Introduce Diagnostics

Author: hamishknight

Sources/_RegexParser/Regex/AST/AST.swift

@@ -15,16 +15,31 @@
 public struct AST: Hashable {
   public var root: AST.Node
   public var globalOptions: GlobalMatchingOptionSequence?
+  public var diags: Diagnostics
 
-  public init(_ root: AST.Node, globalOptions: GlobalMatchingOptionSequence?) {
+  public init(
+    _ root: AST.Node, globalOptions: GlobalMatchingOptionSequence?,
+    diags: Diagnostics
+  ) {
     self.root = root
     self.globalOptions = globalOptions
+    self.diags = diags
   }
 }
 
 extension AST {
   /// Whether this AST tree contains at least one capture nested inside of it.
   public var hasCapture: Bool { root.hasCapture }
+
+  /// Whether this AST tree is either syntactically or semantically invalid.
+  public var isInvalid: Bool { diags.hasAnyError }
+
+  /// If the AST is invalid, throws an error. Otherwise, returns self.
+  @discardableResult
+  public func ensureValid() throws -> AST {
+    try diags.throwAnyError()
+    return self
+  }
 }
 
 extension AST {

Sources/_RegexParser/Regex/Parse/Diagnostics.swift

@@ -242,9 +242,128 @@ extension ParseError: CustomStringConvertible {
   }
 }
 
-// TODO: Fixits, notes, etc.
+/// A fatal error that indicates broken logic in the parser.
+enum FatalParseError: Hashable, Error {
+  case unreachable(String)
+}
+
+extension FatalParseError: CustomStringConvertible {
+  var description: String {
+    switch self {
+    case .unreachable(let str):
+      return "UNREACHABLE: \(str)"
+    }
+  }
+}
+
+// MARK: Diagnostic handling
+
+/// A diagnostic to emit.
+public struct Diagnostic: Hashable {
+  public let behavior: Behavior
+  public let message: String
+  public let location: SourceLocation
+
+  // TODO: Fixits, notes, etc.
+
+  // The underlying ParseError if applicable. This is used for testing.
+  internal let underlyingParseError: ParseError?
+
+  init(_ behavior: Behavior, _ message: String, at loc: SourceLocation,
+       underlyingParseError: ParseError? = nil) {
+    self.behavior = behavior
+    self.message = message
+    self.location = loc
+    self.underlyingParseError = underlyingParseError
+  }
+
+  public var isAnyError: Bool { behavior.isAnyError }
+}
+
+extension Diagnostic {
+  public enum Behavior: Hashable {
+    case fatalError, error, warning
+
+    public var isAnyError: Bool {
+      switch self {
+      case .fatalError, .error:
+        return true
+      case .warning:
+        return false
+      }
+    }
+  }
+}
 
-// TODO: Diagnostics engine, recorder, logger, or similar.
+/// A collection of diagnostics to emit.
+public struct Diagnostics: Hashable {
+  public private(set) var diags = [Diagnostic]()
 
+  public init() {}
+  public init(_ diags: [Diagnostic]) {
+    self.diags = diags
+  }
 
+  /// Add a new diagnostic to emit.
+  public mutating func append(_ diag: Diagnostic) {
+    diags.append(diag)
+  }
 
+  /// Add all the diagnostics of another diagnostic collection.
+  public mutating func append(contentsOf other: Diagnostics) {
+    diags.append(contentsOf: other.diags)
+  }
+
+  /// Add all the new fatal error diagnostics of another diagnostic collection.
+  /// This assumes that `other` was the same as `self`, but may have additional
+  /// diagnostics added to it.
+  public mutating func appendNewFatalErrors(from other: Diagnostics) {
+    let newDiags = other.diags.dropFirst(diags.count)
+    for diag in newDiags where diag.behavior == .fatalError {
+      append(diag)
+    }
+  }
+
+  /// Whether any error is present. This includes fatal errors.
+  public var hasAnyError: Bool {
+    diags.contains(where: { $0.isAnyError })
+  }
+
+  /// Whether any fatal error is present.
+  public var hasFatalError: Bool {
+    diags.contains(where: { $0.behavior == .fatalError })
+  }
+
+  /// If any error diagnostic has been added, throw it as an Error.
+  func throwAnyError() throws {
+    for diag in diags where diag.isAnyError {
+      struct ErrorDiagnostic: Error, CustomStringConvertible {
+        var diag: Diagnostic
+        var description: String { diag.message }
+      }
+      throw ErrorDiagnostic(diag: diag)
+    }
+  }
+}
+
+// MARK: Diagnostic construction
+
+extension Diagnostic {
+  init(_ err: ParseError, at loc: SourceLocation) {
+    self.init(.error, "\(err)", at: loc, underlyingParseError: err)
+  }
+
+  init(_ err: FatalParseError, at loc: SourceLocation) {
+    self.init(.fatalError, "\(err)", at: loc)
+  }
+}
+
+extension Diagnostics {
+  mutating func error(_ err: ParseError, at loc: SourceLocation) {
+    append(Diagnostic(err, at: loc))
+  }
+
+  mutating func fatal(_ err: FatalParseError, at loc: SourceLocation) {
+    append(Diagnostic(err, at: loc))
+  }
+}

