Extending and configuring the discovery process

The discovery process is highly configurable by various plugins which are configured using the Zope Component Architecture.

How to configure plugins

If you look at the configure.zcml file you can see entries such as this one:

<utility
    provides=".interfaces.IDiscoveryService"
    factory=".hostmeta.HostMetaDiscoveryService"
    name="hostmeta"
    />

This entry defines a utility which implements the interface .interfaces.IDiscoveryService. It’s implementation can be found at .hostmeta.HostMetaDiscoveryService and it’s configured under the name hostmeta. Thus we call it a named utility.

Extension points in the main flow

In the main flow you can easily plugin your own discovery plugins by implementing one which provides the pydataportability.discovery.interfaces.IDiscoveryService interface. Give it a separate name and create your own configure.zcml file by copying the original one and adding a new entry for your utility.

Then you can list the name of your IDiscoveryService utility inside the call to the discovery client.

The discovery flow inside the hostmeta discovery service

The named utility of type IDiscoveryService and name hostmeta will first split the URI into it’s elements and will then try to obtain a handler for the URI scheme it found. In the Webfinger case it would find the acct: scheme.

The handler is again obtained by querying for a named utility implementing the necessary logic. In the acct: case a named utility implementing IHostMetaSchemaHandler and with acct as name. This handler is then responsible for extracting the host from the URI and return it to the actual hostmeta handler.

The hostmeta handler will then obtain the host-meta file from this host and try to obtain the location of the Resource Description of the original URI.