undate overview

undate is a python library for working with uncertain or partially known dates.

Warning

This is pre-alpha software and is NOT feature complete! Use with caution. Currently it only supports parsing and formatting dates in ISO8601 format and some portions of EDTF (Extended Date Time Format).

Undate was initially created as part of a DH-Tech hackathon in November 2022.

---

Read Contributors for detailed contribution information.

Example Usage

Often humanities and cultural data include imprecise or uncertain temporal information. We want to store that information but also work with it in a structured way, not just treat it as text for display. Different projects may need to work with or convert between different date formats or even different calendars.

An undate.Undate is analogous to python’s builtin datetime.date object, but with support for varying degrees of precision and unknown information. You can initialize an undate with either strings or numbers for whichever parts of the date are known or partially known. An Undate can take an optional label.

from undate.undate import Undate

november7 = Undate(2000, 11, 7)
november = Undate(2000, 11)
year2k = Undate(2000)
november7_some_year = Undate(month=11, day=7)

partially_known_year = Undate("19XX")
partially_known_month = Undate(2022, "1X")

easter1916 = Undate(1916, 4, 23, label="Easter 1916")

You can convert an Undate to string using a date formatter (current default is ISO8601):

>>> [str(d) for d in [november7, november, year2k, november7_some_year]]
['2000-11-07', '2000-11', '2000', '--11-07']

If enough information is known, an Undate object can report on its duration:

>>> december = Undate(2000, 12)
>>> feb_leapyear = Undate(2024, 2)
>>> feb_regularyear = Undate(2023, 2)
>>> for d in [november7, november, december, year2k, november7_some_year, feb_regularyear, feb_leapyear]:
...    print(f"{d}  - duration in days: {d.duration().days}")
...
2000-11-07  - duration in days: 1
2000-11  - duration in days: 30
2000-12  - duration in days: 31
2000  - duration in days: 366
--11-07  - duration in days: 1
2023-02  - duration in days: 28
2024-02  - duration in days: 29

If enough of the date is known and the precision supports it, you can check if one date falls within another date:

>>> november7 = Undate(2000, 11, 7)
>>> november2000 = Undate(2000, 11)
>>> year2k = Undate(2000)
>>> ad100 = Undate(100)
>>> november7 in november
True
>>> november2000 in year2k
True
>>> november7 in year2k
True
>>> november2000 in ad100
False
>>> november7 in ad100
False

For dates that are imprecise or partially known, undate calculates earliest and latest possible dates for comparison purposes so you can sort dates and compare with equals, greater than, and less than. You can also compare with python datetime.date objects.

>>> november7_2020 = Undate(2020, 11, 7)
>>> november_2001 = Undate(2001, 11)
>>> year2k = Undate(2000)
>>> ad100 = Undate(100)
>>> sorted([november7_2020, november_2001, year2k, ad100])
[<Undate 0100>, <Undate 2000>, <Undate 2001-11>, <Undate 2020-11-07>]
>>> november7_2020 > november_2001
True
>>> year2k < ad100
False
>>> from datetime import date
>>> year2k > date(2001, 1, 1)
False

When dates cannot be compared due to ambiguity or precision, comparison methods raise a NotImplementedError.

>>> november_2020 = Undate(2020, 11)
>>> november7_2020 > november_2020
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Users/rkoeser/workarea/github/undate-python/src/undate/undate.py", line 262, in __gt__
    return not (self < other or self == other)
  File "/Users/rkoeser/workarea/github/undate-python/src/undate/undate.py", line 245, in __lt__
    raise NotImplementedError(
NotImplementedError: Can't compare when one date falls within the other

An UndateInterval is a date range between two Undate objects. Intervals can be open-ended, allow for optional labels, and can calculate duration if enough information is known

>>> from undate.undate import UndateInterval
>>> UndateInterval(Undate(1900), Undate(2000))
<UndateInterval 1900/2000>
>>> UndateInterval(Undate(1900), Undate(2000), label="19th century")
>>> UndateInterval(Undate(1900), Undate(2000), label="19th century").duration().days
36890
<UndateInterval '19th century' (1900/2000)>
>>> UndateInterval(Undate(1900), Undate(2000), label="20th century")
<UndateInterval '20th century' (1900/2000)>
>>> UndateInterval(latest=Undate(2000))  # before 2000
<UndateInterval ../2000>
>>> UndateInterval(Undate(1900))  # after 1900
<UndateInterval 1900/>
>>> UndateInterval(Undate(1900), Undate(2000), label="19th century").duration().days
36890
>>> UndateInterval(Undate(2000, 1, 1), Undate(2000, 1,31)).duration().days
31

You can initialize Undate or UndateInterval objects by parsing a date string with a specific converter, and you can also output an Undate object in those formats. Available converters are "ISO8601" and "EDTF" (but only)

>>> from undate import Undate
>>> Undate.parse("2002", "ISO8601")
<Undate 2002>
>>> Undate.parse("2002-05", "EDTF")
<Undate 2002-05>
>>> Undate.parse("--05-03", "ISO8601")
<Undate --05-03>
>>> Undate.parse("--05-03", "ISO8601").format("EDTF")
'XXXX-05-03'
>>> Undate.parse("1800/1900")
<UndateInterval 1800/1900>

For more examples, refer to the example notebooks included in this repository.

Documentation

Project documentation is available on ReadTheDocs.

For instructions on setting up for local development, see Developer Notes.

License

This software is licensed under the Apache 2.0 License.