CODEWITHSUNDEEP
sparql
Pietro Maria Liuzzo
Pietro Maria Liuzzo

Reputation: 171

SPARQL order by digit part of literal

How do I order the results of my SPARQL query using the digits only in a literal value?

I have a query which returns a variable ?locusFrom which can contain literal values like

I would like to use ORDER BY and sort the results as above using only the numeric part and ignoring the letters.

I have tried 

SELECT ?locusFrom ?locusTo ?locusTarget ?uniCont1
WHERE {
bm:BNFet45 dcterms:hasPart ?uniProd .
?uniProd a SdC:UniProd .
?uniProd dcterms:hasPart ?uniCont .
?uniCont a SdC:UniCont .
?uniCont dcterms:hasPart ?uniCont1 .
?uniCont1 a SdC:UniCont ;
            bm:hasLocus ?locus .
            OPTIONAL{?locus bm:locusTarget ?locusTarget}
            OPTIONAL{?locus bm:locusFrom ?locusFrom}
            OPTIONAL{?locus bm:locusTo ?locusTo}
BIND( (STRBEFORE(?locusFrom,"r") || STRBEFORE(?locusFrom,"v") ) as ?substring )
}
ORDER BY ?substring

and it returns this results (shortened)

<sr:sparql xmlns:sr="http://www.w3.org/2005/sparql-results#">
<sr:head>
<sr:variable name="locusFrom"/>
<sr:variable name="locusTo"/>
<sr:variable name="locusTarget"/>
<sr:variable name="uniCont1"/>
</sr:head>
<sr:results>
<sr:result>
<sr:binding name="locusTarget">
<sr:literal>12</sr:literal>
</sr:binding>
<sr:binding name="uniCont1">
<sr:uri>http://betamasaheft.eu/BNFet45/msitem/f2_i1.1</sr:uri>
</sr:binding>
</sr:result>
<sr:result>
<sr:binding name="locusFrom">
<sr:literal>1r</sr:literal>
</sr:binding>
<sr:binding name="uniCont1">
<sr:uri>http://betamasaheft.eu/BNFet45/msitem/p1_i1.1</sr:uri>
</sr:binding>
</sr:result>
<sr:result>
<sr:binding name="locusFrom">
<sr:literal>18v</sr:literal>
</sr:binding>
<sr:binding name="uniCont1">
<sr:uri>http://betamasaheft.eu/BNFet45/msitem/p1_i1.2</sr:uri>
</sr:binding>
</sr:result>
<sr:result>
<sr:binding name="locusFrom">
<sr:literal>53r</sr:literal>
</sr:binding>
<sr:binding name="uniCont1">
<sr:uri>http://betamasaheft.eu/BNFet45/msitem/p1_i1.3</sr:uri>
</sr:binding>
</sr:result>
<sr:result>
<sr:binding name="locusFrom">
<sr:literal>17r</sr:literal>
</sr:binding>
<sr:binding name="locusTo">
<sr:literal>17v</sr:literal>
</sr:binding>
<sr:binding name="uniCont1">
<sr:uri>http://betamasaheft.eu/BNFet45/msitem/f2_i1.4</sr:uri>
</sr:binding>
</sr:result>
<sr:result>
<sr:binding name="locusFrom">
<sr:literal>18r</sr:literal>
</sr:binding>
<sr:binding name="locusTo">
<sr:literal>23r</sr:literal>
</sr:binding>
<sr:binding name="uniCont1">
<sr:uri>http://betamasaheft.eu/BNFet45/msitem/f2_i1.5</sr:uri>
</sr:binding>
</sr:result>
</sr:results>
</sr:sparql>

where it seams to me that it is starting again to count after the first group. How can I get it to sort all together? Thank you very much!

Upvotes: 1

Views: 578

Answers (1)

Stanislav Kralin
Stanislav Kralin

Reputation: 11469

Basically, REPLACE accepts XPath-style regular expression as its second argument. However, since ?locusFrom is bound within OPTIONAL, you may also need COALESCE.

SELECT * {
   VALUES (?locusFrom) {
       (UNDEF)
       ("")
       ("1")
       ("156vb")
       ("21r")
       ("33vu")
       ("45ra")
       ("ab")
   }
   BIND (xsd:integer(REPLACE(?locusFrom, "\\D+", "")) AS ?number1)
   BIND (COALESCE(?number1, 0) AS ?number2)
} ORDER BY ?number2

Upvotes: 3

Related Questions