Skip to content

Utilities to interact with a SPARQL endpoint in clojure

License

Notifications You must be signed in to change notification settings

ont-app/sparql-endpoint

Repository files navigation

NaturalLexicon logo ont-app/sparql-endpoint

The sparql-endpoint library provides utilities for interfacing with SPARQL 1.1 endpoints in clojure.

Contents

Installation

sparql-endpoint is available as a Maven artifact from clojars.

Clojars Project

Additional documentation is provided at https://cljdoc.org/d/ont-app/sparql-endpoint/0.2.0.

Require thus:

(ns my.namespace
  (:require
    [ont-app.sparql-endpoint.core :as endpoint :refer :all]
    ))

Functions

Functions that interact with SPARQL endpoints

Interacting with a SPARQL endpoint involves POSTs to an update endpoint and GETs to a query endpoint. There are special functions for ASK, SELECT and CONSTRUCT queries. Each of these take mandatory endpoint and query arguments, plus an optional http-req map.

Mandatory arguments: endpoint and query

All of the basic query and update functions take two mandatory arguments:

endpoint is the URL of a SPARQL endpoint

query is a string in an appropriate format for ASK, SELECT, CONSTRUCT, or one of the UPDATE operations.

Optional argument: http-req

HTTP calls are done through the clj-http library. There is a third optional http-req argument which may include additional HTTP request parameters.

For example if endpointrequires authentication, you may specify {:basic-auth "myUserName:myPassword"}

[{:cookie-policy :standard}](https://github.com/dakrone/clj-http#get) is asserted by default, but this can be overridden. The :query-params parameter is reserved, as it is needed to specify the query to the endpoint.

sparql-ask

This function takes an endpoint and a SPARQL ASK query and returns a boolean:

> (sparql-ask 
    "https://query.wikidata.org/bigdata/namespace/wdq/sparql"
    "ASK WHERE {wd:Q5 rdfs:label \"human\"@en}")
true
>

sparql-select

This function takes as its query parameter a SPARQL SELECT query:

> (let [query "# What is the English name for Q5?
               PREFIX rdfs: `
               PREFIX wd: <http://www.wikidata.org/entity/>
               SELECT ?enLabel
               WHERE
               {
                  wd:Q5 rdfs:label ?enLabel.
                  Filter (Lang(?enLabel) = \"en\")
               }
              "
       ]
       (sparql-select 
          "https://query.wikidata.org/bigdata/namespace/wdq/sparql"
          query))
          
[{"enLabel" {"xml:lang" "en", "type" "literal", "value" "human"}}]
>

The bindings returned are direct translations of the JSON returned by the endpoint. These can be mapped by more expressive simplifiers, described below.

sparql-construct

This function takes a SPARQL CONSTRUCT query as its query parameter and returns a string of turtle describing the results.

> (let [query "# Things called 'human'
               PREFIX eg: <http://example.com/>
               PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
               PREFIX wd: <http://www.wikidata.org/entity/>
    
               CONSTRUCT
               {
                 ?human a eg:Human.
               }
               WHERE
               {
                 ?human rdfs:label \"human\"@en.
               }
              "
       ]
       (sparql-construct       
          "https://query.wikidata.org/bigdata/namespace/wdq/sparql"
           query))
   
"
@prefix eg: <http://example.com/> .
@prefix wd: <http://www.wikidata.org/entity/> .
wd:Q823310 a eg:Human .
wd:Q20094897 a eg:Human .
wd:Q26190966 a eg:Human .
wd:Q5 a eg:Human .
"
>

sparql-update

This function POSTS its query parameter (CREATE, INSERT, DELETE, etc) to the specified SPARQL update endpoint, and returns the plain text response.

> (sparql-update "http://localhost:3030/my-dataset/update"
                 "DROP GRAPH <http://myGraph>")
<html>
<head>
</head>
<body>
<h1>Success</h1>
<p>
Update succeeded
</p>
</body>
</html>
> 

This is how you would load a file into a graph

(endpoint/sparql-update
   "http://localhost:3032/my-dataset/update"
   "LOAD <file:test/resources/my-file.ttl>
    INTO GRAPH <http://my-dataset/myGraph>"
   )

Simplifiers

By default the output of sparql-select is parsed from raw JSON output of the endpoint, using the specification described by W3C.

{'value' <value>
 'type' 'uri' | 'literal'
 ;;...maybe...
 'xml:lang' <lang> (if literal)
 'datatype' <datatype> (if literal)
}

It is usually convenient to transform these bindings into simpler representations. Hence the functions simplify and simplifier-for-prologue, described below.

simplify

The function simplify will take a result binding and return a simplified map {<var> <value>...}. This would typically be done in the context of a map function:

(def human-query "PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
             PREFIX wd: <http://www.wikidata.org/entity/>
             SELECT ?enLabel
             WHERE
             {
               wd:Q5 rdfs:label ?enLabel.
               Filter (Lang(?enLabel) = \"en\")
             }")
  (map simplify (sparql-select wikidata-endpoint human-query))
({:enLabel #voc/lstr "human@en"})
;; Compare to [{"enLabel" {"xml:lang" "en", "type" "literal", "value" "human"}}]
>

The #voc/lstr reader macro is defined for literal values with xml:lang tags. Described in more detail in the next sections.

Optional translators argument

simplify optionally takes two arguments, the first of which is translators, a map with four keys: :uri, :lang, :datatype and :bnode. Default values for this map are defined as the value default-translators.

