Merge pull request #393 from hamishknight/stricter-syntax

Author: hamishknight

Sources/_RegexParser/Regex/AST/AST.swift

@@ -52,6 +52,9 @@ extension AST {
     /// Comments, non-semantic whitespace, etc
     case trivia(Trivia)
 
+    /// Intepolation `<{...}>`, currently reserved for future use.
+    case interpolation(Interpolation)
+
     case atom(Atom)
 
     case customCharacterClass(CustomCharacterClass)
@@ -77,6 +80,7 @@ extension AST.Node {
     case let .quantification(v):        return v
     case let .quote(v):                 return v
     case let .trivia(v):                return v
+    case let .interpolation(v):         return v
     case let .atom(v):                  return v
     case let .customCharacterClass(v):  return v
     case let .empty(v):                 return v
@@ -129,7 +133,7 @@ extension AST.Node {
     case .conditional, .customCharacterClass, .absentFunction:
       return true
     case .alternation, .concatenation, .quantification, .quote, .trivia,
-        .empty:
+        .empty, .interpolation:
       return false
     }
   }
@@ -193,6 +197,16 @@ extension AST {
     }
   }
 
+  public struct Interpolation: Hashable, _ASTNode {
+    public let contents: String
+    public let location: SourceLocation
+
+    public init(_ contents: String, _ location: SourceLocation) {
+      self.contents = contents
+      self.location = location
+    }
+  }
+
   public struct Empty: Hashable, _ASTNode {
     public let location: SourceLocation
 

Sources/_RegexParser/Regex/AST/Atom.swift

@@ -819,7 +819,7 @@ extension AST.Node {
     case .alternation, .concatenation, .group,
         .conditional, .quantification, .quote,
         .trivia, .customCharacterClass, .empty,
-        .absentFunction:
+        .absentFunction, .interpolation:
       return nil
     }
   }

Sources/_RegexParser/Regex/Parse/CaptureList.swift

@@ -103,7 +103,7 @@ extension AST.Node {
         break
       }
 
-    case .quote, .trivia, .atom, .customCharacterClass, .empty:
+    case .quote, .trivia, .atom, .customCharacterClass, .empty, .interpolation:
       break
     }
   }

Sources/_RegexParser/Regex/Parse/Diagnostics.swift

@@ -42,6 +42,7 @@ enum ParseError: Error, Hashable {
   case expectedNonEmptyContents
   case expectedEscape
   case invalidEscape(Character)
+  case confusableCharacter(Character)
 
   case cannotReferToWholePattern
 
@@ -128,6 +129,8 @@ extension ParseError: CustomStringConvertible {
       return "expected escape sequence"
     case .invalidEscape(let c):
       return "invalid escape sequence '\\\(c)'"
+    case .confusableCharacter(let c):
+      return "'\(c)' is confusable for a metacharacter; use '\\u{...}' instead"
     case .cannotReferToWholePattern:
       return "cannot refer to whole pattern here"
     case .quantifierRequiresOperand(let q):

Sources/_RegexParser/Regex/Parse/LexicalAnalysis.swift

@@ -589,6 +589,26 @@ extension Source {
     return AST.Quote(str.value, str.location)
   }
 
+  /// Try to consume an interpolation sequence.
+  ///
+  ///     Interpolation -> '<{' String '}>'
+  ///
+  mutating func lexInterpolation() throws -> AST.Interpolation? {
+    let contents = try recordLoc { src -> String? in
+      try src.tryEating { src in
+        guard src.tryEat(sequence: "<{") else { return nil }
+        _ = src.lexUntil { $0.isEmpty || $0.starts(with: "}>") }
+        guard src.tryEat(sequence: "}>") else { return nil }
+
+        // Not currently supported. We error here instead of during Sema to
+        // get a better error for something like `(<{)}>`.
+        throw ParseError.unsupported("interpolation")
+      }
+    }
+    guard let contents = contents else { return nil }
+    return .init(contents.value, contents.location)
+  }
+
   /// Try to consume a comment
   ///
   ///     Comment -> '(?#' [^')']* ')'
@@ -1674,9 +1694,10 @@ extension Source {
         break
       }
 
