cache - reduce number of API calls - provide offline access of ever called data

[femtotrader]

Hello,

It will be nice to provide a cache mechanism to Quandl Python. requests and requests-cache could help. You can see my code here https://github.com/femtotrader/pandas_datareaders/

Kind regards

[jbyck]

Thanks femotrader, but we do not have any plans to introduce a caching layer into the package. We view this package as thin layer around our API, and we are hesitant to add additional complexity, which may be confusing to some users.

[femtotrader]

Is it still something you don't want ?

see http://pandas-datareader.readthedocs.io/en/latest/cache.html

I had a look at https://github.com/quandl/quandl-python/blob/e2540dc8333c03802583344820bb0dd6b93f07df/quandl/connection.py#L38

I think you just need an extra parameter (session for example) session will be set to None by default

You will have

https://github.com/quandl/quandl-python/blob/e2540dc8333c03802583344820bb0dd6b93f07df/quandl/connection.py#L41

    @classmethod
    def execute_request(cls, http_verb, url, **options, session=None):
        try:
            if session is None:
                session = requests.Session()
            func = getattr(session, http_verb)
            response = func(url, **options)

I don't think that's a very invasive modification but can be very useful!

[femtotrader]

in fact... to be even less invasive session could come from options kwargs.

@classmethod
    def execute_request(cls, http_verb, url, **options):
        try:
            if 'session' not in options:
                session = requests.Session()
            else:
                session = options['session']
            func = getattr(session, http_verb)
            response = func(url, **options)

[ericremoreynolds]

One workaround is to use requests_cache.install_cache to monkey patch all HTTP requests, including the ones Quandl produces.

[femtotrader]

Thanks @ericremoreynolds but passing session would be a much cleaner approach than monkey patch !