Support HTTP Proxy on BookWorm

[scottbarnes]

Closes #10230

This commit allows BookWorm to use the HTTP proxy when fetching metadata from Amazon.

BookWorm relies on an Amazon API client that itself relies on urllib3 which explicitly does not and will not support the HTTP_PROXY environment variable.

urllib3 does support HTTP Proxies, but it needs a specific object, and the underlying code for the API does support all of this, but the library provides no way to pass the proxy value through to urllib3.

The workaround here is to create the Configuration object in the Amazon API client, add a proxy object after init, then use that to create a REST client, and then replace the original api_client object with this one, so that there is proxy support.

Note: if trying to create an vendors.AmazonAPI object in a REPL, one will need the 'magic incantation' with infogami._setup() after the config has been loaded, or the http_proxy_url value will be None.

Testing

This isn't easy to test, but the following will work from ol-home0, inside the testing affiliate server container, as this code is currently applied there (once amz_key and amz_secret have the relevant values:

import infogami
from openlibrary.config import load_config
load_config('/olsystem/etc/openlibrary.yml')
infogami._setup()
import os
amz_key=os.getenv("amz_key")
amz_secret=os.getenv("amz_secret")
from openlibrary.core.vendors import AmazonAPI
amazon = AmazonAPI(key=amz_key, secret=amz_secret, tag="internetarchi-20")
amazon.get_product("9351950972")

Without this PR, request for amazon.get_product() will simply time out.

NOTE: this does NOT fix the /isbn endpoint. That still is not working for an unknown reason, and is the next follow up to this.

Stakeholders

@mekarpeles

[mekarpeles]

Thank you for this tip :)