Iterate over object attributes in python
Question:
I have a python object with several attributes and methods. I want to iterate over object attributes.
class my_python_obj(object):
attr1='a'
attr2='b'
attr3='c'
def method1(self, etc, etc):
#Statements
I want to generate a dictionary containing all of the objects attributes and their current values, but I want to do it in a dynamic way (so if later I add another attribute I don’t have to remember to update my function as well).
In php variables can be used as keys, but objects in python are unsuscriptable and if I use the dot notation for this it creates a new attribute with the name of my var, which is not my intent.
Just to make things clearer:
def to_dict(self):
'''this is what I already have'''
d={}
d["attr1"]= self.attr1
d["attr2"]= self.attr2
d["attr3"]= self.attr3
return d
ยท
def to_dict(self):
'''this is what I want to do'''
d={}
for v in my_python_obj.attributes:
d[v] = self.v
return d
Update:
With attributes I mean only the variables of this object, not the methods.
Answers:
Assuming you have a class such as
>>> class Cls(object):
... foo = 1
... bar = 'hello'
... def func(self):
... return 'call me'
...
>>> obj = Cls()
calling dir on the object gives you back all the attributes of that object, including python special attributes. Although some object attributes are callable, such as methods.
>>> dir(obj)
['__class__', '__delattr__', '__dict__', '__doc__', '__format__', '__getattribute__', '__hash__', '__init__', '__module__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__', 'bar', 'foo', 'func']
You can always filter out the special methods by using a list comprehension.
>>> [a for a in dir(obj) if not a.startswith('__')]
['bar', 'foo', 'func']
or if you prefer map/filters.
>>> filter(lambda a: not a.startswith('__'), dir(obj))
['bar', 'foo', 'func']
If you want to filter out the methods, you can use the builtin callable as a check.
>>> [a for a in dir(obj) if not a.startswith('__') and not callable(getattr(obj, a))]
['bar', 'foo']
You could also inspect the difference between your class and its instance object using.
>>> set(dir(Cls)) - set(dir(object))
set(['__module__', 'bar', 'func', '__dict__', 'foo', '__weakref__'])
The correct answer to this is that you shouldn’t. If you want this type of thing either just use a dict, or you’ll need to explicitly add attributes to some container. You can automate that by learning about decorators.
In particular, by the way, method1 in your example is just as good of an attribute.
Objects in python store their atributes (including functions) in a dict called __dict__. You can (but generally shouldn’t) use this to access the attributes directly. If you just want a list, you can also call dir(obj), which returns an iterable with all the attribute names, which you could then pass to getattr.
However, needing to do anything with the names of the variables is usually bad design. Why not keep them in a collection?
class Foo(object):
def __init__(self, **values):
self.special_values = values
You can then iterate over the keys with for key in obj.special_values:
class SomeClass:
x = 1
y = 2
z = 3
def __init__(self):
self.current_idx = 0
self.items = ["x", "y", "z"]
def next(self):
if self.current_idx < len(self.items):
self.current_idx += 1
k = self.items[self.current_idx-1]
return (k, getattr(self, k))
else:
raise StopIteration
def __iter__(self):
return self
then just call it as an iterable
s = SomeClass()
for k, v in s:
print k, "=", v
in general put a __iter__ method in your class and iterate through the object attributes or put this mixin class in your class.
class IterMixin(object):
def __iter__(self):
for attr, value in self.__dict__.iteritems():
yield attr, value
Your class:
>>> class YourClass(IterMixin): pass
...
>>> yc = YourClass()
>>> yc.one = range(15)
>>> yc.two = 'test'
>>> dict(yc)
{'one': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14], 'two': 'test'}
For python 3.6
class SomeClass:
def attr_list(self, should_print=False):
items = self.__dict__.items()
if should_print:
[print(f"attribute: {k} value: {v}") for k, v in items]
return items
As mentioned in some of the answers/comments already, Python objects already store a dictionary of their attributes (methods aren’t included). This can be accessed as __dict__, but the better way is to use vars (the output is the same, though). Note that modifying this dictionary will modify the attributes on the instance! This can be useful, but also means you should be careful with how you use this dictionary. Here’s a quick example:
class A():
def __init__(self, x=3, y=2, z=5):
self.x = x
self._y = y
self.__z__ = z
def f(self):
pass
a = A()
print(vars(a))
# {'x': 3, '_y': 2, '__z__': 5}
# all of the attributes of `a` but no methods!
# note how the dictionary is always up-to-date
a.x = 10
print(vars(a))
# {'x': 10, '_y': 2, '__z__': 5}
# modifying the dictionary modifies the instance attribute
vars(a)["_y"] = 20
print(vars(a))
# {'x': 10, '_y': 20, '__z__': 5}
Using dir(a) is an odd, if not outright bad, approach to this problem. It’s good if you really needed to iterate over all attributes and methods of the class (including the special methods like __init__). However, this doesn’t seem to be what you want, and even the accepted answer goes about this poorly by applying some brittle filtering to try to remove methods and leave just the attributes; you can see how this would fail for the class A defined above.
(using __dict__ has been done in a couple of answers, but they all define unnecessary methods instead of using it directly. Only a comment suggests to use vars).