Each key maps to a function of the form [sparql-binding] -> value

key description default
:uri value is a uri return raw URI string (typically "http://...")
:lang value is literal and has a language tag, e.g. "en" return a LangStr
:datatype value is literal and has an assigned datatype, g.g. "xsd:int" parse XSD values, otherwise return a meta-tagged-literal
:bnode value is a blank node return raw value, typically like "b0"

By default the Jena library is referenced to translate xsd datatypes into instances of an appropriate class. In the following example, Obama's date of birth is translated to an instance of Jena's XSDDateTime, which has a getYears method:

> (let [query "PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
               PREFIX wd: <http://www.wikidata.org/entity/>
               # What is Obama's date of birth?
               SELECT ?dob
               WHERE 
               {
                 wd:Q76 wdt:P569 ?dob.
               } "
       ]
       (-> (sparql-select wikidata-endpoint query)
           (map simplify)
           (first)
           (:dob)
           (.getYears)))
1961
>

Customizing simplifiers

Any of these values can be overridden with custom functions by merging default-translators with an overriding map.

Since ont-app/vocabulary is a dependency of this library, there is an alternative to the default simplifier called simplifier-with-kwis

> (use 'ont-app.vocabulary.wikidata)
> (->> 
    (sparql/sparql-select wikidata-endpoint human-query)
    (map sparql/simplifier-with-kwis)
    (set))
#{{:q :wd/Q823310} {:q :wd/Q5} {:q :wd/Q73755406}}
> (voc/uri-for :wd/Q5)
"http://www.wikidata.org/entity/Q5"

This was defined using the functions update-translators and make-simplifier:

(def simplifier-with-kwis
  (make-simplifier (update-translators default-translators
                                       :uri voc/keyword-for)))

LangStr

LangStr is a type which holds a string and a language tag.

It is defined in the supporting library ont-app.vocabulary.lstr.

Examples:

> (type #voc/lstr "human@en")
ont_app.vocabulary.lstr.LangStr
> (str #voc/lstr "human@en")
human
> (ont-app.vocabulary.lstr/lang  #voc/lstr "human@en")
"en"

See the docs for ont-app/vocabulary for details.

meta-tagged-literal

The function meta-tagged-literal reifies an Object whose toString method is the "value" field in the SPARQL binding map, and whose metadata contains everything else.

This is used In the case of non-xsd datatypes (typically encoded in SPARQL as '""^^MyDatatype'):

> (def my-datum 
   (meta-tagged-literal 
     {"datatype" "MyDatatype" 
      "type" "literal" 
      "value" "<datatype-encoded-with-my-syntax>"}))
my-datum
> 
> my-datum
#object[ont_app.sparql_endpoint.core$meta_tagged_literal$reify__14016 0x2a92b8aa "<datatype-encoded-with-my-syntax>"]
>
> (str my-datum)
"<datatype-encoded-with-my-syntax>"
>
> (meta my-datum)
{"type" "literal", "datatype" "MyDatatype"}
>

simplifier-for-prologue

This function takes a query with a prologue (Including a set of PREFIX declarations) and returns a simplifier function informed by a function which maps full URIs to their corresponding quicknames. It is informed by the function parse-prologue, described below.

Compare this…

(let [query "# Things called 'Barack Obama'
             PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
             PREFIX wd: <http://www.wikidata.org/entity/>
             SELECT *
             WHERE
             {
               ?Q rdfs:label \"Barack Obama\"@en.
             }"
      ]
      (map simplify
           (sparql-select wikidata-endpoint query)))
({:Q "http://www.wikidata.org/entity/Q76"} 
 {:Q "http://www.wikidata.org/entity/Q47513588"}) 
>

… to this …

(let [query "# Things called 'Barack Obama'
             PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
             PREFIX wd: <http://www.wikidata.org/entity/>
             SELECT *
             WHERE
             {
               ?Q rdfs:label \"Barack Obama\"@en.
             }"
     ]
    (map (simplifier-for-prologue query)
         (sparql-select wikidata-endpoint query)))
    
({:Q "wd:Q76"} {:Q "wd:Q47513588"}) 
>

parse-prologue

This function takes a SPARQL query and returns a vector with three values: base, uri-to-quickname, quickname-to-uri.

name description
base The base URI used to resolve relative URIs
uri-to-quickname fn[uri] -> corresponding quickname
quickname-to-uri fn[quickname] -> corresponding full URI

Given a string for which there is no prefix declaration in the query, these last two functions will return their argument unchanged.

> (let [query "BASE <http://example.org/>
               PREFIX eg: <http://example.com/> 
               Select * where {?s ?p ?o.}"
        [base, u-to-q, q-to-u]  (sparql/parse-prologue query)
        ]
  [base,
   (u-to-q "<http://example.com/blah>")
   (q-to-u  "eg:blah")
   (u-to-q "blah")
   (q-to-u "blah")
   ])
   
["http://example.org/"
 "eg:blah"
 "http://example.com/blah"
 "blah"
 "blah"
 ]
>

xsd-type-uri

It is possible to get the URI string for xsd types using the same Jena library that parses them. This works for most of the standard types:

> (sparql/xsd-type-uri 1)
"http://www.w3.org/2001/XMLSchema#long"
>
> (sparql/xsd-type-uri 1.0)
"http://www.w3.org/2001/XMLSchema#double"
>
> (sparql/xsd-type-uri #inst "2020-02-14")
"http://www.w3.org/2001/XMLSchema#dateTime"
>

License

Copyright © 2020-3 Eric D. Scott

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.

Natural Lexicon logo

Natural Lexicon logo - Copyright © 2020 Eric D. Scott. Artwork by Athena M. Scott.

Released under Creative Commons Attribution-ShareAlike 4.0 International license. Under the terms of this license, if you display this logo or derivates thereof, you must include an attribution to the original source, with a link to https://github.com/ont-app, or http://ericdscott.com.

About

Utilities to interact with a SPARQL endpoint in clojure

Resources

License

Stars

Watchers

Forks

Packages

No packages published