From 836661087aaf5c1c37c1d87903c6f9cbabdeee4e Mon Sep 17 00:00:00 2001
From: Jesper Zedlitz <jesper@zedlitz.de>
Date: Wed, 19 Feb 2025 08:06:49 +0100
Subject: [PATCH] prefer the application profile of the transfer option
The metadata has evolved. Now the format of a transfer option can usually be found in its gmd:applicationProfile element.
---
.../opendata/csw2dcat/MDMetadata2Dataset.java | 46 +++++++++++--------
.../de/landsh/opendata/csw2dcat/Mapping.java | 6 +++
.../csw2dcat/MDMetadata2DatasetTests.java | 24 ++++++++++
3 files changed, 56 insertions(+), 20 deletions(-)
diff --git a/src/main/java/de/landsh/opendata/csw2dcat/MDMetadata2Dataset.java b/src/main/java/de/landsh/opendata/csw2dcat/MDMetadata2Dataset.java
index dc61715..9a43c2b 100644
--- a/src/main/java/de/landsh/opendata/csw2dcat/MDMetadata2Dataset.java
+++ b/src/main/java/de/landsh/opendata/csw2dcat/MDMetadata2Dataset.java
@@ -1054,21 +1054,30 @@ public class MDMetadata2Dataset {
return value;
}
- void addDistribution(Resource dataset, Element onlineResource) {
- final Model model = dataset.getModel();
-
+ private static String getGlobalFormatForAllTransferOptions(Element onlineResource) {
final Element distributionElement = onlineResource.getParent().getParent().getParent().getParent();
-
if (!"MD_Distribution".equals(distributionElement.getName())) {
throw new IllegalArgumentException("Could not navigate to MD_Distribution element.");
}
-
final Element distributionFormat = distributionElement.element("distributionFormat");
+ if( distributionFormat == null)
+ return null;
+
+ return getTextOrNull(distributionFormat.selectSingleNode("gmd:MD_Format/gmd:name/gco:CharacterString"));
+ }
+
+ void addDistribution(Resource dataset, Element onlineResource) {
+ final Model model = dataset.getModel();
+
+
+ // there might be a single format specified or all transfer options
+ final String globalFormat = getGlobalFormatForAllTransferOptions(onlineResource);
+
final String url = repairURL(getTextOrNull(onlineResource.selectSingleNode("gmd:linkage/gmd:URL")));
- final String format = distributionFormat == null ? null : getTextOrNull(distributionFormat.selectSingleNode("gmd:MD_Format/gmd:name/gco:CharacterString"));
final String name = getTextOrNull(onlineResource.selectSingleNode("gmd:name/*"));
final String description = getTextOrNull(onlineResource.selectSingleNode("gmd:description/*"));
+ final String applicationProfile = getTextOrNull(onlineResource.selectSingleNode("gmd:applicationProfile/*"));
// Check if the specified URL is a valid URI and IRI.
if (!isValidURLandIRI(url)) {
@@ -1082,22 +1091,19 @@ public class MDMetadata2Dataset {
distribution.addProperty(DCAT.accessURL, model.createResource(url));
Resource formatResource;
- try {
- formatResource = Mapping.mapFormat(format);
- } catch (IllegalArgumentException e) {
- formatResource = null;
- }
- if (format == null) {
- // try to determine the format of the distribution based on gmd:applicationProfile
- final String applicationProfile = getTextOrNull(onlineResource.selectSingleNode("gmd:applicationProfile/*"));
- if (StringUtils.isNotEmpty(applicationProfile)) {
- formatResource = Mapping.mapApplicationProfile(applicationProfile);
+ // Now it seems to be best to evaluate the transfer options' gmd:applicationProfile first.
+ if (StringUtils.isNotEmpty(applicationProfile)) {
+ formatResource = Mapping.mapApplicationProfile(applicationProfile);
+ } else {
+ try {
+ formatResource = Mapping.mapFormat(globalFormat);
+ } catch (IllegalArgumentException e) {
+ formatResource = null;
+ }
+ if (formatResource == null) {
+ formatResource = guessFormat(url);
}
- }
-
- if (formatResource == null) {
- formatResource = guessFormat(url);
}
if (formatResource != null) {
diff --git a/src/main/java/de/landsh/opendata/csw2dcat/Mapping.java b/src/main/java/de/landsh/opendata/csw2dcat/Mapping.java
index f12fd64..96e7a52 100644
--- a/src/main/java/de/landsh/opendata/csw2dcat/Mapping.java
+++ b/src/main/java/de/landsh/opendata/csw2dcat/Mapping.java
@@ -185,6 +185,12 @@ public class Mapping {
* URI
*/
public static Resource mapApplicationProfile(String applicationProfile) {
+ if(StringUtils.startsWith(applicationProfile, "http://publications.europa.eu/resource/authority/file-type/")) {
+ // The specified value is already a URI from the EU publication's office list
+ return ResourceFactory.createResource(applicationProfile);
+ }
+
+
if ("GeoTIFF".equals(applicationProfile)) {
return ResourceFactory.createResource("http://publications.europa.eu/resource/authority/file-type/GEOTIFF");
} else if ("PDF".equals(applicationProfile)) {
diff --git a/src/test/java/de/landsh/opendata/csw2dcat/MDMetadata2DatasetTests.java b/src/test/java/de/landsh/opendata/csw2dcat/MDMetadata2DatasetTests.java
index be227fb..e7c2eeb 100644
--- a/src/test/java/de/landsh/opendata/csw2dcat/MDMetadata2DatasetTests.java
+++ b/src/test/java/de/landsh/opendata/csw2dcat/MDMetadata2DatasetTests.java
@@ -1508,4 +1508,28 @@ public class MDMetadata2DatasetTests {
distributionWMS.getPropertyResourceValue(DCTerms.format).getURI());
}
+ @Test
+ public void convert_variousFormats() throws DocumentException, IOException {
+
+ final Element input = saxReader.read(getClass().getResourceAsStream("/1c82089a-313e-4c25-9389-0b704c885401.xml")).getRootElement();
+
+ Resource dataset = service.convert(input);
+
+ final Map<String, Resource> distributionMap = collectDistributions(dataset);
+ assertEquals(3, distributionMap.size());
+
+ Resource distributionPDF = distributionMap.get("https://www.schleswig-holstein.de/mm/downloads/LFU/Geologie/GUEK250_mit_erlaeuterung.pdf");
+ Resource distributionSHP = distributionMap.get("https://umweltanwendungen.schleswig-holstein.de/data/meta/boden/geokart/guek250/GUEK250_abgabe.zip");
+ Resource distributionJPG = distributionMap.get("https://umweltanwendungen.schleswig-holstein.de/data/meta/boden/geokart/guek250/guek250.jpg");
+
+ assertNotNull(distributionJPG);
+ assertNotNull(distributionSHP);
+ assertNotNull(distributionPDF);
+
+ assertEquals("http://publications.europa.eu/resource/authority/file-type/SHP",
+ distributionSHP.getPropertyResourceValue(DCTerms.format).getURI());
+ assertEquals("http://publications.europa.eu/resource/authority/file-type/PDF",
+ distributionPDF.getPropertyResourceValue(DCTerms.format).getURI());
+ }
+
}
--
GitLab