Parsing Dates, Precisely or Fuzzily






Parsing Dates, Precisely or Fuzzily

Problem

You want to transform a string describing a date or date/time into a Date object. You might not know the format of the string ahead of time.

Solution

The best solution is to pass the date string into Date.parse or DateTime.parse. These methods use heuristics to guess at the format of the string, and they do a pretty good job:

	require 'date'

	Date. 
parse('2/9/2007').to_s
	# => "2007-02-09"

	 
DateTime.parse('02-09-2007 12:30:44 AM').to_s
	# => "2007-09-02T00:30:44Z"

	DateTime. 
parse('02-09-2007 12:30:44 PM EST').to_s
	# => "2007-09-02T12:30:44-0500"

	 
Date.parse('Wednesday, January 10, 2001').to_s
	# => "2001-01-10"

Discussion

The parse methods can save you a lot of the drudgework associated with parsing times in other programming languages, but they don't always give you the results you want. Notice in the first example how Date.parse assumed that 2/9/2007 was an American (month first) date instead of a European (day first) date. parse also tends to misinterpret two-digit years:

	Date.parse('2/9/07').to_s                   # => "0007-02-09"

Let's say that Date.parse doesn't work for you, but you know that all the dates you're processing will be formatted a certain way. You can create a format string using the standard strftime directives, and pass it along with a date string into DateTime.strptime or Date.strptime. If the date string matches up with the format string, you'll get a Date or DateTime object back. You may already be familiar with this technique, since this many languages, as well as the Unix date command, do date formatting this way.

Some common date and time formats include:

	american_date = '%m/%d/%y'
	Date.strptime('2/9/07', american_date).to_s            # => "2007-02-09"
	DateTime.strptime('2/9/05', american_date).to_s        # => "2005-02-09T00:00:00Z"
	Date.strptime('2/9/68', american_date).to_s            # => "2068-02-09"
	Date.strptime('2/9/69', american_date).to_s            # => "1969-02-09"

	european_date = '%d/%m/%y'
	Date.strptime('2/9/07', european_date).to_s            # => "2007-09-02"
	Date.strptime('02/09/68', european_date).to_s          # => "2068-09-02"
	Date.strptime('2/9/69', european_date).to_s            # => "1969-09-02"

	four_digit_year_date = '%m/%d/%Y'
	Date.strptime('2/9/2007', four_digit_year_date).to_s   # => "2007-02-09"
	Date.strptime('02/09/1968', four_digit_year_date).to_s # => "1968-02-09"
	Date.strptime('2/9/69', four_digit_year_date).to_s     # => "0069-02-09"

	date_and_time = '%m-%d-%Y %H:%M:%S %Z'
	DateTime.strptime('02-09-2007 12:30:44 EST', date_and_time).to_s
	# => "2007-02-09T12:30:44-0500"
	DateTime.strptime('02-09-2007 12:30:44 PST', date_and_time).to_s
	# => "2007-02-09T12:30:44-0800"
	DateTime.strptime('02-09-2007 12:30:44 GMT', date_and_time).to_s
	# => "2007-02-09T12:30:44Z"

	twelve_hour_clock_time = '%m-%d-%Y %I:%M:%S %p'
	DateTime.strptime('02-09-2007 12:30:44 AM', twelve_hour_clock_time).to_s
	# => "2007-02-09T00:30:44Z"
	DateTime.strptime('02-09-2007 12:30:44 PM', twelve_hour_clock_time).to_s
	# => "2007-02-09T12:30:44Z"

	word_date = '%A, %B %d, %Y'
	Date.strptime('Wednesday, January 10, 2001', word_date).to_s
	# => "2001-01-10"

If your date strings might be in one of a limited number of formats, try iterating over a list of format strings and attempting to parse the date string with each one in turn. This gives you some of the flexibility of Date.parse while letting you override the assumptions it makes. Date.parse is still faster, so if it'll work, use that.

	Date.parse('1/10/07').to_s                             # => "0007-01-10"
	Date.parse('2007 1 10').to_s
	# ArgumentError: 3 elements of civil date are necessary

	TRY_FORMATS = ['%d/%m/%y', '%Y %m %d']
	def try_to_parse(s)
	  parsed = nil
	  TRY_FORMATS.each do |format|
	    begin
	      parsed = Date.strptime(s, format)
	      break
	    rescue ArgumentError
	    end
	  end
	  return parsed
	end

	try_to_parse('1/10/07').to_s                          # => "2007-10-01"
	try_to_parse('2007 1 10').to_s                        # => "2007-01-10"

Several common date formats cannot be reliably represented by strptime format strings. Ruby defines class methods of Time for parsing these date strings, so you don't have to write the code yourself. Each of the following methods returns a Time object.

Time.rfc822 parses a date string in the format of RFC822/RFC2822, the Internet email standard. In an RFC2822 date, the month and the day of the week are always in English (for instance, "Tue" and "Jul"), even if the locale is some other language.

	require 'time'
	mail_received = 'Tue, 1 Jul 2003 10:52:37 +0200'
	Time.rfc822(mail_received)
	# => Tue Jul 01 04:52:37 EDT 2003

To parse a date in the format of RFC2616, the HTTP standard, use Time.httpdate. An RFC2616 date is the kind of date you see in HTTP headers like Last-Modified. As with RFC2822, the month and day abbreviations are always in English:

	last_modified = 'Tue, 05 Sep 2006 16:05:51 GMT'
	Time.httpdate(last_modified)
	# => Tue Sep 05 12:05:51 EDT 2006

To parse a date in the format of ISO 8601 or XML Schema, use Time.iso8601 or Time.xmlschema:

	timestamp = '2001-04-17T19:23:17.201Z'
	t = Time.iso8601(timestamp)    # => Tue Apr 17 19:23:17 UTC 2001
	t.sec                          # => 17
	t.tv_usec                      # => 201000

Don't confuse these class methods of Time with the instance methods of the same names. The class methods create Time objects from strings. The instance methods go the other way, formatting an existing Time object as a string:

	t = Time.at(1000000000)        # => Sat Sep 08 21:46:40 EDT 2001
	t.rfc822                       # => "Sat, 08 Sep 2001 21:46:40 -0400"
	t.httpdate                     # => "Sun, 09 Sep 2001 01:46:40 GMT"
	t.iso8601                      # => "2001-09-08T21:46:40-04:00"

See Also

  • The RDoc for the Time#strftime method lists most of the supported strftime directives (ri Time#strftime); for a more detailed and complete list, see the table in Recipe 3.3, " Printing a Date"



 Python   SQL   Java   php   Perl 
 game development   web development   internet   *nix   graphics   hardware 
 telecommunications   C++ 
 Flash   Active Directory   Windows