fdc-shared-kernel 0.0.199__tar.gz → 0.0.200__tar.gz

{fdc_shared_kernel-0.0.199 → fdc_shared_kernel-0.0.200}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fdc_shared_kernel
-Version: 0.0.199
+Version: 0.0.200
 Summary: Shared library for microservice
 Author-email: Shikhil S <shikhil.s@dbizsolution.com>, Ahammed Akdham N <ahammedakdham.n@dbizsolution.com>
 Classifier: Programming Language :: Python :: 3

{fdc_shared_kernel-0.0.199 → fdc_shared_kernel-0.0.200}/fdc_shared_kernel.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: fdc_shared_kernel
-Version: 0.0.199
+Version: 0.0.200
 Summary: Shared library for microservice
 Author-email: Shikhil S <shikhil.s@dbizsolution.com>, Ahammed Akdham N <ahammedakdham.n@dbizsolution.com>
 Classifier: Programming Language :: Python :: 3

{fdc_shared_kernel-0.0.199 → fdc_shared_kernel-0.0.200}/fdc_shared_kernel.egg-info/SOURCES.txt

@@ -61,6 +61,7 @@ shared_kernel/data_warehouse_handlers/connections/mysql_warehouse_connection.py
 shared_kernel/data_warehouse_handlers/connections/postgresql_warehouse_connection.py
 shared_kernel/data_warehouse_handlers/connections/redshift_warehouse_connection.py
 shared_kernel/data_warehouse_handlers/connections/snowflake_warehouse_connection.py
+shared_kernel/data_warehouse_handlers/custom_query_builders/athena_query.py
 shared_kernel/data_warehouse_handlers/custom_query_builders/databricks_query.py
 shared_kernel/data_warehouse_handlers/query_executors/csv_query_executor.py
 shared_kernel/data_warehouse_handlers/query_executors/databricks_query_executor.py

{fdc_shared_kernel-0.0.199 → fdc_shared_kernel-0.0.200}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "fdc_shared_kernel"
-version = "0.0.199"
+version = "0.0.200"
 requires-python = ">=3.7"
 readme = "README_pypi.md"
 description = "Shared library for microservice"

{fdc_shared_kernel-0.0.199 → fdc_shared_kernel-0.0.200}/shared_kernel/data_warehouse_handlers/csv_handler.py

@@ -1,6 +1,7 @@
 from pypika import Query
 
 from shared_kernel.data_warehouse_handlers.connections.csv_warehouse_connection import CSVWarehouseConnection
+from shared_kernel.data_warehouse_handlers.custom_query_builders.athena_query import AthenaQuery
 from shared_kernel.data_warehouse_handlers.query_executors.csv_query_executor import CSVQueryExecutor
 from shared_kernel.dataclasses.warehouse_configs import CSVConfig
 from shared_kernel.interfaces.data_warehouse_handler import WarehouseHandler
@@ -34,7 +35,7 @@ class CSVWarehouseHandler(WarehouseHandler):
         return CSVQueryExecutor(warehouse_connection=self._build_connection())
 
     def get_query_object(self) -> Query:
-        return Query
+        return AthenaQuery
 
     def create_records_and_count_query(
         self,
@@ -45,9 +46,10 @@ class CSVWarehouseHandler(WarehouseHandler):
         count_query = f"SELECT COUNT(*) FROM ({query}) AS _count"
 
         if limit:
-            query = f"{query} LIMIT {int(limit)}"
+            # Athena requires OFFSET before LIMIT in the grammar.
             if page:
                 offset = (int(page) - 1) * int(limit)
                 query = f"{query} OFFSET {offset}"
+            query = f"{query} LIMIT {int(limit)}"
 
         return count_query, query

fdc_shared_kernel-0.0.200/shared_kernel/data_warehouse_handlers/custom_query_builders/athena_query.py

@@ -0,0 +1,27 @@
+from pypika import Query
+from pypika.queries import QueryBuilder
+
+
+class AthenaQuery(Query):
+    """
+    Defines a query class for use with Athena (Trino/Presto engine).
+    """
+
+    @classmethod
+    def _builder(cls, **kwargs) -> "AthenaQueryBuilder":
+        return AthenaQueryBuilder(dialect="athena", **kwargs)
+
+
+class AthenaQueryBuilder(QueryBuilder):
+    QUERY_CLS = AthenaQuery
+
+    def _apply_pagination(self, querystring: str) -> str:
+        # Trino/Presto (and therefore Athena) requires OFFSET before LIMIT;
+        # the pypika default emits LIMIT before OFFSET, which Athena rejects.
+        if self._offset:
+            querystring += self._offset_sql()
+
+        if self._limit is not None:
+            querystring += self._limit_sql()
+
+        return querystring