Support reading protobuf definitions contained within a Substreams package

[alex-pakalniskis]

Problem

Even though Substreams packages contain relevant protobuf definitions, Substreams-powered subgraph developers are unable to reference those protobuf definitions when generating AssemblyScript types for their subgraph with protoc. See this example protoc command supporting wildcard matching of protobuf files.

protoc --plugin=./node_modules/protobuf-as/bin/protoc-gen-as --as_out=src/pb/ ./proto/*.proto

protoc does not natively parse Substreams packages, so users are left unable to proceed unless they source the required files. Furthermore, protobuf files are not easily found on the Substreams registry, so users must search GitHub and hope for the best. Locating the exactly correct protobuf definition files for a Substreams package adds unnecessary friction to the Substreams-powered subgraph developer experience.

Potential solution

Enable Graph CLI to generate AssemblyScript types from the *.proto files contained in a Substreams package (perhaps its just a nice wrapper for protoc + protobuf-as)

graph protogen #create a new command specific to protobuf files to look for `*.proto` files in `*.spkg`
# or
graph codegen #extend codegen to look for `*.proto` files in `*.spkg` if Substreams-powered subgraph

[abourget]

The say substreams protogen works, is it writes a buf.gen.yaml file on-the-fly (unless its present), runs buf generate and then removes it (unless it was already present).

It would be neat that this be done in the right place, graph cli would be extraordinary!
At the same time, substreams protogen already does that, and could add that layer, but I'm unsure it's the right tool if someone is just using a downloaded spkg and doesn't have the substreams CLI installed.