Sources/_RegexParser/Regex/Parse/Parse.swift

@@ -165,7 +165,8 @@ extension Parser {
       }
       fatalError("Unhandled termination condition")
     }
-    return .init(ast, globalOptions: opts)
+    // TODO: Record and store diagnostics on the AST.
+    return .init(ast, globalOptions: opts, diags: Diagnostics())
   }
 
   /// Parse a regular expression node. This should be used instead of `parse()`

Sources/_RegexParser/Regex/Printing/PrintAsCanonical.swift

@@ -32,7 +32,7 @@ extension AST.Node {
     showDelimiters delimiters: Bool = false,
     terminateLine: Bool = false
   ) -> String {
-    AST(self, globalOptions: nil).renderAsCanonical(
+    AST(self, globalOptions: nil, diags: Diagnostics()).renderAsCanonical(
       showDelimiters: delimiters, terminateLine: terminateLine)
   }
 }

Sources/_StringProcessing/PrintAsPattern.swift

@@ -56,7 +56,7 @@ extension PrettyPrinter {
   mutating func printBackoff(_ node: DSLTree.Node) {
     precondition(node.astNode != nil, "unconverted node")
     printAsCanonical(
-      .init(node.astNode!, globalOptions: nil),
+      .init(node.astNode!, globalOptions: nil, diags: Diagnostics()),
       delimiters: true)
   }
 

Sources/_StringProcessing/Regex/Core.swift

@@ -37,7 +37,8 @@ public struct Regex<Output>: RegexComponent {
     self.program = Program(ast: ast)
   }
   init(ast: AST.Node) {
-    self.program = Program(ast: .init(ast, globalOptions: nil))
+    self.program = Program(ast:
+        .init(ast, globalOptions: nil, diags: Diagnostics()))
   }
 
   // Compiler interface. Do not change independently.

Sources/_StringProcessing/Regex/DSLTree.swift

@@ -381,16 +381,6 @@ extension DSLTree {
   }
 }
 
-extension DSLTree {
-  var ast: AST? {
-    guard let root = root.astNode else {
-      return nil
-    }
-    // TODO: Options mapping
-    return AST(root, globalOptions: nil)
-  }
-}
-
 extension DSLTree {
   var hasCapture: Bool {
     root.hasCapture

Sources/_StringProcessing/Utility/ASTBuilder.swift

@@ -48,7 +48,7 @@ func empty() -> AST.Node {
 }
 
 func ast(_ root: AST.Node, opts: [AST.GlobalMatchingOption.Kind]) -> AST {
-  .init(root, globalOptions: .init(opts.map { .init($0, .fake) }))
+  .init(root, globalOptions: .init(opts.map { .init($0, .fake) }), diags: Diagnostics())
 }
 
 func ast(_ root: AST.Node, opts: AST.GlobalMatchingOption.Kind...) -> AST {

Tests/RegexTests/ParseTests.swift

@@ -48,8 +48,9 @@ func parseTest(
   line: UInt = #line
 ) {
   parseTest(
-    input, .init(expectedAST, globalOptions: nil), throwsError: errorKind,
-    syntax: syntax, captures: expectedCaptures, file: file, line: line
+    input, .init(expectedAST, globalOptions: nil, diags: Diagnostics()),
+    throwsError: errorKind, syntax: syntax, captures: expectedCaptures,
+    file: file, line: line
   )
 }