From 8fb647b21305e61460193225866e3ad31c7fca35 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Fran=C3=A7ois=20Prunayre?= <fx.prunayre@gmail.com>
Date: Mon, 7 Oct 2024 10:30:49 +0200
Subject: [PATCH] Thesaurus / Date improvements. (#8392)

* Thesaurus / Date parsing fix.

https://github.com/geonetwork/core-geonetwork/pull/6972/files#diff-c7548d94bdb4268915f50f66df5a90f1f7868e4fb1de2f3d2938397e202aaf2fR1057
added month and year format support. This was resolving thesaurus date
to the last matching format ie. 1st January for the default date.

This can be tested using regions thesaurus which will contains 2 dates
for the thesaurus block.

* Thesaurus / Avoid having multiple dates with same type code

https://github.com/geonetwork/core-geonetwork/pull/6972 adds retrieval of `dct:issued|modified|created` dates but multiple dates could be added more than one time with same date type code. Improve the mapping.

* Thesaurus / INSPIRE registry / Only keep one date to avoid validation errors

* Thesaurus / LD registry registry / Only keep one date

No need for duplicated date
---
 .../java/org/fao/geonet/kernel/Thesaurus.java |  11 +-
 .../convert/thesaurus-transformation.xsl      | 135 ++++--------------
 .../convert/thesaurus-transformation.xsl      | 102 +++----------
 .../services/thesaurus/ldregistry-to-skos.xsl |   1 -
 .../services/thesaurus/registry-to-skos.xsl   |   1 -
 5 files changed, 52 insertions(+), 198 deletions(-)

diff --git a/core/src/main/java/org/fao/geonet/kernel/Thesaurus.java b/core/src/main/java/org/fao/geonet/kernel/Thesaurus.java
index efaeaf60a89..91a506b57ab 100644
--- a/core/src/main/java/org/fao/geonet/kernel/Thesaurus.java
+++ b/core/src/main/java/org/fao/geonet/kernel/Thesaurus.java
@@ -1064,12 +1064,11 @@ private Date parseThesaurusDate(Element dateEl) {
 
         StringBuffer errorMsg = new StringBuffer("Error parsing the thesaurus date value: ");
         errorMsg.append(dateVal);
-        boolean success = false;
 
         for (SimpleDateFormat df : dfList) {
             try {
                 thesaurusDate = df.parse(dateVal);
-                success = true;
+                return thesaurusDate;
             } catch (Exception ex) {
                 // Ignore the exception and try next format
                 errorMsg.append("\n  * with format: ");
@@ -1079,11 +1078,9 @@ private Date parseThesaurusDate(Element dateEl) {
             }
         }
         // Report error if no success
-        if (!success) {
-            errorMsg.append("\nCheck thesaurus date in ");
-            errorMsg.append(this.fname);
-            Log.error(Geonet.THESAURUS_MAN, errorMsg.toString());
-        }
+        errorMsg.append("\nCheck thesaurus date in ");
+        errorMsg.append(this.fname);
+        Log.error(Geonet.THESAURUS_MAN, errorMsg.toString());
         return thesaurusDate;
     }
 
diff --git a/schemas/iso19115-3.2018/src/main/plugin/iso19115-3.2018/convert/thesaurus-transformation.xsl b/schemas/iso19115-3.2018/src/main/plugin/iso19115-3.2018/convert/thesaurus-transformation.xsl
index 40e13cd3e65..59a02c55f4d 100644
--- a/schemas/iso19115-3.2018/src/main/plugin/iso19115-3.2018/convert/thesaurus-transformation.xsl
+++ b/schemas/iso19115-3.2018/src/main/plugin/iso19115-3.2018/convert/thesaurus-transformation.xsl
@@ -82,6 +82,12 @@
     <xsl:param name="withThesaurusAnchor"/>
 
