fix: Find substrate port on different log lines

[lexnv]

Updating substrate node to the latest version would result in an infinite loop for our integration testing.

The root cause of this issue is that the substrate switched to json-rpsee.
This included changing the log message we are looking for when finding the port of the substrate.

Ensure that the port can be obtained from both new and old versions of the substrate node.

Testing Done

Before

test client::chain_subscribe_blocks has been running for over 60 seconds
test client::chain_subscribe_finalized_blocks has been running for over 60 seconds
test client::fetch_block has been running for over 60 seconds

After

test client::chain_subscribe_finalized_blocks ... ok
test client::fetch_block ... ok
test client::chain_subscribe_blocks ... ok
test client::fetch_read_proof ... ok

[niklasad1]

you could also do: https://github.com/paritytech/substrate/blob/master/bin/node/cli/tests/common.rs#L164-#L167

then just parse it as SocketAddr then call port on it.

would be better than to assume that the address is 127.0.0.1

[niklasad1]

sorry, I didn't think about making the log easy to parse :)

[jsdw]

For the testing pov I guess it's ok assuming 127.0.0.1 though (ie you could start it on 0.0.0.0, but you couldn't then connect to 0.0.0.0)?

[niklasad1]

it's fair assumption, I was thinking that some weird platforms could actually have some other loopback address than 127.0.0.1. Picking hair here just ignore me but I was more thinking about having to avoid parsing the port manually :P

I think you can connect to 0.0.0.0 but it means all addresses IIRC.

[jsdw]

I think you could simplify this matching code to be:

// does the line contain our port (we expect this specific output from substrate).
let line_end = line
    .rsplit_once("Listening for new connections on 127.0.0.1:")
    .or_else(|| line.rsplit_once("Running JSON-RPC WS server: addr=127.0.0.1:"))
    .map(|(_,port_str)| port_str)?;

[jsdw]

Other than the suggestion LGTM