Date/time parser for Ruby with the following features:
-
Extensible with custom formats and tokens.
-
It’s pretty fast. Up to 60% faster than Time/Date parse method.
-
Control the parser strictness.
-
Control behaviour of ambiguous date formats (US vs European e.g. mm/dd/yy, dd/mm/yy).
-
I18n support (for months), if I18n gem loaded.
-
Fewer WTFs than Time/Date parse method.
-
Has no dependencies.
-
Works with Ruby MRI 1.8.*, 1.9.2, Rubinius and JRuby.
Extracted from the validates_timeliness gem, it has been rewritten cleaner and much faster. It’s most suitable for when you need to control the parsing behaviour. It’s faster than the Time/Date class parse methods, so it has general appeal.
The simplest example is just a straight forward string parse:
Timeliness.parse('2010-09-08 12:13:14') #=> Wed Sep 08 12:13:14 1000 2010 Timeliness.parse('2010-09-08') #=> Wed Sep 08 00:00:00 1000 2010 Timeliness.parse('12:13:14') #=> Sat Jan 01 12:13:14 1100 2000
You can provide a type which will tell the parser that you are only interested in the part of the value for that type.
Timeliness.parse('2010-09-08 12:13:14', :date) #=> Wed Sep 08 00:00:00 1000 2010 Timeliness.parse('2010-09-08 12:13:14', :time) #=> Sat Jan 01 12:13:14 1100 2000 Timeliness.parse('2010-09-08 12:13:14', :datetime) #=> Wed Sep 08 12:13:14 1000 2010 i.e. the whole string is used
Now let’s get strict. Pass the :strict option with true and things get finicky
Timeliness.parse('2010-09-08 12:13:14', :date, :strict => true) #=> nil Timeliness.parse('2010-09-08 12:13:14', :time, :strict => true) #=> nil Timeliness.parse('2010-09-08 12:13:14', :datetime, :strict => true) #=> Wed Sep 08 12:13:14 1000 2010 i.e. the whole string is used
The date and time strings are not accepted for a datetime type. The strict option without a type is ignored.
Notice a time only string will return with a date value. The date value can be configured globally with this setting:
Timeliness.date_for_time_type = [2010, 1, 1]
or using a lambda thats evaluated when parsed
Timeliness.date_for_time_type = lambda { Time.now }
It can also be specified with :now option:
Timeliness.parse('12:13:14', :now => Time.mktime(2010,9,8)) #=> Wed Sep 08 12:13:14 1000 2010
As well conforming to the Ruby Time class style.
Timeliness.parse('12:13:14', Time.mktime(2010,9,8)) #=> Wed Sep 08 12:13:14 1000 2010
To control what zone the time object is returned in, you have two options. Firstly you can set the default zone. Below is the list of options with their effective time creation method call
Timeliness.default_timezone = :local # Time.local(...) Timeliness.default_timezone = :utc # Time.utc(...) Timeliness.default_timezone = :current # Time.zone.local(...). Use current zone. Timeliness.default_timezone = 'Melbourne' # Time.use_zone('Melbourne') { Time.zone.local(...) }. Doesn't change Time.zone.
The last two options require that you have ActiveSupport timezone extension loaded.
You can also use the :zone option to control it for a single parse call:
Timeliness.parse('2010-09-08 12:13:14', :zone => :utc) #=> Wed Sep 08 12:13:14 UTC 2010 Timeliness.parse('2010-09-08 12:13:14', :zone => :local) #=> Wed Sep 08 12:13:14 1000 2010 Timeliness.parse('2010-09-08 12:13:14', :zone => :current) #=> Wed Sep 08 12:13:14 1000 2010, with Time.zone = 'Melbourne' Timeliness.parse('2010-09-08 12:13:14', :zone => 'Melbourne') #=> Wed Sep 08 12:13:14 1000 2010
Remember, you must have ActiveSupport timezone extension loaded to use the last two examples.
To get super finicky, you can restrict the parsing to a single format with the :format option
Timeliness.parse('2010-09-08 12:13:14', :format => 'yyyy-mm-dd hh:nn:ss') #=> Wed Sep 08 12:13:14 UTC 2010 Timeliness.parse('08/09/2010 12:13:14', :format => 'yyyy-mm-dd hh:nn:ss') #=> nil
Sometimes you may want to parse a string with a zone abbreviation (e.g. MST) or the zone offset (e.g. +1000). These values are supported by the parser and will be used when creating the time object. The return value will be in the default timezone or the zone specified with the :zone option.
Timeliness.parse('Wed, 08 Sep 2010 12:13:14 MST') => Thu, 09 Sep 2010 05:13:14 EST 10:00
Timeliness.parse('2010-09-08T12:13:14-06:00') => Thu, 09 Sep 2010 05:13:14 EST 10:00
To enable zone abbreviations to work you must have loaded ActiveSupport.
