My initial thought is I'd rather just have a module that would find project-related .sql files and parse into named blocks based on the comments, treating them as named strings. No db entanglement required, just a simple way to organize queries in sql files. You would use it more like this:

    import sqlite3
    import project_queries as queries  # a module that loads queries from local .sql files as strings in module namespace

    conn = sqlite3.connect('myapp.db')
    
    cursor = conn.cursor(queries.get_all_users)
    #                    ^^^^^^^^^^^^^^^^^^^^^
    
    users = cursor.fetchall()
    # >>> [(1, "nackjicholson", "William", "Vaughn"), (2, "johndoe", "John", "Doe"), ...]

aiosql may add some features to help with passing values to the query. I see `^` used with `:users` in an example but don't quite get it.

    -- name: get-user-by-username^

I often need to generate DDL details like table and column names in the query, which you don't want escaped, along with data like values and ids where you would want escaping (:users or %(users)s).

It works nicely to use string formatting (`CREATE TABLE {table_name} ...`) for DDL along with interpolation (`WHERE user_id IN %(user_ids)s`) in the same query.