Question

Intro: I’m writing web interface with SQLAlchemy reflection that supports multiple databases. It turns out that authors of application defined postgresql with lowercase tables/columns, eg. job.jobstatus while sqlite has mixed case, eg Job.JobStatus. I’m using DeclarativeReflectedBase from examples to combine reflection and declarative style.

The issue: Configure SQLAlchemy to work with tables/columns case insensitive with reflection

I have done so far:

• I have changed DeclarativeReflectedBase.prepare() method to pass quote=False into Table.__init__

What is left to be solved:

• relationship definitions still has to obey case when configuring joins, like primaryjoin=”Job.JobStatus==Status.JobStatus”.
• configure __tablename__ based on engine type

The question: Are my assumptions correct or is there more straightforward way? Maybe I could tell reflection to reflect everything lowercase and all problems are gone.

Answer 1

you’d probably want to look into defining a “.key” on each Column that’s in lower case, that way you can refer to columns as lower case within application code. You should use the column_reflect event (See http://docs.sqlalchemy.org/en/latest/core/events.html#schema-events) to define this key as a lower case version of the .name.

then, when you reflect the table, I’d just do something like this:

def reflect_table(name, engine):
    if engine.dialect.name == 'postgresql':
        name = name.lower()
    return Table(name, autoload=True, autoload_with=engine)

my_table = reflect_table("MyTable", engine)

I think that might cover it.