Python won't eval() code

Question

I have the following eval() thing:

c = Customer()
eval("c.name = row.value('customer', '{c}')".format(c=column_name), { 'c': c, 'row': row})

When I try to run that, I get this:

Traceback (most recent call last):
 File "./import.py", line 19, in <module>
 c = Customer.save_from_row(row)
 File "/home/jason/projects/mcifdjango/mcif/models/customer.py", line 43, in save_from_row
 eval("c.name = row.value('customer', '{c}')".format(c=column_name), { 'c': c, 'row': row})
 File "<string>", line 1
 c.name = row.value('customer', 'name')
 ^
SyntaxError: invalid syntax

What am I doing wrong?

Edit: Because it looks like I didn't explain the context of my problem well enough, here's what I ended up doing, if anyone's curious:

@classmethod
def save_from_row(cls, row):
 c = cls()
 map(lambda column_name: setattr(c, column_name, row.value('customer', column_name)), c.distinguishing_column_names())
 return c.upsert()

Before I found out about setattr() I was separately setting several different attributes on c.

• Mike DeSimone: So... what's wrong with c.name = row.value('customer', column_name)? Why get eval() involved?
• Jason Swett: Because I have a whole list of different column names and properties of Customer that I need to assign.
• S.Lott: Why not use getattr and setattr?
• Jason Swett: Because I didn't know about them. I tried setattr and it did exactly what I wanted. If you post setattr as an answer I'll accept it.

Answer 1

eval evaluates expressions. Assignment is a statement, not an expression.

And don't even get me started on how easily misused and - in 99.99% of all cases - utterly unnecesary eval is. Just refer to the numerous other eval questions, I bet each has at least one such rant in an answer or comment - so I'll save my breath and link to one I like. (That being said, exec works like eval for statements.)

• Jason Swett: Any recommendation for what I might do instead of eval?
• user395760: @Jason: Depends on what you need. In general, if you are emulating PHP's variable variables - i.e. associate values with strings - a dictionary will do work as well while preserving sanity (and be orders of magnitude faster while we're at it).

Answer 2

Wouldn't this do what you need?:

c = Customer()
name_cols = (('name', 'custname'), ('addr', 'cust_addr'))
for name, col in name_cols:
 setattr(c, name, row.value('customer', col))