OOP PYTHON: Using cls() to create multiple constructor without calling init

Question

I have a Python class which takes an url in parameter and launches a crawler on a news website.

Once the creation of the object is finished, the object is stored in a Elasticsearch cluster.

I want to create a method that takes in input the Elasticsearch document, and creates an object from it.

class NewsArticle():

    def __init__(self, url):
        self.url = url
        # Launch a crawler and fill in the other fields like author, date, ect ...

    @classmethod
    def from_elasticsearch(cls, elasticsearch_article):
        document = elasticsearch_article['_source']
        obj = cls(document['url'])
        obj.url = document['url']
        obj.author = document['author']
        .
        .
        .

The problem is, when I'm calling...

# response is my document from elasticsearch
res = NewsArticle.from_elasticsearch(response)

...the method __init__ will be called and will launch my crawler. Is there anyway that it doesn't launch my crawler or call the init method?

So you want to create an object without initializing an object? — Steven Summers, Feb 13 '17 at 10:39
Maybe you should not have that crawler stuff in your `__init__`. — khelwood, Feb 13 '17 at 10:41
@StevenSummers I would like two know if I can have 2 differents constructors — mel, Feb 13 '17 at 10:43
No, `python` doesn't support overloading methods. It does however let you provide optional arguments and you can pass a flag (`bool`) to determine other actions. Or use other methods to set values. Or as khelwood mentioned, re-structure your code so it runs when you call it to. — Steven Summers, Feb 13 '17 at 10:52

score 1 · Accepted Answer · answered Feb 13 '17 at 10:50

1

How about a simple if and a default parameter crawl:

class NewsArticle():

    def __init__(self, url, crawl=True):
        self.url = url
        if crawl:
            # Launch a crawler and fill in the other fields like author, date, ect ...

    @classmethod
    def from_elasticsearch(cls, elasticsearch_article):
        document = elasticsearch_article['_source']
        obj = cls(document['url'], crawl=False)
        obj.url = document['url']
        obj.author = document['author']

answered Feb 13 '17 at 10:50

Mike Müller

82,630
20
166
161

I did what you say and I also took my crawler out of the init in a class method called from_crawled. The if statement will call from_crawled or from_elasticsearch depending of the argument in the init. Is that a good architecture? Should I use @classmethod? – mel Feb 13 '17 at 11:02
Sounds good to me. The use of class method also looks reasonable. – Mike Müller Feb 13 '17 at 11:12
Which do you recommend: using my init method only to define the type of my attributes and using class method to create my objects or putting a if statement in my init, adding the methodology and the response from Elasticsearch as parameters? – mel Feb 13 '17 at 11:16
I would allow to make fully working instance without the need of a class method. So an if in `__init__` seems fine. – Mike Müller Feb 13 '17 at 13:52

OOP PYTHON: Using cls() to create multiple constructor without calling __init__

1 Answers1

OOP PYTHON: Using cls() to create multiple constructor without calling init