search: adding extra metadata field

invenio_geographic_identifiers/contrib/geonames/datastreams.py

@@ -10,6 +10,7 @@
 
 from itertools import chain
 
+import pycountry
 from invenio_access.permissions import system_identity
 from invenio_vocabularies.datastreams.transformers import BaseTransformer
 from invenio_vocabularies.datastreams.writers import ServiceWriter
@@ -25,21 +26,19 @@ class GeoNamesTransformer(BaseTransformer):
 
     def apply(self, stream_entry, *args, **kwargs):
         """Applies the transformation to the entry."""
+        record_country = []
+
+        country_code = stream_entry.entry.get("country_code")
+        country_code = pycountry.countries.get(alpha_2=country_code)
+
+        if country_code:
+            record_country = [country_code.name, country_code.official_name]
+
         stream_entry.entry = {
             "id": f"geonames::{stream_entry.entry['geonameid']}",
             "scheme": "geonames",
-            "name": list(
-                chain(
-                    *list(
-                        filter(
-                            lambda x: x is not None,
-                            map(
-                                lambda y: stream_entry.entry.get(y),
-                                ["asciiname", "name", "alternativenames"],
-                            ),
-                        )
-                    )
-                )
+            "name": (
+                stream_entry.entry.get("name") or stream_entry.entry.get("asciiname")
             ),
             "locations": [
                 {
@@ -52,6 +51,22 @@ def apply(self, stream_entry, *args, **kwargs):
                     }
                 }
             ],
+            "extras": list(
+                chain(
+                    *[
+                        list(
+                            filter(
+                                lambda x: x is not None and x.strip(),
+                                map(
+                                    lambda y: stream_entry.entry.get(y),
+                                    ["asciiname", "name", "alternativenames"],
+                                ),
+                            )
+                        ),
+                        record_country,
+                    ]
+                )
+            ),
         }
 
         return stream_entry

invenio_geographic_identifiers/geoidentifiers/config.py

@@ -21,9 +21,9 @@ class GeographicIdentifiersSearchOptions(SearchOptions):
     suggest_parser_cls = FilteredSuggestQueryParser.factory(
         filter_field="scheme",
         fields=[  # suggest fields
-            "name^100",
-            "name._2gram",
-            "name._3gram",
+            "extras^100",
+            "extras._2gram",
+            "extras._3gram",
         ],
     )
 

invenio_geographic_identifiers/geoidentifiers/jsonschemas/geoidentifiers/geoidentifier-v1.0.0.json

@@ -21,10 +21,7 @@
     },
     "name": {
       "description": "Human readable label.",
-      "type": "array",
-      "items": {
-        "type": "string"
-      }
+      "type": "string"
     },
     "locations": {
       "type": "array",
@@ -34,6 +31,13 @@
           "$ref": "local://records/definitions-v2.0.0.json#/GeoJSON-Geometry"
         }
       }
+    },
+    "extras": {
+      "description": "Extra metadata about the identifier.",
+      "type": "array",
+      "items": {
+        "type": "string"
+      }
     }
   }
 }

invenio_geographic_identifiers/geoidentifiers/mappings/os-v2/geoidentifiers/geoidentifier-v1.0.0.json

@@ -46,6 +46,14 @@
           }
         }
       },
+      "extras": {
+        "type": "text",
+        "fields": {
+          "suggest": {
+            "type": "search_as_you_type"
+          }
+        }
+      },
       "pid": {
         "type": "object",
         "properties": {

invenio_geographic_identifiers/geoidentifiers/schema.py

@@ -55,11 +55,33 @@ class LocationSchema(Schema):
 class GeographicIdentifiersSchema(BaseVocabularySchema):
     """Service schema for geographic identifiers."""
 
-    # following the definition made in the ``invenio-vocabularies
-    # (subjects)``, here the ``id`` is ``required``
+    #
+    # Identifier
+    #
     id = SanitizedUnicode(required=True)
+
+    #
+    # Scheme
+    #
     scheme = SanitizedUnicode(required=True)
-    name = fields.List(required=True, cls_or_instance=SanitizedUnicode(required=True))
+
+    #
+    # Identifier name (e.g., Geonames feature name)
+    #
+    name = SanitizedUnicode(required=True)
+
+    #
+    # Location (e.g., list of points)
+    #
     locations = fields.List(
         required=True, cls_or_instance=fields.Nested(LocationSchema)
     )
+
+    #
+    # Extra metadata (This field is used in the search)
+    #
+    extras = fields.List(
+        required=True,
+        cls_or_instance=SanitizedUnicode(required=True),
+        load_only=True,
+    )

setup.cfg

@@ -35,6 +35,7 @@ packages = find:
 python_requires = >=3.7
 zip_safe = False
 install_requires =
+    pycountry<=23.12.11
     invenio-i18n<=1.3.3
     invenio-vocabularies>=1.0.0,<2.0.0