     <mri:descriptiveKeywords>
+      <xsl:namespace name="mri" select="'http://standards.iso.org/iso/19115/-3/mri/1.0'"/>
+      <xsl:namespace name="mcc" select="'http://standards.iso.org/iso/19115/-3/mcc/1.0'"/>
+      <xsl:namespace name="cit" select="'http://standards.iso.org/iso/19115/-3/cit/2.0'"/>
+      <xsl:namespace name="gco" select="'http://standards.iso.org/iso/19115/-3/gco/1.0'"/>
+      <xsl:namespace name="gcx" select="'http://standards.iso.org/iso/19115/-3/gcx/1.0'"/>
+      <xsl:namespace name="xlink" select="'http://www.w3.org/1999/xlink'"/>
       <xsl:choose>
         <xsl:when test="$withXlink">
           <xsl:variable name="multiple"
@@ -257,20 +263,22 @@
                               codeListValue="{$thesauri/thesaurus[key = $currentThesaurus]/dname}" />
     </mri:type>
     <xsl:if test="$thesaurusInfo">
+      <xsl:variable name="thesaurus"
+                    select="$thesauri/thesaurus[key = $currentThesaurus]"/>
+
       <xsl:variable name="thesaurusInMainLanguage"
-                    select="$thesauri/thesaurus[key = $currentThesaurus]
-                              /multilingualTitles/multilingualTitle[
+                    select="$thesaurus/multilingualTitles/multilingualTitle[
                                 lang = util:twoCharLangCode($mainLanguage, '')]/title"/>
       <xsl:variable name="thesaurusTitle"
                     select="if ($thesaurusInMainLanguage != '')
                             then $thesaurusInMainLanguage
-                            else $thesauri/thesaurus[key = $currentThesaurus]/title"/>
+                            else $thesaurus/title"/>
       <mri:thesaurusName>
         <cit:CI_Citation>
           <cit:title>
             <xsl:choose>
               <xsl:when test="$withThesaurusAnchor = true()">
-                <gcx:Anchor xlink:href="{$thesauri/thesaurus[key = $currentThesaurus]/defaultNamespace}">
+                <gcx:Anchor xlink:href="{$thesaurus/defaultNamespace}">
                   <xsl:value-of select="$thesaurusTitle"/>
                 </gcx:Anchor>
               </xsl:when>
@@ -282,121 +290,30 @@
             </xsl:choose>
           </cit:title>
 
-          <xsl:variable name="thesaurusDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/date)" />
-          <xsl:variable name="thesaurusCreatedDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/createdDate)"/>
-          <xsl:variable name="thesaurusIssuedDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/issuedDate)"/>
-          <xsl:variable name="thesaurusModifiedDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/modifiedDate)"/>
+          <xsl:variable name="thesaurusDates" as="node()*">
+            <publication date="{normalize-space($thesaurus/date)}"/>
+            <creation date="{normalize-space($thesaurus/createdDate)}"/>
+            <publication date="{normalize-space($thesaurus/issuedDate)}"/>
+            <publication date="{normalize-space($thesaurus/modifiedDate)}"/>
+          </xsl:variable>
 
