node package manager
It’s your turn. Help us improve JavaScript. Take the 2017 JavaScript Ecosystem Survey »



Build Status Coverage Status NPM version License AGPL-3.0

An Extended Date Time Format (EDTF) / ISO 8601-2 parser and toolkit for date/time hackers and enthusiasts.


EDTF / ISO 8601-2

EDTF.js fully implements EDTF levels 0, 1, and 2 as specified by WD 2016-02-16 of ISO 8601-2 with the following exceptions (as raised by the EDTF community):

  1. Symbols for unknown and open dates in intervals have been switched: * makes more sense to represent an open date because it is often used as a wildcard to match "all" or "everything". Also, when an interval is blank, it suggessts "incomplete" or "unknown".

  2. "Before or after" is redundant and has been removed. It is covered by "One of a Set", e.g., [1760-12..] which means "December 1760 or some later month."

  3. Seasons in intervals are supported at the experimental/non-standard level 3.


EDTF.js is written as a standard Nodes.js / CommonJS module using many ES6 features. It therefore requires Node.js 6 or later (for Nodes.js 4/5 use the appropriate --harmony flags as necessary).

Bundling using RollUp

To bundle EDTF.js with RollUp you'll need to enable the following plugins

  • rollup-plugin-json
  • rollup-plugin-node-resolve
  • rollup-plugin-commonjs
  • rollup-plugin-node-globals (optional)
  • rollup-plugin-node-builtins (optional)

The last two are optional; you need them only if you want to include Node.js assert in your bundle.


$ npm install edtf


EDTF.js exports a top-level function which takes either a string (with optional parser constraints), a parse result, a regular or any of the extended date objects and returns a new, extended date object as appropriate.

const edtf = require('edtf')

edtf('2016-XX')          #-> returns an edtf.Date
edtf(new Date())         #-> returns an edtf.Date
edtf('2016-04~/2016-05') #-> returns an edtf.Interval

For a list of all types supported by EDTF.js see:

#-> ['Date', 'Year', 'Decade', 'Century', 'Season', 'Interval', 'List', 'Set']

Each type provides at least the following properties: the date's corresponding EDTF string, its minimum and maximum numeric value, its type, as well as its date part values.

edtf('2016?').edtf         #-> '2016?'

edtf('2016-02').min        #-> 1454284800000, i.e. 2016-02-01T00:00:00Z
edtf('2016-02').max        #-> 1456790399999, i.e. 2016-02-29T23:59:59Z

edtf('2016-02-2X').min     #-> 1455926400000, i.e. 2016-02-20T00:00:00Z
edtf('2016-02-2X').max     #-> 1456790399999, i.e. 2016-02-29T23:59:59Z

edtf('[..2016,2017]').min  #-> -Infinity
edtf('[..2016,2017]').max  #-> 1514764799999, i.e. 2017-12-31T23:59:59Z

edtf('2016-34').min        #-> 1459468800000, i.e. 2016-04-01T00:00:00Z
edtf('2016-34').max        #-> 1467331199999, i.e. 2016-06-30T23:59:59Z

edtf('2016?-02').values    #-> [2016, 1]
edtf('2016-05').values     #-> [2016, 4]

A date's min value is also used as its primitive value for numeric coercion. Because this is the case for all EDTF.js classes, comparison semantics tend to align well with common-sense expectations -- but be careful, as always, when type coercion is at play.

Unspecified, uncertain, and approximate dates

EDTF.js keeps track of qualified dates using bitmasks. If you are interested in binary yes or no, you can always convert a bitmask's value to boolean. For more fine-grained information, the edtf.Bitmask class provides a convenient interface for accessing these states:

 edtf('2016-05?').uncertain.value        #-> 63 / yes
 edtf('2016-05?').approximate.value      #-> 0  / no

 edtf('2016-05?')'year')   #-> 15 / yes
 edtf('2016-05?')'month')  #-> 48 / yes
 edtf('2016-05?')'day')    #-> 0  / no

 edtf('2016-?05')'year')   #-> 0  / no
 edtf('2016-?05')'month')  #-> 48 / yes

 edtf('201X-XX').unspecified.value       #-> 56 / yes
 edtf('201X-XX')'year')  #-> 8  / yes
 edtf('201X-XX')'month') #-> 48 / yes
 edtf('201X-XX')'day')   #-> 0  / no

Instead of year, month, and day, you can also query a string-based representation of the bitmask:

 edtf('201X-XX').unspecified.test('XXYYMMDD') #-> 0  / no
 edtf('201X-XX').unspecified.test('YYYXMMDD') #-> 8  / yes

When printing qualified dates, EDTF.js will try to find an optimal rendtition of qualification symbols. For that reason, you can use EDTF.js to normalize EDTF strings:

 edtf('?2016-?05-31').edtf   #-> 2016-05?-31
 edtf('?2016-?05~-31').edtf  #-> 2016-05%-31

Because every extended date object has a min and max value, you can always test if one date covers another one:

 edtf('2016-06/2016-09').covers(edtf('2016-08-24')) #-> true
 edtf('2016-06/2016-09').covers(edtf('2016-05-31')) #-> false

