Add new SpanScanner.eager().

This is more efficient for uses like the YAML parser, which uses the
current line and especially column information frequently while parsing
a file.

R=rnystrom@google.com

Review URL: https://codereview.chromium.org//1318603008 .

CHANGELOG.md

@@ -1,3 +1,8 @@
+## 0.1.4
+
+* Add `new SpanScanner.eager()` for creating a `SpanScanner` that eagerly
+  computes its current line and column numbers.
+
 ## 0.1.3+2
 
 * Fix `LineScanner`'s handling of carriage returns to match that of

lib/src/eager_span_scanner.dart

@@ -0,0 +1,115 @@
+// 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 string_scanner.eager_span_scanner;
+
+import 'package:charcode/ascii.dart';
+
+import 'line_scanner.dart';
+import 'span_scanner.dart';
+
+// TODO(nweiz): Currently this duplicates code in line_scanner.dart. Once
+// sdk#23770 is fully complete, we should move the shared code into a mixin.
+
+/// A regular expression matching newlines across platforms.
+final _newlineRegExp = new RegExp(r"\r\n?|\n");
+
+/// A [SpanScanner] that tracks the line and column eagerly, like [LineScanner].
+class EagerSpanScanner extends SpanScanner {
+  int get line => _line;
+  int _line = 0;
+
+  int get column => _column;
+  int _column = 0;
+
+  LineScannerState get state =>
+      new _EagerSpanScannerState(this, position, line, column);
+
+  bool get _betweenCRLF => peekChar(-1) == $cr && peekChar() == $lf;
+
+  set state(LineScannerState state) {
+    if (state is! _EagerSpanScannerState ||
+        !identical((state as _EagerSpanScannerState)._scanner, this)) {
+      throw new ArgumentError("The given LineScannerState was not returned by "
+          "this LineScanner.");
+    }
+
+    super.position = state.position;
+    _line = state.line;
+    _column = state.column;
+  }
+
+  set position(int newPosition) {
+    var oldPosition = position;
+    super.position = newPosition;
+
+    if (newPosition > oldPosition) {
+      var newlines = _newlinesIn(string.substring(oldPosition, newPosition));
+      _line += newlines.length;
+      if (newlines.isEmpty) {
+        _column += newPosition - oldPosition;
+      } else {
+        _column = newPosition - newlines.last.end;
+      }
+    } else {
+      var newlines = _newlinesIn(string.substring(newPosition, oldPosition));
+      if (_betweenCRLF) newlines.removeLast();
+
+      _line -= newlines.length;
+      if (newlines.isEmpty) {
+        _column -= oldPosition - newPosition;
+      } else {
+        _column = newPosition -
+            string.lastIndexOf(_newlineRegExp, newPosition) - 1;
+      }
+    }
+  }
+
+  EagerSpanScanner(String string, {sourceUrl, int position})
+      : super(string, sourceUrl: sourceUrl, position: position);
+
+  int readChar() {
+    var char = super.readChar();
+    if (char == $lf || (char == $cr && peekChar() != $lf)) {
+      _line += 1;
+      _column = 0;
+    } else {
+      _column += 1;
+    }
+    return char;
+  }
+
+  bool scan(Pattern pattern) {
+    if (!super.scan(pattern)) return false;
+
+    var newlines = _newlinesIn(lastMatch[0]);
+    _line += newlines.length;
+    if (newlines.isEmpty) {
+      _column += lastMatch[0].length;
+    } else {
+      _column = lastMatch[0].length - newlines.last.end;
+    }
+
+    return true;
+  }
+
+  /// Returns a list of [Match]es describing all the newlines in [text], which
+  /// is assumed to end at [position].
+  List<Match> _newlinesIn(String text) {
+    var newlines = _newlineRegExp.allMatches(text).toList();
+    if (_betweenCRLF) newlines.removeLast();
+    return newlines;
+  }
+}
+
+/// A class representing the state of an [EagerSpanScanner].
+class _EagerSpanScannerState implements LineScannerState {
+  final EagerSpanScanner _scanner;
+  final int position;
+  final int line;
+  final int column;
+
+  _EagerSpanScannerState(this._scanner, this.position, this.line, this.column);
+}
+

lib/src/line_scanner.dart

@@ -8,6 +8,8 @@ import 'package:charcode/ascii.dart';
 
 import 'string_scanner.dart';
 
+// Note that much of this code is duplicated in eager_span_scanner.dart.
+
 /// A regular expression matching newlines across platforms.
 final _newlineRegExp = new RegExp(r"\r\n?|\n");
 

lib/src/span_scanner.dart

@@ -6,6 +6,7 @@ library string_scanner.span_scanner;
 
 import 'package:source_span/source_span.dart';
 
+import 'eager_span_scanner.dart';
 import 'exception.dart';
 import 'line_scanner.dart';
 import 'string_scanner.dart';
@@ -56,6 +57,20 @@ class SpanScanner extends StringScanner implements LineScanner {
       : _sourceFile = new SourceFile(string, url: sourceUrl),
         super(string, sourceUrl: sourceUrl, position: position);
 