I have a python object with several attributes and methods. I want to iterate over object attributes.
class my_python_obj(object):
attr1='a'
attr2='b'
attr3='c'
def method1(self, etc, etc):
#Statements
I want to generate a dictionary containing all of the objects attributes and their current values, but I want to do it in a dynamic way (so if later I add another attribute I don’t have to remember to update my function as well).
In php variables can be used as keys, but objects in python are unsuscriptable and if I use the dot notation for this it creates a new attribute with the name of my var, which is not my intent.
Just to make things clearer:
def to_dict(self):
'''this is what I already have'''
d={}
d["attr1"]= self.attr1
d["attr2"]= self.attr2
d["attr3"]= self.attr3
return d
ยท
def to_dict(self):
'''this is what I want to do'''
d={}
for v in my_python_obj.attributes:
d[v] = self.v
return d
Update:
With attributes I mean only the variables of this object, not the methods.
Assuming you have a class such as
>>> class Cls(object):
... foo = 1
... bar = 'hello'
... def func(self):
... return 'call me'
...
>>> obj = Cls()
calling dir on the object gives you back all the attributes of that object, including python special attributes. Although some object attributes are callable, such as methods.
>>> dir(obj)
['__class__', '__delattr__', '__dict__', '__doc__', '__format__', '__getattribute__', '__hash__', '__init__', '__module__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', '__weakref__', 'bar', 'foo', 'func']
You can always filter out the special methods by using a list comprehension.
>>> [a for a in dir(obj) if not a.startswith('__')]
['bar', 'foo', 'func']
or if you prefer map/filters.
>>> filter(lambda a: not a.startswith('__'), dir(obj))
['bar', 'foo', 'func']
If you want to filter out the methods, you can use the builtin callable as a check.
>>> [a for a in dir(obj) if not a.startswith('__') and not callable(getattr(obj, a))]
['bar', 'foo']
You could also inspect the difference between your class and its instance object using.
>>> set(dir(Cls)) - set(dir(object))
set(['__module__', 'bar', 'func', '__dict__', 'foo', '__weakref__'])
The correct answer to this is that you shouldn’t. If you want this type of thing either just use a dict, or you’ll need to explicitly add attributes to some container. You can automate that by learning about decorators.
In particular, by the way, method1 in your example is just as good of an attribute.
Objects in python store their atributes (including functions) in a dict called __dict__. You can (but generally shouldn’t) use this to access the attributes directly. If you just want a list, you can also call dir(obj), which returns an iterable with all the attribute names, which you could then pass to getattr.
However, needing to do anything with the names of the variables is usually bad design. Why not keep them in a collection?
class Foo(object):
def __init__(self, **values):
self.special_values = values
You can then iterate over the keys with for key in obj.special_values:
class SomeClass:
x = 1
y = 2
z = 3
def __init__(self):
self.current_idx = 0
self.items = ["x", "y", "z"]
def next(self):
if self.current_idx < len(self.items):
self.current_idx += 1
k = self.items[self.current_idx-1]
return (k, getattr(self, k))
else:
raise StopIteration
def __iter__(self):
return self
then just call it as an iterable
s = SomeClass()
for k, v in s:
print k, "=", v
in general put a __iter__ method in your class and iterate through the object attributes or put this mixin class in your class.
class IterMixin(object):
def __iter__(self):
for attr, value in self.__dict__.iteritems():
yield attr, value
Your class:
>>> class YourClass(IterMixin): pass
...
>>> yc = YourClass()
>>> yc.one = range(15)
>>> yc.two = 'test'
>>> dict(yc)
{'one': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14], 'two': 'test'}
For python 3.6
class SomeClass:
def attr_list(self, should_print=False):
items = self.__dict__.items()
if should_print:
[print(f"attribute: {k} value: {v}") for k, v in items]
return items
As mentioned in some of the answers/comments already, Python objects already store a dictionary of their attributes (methods aren’t included). This can be accessed as __dict__, but the better way is to use vars (the output is the same, though). Note that modifying this dictionary will modify the attributes on the instance! This can be useful, but also means you should be careful with how you use this dictionary. Here’s a quick example:
class A():
def __init__(self, x=3, y=2, z=5):
self.x = x
self._y = y
self.__z__ = z
def f(self):
pass
a = A()
print(vars(a))
# {'x': 3, '_y': 2, '__z__': 5}
# all of the attributes of `a` but no methods!
# note how the dictionary is always up-to-date
a.x = 10
print(vars(a))
# {'x': 10, '_y': 2, '__z__': 5}
# modifying the dictionary modifies the instance attribute
vars(a)["_y"] = 20
print(vars(a))
# {'x': 10, '_y': 20, '__z__': 5}
Using dir(a) is an odd, if not outright bad, approach to this problem. It’s good if you really needed to iterate over all attributes and methods of the class (including the special methods like __init__). However, this doesn’t seem to be what you want, and even the accepted answer goes about this poorly by applying some brittle filtering to try to remove methods and leave just the attributes; you can see how this would fail for the class A defined above.
(using __dict__ has been done in a couple of answers, but they all define unnecessary methods instead of using it directly. Only a comment suggests to use vars).