kwvanderlinde / python-cached-requests Goto Github PK

The current implementation is not aware of ETag at all, though it naturally stores ETag along with the rest of the headers. The CachedHTTPAdapter should be updated to add an If-None-Match to requests when the current cache entry is stale.

Note that the Cache interface is not sufficient for caches to implement time-based expiry, especially when combined with the desire to use the ETag to validate a stale entry. This is deliberate, as caches should not be altering requests, which is necessary with ETag. Instead, the adapter will check whether the CacheEntry returned by a cache is stale or not, and use the ETag if it is present.

Another option is to enhance the Cache interface. We could add a stale member to CacheEntry to indicate whether the entry is stale or not, as determined by the cache implementation. It would still be up to the adapter to make use of this information properly, and to delete any invalidated entries.

The current state of logging is awful. In some places, we simply print(...) messages, which is unacceptable for a library to do. Elsewhere, we simply have # TODO ... comments instructing us to log certain cases. In other places yet, we have comments that could be replaced with INFO logs that provide the same value, but can produce output on execution.

The HttpAwareCache is missing some crucial parts. In particular, Cache-Control is not processed at all. We should rectify this with an implementation that respects Cache-Control.

When dealing with headers in the cache and adaptor, we do not respect the case-insensitive nature of header names. We should use a strategy that permits proper lookup of header values regardless of case.