GeoNetwork 4.2.14: Thesaurus ConceptScheme generates invalid empty <gco:Date/> in iso19139.ca.HNAP metadata

[erick-ouellette]

Describe the bug

When creating or modifying a local Thesaurus/Theme in GeoNetwork 4.2.14 (schema iso19139.ca.HNAP), GeoNetwork auto-generates invalid <gmd:CI_Citation> date elements in metadata records referencing that thesaurus.

Specifically:

• A gmd:thesaurusName/gmd:CI_Citation block is materialized in metadata.
• GeoNetwork inserts two <gmd:date> elements.
• Each contains <gco:Date/> (empty).
• HNAP validation fails because empty dates are not permitted.

Attempts to manually remove these date elements in metadata do not persist. On save/reopen, GeoNetwork re-inserts the empty <gco:Date/> elements.

This behavior appears tied to thesaurus registry enrichment logic.

To Reproduce

1. In GN 4.2.14, create a new local Theme (Thesaurus) under

config/codelist/local/thesauri/theme

using RDF similar to:

<rdf:Description rdf:about="https://registry.geonetwork-opensource.org/theme/Theme_name">
    <rdf:type rdf:resource="http://www.w3.org/2004/02/skos/core#ConceptScheme"/>
    <dc:title xml:lang="en">Theme_name</dc:title>
    <dc:title xml:lang="fr">Theme_name/dc:title>
    <dc:description xml:lang="en">...</dc:description>
    <dc:identifier>Theme_name.rdf</dc:identifier>
    <dc:type>theme</dc:type>
</rdf:Description>

2. Restart GeoNetwork to load the new thesaurus.
3. Add a keyword from this theme to an iso19139.ca.HNAP metadata record.
4. Save metadata.
5. Open XML view.

GeoNetwork generates:

<gmd:date>
  <gmd:CI_Date>
    <gmd:date>
      <gco:Date/>
    </gmd:date>
    ...

6. Validate record.

Validation fails with:

Invalid or missing value. '' is not a valid value for Date (gco:Date)
Invalid or missing value. '' is not a valid value for (Date_Type)

7. Manually delete the <gmd:date> blocks.
8. Save metadata.
9. Re-open metadata.

The <gmd:date> blocks reappear automatically.

Expected behavior

If a thesaurus does not define publication/creation dates, GeoNetwork should either:

• Not generate <gmd:date> elements, or
• Generate valid ISO date values, or
• Allow citation without date elements when schema permits.

GeoNetwork should not generate invalid ISO elements that immediately fail schema validation.

Additionally:

• Editing RDF thesaurus files should take effect without requiring a full server restart.
• UI editing of Thesaurus should not corrupt or rewrite RDF metadata unexpectedly.
• Metadata edits should not re-materialize invalid citation fragments.

Observed Workaround

The issue can be temporarily resolved by:

1. Editing the RDF ConceptScheme to match minimal structure used by older (working) themes:

<rdf:Description rdf:about="https://registry.geonetwork-opensource.org/theme/Theme_name">
    <dc:identifier>Theme_name.rdf</dc:identifier>
    <dc:description xml:lang="fr">...</dc:description>
    <dc:description xml:lang="en">...</dc:description>
    <dc:title xml:lang="fr">...</dc:title>
    <dc:title xml:lang="en">...</dc:title>
    <rdf:type rdf:resource="http://www.w3.org/2004/02/skos/core#ConceptScheme"/>
</rdf:Description>

2. Performing a full server restart.

Without restart, GeoNetwork continued using cached thesaurus metadata and kept generating invalid date elements.

3. Re-editing affected metadata records.
4. Saving and validating again.
5. After restart and RDF normalization, new metadata saves no longer injected empty <gco:Date/>.

Important Observations

• Thesaurus definitions appear to be cached at startup.
• Editing RDF files does not reliably refresh in-memory registry models.
• Editing thesauri via UI may re-write RDF in unexpected ways.
• Metadata referencing registry-linked themes triggers automatic citation enrichment.
• That enrichment can produce schema-invalid XML.

Impact

• Causes all affected metadata to fail HNAP validation.
• Batch edits do not persist because GN re-materializes invalid elements.
• Requires server restart to apply RDF corrections.
• Risk of metadata corruption if thesaurus edited in UI.
• Operationally significant in multi-environment deployments.

Desktop (please complete the following information):

• Browser: Edge
• GeoNetwork Version: 4.2.14
• Schema: iso19139.ca.HNAP 4.2.14
• Server:
  • Tomcat 9.0.106
  • Java Adoptium 8u462b08
  • ElasticSearch 7.17.15

Additional context

This issue did not occur with themes created in GN <= 4.2.11.

It appears specific to:

• GN 4.2.14
• iso19139.ca.HNAP schema behavior
• Thesaurus registry materialization logic

It would be helpful if:

• Thesaurus reload could be triggered without full restart.
• Citation enrichment did not generate invalid empty ISO elements.
• Validation-aware generation logic was implemented.