The zone abbreviations supported are those defined in the TzInfo gem, used by ActiveSupport. If you find some that are missing you can add more:
Timeliness.timezone_mapping.update( 'ZZZ' => 'Sleepy Town' )
Where ‘Sleepy Town’ is a valid zone name supported by ActiveSupport/TzInfo.
If you would like to get the raw array of values before the time object is created, you can with
Timeliness._parse('2010-09-08 12:13:14.123456 MST') # => [2010, 9, 8, 12, 13, 14, 123456, 'MST']
The last two value are the microseconds, and zone abbreviation or offset. Note: The format for this value is not defined. You can add it yourself, easily.
To make it easier to use the parser in Rails or an app using ActiveSupport, you can add/override the methods for to_time, to_date and to_datetime on a string value. These methods will then use the Timeliness parser for converting a string, instead of the default.
You just need to add this line to an initializer or other application file:
require 'timeliness/core_ext'
The gem has default formats included which can be easily added to using the format syntax. Also formats can be easily removed so that they are no longer considered valid.
Below are the default formats. If you think they are easy to read then you will be happy to know that is exactly the same format syntax you can use to define your own. No complex regular expressions are needed.
m/d/yy h:nn:ss OR d/m/yy hh:nn:ss m/d/yy h:nn OR d/m/yy h:nn m/d/yy h:nn_ampm OR d/m/yy h:nn_ampm yyyy-mm-dd hh:nn:ss yyyy-mm-dd h:nn ddd mmm d hh:nn:ss zo yyyy # Ruby time string yyyy-mm-ddThh:nn:ssZ # ISO 8601 without zone offset yyyy-mm-ddThh:nn:sszo # ISO 8601 with zone offset
NOTE: To use non-US date formats see US/Euro Formats section
yyyy/mm/dd yyyy-mm-dd yyyy.mm.dd m/d/yy OR d/m/yy m\d\yy OR d\m\yy d-m-yy dd-mm-yyyy d.m.yy d mmm yy
NOTE: To use non-US date formats see US/Euro Formats section
hh:nn:ss hh-nn-ss h:nn h.nn h nn h-nn h:nn_ampm h.nn_ampm h nn_ampm h-nn_ampm h_ampm
NOTE: Any time format without a meridian token (the ‘ampm’ token) is considered in 24 hour time.
Here is what each format token means:
Format tokens:
y = year
m = month
d = day
h = hour
n = minute
s = second
u = micro-seconds
ampm = meridian (am or pm) with or without dots (e.g. am, a.m, or a.m.)
_ = optional space
tz = Timezone abbreviation (e.g. UTC, GMT, PST, EST)
zo = Timezone offset (e.g. +10:00, -08:00, +1000)
Repeating tokens:
x = 1 or 2 digits for unit (e.g. 'h' means an hour can be '9' or '09')
xx = 2 digits exactly for unit (e.g. 'hh' means an hour can only be '09')
Special Cases:
yy = 2 or 4 digit year
yyyy = exactly 4 digit year
mmm = month long name (e.g. 'Jul' or 'July')
ddd = Day name of 3 to 9 letters (e.g. Wed or Wednesday)
u = microseconds matches 1 to 3 digits
All other characters are considered literal. For the technically minded, these formats are compiled into a single regular expression
To see all defined formats look at the source code.
The perennial problem for non-US developers or applications not primarily for the US, is the US date format of m/d/yy. This is ambiguous with the European format of d/m/yy. By default the gem uses the US formats as this is the Ruby default when it does date interpretation.
To switch to using the European (or Rest of The World) formats use this setting
Timeliness.use_euro_formats
Now ‘01/02/2000’ will be parsed as 1st February 2000, instead of 2nd January 2000.
You can switch back to US formats with
Timeliness.use_us_formats
Sometimes you may not want certain formats to be valid. You can remove formats for each type and the parser will then not consider that a valid format. To remove a format
Timeliness.remove_formats(:date, 'm\d\yy')
Adding new formats is also simple
Timeliness.add_formats(:time, "d o'clock")
Now “10 o’clock” will be a valid value.
You can embed regular expressions in the format but no guarantees that it will remain intact. If you avoid the use of any token characters, and regexp dots or backslashes as special characters in the regexp, it may work as expected. For special characters use POSIX character classes for safety. See the ISO 8601 datetime for an example of an embedded regular expression.
Because formats are evaluated in order, adding a format which may be ambiguous with an existing format, will mean your format is ignored. If you need to make your new format higher precedence than an existing format, you can include the before option like so
Timeliness.add_formats(:time, 'ss:nn:hh', :before => 'hh:nn:ss')
Now a time of ‘59:30:23’ will be interpreted as 11:30:59 pm. This option saves you adding a new one and deleting an old one to get it to work.
When dealing with 2 digit year values, by default a year is interpreted as being in the last century when at or above 30. You can customize this however
Timeliness.ambiguous_year_threshold = 20
Now you get:
year of 19 is considered 2019 year of 20 is considered 1920
-
Adam Meehan ([email protected], github.com/adzap)
Copyright © 2010 Adam Meehan, released under the MIT license
