Passing a schema into transaction context

Is your feature request related to a problem? Please describe.

I've managed to get multi-tenant per schema working correctly by having 2 separate Tortoise apps, one for 'shared' and one for 'tenant'. The 'shared' app is straight-forward as I can just pass that connection into any transaction context e.g.

async with in_transaction(connection_name: 'shared') as conn:
    shared_widgets = await SharedWidgets.all()

However, the 'tenant' app is a bit of a hack and represents a blueprint of all the tenant-specific tables and models to ensure all of the relations work correctly e.g. it is used to qualify relationships between the tenant-specific models. When a new tenant signs up, I generate a new db schema for that specific tenant (let's say 'tenant_schema_12345') and clone all the 'tenant' app tables into this specific db schema. The term 'schema' here refers literally to a postgres database schema, e.g. the 'tenant_schema' part of select from tenant_schema.widgets.

Then finally, when e.g. user requests data specific to their tenant I do:

async with in_transaction(connection_name: 'tenant') as conn:
    # Set the tenant schema context
    conn.execute_script(f"set search_path to tenant_schema_12345")
    tenant_specific_widgets = await TenantSpecificWidgets.all()

This all works perfectly well, but I just want to add a small quality of life enhancement to pass e.g. 'tenant_schema_12345' into in_transaction(), or create a utility function wrapper for it but I can't figure out how best to do that. I don't want to remember to add the conn.execute_script(f"set search_path to tenant_schema_12345") line each time.

Ideally (from a syntactic point of view) it would be nice to add it as an extension to QuerySet, e.g.

def using_schema(schema_name: str) -> QuerySet:
    // Do something that's used when awaiting the query

QuerySet.using_schema = using_schema
...
...
...
tenant_specific_widgets = await TenantSpecificWidgets.all().using_schema('tenant_schema_12345')

I've found one spot where it might work but can't figure out how to go about extending/monkey-patching Tortoise to get it to use the following:

from tortoise.backends.base.client import TransactionContext


class TenantTransactionContext(TransactionContext):
    async def __aenter__(self):
        await self.ensure_connection()
        await self.lock.acquire()  # type:ignore
        self.token = connections.set(self.connection_name, self.connection)
        await self.connection.start()
        # Set the tenant schema context
        self.connection.execute_script(f"set search_path to {schema_name}")
        return self.connection

I'm happy to dig further but some pointers from someone who knows the Tortoise design well would be much appreciated.

Thanks!

Start with tortoise.backends.base.client.TransactionContext and the in_transaction entry point, then trace how QuerySet execution obtains its connection. Compare the proposed tenant context with the existing connection lifecycle and determine where a schema value could be applied safely. Done means callers can provide a schema without repeating execute_script, while existing transaction behavior remains unchanged.