Provide CSW data as DCAT-AP.de RDF XML catalog

GovDataOfficial · Jul 17, 2020 · 7ff7a18 · 7ff7a18
1 parent 3772f3c
commit 7ff7a18
Show file tree

Hide file tree

Showing 4 changed files with 346 additions and 0 deletions.
diff --git a/src/main/resources/camel-oai-pmh.properties b/src/main/resources/camel-oai-pmh.properties
@@ -1,4 +1,5 @@
 harvester.base.url=${oai-pmh.base.url.external}/harvesters/
+rdf.catalog.base.url=${oai-pmh.base.url.external:http:https://localhost:8080/omdf}/gp-csw/catalog.rdf
 db.item.csw.TYPE=${db.item.csw.TYPE}
 db.item.csw.URL=${db.item.csw.URL}
 db.item.ckan.TYPE=${db.item.ckan.TYPE}

diff --git a/src/main/resources/camel-oai-pmh.xml b/src/main/resources/camel-oai-pmh.xml
@@ -31,10 +31,14 @@
 
  <!-- internal endpoint of processing route for csw catalogs -->
  <endpoint id="csw" uri="direct:csw"/>
+ <!-- internal endpoint of processing route for csw catalogs into DCAT catalogs (instead of OAI-PMH) -->
+ <endpoint id="catalog" uri="direct:catalog"/>
  <!-- internal endpoint of processing route for CKAN catalogs -->
  <endpoint id="ckan" uri="direct:ckan"/>
  <!-- internal endpoint of processing route for INSPIRE catalogs -->
  <endpoint id="inspire" uri="direct:inspire"/>
+ <!-- internal endpoint of processing route for INSPIRE catalogs into DCAT catalogs -->
+ <endpoint id="catalog-inspire" uri="direct:catalog-inspire"/>
  <!-- internal endpoint of processing route for INSPIRE catalogs via SOAP 1.1 -->
  <endpoint id="inspireSoap11" uri="direct:inspireSoap11"/>
  <!-- internal endpoint of processing route for INSPIRE catalogs via SOAP 1.2 -->
@@ -48,6 +52,7 @@
  <endpoint id="cswIds" uri="servlet:https:///gp-csw/listIdentifiers?matchOnUriPrefix=true"/>
  <endpoint id="cswRecords" uri="servlet:https:///gp-csw/listRecords?matchOnUriPrefix=true"/>
  <endpoint id="cswRecord" uri="servlet:https:///gp-csw/getRecord?matchOnUriPrefix=true"/>
+ <endpoint id="dcatCatalog" uri="servlet:https:///gp-csw/catalog.rdf"/>
 
  <!-- endpoint for harvester nrw ckan catalog -->
  <endpoint id="ckanHarvesters" uri="servlet:https:///gp-ckan?matchOnUriPrefix=true"/>
@@ -76,6 +81,8 @@
  <setHeader headerName="response_date">
  <simple>${date:now:yyyy-MM-dd'T'HH:mm:ss'Z'}</simple>
  </setHeader>
+ <removeHeader headerName="page"/>
+ <removeHeader headerName="cookie"/>
  <!-- we need some element for xsl input, but the result only depends on header (request) parameters -->
  <setBody>
  <constant>&lt;dummy/&gt;</constant>
@@ -215,6 +222,69 @@
  <to uri="direct:getCoupledServices"/>
  </route>
 
