From 5b57470eef12a4f344aa58481a24b949de9aff05 Mon Sep 17 00:00:00 2001
From: Luca Fabbri <l.fabbri@bopen.eu>
Date: Wed, 5 Apr 2023 16:16:18 +0200
Subject: [PATCH] make use of BulkTransactionsClient easily overridable

TransactionsClient now exposes a bulk_client_cls, easily editable in subclasses
---
 .../stac_fastapi/sqlalchemy/transactions.py   | 88 ++++++++++---------
 1 file changed, 45 insertions(+), 43 deletions(-)

diff --git a/stac_fastapi/sqlalchemy/stac_fastapi/sqlalchemy/transactions.py b/stac_fastapi/sqlalchemy/stac_fastapi/sqlalchemy/transactions.py
index 644b82f2d..e7baa3017 100644
--- a/stac_fastapi/sqlalchemy/stac_fastapi/sqlalchemy/transactions.py
+++ b/stac_fastapi/sqlalchemy/stac_fastapi/sqlalchemy/transactions.py
@@ -21,6 +21,48 @@
 logger = logging.getLogger(__name__)
 
 
+@attr.s
+class BulkTransactionsClient(BaseBulkTransactionsClient):
+    """Postgres bulk transactions."""
+
+    session: Session = attr.ib(default=attr.Factory(Session.create_from_env))
+    debug: bool = attr.ib(default=False)
+    item_table: Type[database.Item] = attr.ib(default=database.Item)
+    item_serializer: Type[serializers.Serializer] = attr.ib(
+        default=serializers.ItemSerializer
+    )
+
+    def __attrs_post_init__(self):
+        """Create sqlalchemy engine."""
+        self.engine = self.session.writer.cached_engine
+
+    def _preprocess_item(self, item: stac_types.Item) -> stac_types.Item:
+        """Preprocess items to match data model.
+
+        # TODO: dedup with GetterDict logic (ref #58)
+        """
+        db_model = self.item_serializer.stac_to_db(item)
+        return self.item_serializer.row_to_dict(db_model)
+
+    def bulk_item_insert(
+        self, items: Items, chunk_size: Optional[int] = None, **kwargs
+    ) -> str:
+        """Bulk item insertion using sqlalchemy core.
+
+        https://docs.sqlalchemy.org/en/13/faq/performance.html#i-m-inserting-400-000-rows-with-the-orm-and-it-s-really-slow
+        """
+        # Use items.items because schemas.Items is a model with an items key
+        processed_items = [self._preprocess_item(item) for item in items]
+        return_msg = f"Successfully added {len(processed_items)} items."
+        if chunk_size:
+            for chunk in self._chunks(processed_items, chunk_size):
+                self.engine.execute(self.item_table.__table__.insert(), chunk)
+            return return_msg
+
+        self.engine.execute(self.item_table.__table__.insert(), processed_items)
+        return return_msg
+
+
 @attr.s
 class TransactionsClient(BaseTransactionsClient):
     """Transactions extension specific CRUD operations."""
@@ -34,6 +76,7 @@ class TransactionsClient(BaseTransactionsClient):
     collection_serializer: Type[serializers.Serializer] = attr.ib(
         default=serializers.CollectionSerializer
     )
+    bulk_client_cls = attr.ib(default=BulkTransactionsClient)
 
     def create_item(
         self,
@@ -46,7 +89,8 @@ def create_item(
 
         # If a feature collection is posted
         if item["type"] == "FeatureCollection":
-            bulk_client = BulkTransactionsClient(session=self.session)
+            breakpoint()
+            bulk_client = self.bulk_client_cls(session=self.session)
             bulk_client.bulk_item_insert(items=item["features"])
             return None
 
@@ -157,45 +201,3 @@ def delete_collection(
                 raise NotFoundError(f"Collection {collection_id} not found")
             query.delete()
             return self.collection_serializer.db_to_stac(data, base_url=base_url)
-
-
-@attr.s
-class BulkTransactionsClient(BaseBulkTransactionsClient):
-    """Postgres bulk transactions."""
-
-    session: Session = attr.ib(default=attr.Factory(Session.create_from_env))
-    debug: bool = attr.ib(default=False)
-    item_table: Type[database.Item] = attr.ib(default=database.Item)
-    item_serializer: Type[serializers.Serializer] = attr.ib(
-        default=serializers.ItemSerializer
-    )
-
-    def __attrs_post_init__(self):
-        """Create sqlalchemy engine."""
-        self.engine = self.session.writer.cached_engine
-
-    def _preprocess_item(self, item: stac_types.Item) -> stac_types.Item:
-        """Preprocess items to match data model.
-
-        # TODO: dedup with GetterDict logic (ref #58)
-        """
-        db_model = self.item_serializer.stac_to_db(item)
-        return self.item_serializer.row_to_dict(db_model)
-
-    def bulk_item_insert(
-        self, items: Items, chunk_size: Optional[int] = None, **kwargs
-    ) -> str:
-        """Bulk item insertion using sqlalchemy core.
-
-        https://docs.sqlalchemy.org/en/13/faq/performance.html#i-m-inserting-400-000-rows-with-the-orm-and-it-s-really-slow
-        """
-        # Use items.items because schemas.Items is a model with an items key
-        processed_items = [self._preprocess_item(item) for item in items]
-        return_msg = f"Successfully added {len(processed_items)} items."
-        if chunk_size:
-            for chunk in self._chunks(processed_items, chunk_size):
-                self.engine.execute(self.item_table.__table__.insert(), chunk)
-            return return_msg
-
-        self.engine.execute(self.item_table.__table__.insert(), processed_items)
-        return return_msg