Question

employees = []
for i in range(0,10):

    emp = Employee(i)
    emp.first_name = "%s-%s"%("first name", i)
    emp.last_name = "%s-%s"%("last_name", i)
    emp.desgination = "%s-%s"%("engineer", i)

    employees.append(emp)


ids = [e.eid for e in employees]

Following is my class definition:

class Employee:

    _fields = {}

    def __init__(self, eid):
        self.eid = eid

    def __getattr__(self, name):
        return self._fields.get(name)

    def __setattr__(self,name,value):
        self._fields[name] = value

    def __str__(self):
        return str(self._fields)

    def __unicode__(self):
        return str(self._fields)

The issue is that when I print ids, it contains 10 times 9… i.e.

[9, 9, 9, 9, 9, 9, 9, 9, 9, 9]

It seems that the same emp variable is being overwritten. I am not sure what going wrong. Though I am a Java coder but I thought I had a fair idea of Python as well.

Answer 1

The problem is, indeed, your java past ! 🙂

The error is here:

_fields = {}

_fields is a CLASS member ! So each instance of Employee shares the same _fields var, and whenever you modify one, you modify all other objects.
You must move the _fields = {} part into the __init__ function :

self._fields={}

But then, you will run into another problem: self.eid = xx invokes the __setattr__ method! (and so does self._fields !)

The solution is to use self.__dict__['_fields'] when you need to access to the instance’s ‘_fields’ member instead of self._fields. It will directly access the member, instead of going through __getattr__ and an infinite reccursion, with a nice stack overflow in the end.