Question

I have a web-app and initially I used feedparser to pull data from my blog. Now I thought to use Google API for extending various other features.

As I didn’t want to break my old code and want to add the new API as an additional add-on type thing, I thought to setup a structure very similar to BeautifulSoup where it provides a choice to select an HTML/ XML parser depending upon the situation

BeautifulSoup(markup, "html.parser") 
BeautifulSoup(markup, "lxml")
# They still support the same methods mentioned in the documentation.

This is what I did:

from apis import api1, api2
from dev_user import my_api_choice
select = { "api1" : api1, "api2": api2 }

obj = pservice (select["my_api_choice"] )
obj.method1() 
# invokes method1 of the inherited class.

pservice is very close to wrap-up class. I initially thought to write it because I could have a chance to use Google API in it.

I even made sure that api2 (which provides Google API functionality) provides same kind of methods as that of api1 (even their method names are same).

So, is this a good way to do it? or else, how can I possibly do it in the other way..

Answer 1

You may create an abstract class API with some abstract method like fetch or pull and inherit it and build concrete implementations under classes like API1 / API2 / etc…

For example implementation of fetch function for Google API would be done for HTML and same way fetch function for feedparser would be implemented using XML…

Now based on your requirement or current configuration or whatever input; make a decision which API you want to use and create dynamic instance for the same, now as method names would be same in all classes you just need to fire appropriate function for that instance and you’re done.

Further more, afterwards you can introduce API3 and update your configuration – to use API3 class instance as another content provider.