Communication
This communication is currently implemented by using htsmsg's. All strings are encoded as UTF-8.
There are two distinct ways for communication within HTSP.
Apart from this there is a number of messages that needs to be exchanged during login, see the login section below.
RPC Communication
There is a normal RPC way of doing things. I.e. the client sends a request and the server responds with a reply. All the RPC methods are listed below as the 'Client to Server' methods. Apart from all message fields listed within each message type the client can add an additional field:
RPC request extra fields:
The followings field should be used by the client to match the reply with the request. All replies are guaranteed to arrive in the same order as the requests. Even so, probably the best way to implement the request-reply client is by taking advantage of the 'seq' field.
RPC reply extra fields:
Streaming Communication
For streaming of live TV and various related messages the server will continuously push data to the client. These messages are referred to as asynchronous messages and always have the 'method' field set and never have the 'seq' field set. Also, the client can enable an additional asyncMetadata mode and by doing so it will be notified by the server when meta data changes. (EPG updates, creation of channels and tags, etc).
Authentication
In Tvheadend, each method has an associated access restriction. Currently there is only one restriction (Streaming). However, this may change in the future.
Privileges for these restrictions may be granted in two ways: Username + Password and/or Source IP address. Therefore it is possible to gain permissions to the system without entering a username and password. While this is really useful it also complicates the authentication schema a bit. Upon connect the initial privileges will be raised based on the source address.
Before any username / password based authentication has taken place the client must have obtained a challenge (which stays fixed for the session). This is done via the 'hello' method.
In principle it's possible to use two different authentication idioms with HTSP. Depending on how your application works one or another may be more suitable. While they do not really differ from a protocol point of view it's worth mentioning a bit about them here:
Initial Login Authentication
The client performs all of its authentication using the 'login' method.
It may choose to send:
Username and password: Privileges will be raised based on these credentials.
Username only: Privileges will be based on just the source address. The username will be used for various logging purposes.
Nothing: Privileges will be based on just the source address.
If no privileges are granted after the login message has been received by the server (i.e. both network and username + password based) the server will reply with 'noaccess' set to 1. A client that only employs initial login should honor this flag and ask the user for a username + password and retry by using the 'authenticate' method. I.e. it should not send the 'login' method again.
On-Demand Authentication
The client performs all of its authentication when it needs to.
When using this method, the client will check every RPC reply for the 'noaccess' field. If it set to 1 it whould ask the user for username + password and retry the request but also add 'username' and 'digest' to the original message. (See RPC request extra fields above)
Typically it would not send a username or digest during login.
Last updated