Re-introduce DSLTree.Atom.any

This time as a "true any" that matches any
character, including newlines.

Author: hamishknight

Sources/_StringProcessing/ByteCodeGen.swift

@@ -55,6 +55,9 @@ fileprivate extension Compiler.ByteCodeGen {
       }
     }
     switch a {
+    case .any:
+      emitAny()
+
     case .dot:
       emitDot()
 
@@ -320,23 +323,31 @@ fileprivate extension Compiler.ByteCodeGen {
     builder.buildMatch(c, isCaseInsensitive: false)
   }
 
-  mutating func emitDot() {
-    switch (options.semanticLevel, options.dotMatchesNewline) {
-    case (.graphemeCluster, true):
+  mutating func emitAny() {
+    switch options.semanticLevel {
+    case .graphemeCluster:
       builder.buildAdvance(1)
-    case (.graphemeCluster, false):
+    case .unicodeScalar:
+      // TODO: builder.buildAdvanceUnicodeScalar(1)
       builder.buildConsume { input, bounds in
-        input[bounds.lowerBound].isNewline
-        ? nil
-        : input.index(after: bounds.lowerBound)
+        input.unicodeScalars.index(after: bounds.lowerBound)
       }
+    }
+  }
 
-    case (.unicodeScalar, true):
-      // TODO: builder.buildAdvanceUnicodeScalar(1)
+  mutating func emitDot() {
+    if options.dotMatchesNewline {
+      emitAny()
+      return
+    }
+    switch options.semanticLevel {
+    case .graphemeCluster:
       builder.buildConsume { input, bounds in
-        input.unicodeScalars.index(after: bounds.lowerBound)
+        input[bounds.lowerBound].isNewline
+        ? nil
+        : input.index(after: bounds.lowerBound)
       }
-    case (.unicodeScalar, false):
+    case .unicodeScalar:
       builder.buildConsume { input, bounds in
         input[bounds.lowerBound].isNewline
         ? nil

Sources/_StringProcessing/ConsumerInterface.swift

@@ -128,7 +128,7 @@ extension DSLTree.Atom {
       // can match a single scalar in scalar semantic mode.
       return try Character(s).generateConsumer(opts)
 
-    case .dot:
+    case .any, .dot:
       // FIXME: Should this be a total ordering?
       if opts.semanticLevel == .graphemeCluster {
         return { input, bounds in

Sources/_StringProcessing/PrintAsPattern.swift

@@ -1124,6 +1124,9 @@ extension DSLTree.Atom {
     _ printer: inout PrettyPrinter
   ) -> (String, canBeWrapped: Bool)? {
     switch self {
+    case .any:
+      return (".any", true)
+
     case .dot:
       // FIXME: This is wrong, the DSL doesn't have an equivalent to .dot.
       return (".any", true)
@@ -1167,6 +1170,9 @@ extension DSLTree.Atom {
 
   var _regexBase: String {
     switch self {
+    case .any:
+      return "(?s:.)"
+
     case .dot:
       return "."
 

Sources/_StringProcessing/Regex/DSLTree.swift

@@ -166,6 +166,9 @@ extension DSLTree {
     case char(Character)
     case scalar(Unicode.Scalar)
 
+    /// Any character, including newlines.
+    case any
+
     /// The DSL representation of '.' in a regex literal. This does not match
     /// newlines unless single line mode is enabled.
     case dot
@@ -780,7 +783,8 @@ extension DSLTree.Atom {
     switch self {
     case .changeMatchingOptions, .assertion:
       return false
-    case .char, .scalar, .dot, .backreference, .symbolicReference, .unconverted:
+    case .char, .scalar, .any, .dot, .backreference, .symbolicReference,
+        .unconverted:
       return true
     }
   }