-          <xsl:if test="$thesaurusDate != ''">
+          <xsl:for-each-group select="$thesaurusDates[@date != '']" group-by="@date">
+            <xsl:sort select="@date" order="descending"/>
             <cit:date>
               <cit:CI_Date>
                 <cit:date>
-                  <xsl:choose>
-                    <xsl:when test="contains($thesaurusDate, 'T')">
-                      <gco:DateTime>
-                        <xsl:value-of select="$thesaurusDate" />
-                      </gco:DateTime>
-                    </xsl:when>
-                    <xsl:otherwise>
-                      <gco:Date>
-                        <xsl:value-of select="$thesaurusDate" />
-                      </gco:Date>
-                    </xsl:otherwise>
-                  </xsl:choose>
+                  <xsl:element name="{if (contains(current-grouping-key(), 'T')) then 'gco:DateTime' else 'gco:Date'}">
+                    <xsl:value-of select="current-grouping-key()" />
+                  </xsl:element>
                 </cit:date>
                 <cit:dateType>
                   <cit:CI_DateTypeCode
-                          codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                          codeListValue="publication" />
+                    codeList="http://standards.iso.org/iso/19115/resources/Codelists/cat/codelists.xml#CI_DateTypeCode"
+                    codeListValue="{name(current-group()[1])}"/>
                 </cit:dateType>
               </cit:CI_Date>
             </cit:date>
-          </xsl:if>
-
-          <!-- Publication Date-->
-          <xsl:choose>
-            <xsl:when test="$thesaurusIssuedDate != ''">
-              <cit:date>
-                <cit:CI_Date>
-                  <cit:date>
-                    <xsl:choose>
-                      <xsl:when test="contains($thesaurusIssuedDate, 'T')">
-                        <gco:DateTime>
-                          <xsl:value-of select="$thesaurusIssuedDate" />
-                        </gco:DateTime>
-                      </xsl:when>
-                      <xsl:otherwise>
-                        <gco:Date>
-                          <xsl:value-of select="$thesaurusIssuedDate" />
-                        </gco:Date>
-                      </xsl:otherwise>
-                    </xsl:choose>
-                  </cit:date>
-                  <cit:dateType>
-                    <cit:CI_DateTypeCode
-                      codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                      codeListValue="publication" />
-                  </cit:dateType>
-                </cit:CI_Date>
-              </cit:date>
-            </xsl:when>
-            <xsl:otherwise>
-              <cit:date>
-                <cit:CI_Date>
-                  <cit:date>
-                    <xsl:choose>
-                      <xsl:when test="contains($thesaurusDate, 'T')">
-                        <gco:DateTime>
-                          <xsl:value-of select="$thesaurusDate" />
-                        </gco:DateTime>
-                      </xsl:when>
-                      <xsl:otherwise>
-                        <gco:Date>
-                          <xsl:value-of select="$thesaurusDate" />
-                        </gco:Date>
-                      </xsl:otherwise>
-                    </xsl:choose>
-                  </cit:date>
-                  <cit:dateType>
-                    <cit:CI_DateTypeCode
-                      codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                      codeListValue="publication" />
-                  </cit:dateType>
-                </cit:CI_Date>
-              </cit:date>
-            </xsl:otherwise>
-          </xsl:choose>
-
-          <!--Creation Date-->
-          <xsl:if test="$thesaurusCreatedDate != ''">
-            <cit:date>
-              <cit:CI_Date>
-                <cit:date>
-                  <xsl:choose>
-                    <xsl:when test="contains($thesaurusCreatedDate, 'T')">
-                      <gco:DateTime>
-                        <xsl:value-of select="$thesaurusCreatedDate" />
-                      </gco:DateTime>
-                    </xsl:when>
-                    <xsl:otherwise>
-                      <gco:Date>
-                        <xsl:value-of select="$thesaurusCreatedDate" />
-                      </gco:Date>
-                    </xsl:otherwise>
-                  </xsl:choose>
-                </cit:date>
-                <cit:dateType>
-                  <cit:CI_DateTypeCode
-                    codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                    codeListValue="creation" />
-                </cit:dateType>
-              </cit:CI_Date>
-            </cit:date>
-          </xsl:if>
+          </xsl:for-each-group>
 
           <xsl:if test="$withThesaurusAnchor">
             <cit:identifier>
diff --git a/schemas/iso19139/src/main/plugin/iso19139/convert/thesaurus-transformation.xsl b/schemas/iso19139/src/main/plugin/iso19139/convert/thesaurus-transformation.xsl
index a27b85bcb18..9e3103d71c4 100644
--- a/schemas/iso19139/src/main/plugin/iso19139/convert/thesaurus-transformation.xsl
+++ b/schemas/iso19139/src/main/plugin/iso19139/convert/thesaurus-transformation.xsl
@@ -136,6 +136,10 @@
     <xsl:param name="withThesaurusAnchor"/>
 
