Add a scanner and parser for platform selectors.

lib/src/backend/platform_selector/scanner.dart

Index: lib/src/backend/platform_selector/scanner.dart
diff --git a/lib/src/backend/platform_selector/scanner.dart b/lib/src/backend/platform_selector/scanner.dart
new file mode 100644
index 0000000000000000000000000000000000000000..fac4925b920f422e494f12e4d801e25426da90b3
--- /dev/null
+++ b/lib/src/backend/platform_selector/scanner.dart
@@ -0,0 +1,144 @@
+// Copyright (c) 2015, the Dart project authors.  Please see the AUTHORS file
+// for details. All rights reserved. Use of this source code is governed by a
+// BSD-style license that can be found in the LICENSE file.
+
+library unittest.backend.platform_selector.scanner;
+
+import 'package:string_scanner/string_scanner.dart';
+
+import 'token.dart';
+
+/// A regular expression matching both whitespace and single-line comments.
+///
+/// This will only match if consumes at least one character.
+final _whitespaceAndSingleLineComments =
+    new RegExp(r"([ \t\n]+|//[^\n]*(\n|$))+");
+
+/// A regular expression matching the body of a multi-line comment, after `/*`
+/// but before `*/` or a nested `/*`.
+///
+/// This will only match if it consumes at least one character.
+final _multiLineCommentBody = new RegExp(r"([^/*]|/[^*]|\*[^/])+");
+
+/// A regular expression matching an identifier.
+///
+/// Unlike standard Dart identifiers, platform selector identifiers may
+/// contain dashes for consistency with command-line arguments.
+final _identifier = new RegExp(r"[a-zA-Z_-][a-zA-Z0-9_-]*");
+
+/// A scanner that converts a platform selector string into a stream of
+/// tokens.
+class Scanner {
+  /// The underlying string scanner.
+  final SpanScanner _scanner;
+
+  /// The next token to emit.
+  Token _next;
+
+  /// Whether the scanner has emitted a [TokenType.endOfFile] token.
+  bool _endOfFileEmitted = false;
+
+  Scanner(String selector)
+      : _scanner = new SpanScanner(selector);
+
+  /// Returns the next token that will be returned by [next].
+  ///
+  /// Throws a [StateError] if [next] has already returned a
+  /// [TokenType.endOfFile] token.
+  Token peek() {
+    if (_next == null) _next = _getNext();
+    return _next;
+  }
+
+  /// Consumes and returns the next token in the stream.
+  ///
+  /// Throws a [StateError] if this has already returned a
+  /// [TokenType.endOfFile] token.
+  Token next() {
+    var token = _next == null ? _getNext() : _next;
+    _endOfFileEmitted = token.type == TokenType.endOfFile;
+    _next = null;
+    return token;
+  }
+
+  /// Scan and return the next token in the stream.
+  Token _getNext() {
+    if (_endOfFileEmitted) {
+      throw new StateError("No more tokens.");

[Bob Nystrom, 2015/03/11 20:05:52]

Nit: you could make this a one-line if if you want.

[nweiz, 2015/03/12 19:48:57]

Done.

+    }
+
+    _consumeWhitespace();
+    if (_scanner.isDone) {
+      return new Token(
+          TokenType.endOfFile, _scanner.spanFrom(_scanner.state));
+    }
+
+    switch (_scanner.peekChar()) {
+      case 0x28 /* ( */: return _scanOperator(TokenType.leftParen);
+      case 0x29 /* ) */: return _scanOperator(TokenType.rightParen);
+      case 0x3F /* ? */: return _scanOperator(TokenType.questionMark);
+      case 0x3A /* : */: return _scanOperator(TokenType.colon);
+      case 0x21 /* ! */: return _scanOperator(TokenType.not);
+      case 0x7C /* | */: return _scanOr();
+      case 0x26 /* & */: return _scanAnd();
+      default: return _scanIdentifier();
+    }
+  }
+
+  /// Scans a single-character operator and returns a token of type [type].
+  ///
+  /// This assumes that the caller has already verified that the next character
+  /// is correct for the given operator.
+  Token _scanOperator(TokenType type) {
+    var start = _scanner.state;
+    _scanner.readChar();
+    return new Token(type, _scanner.spanFrom(start));
+  }
+
+  /// Scans a `||` operator and returns the appropriate token.
+  ///
+  /// This validates that the next two characters are `||`.
+  Token _scanOr() {
+    var start = _scanner.state;
+    _scanner.expect("||");
+    return new Token(TokenType.or, _scanner.spanFrom(start));
+  }
+
+  /// Scans a `&&` operator and returns the appropriate token.
+  ///
+  /// This validates that the next two characters are `&&`.
+  Token _scanAnd() {
+    var start = _scanner.state;
+    _scanner.expect("&&");
+    return new Token(TokenType.and, _scanner.spanFrom(start));
+  }
+
+  /// Scans and returns an identifier token.
+  Token _scanIdentifier() {
+    _scanner.expect(_identifier, name: "expression");
+    return new IdentifierToken(_scanner.lastMatch[0], _scanner.lastSpan);
+  }
+
+  /// Consumes all whitespace and comments immediately following the cursor's
+  /// current position.
+  void _consumeWhitespace() {
+    while (_scanner.scan(_whitespaceAndSingleLineComments) ||
+        _multiLineComment()) {
+      // Do nothing.
+    }
+  }
+
+  /// Consumes a single multi-line comment.
+  ///
+  /// Returns whether or not a comment was consumed.
+  bool _multiLineComment() {
+    if (!_scanner.scan("/*")) return false;
+
+    while (_scanner.scan(_multiLineCommentBody) || _multiLineComment()) {
+      // Do nothing.
+    }
+    _scanner.expect("*/");
+
+    return true;
+  }
+}