Feature/geodata

Makefile

@@ -1,3 +1,4 @@
+
 source/cc-licenses.ttl: source/construct-cc-license-data.rq
 	cat source/cc-licenses-header.txt > $@
 	(echo https://creativecommons.org/publicdomain/mark/1.0/ && echo https://creativecommons.org/publicdomain/mark/1.0/deed.sv && curl -sL https://creativecommons.org/about/cclicenses/ | sed -nE 's!.*<a href="([^"]+)"><img .*!\1\n\1deed.sv!p') | xargs python3 scripts/construct.py $^ >> $@
@@ -8,3 +9,10 @@ source/sab.ttl: scripts/extract_sab_data_from_docx.py cache/esab-2015_1.docx
 	# ../librisxl/whelk-core/src/main/java/se/kb/libris/export/dewey/dewey_sab.txt
 	# TODO 2: In XL, add precomposed usages (extract from usage in records)? See:
 	# ../librisxl/marc_export/src/main/resources/se/kb/libris/export/sabrub.txt # precomposed
+
+cache/geocore.ttl: cache/remote-geocore.ttl
+	cat $^ | scripts/fmt.sh > $@
+	#scripts/construct.py source/geo/modify-geocore.rq $^ > $@
+
+cache/remote-geocore.ttl: source/geo/construct-geocore.rq
+	scripts/rq.sh https://query.wikidata.org/sparql $^ > $@

lxltools/datacompiler.py

@@ -178,8 +178,12 @@ def _compile_dataset(self, name, result):
 
             meta = node.pop('meta', None)
             if meta:
+                if meta.get('@id', "").startswith('_:'):
+                    del meta['@id']
+
                 if 'created' in meta:
                     created_ms = timeutil.w3c_dtz_to_ms(meta.pop('created'))
+
                 if 'modified' in meta:
                     modified_ms = timeutil.w3c_dtz_to_ms(meta.pop('modified'))
 

lxltools/timeutil.py

@@ -25,4 +25,4 @@ def to_w3c_dtz(ms: float):
 
 
 def to_http_date(s: float):
-    return datetime.utcfromtimestamp(s).strftime('%a, %d %b %Y %H:%M:%S GMT')
+    return datetime.fromtimestamp(s, tz=timezone.utc).strftime('%a, %d %b %Y %H:%M:%S GMT')

scripts/fmt.sh

@@ -0,0 +1,2 @@
+#!/bin/bash
+trld -ittl -e -f -c $(dirname $0)/../build/sys/context/kbv.jsonld -B -ottl | sed 's/rdf:type/a/'

scripts/rq.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+endpoint=$1
+queryfile=$2
+
+curl -s $endpoint -HAccept:text/turtle --data-urlencode "query@$queryfile"

source/construct-libraries.rq

@@ -8,85 +8,86 @@ base <https://libris.kb.se/library/>
 construct {
 
     ?library a ?libtype ;
-        :meta _:meta ;
+        :meta ?meta ;
         owl:sameAs ?bdb_id, ?sameas ;
         :sigel ?sigel ;
         :name ?name ;
         :url ?url ;
         :qualifier ?dept ;
-        :category ?registranturl .
-        #:place ?place .
-
-    _:meta :created ?created ; :modified ?modified .
-
-    # TODO: Model a proper place/adress relation.
-    # ?place a :Place ;
-    #     :label ?region ;
-    #     :code ?municipality_code ;
-    #     :isPartOf ?country ;
-    #     :latitude ?lat ;
-    #     :longitude ?long .
-    # 
-    # ?country a :Country ;
-    #     :code ?country_code .
+        :category ?registranturl ;
+        :locatedIn ?municipality ;
+        :geo ?geo ;
+        :country ?country .
+
+    ?meta :created ?created ; :modified ?modified .
+
+    ?geo :latitude ?lat ; :longitude ?long .
 
 } where {
 
     ?bdb_id
         bibdb:sigel ?sigel ;
         sdo:name ?name .
-        # :organisation, sdo:address
 
-    optional {
-        ?bdb_id bibdb:libris_reg ?reg . filter ( ?reg = true )
-        bind (iri('https://id.kb.se/term/bibdb/Registrant') as ?registranturl)
-    }
+    bind(encode_for_uri(replace(str(?sigel), "\\s+", "")) as ?sigelslug)
 
-    optional {
-        ?bdb_id bibdb:dept ?dept
-    }
+    optional { ?bdb_id a ?type }
+    bind(if(?type = sdo:Library, :Library, :Bibliography) as ?libtype)
 
-    optional {
-        ?bdb_id bibdb:date_created ?raw_created .
-        bind(concat(?raw_created, '.000Z') as ?created)
-    }
-    optional {
-        ?bdb_id bibdb:date_modified ?raw_modified .
-        bind(concat(?raw_modified, '.000Z') as ?modified)
-    }
+    bind(iri(concat(str(</library/>), ?sigelslug)) as ?library)
 