     <gmd:descriptiveKeywords>
+      <xsl:namespace name="gmd" select="'http://www.isotc211.org/2005/gmd'"/>
+      <xsl:namespace name="gco" select="'http://www.isotc211.org/2005/gco'"/>
+      <xsl:namespace name="gmx" select="'http://www.isotc211.org/2005/gmx'"/>
+      <xsl:namespace name="xlink" select="'http://www.w3.org/1999/xlink'"/>
       <xsl:choose>
         <xsl:when test="$withXlink">
           <xsl:variable name="isLocalXlink"
@@ -314,22 +318,24 @@
         codeListValue="{$thesauri/thesaurus[key = $currentThesaurus]/dname}"/>
     </gmd:type>
     <xsl:if test="$thesaurusInfo">
+      <xsl:variable name="thesaurus"
+                    select="$thesauri/thesaurus[key = $currentThesaurus]"/>
 
       <xsl:variable name="thesaurusInMainLanguage"
-                    select="$thesauri/thesaurus[key = $currentThesaurus]
+                    select="$thesaurus
                               /multilingualTitles/multilingualTitle[
                                 lang = util:twoCharLangCode($mainLanguage, '')]/title"/>
       <xsl:variable name="thesaurusTitle"
                     select="if ($thesaurusInMainLanguage != '')
                             then $thesaurusInMainLanguage
-                            else $thesauri/thesaurus[key = $currentThesaurus]/title"/>
+                            else $thesaurus/title"/>
 
       <gmd:thesaurusName>
         <gmd:CI_Citation>
           <gmd:title>
             <xsl:choose>
               <xsl:when test="$withTitleAnchor = true()">
-                <gmx:Anchor xlink:href="{$thesauri/thesaurus[key = $currentThesaurus]/defaultNamespace}">
+                <gmx:Anchor xlink:href="{$thesaurus/defaultNamespace}">
                   <xsl:value-of select="$thesaurusTitle"/>
                 </gmx:Anchor>
               </xsl:when>
@@ -341,95 +347,31 @@
             </xsl:choose>
           </gmd:title>
 
-          <xsl:variable name="thesaurusDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/date)"/>
-          <xsl:variable name="thesaurusCreatedDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/createdDate)"/>
-          <xsl:variable name="thesaurusIssuedDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/issuedDate)"/>
-          <xsl:variable name="thesaurusModifiedDate"
-                        select="normalize-space($thesauri/thesaurus[key = $currentThesaurus]/modifiedDate)"/>
 
-          <!-- Publication Date-->
-          <xsl:choose>
-            <xsl:when test="$thesaurusIssuedDate != ''">
-              <gmd:date>
-                <gmd:CI_Date>
-                  <gmd:date>
-                    <xsl:choose>
-                      <xsl:when test="contains($thesaurusIssuedDate, 'T')">
-                        <gco:DateTime>
-                          <xsl:value-of select="$thesaurusIssuedDate"/>
-                        </gco:DateTime>
-                      </xsl:when>
-                      <xsl:otherwise>
-                        <gco:Date>
-                          <xsl:value-of select="$thesaurusIssuedDate"/>
-                        </gco:Date>
-                      </xsl:otherwise>
-                    </xsl:choose>
-                  </gmd:date>
-                  <gmd:dateType>
-                    <gmd:CI_DateTypeCode
-                      codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                      codeListValue="publication"/>
-                  </gmd:dateType>
-                </gmd:CI_Date>
-              </gmd:date>
-            </xsl:when>
-            <xsl:otherwise>
-              <gmd:date>
-                <gmd:CI_Date>
-                  <gmd:date>
-                    <xsl:choose>
-                      <xsl:when test="contains($thesaurusDate, 'T')">
-                        <gco:DateTime>
-                          <xsl:value-of select="$thesaurusDate"/>
-                        </gco:DateTime>
-                      </xsl:when>
-                      <xsl:otherwise>
-                        <gco:Date>
-                          <xsl:value-of select="$thesaurusDate"/>
-                        </gco:Date>
-                      </xsl:otherwise>
-                    </xsl:choose>
-                  </gmd:date>
-                  <gmd:dateType>
-                    <gmd:CI_DateTypeCode
-                      codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                      codeListValue="publication"/>
-                  </gmd:dateType>
-                </gmd:CI_Date>
-              </gmd:date>
-            </xsl:otherwise>
-          </xsl:choose>
+          <xsl:variable name="thesaurusDates" as="node()*">
+            <publication date="{normalize-space($thesaurus/date)}"/>
+            <creation date="{normalize-space($thesaurus/createdDate)}"/>
+            <publication date="{normalize-space($thesaurus/issuedDate)}"/>
+            <publication date="{normalize-space($thesaurus/modifiedDate)}"/>
+          </xsl:variable>
 
