nwaku/tests/test_helpers.nim
Hanno Cornelius 16c85db43c
chore: update submodules and bump Nim to 1.6.6 (#1307)
* chore: update submodules

* fix: libp2p now provides callback to update peer addrs

* fix: breaking change in EthereumNode constructor

* fix: contentType type has changed (again)

* fix: explicit future type

* fix: nim 1.6.6 error handling requirements

* fix: missed a spot - peer info addrs in sim2

* fix: help compiler a bit here
2022-10-28 12:51:46 +03:00

46 lines
1.5 KiB
Nim

import
chronos, bearssl/rand,
eth/[keys, p2p]
import libp2p/crypto/crypto
var nextPort = 30303
proc localAddress*(port: int): Address =
let port = Port(port)
result = Address(udpPort: port, tcpPort: port,
ip: parseIpAddress("127.0.0.1"))
proc setupTestNode*(
rng: ref HmacDrbgContext,
capabilities: varargs[ProtocolInfo, `protocolInfo`]): EthereumNode =
let
keys1 = keys.KeyPair.random(rng[])
address = localAddress(nextPort)
result = newEthereumNode(keys1, address, NetworkId(1),
addAllCapabilities = false,
bindUdpPort = address.udpPort, # Assume same as external
bindTcpPort = address.tcpPort, # Assume same as external
rng = rng)
nextPort.inc
for capability in capabilities:
result.addCapability capability
# Copied from here: https://github.com/status-im/nim-libp2p/blob/d522537b19a532bc4af94fcd146f779c1f23bad0/tests/helpers.nim#L28
type RngWrap = object
rng: ref rand.HmacDrbgContext
var rngVar: RngWrap
proc getRng(): ref rand.HmacDrbgContext =
# TODO if `rngVar` is a threadvar like it should be, there are random and
# spurious compile failures on mac - this is not gcsafe but for the
# purpose of the tests, it's ok as long as we only use a single thread
{.gcsafe.}:
if rngVar.rng.isNil:
rngVar.rng = crypto.newRng()
rngVar.rng
template rng*(): ref rand.HmacDrbgContext =
getRng()