+ <!-- route for INSPIRE catalogs output as RDF catalog -->
+ <route>
+ <from ref="catalog-inspire"/>
+ <setHeader headerName="response_date">
+ <simple>${date:now:yyyy-MM-dd'T'HH:mm:ss'Z'}</simple>
+ </setHeader>
+ <setHeader headerName="baseCatalogUri">
+ <simple>{{rdf.catalog.base.url}}</simple>
+ </setHeader>
+ <removeHeader headerName="resumptionToken"/>
+ <removeHeader headerName="cookie"/>
+ <!-- we need some element for xsl input, but the result only depends on header (request) parameters -->
+ <setBody>
+ <constant>&lt;dummy/&gt;</constant>
+ </setBody>
+ <to uri="xslt:inspireRequest.xsl"/>
+ <to uri="log:input"/>
+ <choice>
+ <when>
+ <!-- Only send a GetRecords request to the target catalog if we have not already created an OAI-PMH response -->
+ <xpath>count(/oai:OAI-PMH) = 0</xpath>
+ <setHeader headerName="CamelHttpMethod">
+ <constant>POST</constant>
+ </setHeader>
+ <choice>
+ <when>
+ <simple>${header.soapVersion} == '1.1'</simple>
+ <setHeader headerName="Content-Type">
+ <constant>text/xml</constant>
+ </setHeader>
+ </when>
+ <when>
+ <simple>${header.soapVersion} == '1.2'</simple>
+ <setHeader headerName="Content-Type">
+ <constant>application/soap+xml</constant>
+ </setHeader>
+ </when>
+ <otherwise>
+ <setHeader headerName="Content-Type">
+ <constant>application/xml</constant>
+ </setHeader>
+ </otherwise>
+ </choice>
+ <!-- Replaced dynamically with the target catalog URL -->
+ <to uri="http:https://dummy?httpClient.soTimeout=180000"/>
+ <!-- transform the GetRecordsResponse -->
+ <choice>
+ <when>
+ <simple>${header.out} == null</simple>
+ <to uri="xslt:inspireResponseDcatCatalog.xsl?uriResolverFactory=uriResolverFactory"/>
+ </when>
+ <when>
+ <simple>${header.out} != 'csw'</simple>
+ <to uri="xslt:inspireResponseDcatCatalog.xsl?uriResolverFactory=uriResolverFactory"/>
+ </when>
+ </choice>
+ </when>
+ </choice>
+ <setHeader headerName="Content-Type">
+ <constant>application/rdf+xml</constant>
+ </setHeader>
+ </route>
+
  <!-- management route for harvester types -->
  <route>
  <from ref="mgmtHarvesterTypes"/>
@@ -248,6 +318,30 @@
  </recipientList>
  </route>
 
+ <route>
+ <from ref="catalog"/>
+ <setHeader headerName="CamelHttpUri">
+ <simple>{{db.item.csw.URL}}</simple>
+ </setHeader>
+ <setHeader headerName="CamelHttpPath">
+ <constant/>
+ </setHeader>
+ <choice>
+ <when>
+ <simple>${header.metadataPrefix} == null</simple>
+ <setHeader headerName="metadataPrefix">
+ <simple>dcat_ap</simple>
+ </setHeader>
+ </when>
+ </choice>
+ <recipientList>
+ <simple>catalog-inspire</simple>
+ <!-- If SOAP-specific settings are needed, add endpoints above and replace with:
+ <simple>catalog-{{db.item.csw.TYPE}}</simple>
+ -->
+ </recipientList>
+ </route>
+
  <route>
  <from ref="cswHarvesters"/>
  <to ref="csw"/>
@@ -277,6 +371,14 @@
  <to ref="csw"/>
  </route>
 
+ <route>
+ <from ref="dcatCatalog"/>
+ <setHeader headerName="verb">
+ <simple>ListRecords</simple>
+ </setHeader>
+ <to ref="catalog"/>
+ </route>
+
  <route>
  <from ref="ckan"/>
  <setHeader headerName="response_date">

diff --git a/src/main/resources/inspireRequest.xsl b/src/main/resources/inspireRequest.xsl
@@ -17,6 +17,7 @@
  <xsl:param name="until"/>
  <xsl:param name="set"/>
  <xsl:param name="resumptionToken"/>
+ <xsl:param name="page"/>
  <xsl:param name="soapVersion"/>
 
  <xsl:param name="response_date"/>
@@ -137,6 +138,14 @@
  </xsl:otherwise>
  </xsl:choose>
  </xsl:when>
+ <xsl:when test="$page and number($page) > 1">
+ <xsl:variable name="startPosition" select="((number($page) - 1) * $recordsPerPage) + 1"/>
+ <xsl:call-template name="getRecords">
+ <xsl:with-param name="startPosition" select="$startPosition"/>
+ <xsl:with-param name="fromParam" select="$from"/>
+ <xsl:with-param name="untilParam" select="$until"/>
+ </xsl:call-template>
+ </xsl:when>
  <xsl:otherwise>
  <xsl:call-template name="getRecords">
  <xsl:with-param name="fromParam" select="$from"/>