Harvester / Add support for Zenodo / Clean up HTML from abstract.

geonetwork · Jan 6, 2025 · 2510e0e · 2510e0e
1 parent bc00549
commit 2510e0e
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 9 deletions.
diff --git a/schemas/iso19115-3.2018/pom.xml b/schemas/iso19115-3.2018/pom.xml
@@ -34,6 +34,11 @@
       <groupId>org.xmlunit</groupId>
       <artifactId>xmlunit-matchers</artifactId>
     </dependency>
+    <dependency>
+      <groupId>org.jsoup</groupId>
+      <artifactId>jsoup</artifactId>
+      <scope>test</scope>
+    </dependency>
   </dependencies>
 
   <build>

diff --git a/schemas/iso19115-3.2018/src/main/plugin/iso19115-3.2018/convert/fromZenodo.xsl b/schemas/iso19115-3.2018/src/main/plugin/iso19115-3.2018/convert/fromZenodo.xsl
@@ -12,7 +12,7 @@
                 xmlns:gcx="http://standards.iso.org/iso/19115/-3/gcx/1.0"
                 xmlns:xlink="http://www.w3.org/1999/xlink"
                 xmlns:gml="http://www.opengis.net/gml/3.2"
-                xmlns:md="http://www.pangaea.de/MetaData"
+                xmlns:util="java:org.fao.geonet.util.XslUtil"
                 exclude-result-prefixes="#all">
 
   <!--
@@ -176,7 +176,7 @@
 
           <mri:abstract>
             <gco:CharacterString>
-              <xsl:value-of select="metadata/description"/>
+              <xsl:value-of select="util:html2text(metadata/description, true())"/>
             </gco:CharacterString>
           </mri:abstract>
 
@@ -296,7 +296,7 @@
                 <mri:associationType>
                   <mri:DS_AssociationTypeCode
                     codeList="http://standards.iso.org/iso/19115/resources/Codelists/cat/codelists.xml#DS_AssociationTypeCode"
-                    codeListValue="largerWorkCitation"/>
+                    codeListValue="partOfSeamlessDatabase"/>
                 </mri:associationType>
                 <mri:metadataReference xlink:href="{.}" xlink:title="{.}"/>
               </mri:MD_AssociatedResource>

diff --git a/schemas/iso19115-3.2018/src/test/java/org/fao/geonet/util/XslUtil.java b/schemas/iso19115-3.2018/src/test/java/org/fao/geonet/util/XslUtil.java
@@ -22,11 +22,29 @@
  */
 package org.fao.geonet.util;
 
+import org.jsoup.Jsoup;
+
+/**
+ * Copy of some utility functions required for testing XSL files.
+ * Dependency to XslUtil in core would create a circular dependency.
+ */
 public class XslUtil {
     public static String twoCharLangCode(String iso3code) {
         return iso3code.substring(0, 2);
     }
     public static String threeCharLangCode(String iso2code) {
         return "fre";
     }
+    public static String htmlElement2textReplacer(String html) {
+        return html
+            .replaceAll("<br */?>", System.getProperty("line.separator"))
+            .replaceAll("<li>(.*)</li>", System.getProperty("line.separator") + "* $1");
+    }
+    public static String html2text(String html) {
+        return Jsoup.parse(html).wholeText();
+    }
+    public static String html2text(String html, boolean substituteHtmlToTextLayoutElement) {
+        return html2text(
+            substituteHtmlToTextLayoutElement ? htmlElement2textReplacer(html) : html);
+    }
 }
diff --git a/schemas/iso19115-3.2018/src/test/resources/zenodo.xml b/schemas/iso19115-3.2018/src/test/resources/zenodo.xml
@@ -154,14 +154,15 @@
         </cit:CI_Citation>
       </mri:citation>
       <mri:abstract>
-        <gco:CharacterString>&lt;p&gt;Argo is a real-time global ocean in situ observing system. It provides thousands
+        <gco:CharacterString>Argo is a real-time global ocean in situ observing system. It provides thousands
           of highly accurate ocean measurements every day. The Argo dataset has now accumulated more than 2.3 million
-          vertical ocean profiles and accessing it for scientific analysis remains a challenge.&lt;/p&gt; &lt;p&gt;The
+          vertical ocean profiles and accessing it for scientific analysis remains a challenge.
+          The
           Argo expert community, focused on delivering a curated dataset of the best scientific quality possible, has
           never provided its user base with a Python software package to easily access and manipulate Argo measurements:
-          the argopy software aims to fill this gap. &lt;strong&gt;The argopy software can be used to easily fetch and
-          manipulate measurements from Argo floats&lt;/strong&gt;. It is dedicated to scientists without knowledge of
-          the Argo data management system but is also designed to accommodate expert requirements.&lt;/p&gt;
+          the argopy software aims to fill this gap. The argopy software can be used to easily fetch and
+          manipulate measurements from Argo floats. It is dedicated to scientists without knowledge of
+          the Argo data management system but is also designed to accommodate expert requirements.
         </gco:CharacterString>
       </mri:abstract>
       <mri:pointOfContact>
@@ -331,7 +332,7 @@
           <mri:associationType>
             <mri:DS_AssociationTypeCode
               codeList="http://standards.iso.org/iso/19115/resources/Codelists/cat/codelists.xml#DS_AssociationTypeCode"
-              codeListValue="largerWorkCitation"/>
+              codeListValue="partOfSeamlessDatabase"/>
           </mri:associationType>
           <mri:metadataReference xlink:href="https://doi.org/10.5281/zenodo.4009263"
                                  xlink:title="https://doi.org/10.5281/zenodo.4009263"/>