IFRS: prototype migration framework and object_size migration (GSI-1271)

services/ifrs/src/ifrs/migration_logic/_utils.py

@@ -80,8 +80,13 @@ def __init__(
         self._staged_collections: set[str] = set()
 
     @asynccontextmanager
-    async def auto_finalize(self, coll_names: str | list[str], copy_indexes: bool):
-        """Use within `apply()` or `unapply()` to
+    async def auto_finalize(
+        self, coll_names: str | list[str], copy_indexes: bool = False
+    ):
+        """Use within `apply()` or `unapply()` as a context manager to automatically
+        stage the temporary migrated collections for the specified collection names and
+        then drop the old collections. Set `copy_indexes` to True if the indexes are
+        expected to be identical between the old and new collection versions.
 
         Should be used for most migrations, but complex migrations might need to take
         a more manual approach. For that reason, this context manager is optional.
@@ -97,7 +102,7 @@ async def auto_finalize(self, coll_names: str | list[str], copy_indexes: bool):
 
             # copy indexes if needed (not implemented yet)
             if copy_indexes:
-                pass
+                raise NotImplementedError("Index copying is not yet implemented")
             # Drop old collections. Don't do the index copy check unless we perform the
             #  index copying via this method. Otherwise we can't be sure it wasn't
             #  handled some other way
@@ -190,8 +195,8 @@ async def stage_collection(self, original_coll_name: str):
 
         # Rename the old collection by giving it a prefix
         # e.g. "users" -> "tmp_v7_old_users"
-        old_collection = self._db[original_coll_name]
         temp_old_coll_name = self.old_temp_name(original_coll_name)
+        old_collection = self._db[original_coll_name]
         await old_collection.rename(temp_old_coll_name)
 
         # Rename the new, temp collection by removing its prefix
@@ -256,7 +261,9 @@ async def drop_old_collections(self, *, enforce_indexes: bool):
             old_temp_name = self.old_temp_name(coll_to_drop)
             collection = self._db[old_temp_name]
             await collection.drop()
-            log.debug("Dropped old collection")
+            log.debug(
+                "Dropped old collection for '%s' ('%s')", coll_to_drop, old_temp_name
+            )
             self._staged_collections.remove(coll_to_drop)
 
     @abstractmethod

services/ifrs/src/ifrs/migration_logic/ifrs_migrations.py

@@ -51,7 +51,7 @@ async def remove_object_size(self, doc: Document) -> Document:
 
     async def apply(self):
         """Populate `object_size` field on docs in the file_metadata collection"""
-        async with self.auto_finalize(METADATA_COLLECTION, copy_indexes=False):
+        async with self.auto_finalize(METADATA_COLLECTION):
             await self.migrate_docs_in_collection(
                 coll_name=METADATA_COLLECTION,
                 change_function=self.add_object_size,
@@ -61,7 +61,7 @@ async def apply(self):
 
     async def unapply(self):
         """Remove `object_size`"""
-        async with self.auto_finalize(METADATA_COLLECTION, copy_indexes=False):
+        async with self.auto_finalize(METADATA_COLLECTION):
             await self.migrate_docs_in_collection(
                 coll_name=METADATA_COLLECTION,
                 change_function=self.remove_object_size,