Iterable dates offer an includes() test as well which returns true only if a given date is part of the iteration. For example:

 edtf('2016-06/2016-09').includes(edtf('2016-08-24')) #-> false

August 24, 2016 is covered by the interval '2016-06/2016-09' but is not included in it, because the interval has month precision and can be iterated as:

 #-> [2016-06, 2016-07, 2016-08, 2016-09]

Enumerating dates

EDTF.js dates offer iterators to help measure the duration between two dates. These iterators are dependent on a date's precision:

 edtf('2016').next()        #-> 2017
 edtf('2016-05').next()     #-> 2016-06
 edtf('2016-02-28').next()  #-> 2016-02-29
 edtf('2017-02-28').next()  #-> 2017-03-01

Careful, if your date has no precision, next will return the next second!

The generator *between() returns all the dates, by precision, between two dates; similarly, *until() returns all dates in between and includes the first of the two dates; to generate the full range inluding both dates, use *through().

 #-> [2016-06]

 #-> [2016-05, 2016-06]

 #-> [2016-05, 2016-06, 2016-07]

Note, that all range generators also work in reverse order:

 #-> [2016-07, 2016-06, 2016-05]

 #-> [2016-07, 2016-06]

 #-> [2016-06]


The EDTF.js classes Date, Interval, List, and Set (lists model EDTF's 'multiple dates', while sets model 'one of a set') are iterable. Dates are iterable.

 #-> [2015, 2016, 2017, 2018]

Note that this also works with varying precisions:

  #-> [2015-10, 2015-11, 2015-12, 2016-01, 2016-02, 2016]

Consecutive dates in lists and sets are expanded during an iteration:

 #-> [2015, 2018, 2019, 2020]


To use EDTF.js' date parser directly, call edtf.parse() with an input string and optional parser constraints. The parser will always return plain JavaScript objects which you can pass to edtf() for conversion to extended date object, or to your own post-processing.

#-> { type: 'Date', level: 0, values: [2016, 1] }

As you can see, the parser output includes the compatibility level of the date parsed; the values array contains the individual date parts in a format compatible with JavaScript's Date semantics (months are a zero-based index).

Unspecified, uncertain, or approximate dates are returned as a numeric value representing a bitmask; refer to the edtf.Bitmask class for details.

#-> { type: 'Date', level: 2, values: [2016, 1], uncertain: 15, approximate: 48 }

#-> { type: 'Date', level: 2, values: [2000, 1], unspecified: 12 }

Note that unspecified date values will always return the least possible value, e.g., '2000' for '20XX'. Note, also, that EDTF.js will not parse impossible unspecified dates. For instance, none of the following examples can be valid dates:

edtf.parse('2016-02-3X') #-> A day in February cannot start with a 3
edtf.parse('2016-2X-XX') #-> There are only 12 months
edtf.parse('2016-XX-32') #-> No month has 32 days

Intervals, Sets, and Lists will contain their parsed constituent dates in the values array:

#-> { type: 'Interval', level: 0, values: [{..}, {..}] }

By passing level or types constraints to the parser, you can ensure EDTF.js will accept only dates supported by your application.

edtf.parse('2016?', { level: 0 }) #-> parse error
edtf.parse('2016?', { level: 1 }) #-> ok

edtf.parse('2016?-02', { level: 1 }) #-> parse error
edtf.parse('2016?-02', { level: 2 }) #-> ok

edtf.parse('2016-21', { types: ['Date'] })           #-> parse error
edtf.parse('2016-21', { types: ['Date', 'Season'] }) #-> ok

edtf.parse('2016?', { level: 0, types: ['Date'] })   #-> parse error
edtf.parse('2016?', { level: 1, types: ['Date'] })   #-> ok


EDTF.js can generate random EDTF strings for you. Simply call edtf.sample() to create a new iterator:

let it = edtf.sample() #-> { value: '0097-26', done: false } #-> { value: '0000-09-30T22:50:54-07', done: false }

For a finite iterator, simply pass a count:

[...edtf.sample({ count: 3 }]
#-> ['-003%', '-0070-07-31%', '[-0080-10..]']

You can also generate strings at a given compatibility level:

[...edtf.sample({ count: 3, level: 0 }]
#-> ['0305/0070-04-30', '-07', '0000/0013']

[...edtf.sample({ count: 3, level: 1 }]
#-> ['00XX', 'Y80105', '0000~']

[...edtf.sample({ count: 3, level: 2 }]
#-> ['Y1E30', '-8110S2', '{%0401}']

Note that some grammar rules at levels 1 and 2 may, potentially, generate strings at a lower level (but never higher).

Finally, at each level you can also limit the generated strings to a given type (you must specify a level for this to work):

[...edtf.sample({ count: 3, level: 2, type: 'Decade' }]
#-> ['003', '030~', '000']


The EDTF.js parser is based on the awesome nearley parser generator.

The EDTF.js generator uses the ingenious randexp.