capture concept-scheme structure in database field (#812)

* proposal to capture full concept-scheme structure in database field, to facilitate inclusion of themes in ogcapi-records

* use the new structure of ogcapi-records for themes

uses identification.keywords2 and keyword_object awaiting release of owslib

etc/mappings.py

@@ -52,6 +52,7 @@
         'pycsw:Abstract': 'abstract',
         'pycsw:Keywords': 'keywords',
         'pycsw:KeywordType': 'keywordstype',
+        'pycsw:Themes': 'themes',
         'pycsw:Format': 'format',
         'pycsw:Source': 'source',
         'pycsw:Date': 'date',

pycsw/core/admin.py

@@ -144,6 +144,7 @@ def setup_db(database, table, home, create_sfsql_tables=True, create_plpythonu_f
         Column('edition', Text, index=True),
         Column('keywords', Text, index=True),
         Column('keywordstype', Text, index=True),
+        Column('themes', Text, index=True),
         Column('parentidentifier', Text, index=True),
         Column('relation', Text, index=True),
         Column('time_begin', Text, index=True),

pycsw/core/config.py

@@ -125,6 +125,7 @@ def __init__(self, prefix='csw30'):
                 'pycsw:Edition': 'edition',
                 'pycsw:Keywords': 'keywords',
                 'pycsw:KeywordType': 'keywordstype',
+                'pycsw:Themes': 'themes',
                 'pycsw:Format': 'format',
                 'pycsw:Source': 'source',
                 'pycsw:Date': 'date',

pycsw/core/metadata.py

@@ -1365,6 +1365,10 @@ def _parse_iso(context, repos, exml):
             all_keywords = [item for sublist in md.identification.keywords for item in sublist['keywords'] if item is not None]
             _set(context, recobj, 'pycsw:Keywords', ','.join(all_keywords))
             _set(context, recobj, 'pycsw:KeywordType', md.identification.keywords[0]['type'])
+            _set(context, recobj, 'pycsw:Themes', 
+                 # OWSlib currently stores keywords extended with thesaurus in keywords2, see https://github.com/geopython/OWSLib/issues/301
+                 json.dumps([t for t in md.identification.keywords2 if t.thesaurus is not None], 
+                            default=lambda o: o.__dict__))
 
         if (hasattr(md.identification, 'creator') and
             len(md.identification.creator) > 0):

pycsw/ogc/api/records.py

@@ -1035,6 +1035,17 @@ def record2json(record, url, collection, stac_item=False):
             })
         record_dict['properties']['providers'] = rcnt
 
+    if record.themes:
+        ogcapiThemes = []
+        # For a scheme, prefer uri over label
+        # OWSlib currently uses .keywords_object for keywords with url, see https://github.com/geopython/OWSLib/pull/765
+        for theme in json.loads(record.themes):
+            ogcapiThemes.append({
+                'scheme': theme['thesaurus'].get('url',theme['thesaurus'].get('title','')),
+                'concepts': [c for c in theme.get('keywords_object',[]) if c not in [None,""]]
+            })
+        record_dict['properties']['themes'] = ogcapiThemes
+
     if record.links:
         rdl = record_dict['links']
 

pycsw/ogc/api/templates/item.html

@@ -50,16 +50,24 @@ <h2>{{ data['id'] }}</h2>
                 {% endfor %}
                 </ul>
               </td>
-            {% elif key == 'themes' %}
-              <td>
-                {% for theme in value %}
-                  <b>{{ theme['scheme'] }}</b>
-                  <ul>
-                  {% for concept in theme['concepts'] %}
-                    <li>{{ concept }}</li>
-                  {% endfor %}
-                  </ul>
+
+            {% elif key == 'themes' %} 
+              <td> 
+              {% for theme in value %}
+
+                <b>{{ theme['scheme'] }}</b>
+                <ul>
+                {% for concept in theme['concepts'] %}
+                  <li>
+                    {% if concept['url'] %}
+                      <a href="{{ concept['url'] }}">{{ concept['name'] or concept['url'] }}</a>
+                    {% else %}
+                      {{ concept['name'] }}
+                    {% endif %}
+                  </li>
                 {% endfor %}
+                </ul>
+              {% endfor %}
               </td>
             {% elif key == 'externalIds' %}
               <td>