Jon,

A dateTime might look like: "2008-10-09T14:15:32+0100"

"1215-03-21" isn't the right form for a dateTime and so in the cast in the FILTER

xsd:dateTime("1215-10-15")

it should throw an error because the cast is of a lexical form that is not legal for a dateTime. Errors will cause the current row to be rejected.

Just xsd:date("1215-02-01") should work if the cast is supported (rdflib supports xsd:date IIRC)

It can also be written:  "1215-02-01"^^xsd:date


> EXCEPT in the case of a duration that starts 1215-10-14 and ends 1215-11-01.

This query explicitly tests for specific start and finish:

PREFIX  rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX  xsd:  <http://www.w3.org/2001/XMLSchema#>
PREFIX  loc:  <http://simile.mit.edu/2005/05/ontologies/location#>
PREFIX  crm:  <http://cidoc.ics.forth.gr/rdfs/cidoc_v4.2.rdfs#>
PREFIX  place: <http://jjcrump.itinerary.king.john/place#>

SELECT DISTINCT *
WHERE
{
   ?event  crm:P116F.starts  ?start .
   ?event  crm:P115F.finishes  ?finish .
    FILTER ( ?start = "1208-01-05"^^xsd:date &&
             ?finish  = "1208-01-06"^^xsd:date)
    }

This pattern grabs ?start and ?finish for the event.

You can negate the filter

    FILTER (  ! ( ?start = "1208-01-05"^^xsd:date &&
                 ?finish  = "1208-01-06"^^xsd:date) )

operators < <= etc should work.

>       {?event crm:P117F.occurs_during ?date .}
> UNION {?event crm:P116F.starts ?date .}
> UNION {?event crm:P115F.finishes ?date .}

This pattern will not make the start and finish available to a filter in the same row. It create a table of 2 columns, "event" and "date" and puts the "starts" date in a different row to "finishes" date and the FILTER sees rows one at a time (no cross row comparison).

Did you mean something like:

        {?event crm:P117F.occurs_during ?date .}
  UNION { ?event crm:P116F.starts ?start ;
                 crm:P115F.finishes ?finish .
        }
(maybe decide which of start or finish you might use as "the" date and place ?date in one of the positions. )

The filter can then go inside the


        {?event crm:P117F.occurs_during ?date .}
  UNION { ?event crm:P116F.starts ?start ;
                 crm:P115F.finishes ?finish .
          FILTER (  ! ( ?start = "1208-01-05"^^xsd:date &&
                 ?finish  = "1208-01-06"^^xsd:date) )
        }

as it only applies to the start/finish case.

Hope that helps and I haven't mangled your question too badly,

        Andy



Jon Crump wrote:
All,

With limited programming experience, I've been using the python module rdflib to familiarize myself with SPARQL and have what is probably a naive question, I hope you'll bear with me.

I have an RDF graph that looks, in part, like this:

<E5.Event rdf:about="http://jjcrump.itinerary.king.john/Event#75599bf70a3a036ef0b6c8b1ef331c2704fe648e";> <P117F.occurs_during rdf:datatype="http://www.w3.org/2001/XMLSchema#date";>1215-03-21</P117F.occurs_during> <P11F.had_participant rdf:resource="http://jjcrump.itinerary.king.john/person/king_john"/> <P7F.took_place_at rdf:resource="http://jjcrump.itinerary.king.john/place#geddington_northamptonshire"/>
</E5.Event>

<E5.Event rdf:about="http://jjcrump.itinerary.king.john/Event#7577d039daff5397bc5c4d8c671432fa0cdedaa1";> <P115F.finishes rdf:datatype="http://www.w3.org/2001/XMLSchema#date";>1208-01-06</P115F.finishes> <P116F.starts rdf:datatype="http://www.w3.org/2001/XMLSchema#date";>1208-01-05</P116F.starts> <P11F.had_participant rdf:resource="http://jjcrump.itinerary.king.john/person/king_john"/> <P7F.took_place_at rdf:resource="http://jjcrump.itinerary.king.john/place#burbage_wiltshire"/>
</E5.Event>

ie. Instants have a single property for date, durations have start/end properties. I'm trying to construct a query that returns all events that fall within, or intersect with a date range.

This, for example, works:

PREFIX loc: <http://simile.mit.edu/2005/05/ontologies/location#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
PREFIX place: <http://jjcrump.itinerary.king.john/place#>
PREFIX crm: <http://cidoc.ics.forth.gr/rdfs/cidoc_v4.2.rdfs#>
PREFIX xsd: <http://www.w3.org/2001/XMLSchema#>

SELECT DISTINCT ?event ?place
WHERE {{?event crm:P117F.occurs_during ?date .} UNION {?event crm:P116F.starts ?date .}
UNION {?event crm:P115F.finishes ?date .}
OPTIONAL { ?event crm:P7F.took_place_at ?place .}
FILTER (xsd:dateTime(?date) >= xsd:dateTime("1215-10-15")
&& xsd:dateTime(?date) <= xsd:dateTime("1215-10-31") )}

EXCEPT in the case of a duration that starts 1215-10-14 and ends 1215-11-01.

Is there any kind of arithmetic that can be done in the filter clause that will catch this case as well?

Many thanks for your patience,
Jon

--
Hewlett-Packard Limited
Registered Office: Cain Road, Bracknell, Berks RG12 1HN
Registered No: 690597 England

Reply via email to