6

The datetime module does date validation and math which is fine when you care about reality.

I need an object that holds dates generated even if they were invalid. Date time is way too strict as sometimes I know year only or year and month only and sometimes I have a date like 2011-02-30.

Is there a module out there that is like datetime but that can handle invalid dates?

If not, what's the best way to handle this while duplicating as little functionality as possible and still allowing date math when it is possible to perform?

UPDATE

Motivation for this is integration with multiple systems that use dates and don't care about invalid dates (mysql and perl) in addition to wanting the ability to tell basic general ranges of time. For fuzzy date math start from the beginning of the known unit of time (if I know year and month but not day, use the first, if i know year but no month or day, use january first). This last bit is not necessary but would be nice and I get why it is not common as people who need special case date math will probably build it themselves.

One of the major issues I have is loading dates from mysql into python using sqlalchemy and mysqldb -- if you load a value from a date column im mysql that looks like '2011-01-00' in mysql, you get None in python. This is not cool by any stretch.

Improve this question
4
  • 1
    Won't date math be hard when you don't know how many days there are in say, each month? Maybe I don't quite understand what you are looking for. Commented Jul 14, 2011 at 19:03
  • @Pat B My guess would be: assume real life until there's a departure. Also, @SynapticUnderrun You might get some more responses if you explained why you want/need this (Also I'm just interested). Commented Jul 14, 2011 at 20:04
  • 1
    If you don't care whether the date is valid or not, store it as a string. The datetime module uses epoch time internally so it will follow the calendar. Also check out the pytz module for time zone support if that is the reason you're concerned about tolerance. Commented Jul 14, 2011 at 20:19
  • You should get a None, because it's not a date, it's a bunch of numbers joined by -. Commented Jul 14, 2011 at 21:09

2 Answers 2

Reset to default
2

If the datetime module is too strict, store it as a string or a tuple padded with zero's for unknown values.

Edit if you're having trouble translating from mysql to python, write yourself an adapter class to do the conversion in a safe way (See Changing attribute behaviour).

Edit2 a simple alternative might be to write a short script to run through all your SQL dates and do the corrections there i.e. change 2011-01-00 to 2011-01-01

Edit3 The docs suggest you can override reflected columns so you could treat your timestamp columns as strings ...or in later versions of SQLAlchemy, you can use the column_property for overrides.

Improve this answer
Sign up to request clarification or add additional context in comments.

2 Comments

how do i get mysqldb to use my class instead of datetime?
adapter class doesn't sound fun, altering invalid dates is a change to the actual data that is there (not allowable within the scope of my project, but the column_property sqlalchemy option looks like what i want (property that is a cast() of my column).
0

I haven't heard of such module out there and don't think there is one.

I would probably end up storing two dates for every instance: 1. the original input as a string, which could contain anything, even "N/A", just for showing back the original value, and 2. parsed and "normalized" datetime object which is the closest representation of the input. Depending on the purpose I would allow Null/None objects where it really couldn't be estimated (like the mentioned "N/A" case) or not. This solution will allow you to revert/change the "estimation" as you do not lose any information.

If you don't care about it so much, SQLAlchemy allows declaring your own column and data types for transparently converting such values back and forth into a string column in the DB.

Improve this answer

Comments

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.