how to dynamically create SQLAlchemy columns

Question

I have a csv file with first line as fields and remaining lines as data. With this file I would like to create a table. Since the fields are quite long and may vary, I want to dynamically create it. After several tries and searches, I figured I could maybe generate a Class from factory and then assign the attributes to the new class could generate dynamic columns. But it fails. And based on the error, I tried some other methods, all failed. Please help. I'm considering whether it's doable for such an idea.

Reader = csv.reader(open('Book1.csv', 'rb'), delimiter=',', quotechar='|')
TableItem = Reader.next()[0:]
def Factory(*args, **kwargs):
    args=args[0]
    def init(self, *iargs, **ikwargs):
        #setattr(self,__tablename__,ikwargs[__tablename__])
        for k,v in kwargs.items():
            setattr(self, k, v)
        for i in range(len(iargs)):
            setattr(self, args[i], iargs[i])
        for k,v in ikwargs.items():
            setattr(self, k, v)

    name = kwargs.pop("name", "myFactory")
    kwargs.update(dict((k, None) for k in args))
    return type(name, (object,), {'__init__': init})

LIS=Factory(TableItem)
class newLIS(LIS,Base):
    __tablename__='testonly'
    pass
mytest=[]
for row in Reader:
    mytest.append(newLIS(row))

The error is like:

sqlalchemy.exc.ArgumentError: Mapper Mapper|newLIS|testonly could not assemble
any primary key columns for mapped table 'testonly'

Answer 1

I'm the author of the related answer:

sqlalchemy dynamic mapping

I have this answer to another question, which I think is even more related - it could even be a duplicate:

Database on the fly with scripting languages

As you can see, to make it you need to create a table object and map it against your class. Also, mapped classes in sqlalchemy need a primary key on the table, so you have to add one. Are you having any other problem besides that? If so, can you paste your updated code?