fix: Asynchronous connection error when using databases other than Postgres or SQLite

[MarceloNunesAlves]

This PR introduces a new environment variable, LANGFLOW_DATABASE_DRIVER_ASYNC, to specify the database driver used for asynchronous database connections. This allows for greater flexibility in supporting different database systems without requiring code changes.

Motivation:

Currently, asynchronous database connections were limited to Postgres and SQLite. This restriction made it difficult to use Langflow with other database systems.

Below is the error message:

Changes:

• Added the LANGFLOW_DATABASE_DRIVER_ASYNC environment variable.

• Updated the asynchronous connection logic to dynamically select the appropriate database driver based on the value of this environment variable.

• Added error handling for unsupported database drivers.

Example:

To connect to a MySQL database asynchronously, set the environment variable as follows:

LANGFLOW_DATABASE_DRIVER_ASYNC=mysql+aiomysql

[codspeed-hq]

CodSpeed Performance Report

Merging #5555 will degrade performances by 58.3%

_{Comparing MarceloNunesAlves:fix/db_async_driver (fc82703) with main (16ff8eb)}

Summary

⚡ 1 improvements
❌ 2 regressions
✅ 12 untouched benchmarks

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Benchmarks breakdown

	Benchmark	main	MarceloNunesAlves:fix/db_async_driver	Change
❌	test_get_and_cache_all_types_dict	1 ms	2.5 ms	-58.3%
⚡	test_successful_run_with_input_type_text	271.3 ms	208 ms	+30.44%
❌	test_successful_run_with_output_type_any	260.3 ms	294.5 ms	-11.61%

[cbornet]

Does it work ? I think to use mysql+aiomysql, you need the aiomysql package to be installed which is not the case in the default LangFlow.

[MarceloNunesAlves]

Does it work ? I think to use mysql+aiomysql, you need the aiomysql package to be installed which is not the case in the default LangFlow.

Yes! You need to install the driver of your choice. And I did it this way because it is generic for other DB, such as Oracle and SQLServer.
I think the idea of the project has always been to allow the use of any relational database. And frankly, the way this method works with these different conditional structures, I don't think it's the most appropriate. but I kept it so as not to harm others who already use this version in production.