parsing dates

[Joe Wicentowski]

James,

I've encountered the same issue with dates in TEI documents.  What
follows are some thoughts/suggestions about using TEI dates in eXist.

I think the issue is The TEI Guidelines allows/encourages one to mix
date datatypes, thereby conflating data with precision in dates
(http://www.tei-c.org/release/doc/tei-p5-doc/en/html/CO.html#CONADA):

------
"Partial dates or times (e.g. 1990, September 1990, twelvish) can be
expressed in the when attribute by simply omitting a part of the value
supplied. Imprecise dates or times (for example early August, some
time after ten and before twelve) may be expressed as date or time
ranges...  The @when attribute always supplies a normalized
representation of the date given as content of the date element. The
format used should be a valid W3C schema datatype. Some typical
examples follow:

* <date when="2001">The year 2001</date>
* <date when="2001-09">September 2001</date>
* <date when="2001-09-11">11 Sept 01</date>
* <date when="--09-11">9/11</date>
* <date when="--09">September</date>
* <date when="---11">Eleventh of the month</date>
* <time when="08:48:00">8:48</time>
* <date when="2001-09-11T12:48:00">Sept 11th, 12 minutes before 9 am</date>

Note in the last example the use of a normalized representation for
the date string which includes a time: this example could thus equally
well be tagged using the time element.
The following examples demonstrate the use of the date element to mark
a period of time:
* <p>Those five years — <date from="1918" to="1923">1918 to
1923</date> — had been, he suspected, somehow very important.</p>"
-----

If you mix 2+ date datatypes, you will encounter some difficulties
when doing what you asked about.  Setting range indexes in eXist is
impossible if you mix datatypes, as far as I know.  I think we might
encode TEI texts more effectively to make processing easier by (1)
either using a consistent datatype or (2) replacing the generic @when
with a new way of encoding which implies the datatype.

As far as your specific question, you could run a series of "if ($date
castable as ...) then ... else" statements on the following datatypes:

* date
* gYear
* gMonth
* gDay
* gYearMonth
* gMonthDay
* time
* dateTime

As an alternative to conditional expressions, I'm not sure if
typeswitch would work for this?

- Joe



On Mon, Oct 5, 2009 at 7:05 PM, James Cummings
<http://x-query.com/mailman/listinfo/talk> wrote:
> Hi there,
>
> I've a silly question about the best way to parse dates.  I'm passing
> a parameter (in eXist) that is part of the URL for an xquery and I
> want to separate it into its individual parts of year month and day. I
> could test the string and do substring-before and after and such, or
> create my own function, but it seemed better practice to use the
> functions like year-from-date and month-from-date.  So I tried:
>
> let $date := request:get-parameter("date", ("1789-01-01")) cast as xs:date
> let $year := fn:year-from-date($date)
> let $month := fn:month-from-date($date)
> let $day := fn:day-from-date($date)
>
> which works fine when the date is YYYY-MM-DD.  But what I want to
> achieve is being able to handle YYYY-MM-DD, YYYY-MM or just YYYY.  The
> last two give me an error of "xs:date must not have hour, minute or
> second fields set" (erm... which it doesn't, which is why I'm
> confused). I'm happy for $month and $day to be empty if they don't
> exist.
>
> I'm sure it is something terribly obvious with the datatype.  Should I
> just be breaking it down to its component parts myself?  It just
> seemed like I was reinventing the wheel.
>
> Thanks for any suggestions.
>
> -James
> _______________________________________________
> http://x-query.com/mailman/listinfo/talk
> http://x-query.com/mailman/listinfo/talk
>