-    bind(encode_for_uri(replace(str(?sigel), "\\s+", "")) as ?sigelslug)
+    {
 
-    optional { ?bdb_id a ?type }
+        optional { ?bdb_id :organisation ?org }
+        optional { ?bdb_id bibdb:dept ?dept }
+        optional {
+            ?bdb_id bibdb:libris_reg ?reg . filter ( ?reg = true )
+            bind (<https://id.kb.se/term/bibdb/Registrant> as ?registranturl)
+        }
+
+        optional {
+            ?bdb_id bibdb:date_created ?raw_created .
+            bind(concat(?raw_created, '.000Z') as ?created)
+        }
+        optional {
+            ?bdb_id bibdb:date_modified ?raw_modified .
+            bind(concat(?raw_modified, '.000Z') as ?modified)
+        }
+        bind(bnode(coalesce(?created, ?modified)) as ?meta)
+
+        optional {
+            ?bdb_id sdo:url ?url .
+            FILTER(?url != "" && ?url != "http://")
+        }
+
+        optional {
+          ?bdb_id bibdb:country_code ?country_code .
+          FILTER(?country_code != "")
+        }
+        # TODO: tr, fi, ...
+        bind(if(?country_code = 'se', <https://id.kb.se/country/sw>, ?NO_country) as ?country)
+
+    } union {
+
+        ?bdb_id sdo:address [
+            sdo:streetAddress ?streetAddress ;
+            sdo:addressLocality ?city ;
+            sdo:postalCode ?zipCode
+            ] .
+        FILTER(!STRSTARTS(?streetAddress, "FE ") && !STRSTARTS(?streetAddress, "FE "))
+
+    } union {
+
+        ?bdb_id bibdb:municipality_code ?municipality_code .
+        FILTER(?municipality_code not in ('', '-'))
+        bind(IRI(CONCAT(STR(</dataset/geo/sw/municipality/>), ?municipality_code)) as ?municipality)
+
+    } union {
+
+        ?bdb_id sdo:latitude ?lat ; sdo:longitude ?long .
+        FILTER(?lat > 0 && ?long > 0)
+        bind(concat('POINT(', STR(?long), ' ', STR(?lat), ')') as ?geo)
 
-    optional {
-        ?bdb_id sdo:url ?url .
-        FILTER(?url != "" && ?url != "http://")
     }
-    bind(if(?type = sdo:Library, :Library, :Bibliography) as ?libtype)
 
-    # optional {
-    #     ?bdb_id bibdb:country_code ?country_code .
-    #     FILTER(?country_code != "")
-    # }
-    # 
-    # optional {
-    #     ?bdb_id bibdb:municipality_code ?municipality_code .
-    #     FILTER(?municipality_code != "")
-    # }
-    # 
-    # optional {
-    #     ?bdb_id bibdb:region ?region
-    # }
-    # 
-    # optional {
-    #     ?bdb_id sdo:latitude ?lat ; sdo:longitude ?long .
-    #     FILTER(?lat > 0 && ?long > 0)
-    # }
-
-    # TODO: coalesce should not be necessary here, due to the else clause. RDFLib bug?
-    bind(iri(concat(str(coalesce(?uribase, </library/>)), ?sigelslug)) as ?library)
-
-    # bind(if(bound(?region) || bound(?lat), bnode(), ?NO_place) as ?place)
-    # bind(if(bound(?place) && bound(?country_code), bnode(), ?NO_country) as ?country)
 }

source/datasets/geo.ttl

@@ -0,0 +1,8 @@
+prefix xsd: <http://www.w3.org/2001/XMLSchema#>
+prefix : <https://id.kb.se/vocab/>
+base <https://libris.kb.se/dataset/>
+
+<geo/core> a :Dataset ;
+    :sourceData [ :uri "cache/geocore.ttl" ] ;
+    :uriSpace "https://libris.kb.se/dataset/geo/" ;
+    :created "2024-03-07T12:09:24Z"^^xsd:dateTime .

