODO: Still need to convert atoms
+    case members([AST.Atom])
+
+    indirect case intersection(Self, Self)
+    indirect case subtraction(Self, Self)
+    indirect case symmetricDifference(Self, Self)
   }
 }
 
@@ -99,8 +118,8 @@ extension DSLTree.Node {
     case let .conditional(_, t, f): return [t,f]
 
     case .trivia, .empty, .stringLiteral, .regexLiteral,
-        .consumer, .consumerValidator,
-        .customCharacterClass:
+        .consumer, .consumerValidator, .characterPredicate,
+        .customCharacterClass, .unconvertedRegexLiteral:
       return []
     }
   }
@@ -109,7 +128,8 @@ extension DSLTree.Node {
 extension DSLTree.Node {
   var ast: AST? {
     switch self {
-    case let .regexLiteral(ast): return ast
+    case let .regexLiteral(ast):             return ast
+    case let .unconvertedRegexLiteral(ast):  return ast
     case let .convertedRegexLiteral(_, ast): return ast
     default: return nil
     }
@@ -180,6 +200,7 @@ extension AST {
             children.append(.stringLiteral(str))
           } else {
             children.append(astChildren[curIdx].dslTreeNode)
+            children.formIndex(after: &curIdx)
           }
         }
         return .concatenation(children)
@@ -210,15 +231,13 @@ extension AST {
           return .stringLiteral(str)
         }
 
-        // TODO: Should we bottom out here or do something
-        // else? We probably want to rewrite built-in
-        // character classes, etc
+        // TODO: More conversions
+        return .unconvertedRegexLiteral(self)
 
-        return .regexLiteral(self)
+      case .customCharacterClass(_):
 
-      case let .customCharacterClass(v):
-        // FIXME: What should we do here?
-        fatalError("TODO \(v)")
+        // TODO: More conversions
+        return .unconvertedRegexLiteral(self)
 
       case .empty(_):
         return .empty

Sources/_MatchingEngine/Regex/Printing/PrintAsPattern.swift

@@ -102,8 +102,11 @@ extension PrettyPrinter {
 
     case let .stringLiteral(v):
       print(v._quoted)
+
     case let .regexLiteral(v):
       printAsCanonical(v, delimiters: true)
+    case let .unconvertedRegexLiteral(v):
+      printAsCanonical(v, delimiters: true)
 
     case let .convertedRegexLiteral(n, _):
       // FIXME: This recursion coordinates with back-off
@@ -112,14 +115,18 @@ extension PrettyPrinter {
       // label is a lie.
       printAsPattern(convertedFromAST: n)
 
+    case .customCharacterClass:
+      print("/* TODO: custom character class */")
+
     case .groupTransform:
       print("/* TODO: group transforms */")
     case .consumer:
       print("/* TODO: consumers */")
     case .consumerValidator:
       print("/* TODO: consumer validators */")
-    case .customCharacterClass:
-      print("/* TODO: custom character class */")
+    case .characterPredicate:
+      print("/* TODO: character predicates */")
+
     }
   }