+  /// Creates a new [SpanScanner] that eagerly computes line and column numbers.
+  ///
+  /// In general [new SpanScanner] will be more efficient, since it avoids extra
+  /// computation on every scan. However, eager scanning can be useful for
+  /// situations where the normal course of parsing frequently involves
+  /// accessing the current line and column numbers.
+  ///
+  /// Note that *only* the `line` and `column` fields on the `SpanScanner`
+  /// itself and its `LineScannerState` are eagerly computed. To limit their
+  /// memory footprint, returned spans and locations will still lazily compute
+  /// their line and column numbers.
+  factory SpanScanner.eager(String string, {sourceUrl, int position}) =
+      EagerSpanScanner;
+
   /// Creates a [FileSpan] representing the source range between [startState]
   /// and the current position.
   FileSpan spanFrom(LineScannerState startState, [LineScannerState endState]) {

pubspec.yaml

@@ -1,5 +1,5 @@
 name: string_scanner
-version: 0.1.3+2
+version: 0.1.4
 author: "Dart Team <misc@dartlang.org>"
 homepage: https://github.com/dart-lang/string_scanner
 description: >

test/span_scanner_test.dart

@@ -8,53 +8,63 @@ import 'package:string_scanner/string_scanner.dart';
 import 'package:test/test.dart';
 
 void main() {
-  var scanner;
-  setUp(() {
-    scanner = new SpanScanner('foo\nbar\nbaz', sourceUrl: 'source');
+  testForImplementation("lazy", () {
+    return new SpanScanner('foo\nbar\nbaz', sourceUrl: 'source');
   });
 
-  test("tracks the span for the last match", () {
-    scanner.scan('fo');
-    scanner.scan('o\nba');
+  testForImplementation("eager", () {
+    return new SpanScanner.eager('foo\nbar\nbaz', sourceUrl: 'source');
+  });
+}
 
-    var span = scanner.lastSpan;
-    expect(span.start.offset, equals(2));
-    expect(span.start.line, equals(0));
-    expect(span.start.column, equals(2));
-    expect(span.start.sourceUrl, equals(Uri.parse('source')));
+void testForImplementation(String name, SpanScanner create()) {
+  group("for a $name scanner", () {
+    var scanner;
+    setUp(() => scanner = create());
 
-    expect(span.end.offset, equals(6));
-    expect(span.end.line, equals(1));
-    expect(span.end.column, equals(2));
-    expect(span.start.sourceUrl, equals(Uri.parse('source')));
+    test("tracks the span for the last match", () {
+      scanner.scan('fo');
+      scanner.scan('o\nba');
 
-    expect(span.text, equals('o\nba'));
-  });
+      var span = scanner.lastSpan;
+      expect(span.start.offset, equals(2));
+      expect(span.start.line, equals(0));
+      expect(span.start.column, equals(2));
+      expect(span.start.sourceUrl, equals(Uri.parse('source')));
 
-  test(".spanFrom() returns a span from a previous state", () {
-    scanner.scan('fo');
-    var state = scanner.state;
-    scanner.scan('o\nba');
-    scanner.scan('r\nba');
+      expect(span.end.offset, equals(6));
+      expect(span.end.line, equals(1));
+      expect(span.end.column, equals(2));
+      expect(span.start.sourceUrl, equals(Uri.parse('source')));
 
-    var span = scanner.spanFrom(state);
-    expect(span.text, equals('o\nbar\nba'));
-  });
+      expect(span.text, equals('o\nba'));
+    });
+
+    test(".spanFrom() returns a span from a previous state", () {
+      scanner.scan('fo');
+      var state = scanner.state;
+      scanner.scan('o\nba');
+      scanner.scan('r\nba');
+
+      var span = scanner.spanFrom(state);
+      expect(span.text, equals('o\nbar\nba'));
+    });
 
-  test(".emptySpan returns an empty span at the current location", () {
-    scanner.scan('foo\nba');
+    test(".emptySpan returns an empty span at the current location", () {
+      scanner.scan('foo\nba');
 
-    var span = scanner.emptySpan;
-    expect(span.start.offset, equals(6));
-    expect(span.start.line, equals(1));
-    expect(span.start.column, equals(2));
-    expect(span.start.sourceUrl, equals(Uri.parse('source')));
+      var span = scanner.emptySpan;
+      expect(span.start.offset, equals(6));
+      expect(span.start.line, equals(1));
+      expect(span.start.column, equals(2));
+      expect(span.start.sourceUrl, equals(Uri.parse('source')));
 
-    expect(span.end.offset, equals(6));
-    expect(span.end.line, equals(1));
-    expect(span.end.column, equals(2));
-    expect(span.start.sourceUrl, equals(Uri.parse('source')));
+      expect(span.end.offset, equals(6));
+      expect(span.end.line, equals(1));
+      expect(span.end.column, equals(2));
+      expect(span.start.sourceUrl, equals(Uri.parse('source')));
 
-    expect(span.text, equals(''));
+      expect(span.text, equals(''));
+    });
   });
 }