Python trick: How to make lazy objects?

Reposting from the old blog.

This time I’m about to talk about lazy object, I called like that because they are more or less similar to what we use in lazy treeviews (In Gtk) where the Tree shows expanders but didn’t load the child nodes until you expand the node). What I want to do is create objects and count them, but don’t load any data until I need it.

Why would I need this?. Well. I’m currently writing a program for my employer, this program load a list of customers and shows how many accounts it has. As this accounts are kind of complex I create an object for them. This object is responsible for load/save the data for it, and any data you need is accessed via properties.

In my first approach, the object was loading the data as soon as the constructor was called, but sometimes I never access to any info in that object, remember, a customer may have many (and I mean really many) accounts, then why load all the data if there is a chance that I’ll never use it?. Well, that is because there is a chance that I have to use that data.

Then how could I create the object and then load the data when somebody require it?. That may be easy, may not. :-). One approach is making use of python properties. Just create the object and when someone want to use a “value” from the object use the property to retrieve the data, process it and then return a value. A simple property may be defined like this:

I just create a class “demo” and the property “some_property”, every time you use this:

Python will call the method get_some_property allowing you to do your trick. But if you have an object with more and more properties is better to get all data at one time and then just use properties to return a piece of it. You could call a method on every property if the data container is empty but for me it is not the best.

What you could do, is get the data the first time you need it. But, how could I know when a property/variable is requested on my object?

Well, Python classes have several reserved methods, and one of them is __getattribute__. This method let you handle every request to any property/class variable in your object.

With this you can dig in your object to know if the container is empty, and if it is empty fill it with data before returning the property value :-).

Lets say that the demo object have a __data dictionary containing the data. and propeties look into it for the right value, it also have a reference to the storage layer called storage.

Pay attention that I used object.getattribute instead self.property, if I use self.property will call self.getattribute and will cause a recursion exception.

With this, we can create a great number of objects quickly and reducing the memory footprint. The downside is that getting the data will be a lot expensive if we require the data of all (or several) objects at once.

 

How would you do it?, let me know in the comments.

4,008 total views, 1 views today

Software Developer, Python lover and a Free Software enthusiast

%d bloggers like this:
Skip to toolbar