-      // We only allow unknown escape sequences for non-letter ASCII, and
-      // non-ASCII whitespace.
-      guard (char.isASCII && !char.isLetter) ||
+      // We only allow unknown escape sequences for non-letter non-number ASCII,
+      // and non-ASCII whitespace.
+      // TODO: Once we have fix-its, suggest a `0` prefix for octal `[\7]`.
+      guard (char.isASCII && !char.isLetter && !char.isNumber) ||
               (!char.isASCII && char.isWhitespace)
       else {
         throw ParseError.invalidEscape(char)
@@ -1981,10 +2002,21 @@ extension Source {
 
       case "]":
         assert(!customCC, "parser should have prevented this")
-        fallthrough
+        break
 
-      default: return .char(char)
+      default:
+        // Reject non-letter non-number non-`\r\n` ASCII characters that have
+        // multiple scalars. These may be confusable for metacharacters, e.g
+        // `[\u{301}]` wouldn't be interpreted as a custom character class due
+        // to the combining accent (assuming it is literal, not `\u{...}`).
+        let scalars = char.unicodeScalars
+        if scalars.count > 1 && scalars.first!.isASCII && char != "\r\n" &&
+            !char.isLetter && !char.isNumber {
+          throw ParseError.confusableCharacter(char)
+        }
+        break
       }
+      return .char(char)
     }
     guard let kind = kind else { return nil }
     return AST.Atom(kind.value, kind.location)

Sources/_RegexParser/Regex/Parse/Parse.swift

@@ -222,6 +222,13 @@ extension Parser {
         result.append(.quote(quote))
         continue
       }
+
+      // Interpolation -> `lexInterpolation`
+      if let interpolation = try source.lexInterpolation() {
+        result.append(.interpolation(interpolation))
+        continue
+      }
+
       //     Quantification  -> QuantOperand Quantifier?
       if let operand = try parseQuantifierOperand() {
         if let (amt, kind, trivia) =

Sources/_RegexParser/Regex/Parse/Sema.swift

@@ -395,6 +395,11 @@ extension RegexValidator {
       // These are Oniguruma specific.
       throw error(.unsupported("absent function"), at: a.location)
 
+    case .interpolation(let i):
+      // This is currently rejected in the parser for better diagnostics, but
+      // reject here too until we get runtime support.
+      throw error(.unsupported("interpolation"), at: i.location)
+
     case .quote, .trivia, .empty:
       break
     }

Sources/_RegexParser/Regex/Printing/DumpAST.swift

@@ -101,6 +101,10 @@ extension AST.Trivia {
   }
 }
 
+extension AST.Interpolation {
+  public var _dumpBase: String { "interpolation <\(contents)>" }
+}
+
 extension AST.Empty {
   public var _dumpBase: String { "" }
 }

Sources/_RegexParser/Regex/Printing/PrintAsCanonical.swift

@@ -97,6 +97,9 @@ extension PrettyPrinter {
     case let .trivia(t):
       output(t._canonicalBase)
 
+    case let .interpolation(i):
+      output(i._canonicalBase)
+
     case let .atom(a):
       output(a._canonicalBase)
 
@@ -178,6 +181,12 @@ extension AST.Quote {
   }
 }
 
+extension AST.Interpolation {
+  var _canonicalBase: String {
+    "<{\(contents)}>"
+  }
+}
+
 extension AST.Group.Kind {
   var _canonicalBase: String {
     switch self {

Sources/_StringProcessing/Regex/ASTConversion.swift

@@ -137,6 +137,9 @@ extension AST.Node {
       case let .trivia(v):
         return .trivia(v.contents)
 
+      case .interpolation:
+        throw Unsupported("TODO: interpolation")
+
       case let .atom(v):
         switch v.kind {
         case .scalarSequence(let seq):