[WIP] Update to support new OPS metadata API

src/xopr/opr_access.py

@@ -341,18 +341,38 @@ def load_frame_url(self, url: str) -> xr.Dataset:
                     warnings.warn(f"Warning: Unexpected result from ops_api: {result['data']}", UserWarning)
                 else:
                     result_data = {}
-                    for key, value in result['data'].items():
-                        if len(value) == 1:
+                    result_data = result['data']
+                    if 'metadata' in result_data:
+                        result_data = result_data['metadata'] # New style response from OPS API
+                        new_api = True
+                    else:
+                        new_api = False
+
+                    # Flatten lists with single items to just the item, and convert lists with multiple items to sets
+                    for key, value in result_data.items():
+                        if len(value) == 0:
+                            result_data[key] = None
+                        elif len(value) == 1:
                             result_data[key] = value[0]
                         elif len(value) > 1:
                             result_data[key] = set(value)
 
-                    if 'dois' in result_data:
-                        ds.attrs['doi'] = result_data['dois']
-                    if 'rors' in result_data:
-                        ds.attrs['ror'] = result_data['rors']
-                    if 'funding_sources' in result_data:
-                        ds.attrs['funder_text'] = result_data['funding_sources']
+                    # These match on the old style keys from the OPS API for backwards compatibility
+                    if new_api:
+                        for k, v in result_data.items():
+                            if k in ds.attrs:
+                                if ds.attrs[k] != v:
+                                    warnings.warn(f"Warning: Conflicting value for attribute '{k}': existing value '{ds.attrs[k]}', new value '{v}'", UserWarning)
+                            ds.attrs[k] = v
+                    else:
+                        if 'dois' in result_data:
+                            ds.attrs['DOI'] = result_data['dois']
+                        if 'rors' in result_data:
+                            ds.attrs['Organization'] = result_data['rors']
+                        if 'funding_sources' in result_data:
+                            ds.attrs['funder_text'] = result_data['funding_sources']
+
+
 
         # Add the rest of the Matlab parameters
         if filetype == 'hdf5':

src/xopr/opr_tools.py

@@ -76,22 +76,40 @@ def generate_citation(ds : xr.Dataset) -> str:
 
         citation_string += "== Data Citation ==\n"
 
-        if 'ror' in ds.attrs and ds.attrs['ror']:
+        # Find the right key for ROR
+        ror_key = None
+        if 'ror' in ds.attrs:
+            ror_key = 'ror'
+        elif 'Organization' in ds.attrs:
+            ror_key = 'Organization'
+
+        if ror_key and ds.attrs[ror_key]:
             any_citation_info = True
-            if isinstance(ds.attrs['ror'], (set, list)):
-                institution_name = ', '.join([get_ror_display_name(ror) for ror in ds.attrs['ror']])
+            if isinstance(ds.attrs[ror_key], (set, list)):
+                institution_name = ', '.join([get_ror_display_name(ror) for ror in ds.attrs[ror_key]])
             else:
-                institution_name = get_ror_display_name(ds.attrs['ror'])
+                institution_name = get_ror_display_name(ds.attrs[ror_key])
 
             citation_string += f"This data was collected by {institution_name}.\n"
 
-        if 'doi' in ds.attrs and ds.attrs['doi']:
+        # Find the right key for DOI
+        doi_key = None
+        if 'doi' in ds.attrs:
+            doi_key = 'doi'
+        elif 'DOI' in ds.attrs:
+            doi_key = 'DOI'
+
+        if doi_key and ds.attrs[doi_key]:
             any_citation_info = True
-            citation_string += f"Please cite the dataset DOI: https://doi.org/{ds.attrs['doi']}\n"
+            citation_string += f"Please cite the dataset DOI: https://doi.org/{ds.attrs[doi_key]}\n"
 
-        if 'funder_text' in ds.attrs and ds.attrs['funder_text']:
+        if 'funder_text' in ds.attrs and ds.attrs['funder_text']: # Old style funding acknowledgement
             any_citation_info = True
             citation_string += f"Please include the following funder acknowledgment:\n{ds.attrs['funder_text']}\n"
+
+        if 'Funding' in ds.attrs and ds.attrs['Funding']: # New style funding acknowledgement
+            any_citation_info = True
+            citation_string += f"Data collection was funded by: {ds.attrs['Funding']}\n"
 
         if not any_citation_info:
             citation_string += "No specific citation information was retrieved for this dataset. By default, please cite:\n"

src/xopr/stac/metadata.py

@@ -168,7 +168,11 @@ def extract_item_metadata(
     center_freq = float((low_freq + high_freq) / 2)
 
     # Extract science metadata
-    doi = ds.attrs.get('doi', None)
+    if 'DOI' in ds.attrs:
+        doi = ds.attrs['DOI']
+    else:
+        doi = ds.attrs.get('doi', None)
+
     cite = ds.attrs.get('funder_text', None)
     mime = ds.attrs['mimetype']
 

src/xopr/tests/test_ops_api.py

@@ -51,7 +51,12 @@ def test_get_segment_metadata_valid_flights(season, flight_id):
     data = result['data']
     assert data is not None, f"Expected non-null data for {season}/{flight_id}"
 
-    expected_keys = ['dois', 'funding_sources', 'rors']
+    if 'metadata' in data:
+        # New style response from OPS API
+        data = data['metadata']
+        expected_keys = ['DOI', 'Funding', 'Organization']
+    else:
+        expected_keys = ['dois', 'funding_sources', 'rors']
 
     for key in expected_keys:
         assert key in data, f"Expected key '{key}' in data for {season}/{flight_id}"