improve DateTime parsing a UTC timezone offset

sdk/lib/core/date_time.dart

 // Copyright (c) 2011, 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.
 
 part of dart.core;
 
 /**
  * An instant in time, such as July 20, 1969, 8:18pm PST.
  *
  * Create a DateTime object by using one of the constructors
@@ skipping 30 matching lines @@
  *
  * Day and month values begin at 1, and the week starts on Monday.
  * That is, the constants [JANUARY] and [MONDAY] are both 1.
  *
  * ## Working with UTC and local time
  *
  * A DateTime object is in the local time zone
  * unless explicitly created in the UTC time zone.
  *
  *     DateTime dDay = new DateTime.utc(1944, 6, 6);
- *    
+ *
  * Use [isUtc] to determine whether a DateTime object is based in UTC.
  * Use the methods [toLocal] and [toUtc]
  * to get the equivalent date/time value specified in the other time zone.
  * Use [timeZoneName] to get an abbreviated name of the time zone
  * for the DateTime object.
  * To find the difference
  * between UTC and the time zone of a DateTime object
  * call [timeZoneOffset].
  *
  * ## Comparing DateTime objects
@@ skipping 130 matching lines @@
    *
    * The function parses a subset of ISO 8601. Examples of accepted strings:
    *
    * * `"2012-02-27 13:27:00"`
    * * `"2012-02-27 13:27:00.123456z"`
    * * `"20120227 13:27:00"`
    * * `"20120227T132700"`
    * * `"20120227"`
    * * `"+20120227"`
    * * `"2012-02-27T14Z"`
+   * * `"2012-02-27T14+00:00"`
    * * `"-123450101 00:00:00 Z"`: in the year -12345.
    */
   // TODO(floitsch): specify grammar.
   static DateTime parse(String formattedString) {
     final RegExp re = new RegExp(
         r'^([+-]?\d?\d\d\d\d)-?(\d\d)-?(\d\d)'  // The day part.
-        r'(?:[ T](\d\d)(?::?(\d\d)(?::?(\d\d)(.\d{1,6})?)?)? ?([zZ])?)?$');
+        r'(?:[ T](\d\d)(?::?(\d\d)(?::?(\d\d)(.\d{1,6})?)?)?' // The time part
+        r' ?([zZ]|(?:\+00(?::?00)?)?))?$'); // The timezone part

[Lasse Reichstein Nielsen, 2013/10/29 08:48:45]

This seemns to accept a single terminating space (from the beginning of the
timezone part, even if the timezone part is omitted).
Maybe wrap that in (?:...) too, so the '?' for the timezone part includes the
space.

That is, a grammar something like:

  date ::= yeardate time_opt timezone_opt
  yeardate ::= year colon_opt month colon_opt day
  year ::= sign_opt digit{4,5}
  colon_opt :: <empty> | ':'
  sign_opt ::=  <empty> | '+' | '-'
  month ::= digit{2}
  day ::= digit{2}
  time_opt ::= <empty> | (' ' | 'T') hour minutes_opt
  minutes_opt ::= <empty> | ':' digit{2} seconds_opt
  seconds_opt ::= <empty> | ':' digit{2} millis_opt
  millis_opt ::= <empty> | '.' digit{1,6}
  timezone_opt ::= <empty> | space_opt timezone
  space_opt :: ' ' | <empty>
  timezone ::= 'z' | 'Z' | '+' '0' '0' timezonemins_opt
  timezonemins_opt ::= <empty> | colon_opt '0' '0'
  
(Does this look right?)

It's a pet peeve of mine to not create complex regexps without a specification
of what it should do. It's hard enough to read a regexp to see whether it does
the correct thing, it's pretty much impossible if you don't even know what
correct is :).

[floitsch, 2013/10/30 23:17:34]

Done.

+
     Match match = re.firstMatch(formattedString);
     if (match != null) {
       int parseIntOrZero(String matched) {
         if (matched == null) return 0;
         return int.parse(matched);
       }
 
       double parseDoubleOrZero(String matched) {
         if (matched == null) return 0.0;
         return double.parse(matched);
@@ skipping 337 matching lines @@
    * In accordance with ISO 8601
    * a week starts with Monday, which has the value 1.
    *
    *     DateTime moonLanding = DateTime.parse("1969-07-20 20:18:00");
    *     assert(moonLanding.weekday == 7);
    *     assert(moonLanding.weekday == DateTime.SUNDAY);
    *
    */
   external int get weekday;
 }