joda - parsing time zone of pubDate in RSS item

Question

I'm parsing pubDate in RSS item using Joda. The date have to be in RFC-822 format: http://feed2.w3.org/docs/error/InvalidRFC2822Date.html

The problem is that when there is a date like: Wed, 02 Oct 2002 13:00:00 GMT I have to use pattern:

DateTimeFormat.forPattern("EEE, dd MMM yyyy HH:mm:ss ZZZ").withLocale(Locale.ENGLISH).withOffsetParsed();

But it can be also date like: Wed, 02 Oct 2002 15:00:00 +0200. In this case ZZZ dosen't work, I have to use one Z:

DateTimeFormat.forPattern("EEE, dd MMM yyyy HH:mm:ss Z").withLocale(Locale.ENGLISH).withOffsetParsed();

How to create universal solution?

Answer 1

I've made tests with JodaTime 2.7 and found 2 ways to do it:

1. Use DateTimeFormatterBuilder's optional parsers:

   // create parser for "GMT"
   DateTimeParser gmtParser = DateTimeFormat.forPattern("ZZZ").getParser();
   
   // create parser for "+0200"
   DateTimeParser offsetParser = DateTimeFormat.forPattern("Z").getParser();
   
   DateTimeFormatter formatter = new DateTimeFormatterBuilder()
       .appendPattern("EEE, dd MMM yyyy HH:mm:ss ") // common pattern
       .appendOptional(gmtParser)    // optional parser for GMT
       .appendOptional(offsetParser) // optional parser for +0200
       .toFormatter().withLocale(Locale.ENGLISH).withOffsetParsed();
   

2. DateTimeFormatterBuilder can receive an array of parsers that can be used to parse different inputs:

   // create array with all possible patterns
   DateTimeParser[] parsers = {
       DateTimeFormat.forPattern("EEE, dd MMM yyyy HH:mm:ss Z").getParser(),
       DateTimeFormat.forPattern("EEE, dd MMM yyyy HH:mm:ss ZZZ").getParser()
   };
   
   // create a formatter using the parsers array
   DateTimeFormatter formatter = new DateTimeFormatterBuilder()
       .append(null, parsers) // use parsers array
       .toFormatter().withLocale(Locale.ENGLISH).withOffsetParsed();
   

Using any of the solutions above, the formatter will work with both inputs:

System.out.println(formatter.parseDateTime("Wed, 02 Oct 2002 13:00:00 GMT"));
System.out.println(formatter.parseDateTime("Wed, 02 Oct 2002 15:00:00 +0200"));

The output will be:

2002-10-02T13:00:00.000Z
2002-10-02T15:00:00.000+02:00

---

Note: I believe the first solution is better if you have a common part among all patterns and little variation between them. The second solution is better if the patterns are very different from each other. But I also believe it's a matter of opinion and it's up to you to choose.

Answer 2

Workaround solution:

    dateTimeFormat = DateTimeFormat.forPattern("EEE, dd MMM yyyy HH:mm:ss ZZZ").withLocale(Locale.ENGLISH).withOffsetParsed();
    dateTimeFormatOneZ = DateTimeFormat.forPattern("EEE, dd MMM yyyy HH:mm:ss Z").withLocale(Locale.ENGLISH).withOffsetParsed();

private DateTime convertToDate(String pubDate) {

    try {
        return dateTimeFormat.parseDateTime(pubDate);
    } catch (Exception e) {
        try {
            return dateTimeFormatOneZ.parseDateTime(pubDate);
        } catch (Exception e1) {
            return null;
        }
    }
}