rpj / rhp

Redis-to-HTTP proxy

Redis-to-HTTP proxy

Currently there are only two distinct pieces functionality implemented, specifically proxying:

• of pub/sub subscriptions onto a websocket
• the LRANGE command in both standard and domain-specific ways

However, combined these allow relatively powerful client implementations, such as this weather data dashboard.

WebSocket pub/sub subscription proxying

• specify channel in path to GET request /sub/[channel], providing valid credentials as Basic authorization
  • credentials are matched against those stored in users.json, a simple username-as-key to client-as-password mapping
• if /sub/... request was valid & authorized, it will return a one-time token SUB_TOKEN
• use the token as the sole query parameter to the websocket endpoint to start receiving data, e.g. /ws/sub?SUB_TOKEN

The entirety of the aforementioned procedure can be perfomed on the CLI via wscat and curl:

wscat -c ws://HOST:PORT/ws/sub?`curl -H "Authorization: Basic B64_CRED" http://HOST:PORT/sub/CHANNEL`

where HOST, PORT, B64_CRED & CHANNEL should be replaced with appropriate values.

LRANGE proxying

Make GET requests to /list/[key]?<options>, where <options> may be composed of:

Name	Description
start	directly corresponds to the start argument of LRANGE
end	directly corresponds to the stop argument of LRANGE

The rpjios plugin enables the following options when loaded:

Name	Description
back	how many minutes backward from the current time to gather data (iteratively in chunks)
cad	the downsampling "cadence", in seconds; e.g. 600 returns samples separated by at least 600 seconds (10 minutes)

Processing is short-circuited such that only one of the entities above can respond to any given request. Accordingly, only one set of options can be used in conjunction.

For example, back and start should not be combined: if they are (and assuming the value given for back is valid), the rpjios plugin will handle the request and the value given for start will be ignored.