minor STAC API updates (#991)

* minor STAC API updates
* add tests for required link properties
* STAC API: add bbox is geometry is null (#847)
* safeguard null geometry handling on STAC parsing (#847)
* update geometry/bbox handling

pycsw/core/metadata.py

@@ -1769,7 +1769,8 @@ def _parse_oarec_record(context, repos, record):
     if links:
         _set(context, recobj, 'pycsw:Links', json.dumps(links))
 
-    _set(context, recobj, 'pycsw:BoundingBox', util.bbox2wktpolygon(util.geojson_geometry2bbox(record['geometry'])))
+    if record.get('geometry') is not None:
+        _set(context, recobj, 'pycsw:BoundingBox', util.bbox2wktpolygon(util.geojson_geometry2bbox(record['geometry'])))
 
     if 'temporal' in record['properties'].get('extent', []):
         _set(context, recobj, 'pycsw:TempExtent_begin', record['properties']['extent']['temporal']['interval'][0])
@@ -1784,6 +1785,7 @@ def _parse_stac_resource(context, repos, record):
     recobj = repos.dataset()
     keywords = []
     links = []
+    bbox_wkt = None
 
     stac_type = record.get('type', 'Feature')
     if stac_type == 'Feature':
@@ -1793,7 +1795,8 @@ def _parse_stac_resource(context, repos, record):
         stype = 'item'
         title = record['properties'].get('title')
         abstract = record['properties'].get('description')
-        bbox_wkt = util.bbox2wktpolygon(util.geojson_geometry2bbox(record['geometry']))
+        if record.get('geometry') is not None:
+            bbox_wkt = util.bbox2wktpolygon(util.geojson_geometry2bbox(record['geometry']))
     elif stac_type == 'Collection':
         LOGGER.debug('Parsing STAC Collection')
         conformance = 'https://github.com/radiantearth/stac-spec/tree/master/collection-spec/collection-spec.md'
@@ -1804,8 +1807,6 @@ def _parse_stac_resource(context, repos, record):
         if 'extent' in record and 'spatial' in record['extent']:
             bbox_csv = ','.join(str(t) for t in record['extent']['spatial']['bbox'][0])
             bbox_wkt = util.bbox2wktpolygon(bbox_csv)
-        else:
-            bbox_wkt = None
         if 'extent' in record and 'temporal' in record['extent'] and 'interval' in record['extent']['temporal']:
             _set(context, recobj, 'pycsw:TempExtent_begin', record['extent']['temporal']['interval'][0][0])
             _set(context, recobj, 'pycsw:TempExtent_end', record['extent']['temporal']['interval'][0][1])
@@ -1816,7 +1817,6 @@ def _parse_stac_resource(context, repos, record):
         stype = 'catalog'
         title = record.get('title')
         abstract = record.get('description')
-        bbox_wkt = None
 
     _set(context, recobj, 'pycsw:Identifier', record['id'])
     _set(context, recobj, 'pycsw:Typename', typename)

pycsw/ogc/api/records.py

@@ -1328,11 +1328,15 @@ def record2json(record, url, collection, mode='ogcapi-records'):
             }
 
     if mode == 'stac-api':
-        record_dict['properties']['datetime'] = to_rfc3339(record.date)
+        date_, date_type = to_rfc3339(record.date)
+        record_dict['properties']['datetime'] = date_
 
         if None not in [record.time_begin, record.time_end]:
-            record_dict['properties']['start_datetime'] = to_rfc3339(record.time_begin)
-            record_dict['properties']['end_datetime'] = to_rfc3339(record.time_end)
+            start_date, start_date_type = to_rfc3339(record.time_begin)
+            end_date, end_date_type = to_rfc3339(record.time_end)
+
+            record_dict['properties']['start_datetime'] = start_date
+            record_dict['properties']['end_datetime'] = end_date
 
     return record_dict
 

pycsw/ogc/api/util.py

@@ -91,7 +91,10 @@ def json_serial(obj):
     """
 
     if isinstance(obj, (datetime, date, time)):
-        return obj.isoformat() + 'Z'
+        if isinstance(obj, date):
+            return obj.strftime('%Y-%m-%d')
+        else:
+            return obj.isoformat() + 'Z'
     elif isinstance(obj, bytes):
         try:
             LOGGER.debug('Returning as UTF-8 decoded bytes')

pycsw/stac/api.py

@@ -38,6 +38,7 @@
 from pycsw.ogc.api.oapi import gen_oapi
 from pycsw.ogc.api.records import API
 from pycsw.core.pygeofilter_evaluate import to_filter
+from pycsw.core.util import geojson_geometry2bbox
 
