Feature/generate query for benchmarking

src/matchbox/server/postgresql/benchmark/generate_queries.py

@@ -0,0 +1,26 @@
+from matchbox.client.helpers.selector import query, selector
+
+
+# Generates a query that can be used for benchmarking purposes
+def generate_query(crn, fields, backend, resolution, return_type):
+    select_crn = selector(
+        table=str(crn),
+        fields=fields,
+        engine=crn.database.engine,
+    )
+
+    query_string = query(
+        selector=select_crn,
+        backend=backend,
+        resolution=resolution,
+        return_type=return_type,
+    )
+
+    return query_string
+
+
+if __name__ == "__main__":
+    query = ""
+    # vars: crn, fields, backend, resolution, return_type
+    # query = generate_query(crn, fields, backend, resolution, return_type)
+    print(query)

src/matchbox/server/postgresql/utils/query.py

@@ -355,6 +355,8 @@ def query(
             # Get cluster assignments
             mb_ids = sql_to_df(id_query, engine, return_type="arrow")
 
+            query_string = str(id_query)
+
             # Get source data
             raw_data = source.to_arrow(
                 fields=set([source.db_pk] + fields),
@@ -379,14 +381,14 @@ def query(
 
     # Return in requested format
     if return_type == "arrow":
-        return result
+        return result, query_string
     elif return_type == "pandas":
         return result.to_pandas(
             use_threads=True,
             split_blocks=True,
             self_destruct=True,
             types_mapper=ArrowDtype,
-        )
+        ), query_string
     else:
         raise ValueError(f"return_type of {return_type} not valid")