-          <!--Creation Date-->
-          <xsl:if test="$thesaurusCreatedDate != ''">
+          <xsl:for-each-group select="$thesaurusDates[@date != '']" group-by="@date">
+            <xsl:sort select="@date" order="descending"/>
             <gmd:date>
               <gmd:CI_Date>
                 <gmd:date>
-                  <xsl:choose>
-                    <xsl:when test="contains($thesaurusCreatedDate, 'T')">
-                      <gco:DateTime>
-                        <xsl:value-of select="$thesaurusCreatedDate"/>
-                      </gco:DateTime>
-                    </xsl:when>
-                    <xsl:otherwise>
-                      <gco:Date>
-                        <xsl:value-of select="$thesaurusCreatedDate"/>
-                      </gco:Date>
-                    </xsl:otherwise>
-                  </xsl:choose>
+                  <xsl:element name="{if (contains(current-grouping-key(), 'T')) then 'gco:DateTime' else 'gco:Date'}">
+                    <xsl:value-of select="current-grouping-key()" />
+                  </xsl:element>
                 </gmd:date>
                 <gmd:dateType>
                   <gmd:CI_DateTypeCode
                     codeList="http://standards.iso.org/iso/19139/resources/gmxCodelists.xml#CI_DateTypeCode"
-                    codeListValue="creation"/>
+                    codeListValue="{name(current-group()[1])}"/>
                 </gmd:dateType>
               </gmd:CI_Date>
             </gmd:date>
-          </xsl:if>
+          </xsl:for-each-group>
 
           <!--
               You can pull in the publisher from the Thesaurus XML.  See Metadata101/iso19139.ca.HNAP
diff --git a/web/src/main/webapp/xslt/services/thesaurus/ldregistry-to-skos.xsl b/web/src/main/webapp/xslt/services/thesaurus/ldregistry-to-skos.xsl
index 1cec16ae730..02cdb7aa395 100644
--- a/web/src/main/webapp/xslt/services/thesaurus/ldregistry-to-skos.xsl
+++ b/web/src/main/webapp/xslt/services/thesaurus/ldregistry-to-skos.xsl
@@ -71,7 +71,6 @@
 
 
         <dcterms:issued><xsl:value-of select="if ($thesaurusDate != '') then $thesaurusDate else $now"/></dcterms:issued>
-        <dcterms:modified><xsl:value-of select="if ($thesaurusDate != '') then $thesaurusDate else $now"/></dcterms:modified>
 
         <xsl:for-each select="distinct-values($concepts/*[skos:narrower and not(skos:broader)]/@rdf:about)">
           <skos:hasTopConcept rdf:resource="{.}"/>
diff --git a/web/src/main/webapp/xslt/services/thesaurus/registry-to-skos.xsl b/web/src/main/webapp/xslt/services/thesaurus/registry-to-skos.xsl
index 10ee275efd3..8f12f4bedbc 100644
--- a/web/src/main/webapp/xslt/services/thesaurus/registry-to-skos.xsl
+++ b/web/src/main/webapp/xslt/services/thesaurus/registry-to-skos.xsl
@@ -126,7 +126,6 @@
 
 
         <dcterms:issued><xsl:value-of select="if ($thesaurusDate != '') then $thesaurusDate else $now"/></dcterms:issued>
-        <dcterms:modified><xsl:value-of select="if ($thesaurusDate != '') then $thesaurusDate else $now"/></dcterms:modified>
 
         <!-- Add top concepts for all items with no parent -->
         <xsl:if test="$hasBroaderNarrowerLinks">