 LOGGER = logging.getLogger(__name__)
 
@@ -264,6 +265,9 @@ def collections(self, headers_, args):
             'href': self.config['server']['url']
         }]
 
+        response['numberMatched'] = len(response['collections'])
+        response['numberReturned'] = len(response['collections'])
+
         return self.get_response(200, headers_, response)
 
     def collection(self, headers_, args, collection='metadata:main'):
@@ -345,7 +349,7 @@ def items(self, headers_, json_post_data, args, collection='metadata:main'):
         headers, status, response = super().items(headers_, json_post_data, args, collection)
 
         if collection not in self.get_all_collections():
-            msg = f'Invalid collection'
+            msg = 'Invalid collection'
             LOGGER.exception(msg)
             return self.get_exception(400, headers_, 'InvalidParameterValue', msg)
 
@@ -373,6 +377,18 @@ def items(self, headers_, json_post_data, args, collection='metadata:main'):
             else:
                 response2['features'].append(record)
 
+            if record.get('bbox') is None:
+                geometry = record.get('geometry')
+                if geometry is not None:
+                    LOGGER.debug('Calculating bbox from geometry')
+                    bbox = geojson_geometry2bbox(geometry)
+                    record['bbox'] = [float(t) for t in bbox.split(',')]
+
+            for link in record['links']:
+                if link.get('rel') is None:
+                    LOGGER.debug('Missing link relation; adding rel=related')
+                    link['rel'] = 'related'
+
         for count, value in enumerate(response2['links']):
             if value['rel'] == 'alternate':
                 response2['links'].pop(count)
@@ -402,7 +418,7 @@ def item(self, headers_, args, collection, item):
         headers, status, response = super().item(headers_, args, collection, item)
 
         if collection not in self.get_all_collections():
-            msg = f'Invalid collection'
+            msg = 'Invalid collection'
             LOGGER.exception(msg)
             return self.get_exception(400, headers_, 'InvalidParameterValue', msg)
 

tests/functionaltests/suites/stac_api/test_stac_api_functional.py

@@ -83,6 +83,19 @@ def test_conformance(config):
         assert conformance in content['conformsTo']
 
 
+def test_collections(config):
+    api = STACAPI(config)
+    headers, status, content = api.collections({}, {'f': 'json'})
+    content = json.loads(content)
+
+    assert headers['Content-Type'] == 'application/json'
+    assert status == 200
+    assert len(content['links']) == 3
+
+    assert len(content['collections']) == 1
+    assert len(content['collections']) == content['numberMatched']
+    assert len(content['collections']) == content['numberReturned']
+
 def test_queryables(config):
     api = STACAPI(config)
     headers, status, content = api.queryables({}, {})
@@ -118,6 +131,15 @@ def test_items(config):
     assert record['stac_version'] == '1.0.0'
     assert record['collection'] == 'metadata:main'
 
+    for feature in content['features']:
+        if feature.get('geometry') is not None:
+            assert 'bbox' in feature
+            assert isinstance(feature['bbox'], list)
+
+        for link in feature['links']:
+            assert 'href' in link
+            assert 'rel' in link
+
     # test GET KVP requests
     content = json.loads(api.items({}, None, {'bbox': '-180,-90,180,90'})[2])
     assert len(content['features']) == 3
@@ -194,5 +216,9 @@ def test_item(config):
     assert content['stac_version'] == '1.0.0'
     assert content['collection'] == 'metadata:main'
 
+    for link in content['links']:
+        assert 'href' in link
+        assert 'rel' in link
+
     headers, status, content = api.item({}, {}, 'foo', item)
     assert status == 400