Accommodate NCBI API Keys

[MikeTrizna]

NCBI will start requiring API Keys beginning May 1, 2018: https://ncbiinsights.ncbi.nlm.nih.gov/2017/11/02/new-api-keys-for-the-e-utilities/

I author software that accesses the E-utilities, and this software is used by other people. Do those users need keys?

Yes, but only if your software allows them to post more than 3 requests/second. For such cases, we recommend that you allow users to enter their own API key as a setting in the software.

I'll need to modify the code to allow for users to provide their own API key.

I should also utilize https://github.com/SerpentAI/requests-respectful to limit requests to users who haven't provided API keys.

[MikeTrizna]

https://github.com/SerpentAI/requests-respectful requires installation of Redis, which seems like overkill.

Two possible solutions I've found in the meantime:

• The Planet API documentation documents how to deal with a rate limit by simply retrying requests until they work: https://www.planet.com/docs/api-quickstart-examples/large_aoi_best_practices/#rate
• This blog post describes how to use the token bucket algorithm to rate limit async requests (which could really speed things up): https://quentin.pradet.me/blog/how-do-you-rate-limit-calls-with-aiohttp.html
• This Python package uses a decorator to pass in provide a rate limit: https://github.com/tomasbasham/ratelimit.

The ratelimit package looks like the easiest to implement first. So start with that.