Make creation of default index configurable

optimade/server/config.py

@@ -180,6 +180,13 @@ class ServerConfig(BaseSettings):
         ),
     ] = None
 
+    create_default_index: Annotated[bool, Field(
+        description=(
+            "Whether to create a set of default indices "
+            "for supporting databases after inserting JSONL data.")
+        ),
+    ] = False
+
     use_real_mongo: Annotated[
         bool | None,
         Field(description="DEPRECATED: force usage of MongoDB over any other backend."),

optimade/server/entry_collections/entry_collections.py

@@ -275,15 +275,29 @@ def all_fields(self) -> set[str]:
         return self._all_fields
 
     def create_index(self, field: str, unique: bool = False) -> None:
-        """Create an index on the given field.
+        """Create an index on the given field, as stored in the database.
 
         Arguments:
-            field: The field to index.
+            field: The database field to index (i.e., if different from the OPTIMADE field,
+                the mapper should be used to convert between the two).
             unique: Whether or not the index should be unique.
 
         """
         raise NotImplementedError
 
+    def create_default_index(self) -> None:
+        """Create the default index for the collection.
+
+        For example, a database backend could override this method to
+        create a unique index on the `id` field, so that it can be called
+        on server startup.
+
+        This method should use a mapper to convert any OPTIMADE field names
+        to the corresponding stored names in the database.
+
+        """
+        raise NotImplementedError
+
     def get_attribute_fields(self) -> set[str]:
         """Get the set of attribute fields
 

optimade/server/entry_collections/mongo.py

@@ -107,15 +107,29 @@ def insert(self, data: list[EntryResource]) -> None:
             pass
 
     def create_index(self, field: str, unique: bool = False) -> None:
-        """Create an index on the given field.
+        """Create an index on the given field, as stored in the database.
+
+        If any error is raised during index creation, this method should faithfully
+        return it, except for the simple case where an identical index already exists.
 
         Arguments:
-            field: The field to index.
+            field: The database field to index (i.e., if different from the OPTIMADE field,
+                the mapper should be used to convert between the two).
             unique: Whether or not the index should be unique.
 
         """
         self.collection.create_index(field, unique=unique, background=True)
 
+    def create_default_index(self) -> None:
+        """Create the default index for the collection.
+
+        For MongoDB, the default is to create a unique index
+        on the `id` field. This method should obey any configured
+        mappers.
+
+        """
+        self.create_index(self.resource_mapper.get_backend_field("id"), unique=True)
+
     def handle_query_params(
         self, params: EntryListingQueryParams | SingleEntryQueryParams
     ) -> dict[str, Any]:

optimade/server/main.py

@@ -123,7 +123,7 @@ def load_entries(endpoint_name: str, endpoint_collection: EntryCollection):
                 f"Requested JSONL file does not exist: {jsonl_path}. Please specify an absolute group."
             )
 
-        insert_from_jsonl(jsonl_path)
+        insert_from_jsonl(jsonl_path, create_default_index=CONFIG.create_default_index)
 
         LOGGER.debug("Inserted data from JSONL file: %s", jsonl_path)
         if CONFIG.insert_test_data:

optimade/utils.py

@@ -24,7 +24,7 @@
 )
 
 
-def insert_from_jsonl(jsonl_path: Path, make_default_index: bool = False) -> None:
+def insert_from_jsonl(jsonl_path: Path, create_default_index: bool = False) -> None:
     """Insert OPTIMADE JSON lines data into the database.
 
     Arguments:
@@ -101,10 +101,11 @@ def insert_from_jsonl(jsonl_path: Path, make_default_index: bool = False) -> Non
             ENTRY_COLLECTIONS[entry_type].insert(batch[entry_type])
             batch[entry_type] = []
 
-        if make_default_index:
+        # If the chosen database backend supports it, make the default indices
+        if create_default_index:
             for entry_type in ENTRY_COLLECTIONS:
                 try:
-                    ENTRY_COLLECTIONS[entry_type].create_index("id", unique=True)
+                    ENTRY_COLLECTIONS[entry_type].create_default_index()
                 except NotImplementedError:
                     pass
 

tests/test_config.json

@@ -17,6 +17,7 @@
     "mongo_count_timeout": 0,
     "index_base_url": "http://localhost:5001",
     "insert_from_jsonl": "optimade/server/data/test_data.jsonl",
+    "create_default_index": true,
     "provider_fields": {
         "structures": [
             "band_gap",