source/geo/construct-geocore.rq

@@ -0,0 +1,86 @@
+BASE <https://libris.kb.se/dataset/geo/sw/>
+PREFIX : <https://id.kb.se/vocab/>
+PREFIX idkbse: <https://id.kb.se/>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX wdt: <http://www.wikidata.org/prop/direct/>
+PREFIX wd: <http://www.wikidata.org/entity/>
+PREFIX p: <http://www.wikidata.org/prop/>
+PREFIX psv: <http://www.wikidata.org/prop/statement/value/>
+PREFIX wikibase: <http://wikiba.se/ontology#>
+
+CONSTRUCT {
+
+    ?type a :Concept ; :label ?typelabel .
+
+    ?place a :Place ;
+        :sameAs ?sameAs ;
+        :label ?label ;
+        :category ?type ;
+        :code ?swedishMunicipalityCode ;
+        :geo ?geo ;
+        :locatedIn ?locatedIn ;
+        :principalPlace ?capital ;
+        :country <https://id.kb.se/country/sw> ;
+        :startDate ?foundingDate ;
+        :image ?image .
+
+    ?geo a :EarthGlobeCoordinates ;
+      :value ?geocode ;
+      :geoLatitude ?lat ;
+      :geoLongitude ?long ;
+      :geoPrecision ?geoPrecision .
+
+} WHERE {
+    ?wdplace a|wdt:P31 ?type ;
+        wdt:P17 wd:Q34 .  # :country Sweden
+
+    FILTER EXISTS {
+      {
+        ?type wdt:P279* wd:Q914262  # administrative territorial entity of Sweden
+      } UNION {
+        ?type wdt:P279* wd:Q12813115  # urban area of Sweden
+      }
+    }
+
+    ?wdplace rdfs:label ?label .
+
+    FILTER NOT EXISTS { ?wdplace wdt:P576 ?dissolutionDate }
+    OPTIONAL { ?wdplace wdt:P571 ?foundingDate }
+
+    ?type rdfs:label ?typelabel . FILTER(lang(?typelabel) IN ('sv', 'en'))
+
+    FILTER(lang(?label) IN ('sv', 'en'))
+
+    OPTIONAL {
+      ?wdplace wdt:P625 ?geocode .  # ^^:wktLiteral
+      OPTIONAL {
+        ?wdplace p:P625 ?qualifiedgeo .
+        ?qualifiedgeo psv:P625 [
+            wikibase:geoLatitude ?lat ;
+            wikibase:geoLongitude ?long ;
+            wikibase:geoPrecision ?geoPrecision ] .
+      }
+      BIND(BNODE(STR(COALESCE(?qualifiedgeo, ?geocode))) AS ?geo)
+    }
+
+    ?wdplace wdt:P131 ?wdLocatedIn .
+    OPTIONAL {
+      #?wdplace wdt:P525 ?swedishMunicipalityCode .
+      ?wdplace p:P525 ?qualifiedP525 .
+      ?qualifiedP525 ps:P525 ?swedishMunicipalityCode .
+      FILTER NOT EXISTS { ?qualifiedP525 pq:P582 ?endDate }
+      BIND(IRI(CONCAT(STR(<municipality/>), ENCODE_FOR_URI(?swedishMunicipalityCode))) AS ?dsId)
+      BIND(?wdplace AS ?sameAs)
+    }
+
+    OPTIONAL {
+      ?wdplace wdt:P36 ?capital .
+      # :capital :label "centralort"@sv
+      # (Cf. wdt:P1376 == huvudstad.)
+    }
+
+    BIND(IF(?wdLocatedIn = wd:Q34, idkbse:country\/sw, ?wdLocatedIn) AS ?locatedIn)
+    BIND(IF(BOUND(?dsId), ?dsId, ?wdplace) AS ?place)
+
+    OPTIONAL { ?wdplace wdt:P18 ?image }
+}

source/vocab/details.ttl

@@ -21,6 +21,8 @@
 @prefix rdaz: <http://rdaregistry.info/Elements/z/> .
 @prefix rdau: <http://rdaregistry.info/Elements/u/> .
 
+@prefix wikibase: <http://wikiba.se/ontology#> .
+
 @prefix : <https://id.kb.se/vocab/> .
 @prefix kbrel: <https://id.kb.se/relator/> .
 
@@ -1539,6 +1541,31 @@
     rdfs:label "Geographic coverage"@en, "Geografisk täckning"@sv;
     owl:equivalentClass bf2:GeographicCoverage .
 
+:geo a owl:ObjectProperty;
+    rdfs:label "Geographic coordinates"@en, "Geografiska koordinater"@sv;
+    sdo:domainIncludes :Place, :Library;
+    rdfs:range :GlobeCoordinates;
+    owl:equivalentProperty sdo:geo .
+
+:GlobeCoordinates a owl:Class ;
+  rdfs:subClassOf sdo:GeoCoordinates, wikibase:GlobecoordinateValue .
+
+:geoLatitude owl:equivalentProperty sdo:latitude, wikibase:geoLatitude .
+:geoLongitude owl:equivalentProperty sdo:longitude, wikibase:geoLongitude .
+:geoPrecision owl:equivalentProperty wikibase:geoPrecision .
+
+:EarthGlobeCoordinates a owl:Class ;
+    rdfs:label "Earth globe coordinates"@en, "jordglobskoordinater"@sv;
+    rdfs:subClassOf :GlobeCoordinates, [ a owl:Restriction;
+        owl:onProperty wikibase:geoGlobe ;
+        owl:hasValue <http://www.wikidata.org/entity/Q2> ] .
+
+:principalPlace a owl:ObjectProperty;
+    rdfs:label "Principal place"@en, "Centralort"@sv;
+    rdfs:domain :Place;
+    rdfs:range :Place .
+    #owl:equivalentProperty wdt:P36 .
+
 :geographicCoverage a owl:ObjectProperty;
     rdfs:label "Geographic coverage"@en, "Geografisk täckning"@sv;
     rdfs:domain :Creation;

source/vocab/display.jsonld

@@ -822,7 +822,7 @@
           "@id": "Place-cards",
           "@type": "fresnel:Lens",
           "classLensDomain": "Place",
-          "showProperties": [ "prefLabel", "locatedIn", "country", "description" ]
+          "showProperties": [ {"alternateProperties": ["prefLabel", "label"]}, "locatedIn", "country", "description" ]
         },
         "DescriptionConventions": {
           "@id": "DescriptionConventions-cards",