check online resource url for duplications (#369)

* check online resource url for duplications * Revert "check online resource url for duplications" This reverts commit 8f9510d * refactor resource duplication check by xsl in schematron * change error message * Error message with lang, url information * review * update getDuplicatedResource validation function
metadata101 · Apr 19, 2024 · 002796e · 002796e
1 parent d5798a0
commit 002796e
Show file tree

Hide file tree

Showing 3 changed files with 40 additions and 4 deletions.
diff --git a/src/main/plugin/iso19139.ca.HNAP/loc/eng/schematron-rules-common.xml b/src/main/plugin/iso19139.ca.HNAP/loc/eng/schematron-rules-common.xml
@@ -56,7 +56,6 @@
   <ServiceNamespace>srv:SV_ServiceIdentification is misplaced for gmd:MD_ScopeCode: </ServiceNamespace>
   <MissingContactMail>Contact - Electronic mail address is required</MissingContactMail>
 
-  <requireCitation>Citation</requireCitation>
-  <alert.requiredCitation>Data Identification Citation is a required element</alert.requiredCitation>
+  <hasDuplicatedOnlineResource>Duplicated Online Resource URLs for the same language are not allowed. The following duplicate(s) (language; URL) were identified:</hasDuplicatedOnlineResource>
 
 </strings>
diff --git a/src/main/plugin/iso19139.ca.HNAP/loc/fre/schematron-rules-common.xml b/src/main/plugin/iso19139.ca.HNAP/loc/fre/schematron-rules-common.xml
@@ -56,7 +56,6 @@
   <ServiceNamespace>srv:SV_ServiceIdentification est mal placé pour gmd:MD_ScopeCode: </ServiceNamespace>
   <MissingContactMail>Contact - Adresse de courrier électronique obligatoire</MissingContactMail>
 
-  <requireCitation>Citation</requireCitation>
-  <alert.requiredCitation>La citation d’identification des données est un élément obligatoire</alert.requiredCitation>
+  <hasDuplicatedOnlineResource>Les URL de ressources en ligne dupliquées pour la même langue ne sont pas autorisées. Les doublons suivants (langue ; URL) ont été identifiés :</hasDuplicatedOnlineResource>
 
 </strings>
diff --git a/src/main/plugin/iso19139.ca.HNAP/schematron/schematron-rules-common.sch b/src/main/plugin/iso19139.ca.HNAP/schematron/schematron-rules-common.sch
@@ -57,6 +57,37 @@
     </xsl:for-each>
   </xsl:function>
 
+
+  <xsl:function xmlns:sch="http://purl.oclc.org/dsdl/schematron"
+                  name="geonet:getDuplicateResources"
+                  as="xs:string">
+        <xsl:param name="elements" as="element()*"/>
+
+        <!-- Find groups of elements with duplicates -->
+        <xsl:variable name="duplicateResourcesElements">
+            <xsl:for-each-group select="$elements"
+                                group-by="concat(gmd:linkage/gmd:URL/text(), '|', tokenize(gmd:description/gco:CharacterString/text(), ';')[3])">
+                <xsl:if test="count(current-group()) &gt; 1">
+                    <xsl:sequence select="."/>
+                </xsl:if>
+            </xsl:for-each-group>
+        </xsl:variable>
+
+        <!-- Iterate over duplicate groups and generate the list of duplicates -->
+        <xsl:variable name="duplicateResources">
+            <xsl:for-each select="$duplicateResourcesElements/*">
+                    <xsl:if test="position() &gt; 1">
+                        <xsl:text>, </xsl:text>
+                    </xsl:if>
+                    <xsl:value-of select="concat(' ', tokenize(gmd:description/gco:CharacterString/text(), ';')[3], ';')"/>
+                    <xsl:value-of select="gmd:linkage/gmd:URL/text()"/>
+            </xsl:for-each>
+        </xsl:variable>
+
+        <!-- Return the results -->
+        <xsl:value-of select="$duplicateResources"/>
+  </xsl:function>
+
   <!--- Metadata pattern -->
   <sch:pattern>
     <sch:title>$loc/strings/Metadata</sch:title>
@@ -520,6 +551,8 @@
       <sch:let name="mapRESTCount" value="count(gmd:transferOptions/gmd:MD_DigitalTransferOptions/gmd:onLine[@xlink:role='urn:xml:lang:eng-CAN' and translate(gmd:CI_OnlineResource/gmd:protocol/gco:CharacterString, $uppercase, $smallcase) = 'esri rest: map service']) +
                 count(gmd:transferOptions/gmd:MD_DigitalTransferOptions/gmd:onLine[@xlink:role='urn:xml:lang:fra-CAN' and translate(gmd:CI_OnlineResource/gmd:protocol/gco:CharacterString, $uppercase, $smallcase) = 'esri rest: map service'])" />
 
+      <sch:let name="onlineResources" value="gmd:transferOptions/gmd:MD_DigitalTransferOptions/gmd:onLine/gmd:CI_OnlineResource" />
+
       <sch:assert test="$mapRESTCount &lt;= 2">$loc/strings/MapResourcesRESTNumber</sch:assert>
       <sch:assert test="$mapRESTCount = 0 or $mapRESTCount = 2 or $mapRESTCount &gt; 2">$loc/strings/MapResourcesREST</sch:assert>
 
@@ -528,6 +561,11 @@
 
       <sch:assert test="$mapWMSCount &lt;= 2">$loc/strings/MapResourcesWMSNumber</sch:assert>
       <sch:assert test="$mapWMSCount = 0 or $mapWMSCount = 2 or $mapWMSCount &gt; 2">$loc/strings/MapResourcesWMS</sch:assert>
+
+      <sch:let name="duplicatedResource" value="geonet:getDuplicateResources($onlineResources)" />
+      <sch:let name="locMsg" value="geonet:appendLocaleMessage($loc/strings/hasDuplicatedOnlineResource, $duplicatedResource)" />
+
+      <sch:assert test="not(string($duplicatedResource))">$locMsg</sch:assert>
     </sch:rule>
 
     <!-- Distribution - Format -->