mirror of
https://github.com/vitorpamplona/amethyst.git
synced 2026-10-05 19:28:25 +00:00
Claude
9e41858c8f
Six new harness tests against MDK, all green in one clean run: 20 avatar-url amy->wn 21 avatar-url wn->amy 22 edit amy->wn 23 deletion amy->wn 24 media-v2 amy->wn 25 media-v2 wn->amy The media pair needed a blob store, so the harness now runs a loopback Blossom server of its own (`blossom-server.py`, PUT /upload + GET /<sha256>). It holds nothing but ciphertext — the file key comes from each group's MLS exporter — so a download that hashes back to the original bytes is proof both implementations derived the same key. That is the whole point of tests 24 and 25, and they pass in both directions. Test 20 sends a URL that is deliberately NOT normalized (`https://Example.COM:443/a/./avatars/../pic.png`) and asserts both what we store and what wn reads back. Normalization is the wire format for this component — a decoder rejects bytes that differ from its own serialization — so a disagreement here is a group the other side cannot read at all, not a cosmetic difference. Tests 22 and 23 assert what the protocol actually says rather than what a renderer happens to do. For the edit that means a well-formed kind:1009 reaching wn (one `e` tag naming the target, the replacement as its body, the right author) plus our own reader applying the overlay — MDK's storage deliberately leaves the original row's body alone and lets the client compute the chain, so asserting on painted text would be testing its TUI. For the deletion it means the `deleted` flag on wn's materialized timeline, which is its user-visible truth. Two harness bugs surfaced while getting there, both of the kind that make a failure unreadable rather than wrong. `run.env` — where tests hand each other group ids — survived the per-run state wipe, so a `--tests` subset that consumed without re-creating failed on "not a member" for a group id from a previous run. And `amy_json` read `$?` inside `if ! cmd`, where it is the status of the negation, so every failure reported "exit 0". Push (kind 451) has no cross-implementation test here and cannot: the owner proof is an UNPUBLISHED event handed to a push service, the reference CLI exposes no command that emits one, and the harness runs no push service. There is nothing for two implementations to disagree about on the wire. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016kCuA6tc4JQzHPCDd39GHq
224 lines
8.0 KiB
Bash
Executable File
224 lines
8.0 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
#
|
|
# marmot-interop-headless.sh — zero-prompt, zero-internet interop harness.
|
|
#
|
|
# Drives Identity A via the `amy` CLI (./gradlew :cli:installDist) and
|
|
# Identities B/C via MDK's `wn`/`wnd`. Spins up a local
|
|
# nostr-rs-relay on ws://127.0.0.1:$RELAY_PORT so nothing ever leaves the
|
|
# machine. Matches the 13 test scenarios in marmot-interop.sh but without
|
|
# any human prompts — all checks run to completion and the exit code
|
|
# reflects pass/fail totals.
|
|
#
|
|
# Usage: ./marmot-interop-headless.sh [--port N] [--no-build] [--reuse-state]
|
|
# [--tests "name ..."]
|
|
#
|
|
set -uo pipefail
|
|
|
|
SCRIPT_DIR="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)"
|
|
REPO_ROOT="$(cd -- "$SCRIPT_DIR/../../.." && pwd)"
|
|
TESTS_DIR="$(cd -- "$SCRIPT_DIR/.." && pwd)"
|
|
STATE_DIR="$SCRIPT_DIR/state-headless"
|
|
LOG_DIR="$STATE_DIR/logs"
|
|
# A is the amy account inside the fake $HOME=$STATE_DIR layout. B and
|
|
# C are wnd (whitenoise) state dirs — different binary, separate
|
|
# convention, so they stay as plain $STATE_DIR siblings.
|
|
A_DIR="$STATE_DIR/.amy/A"
|
|
B_DIR="$STATE_DIR/B"
|
|
C_DIR="$STATE_DIR/C"
|
|
B_SOCKET="$B_DIR/wnd.sock"
|
|
C_SOCKET="$C_DIR/wnd.sock"
|
|
|
|
RUN_TS="$(date +%Y%m%d-%H%M%S)"
|
|
LOG_FILE="$LOG_DIR/run-$RUN_TS.log"
|
|
RESULTS_FILE="$STATE_DIR/results-$RUN_TS.tsv"
|
|
|
|
WN_REPO="${WN_REPO:-$SCRIPT_DIR/state/mdk}"
|
|
WN_BIN="$WN_REPO/target/release/wn"
|
|
WND_BIN="$WN_REPO/target/release/wnd"
|
|
AMY_BIN="$REPO_ROOT/cli/build/install/amy/bin/amy"
|
|
|
|
# Local relay wiring — cloned + built during preflight, started on
|
|
# $RELAY_PORT. The harness never touches the public internet for test
|
|
# traffic; wn/wnd/amy all point at this one loopback endpoint.
|
|
#
|
|
# Bind to 127.0.0.2 rather than 127.0.0.1: Quartz's RelayUrlNormalizer
|
|
# strips literal 127.0.0.1 / localhost / 192.168.* out of NIP-17 inbox
|
|
# (kind:10050) and KeyPackage (kind:10051) relay-list events as a
|
|
# privacy guard, which would silently leave the harness publishing to
|
|
# Amethyst's public defaults instead of the loopback. 127.0.0.2 is
|
|
# still pure loopback (no network traffic) but isn't on the strip list.
|
|
RELAY_HOST="${RELAY_HOST:-127.0.0.2}"
|
|
RELAY_REPO="${RELAY_REPO:-$STATE_DIR/nostr-rs-relay}"
|
|
RELAY_BIN="$RELAY_REPO/target/release/nostr-rs-relay"
|
|
RELAY_DATA="$STATE_DIR/relay"
|
|
RELAY_PORT="${RELAY_PORT:-8080}"
|
|
RELAY_URL="ws://$RELAY_HOST:$RELAY_PORT"
|
|
|
|
# MDK's reference QUIC broker, for the agent-text-stream tests. Loopback like
|
|
# everything else; the tests skip when the binary was never built.
|
|
BROKER_BIN="$WN_REPO/target/release/marmot-quic-broker"
|
|
BROKER_HOST="${BROKER_HOST:-127.0.0.1}"
|
|
BROKER_PORT="${BROKER_PORT:-4455}"
|
|
BROKER_URI="quic://$BROKER_HOST:$BROKER_PORT"
|
|
BROKER_PID=""
|
|
|
|
# A loopback Blossom blob store for the encrypted-media tests. Ciphertext only:
|
|
# the file key comes from each group's MLS exporter and never reaches it.
|
|
BLOSSOM_HOST="${BLOSSOM_HOST:-127.0.0.1}"
|
|
BLOSSOM_PORT="${BLOSSOM_PORT:-8456}"
|
|
BLOSSOM_URL="http://$BLOSSOM_HOST:$BLOSSOM_PORT"
|
|
BLOSSOM_PID=""
|
|
|
|
NO_BUILD=0
|
|
# Every run starts from empty stores. wnd already wipes B's and C's data dirs
|
|
# on each start, but A's amy home and the relay's SQLite file used to survive,
|
|
# and the leftovers are not inert: a KeyPackage A published in an earlier run
|
|
# is still on the relay for B to invite with, an old group's kind:445 events
|
|
# still arrive and fail to decrypt, and A's cursors still say it has seen them.
|
|
# That drift is what made tests 03 and 08 fail on a dirty tree and pass on a
|
|
# clean one. Pass --reuse-state when you are deliberately debugging carry-over.
|
|
RESET_STATE=1
|
|
# Space-separated test function names; empty means the full suite below.
|
|
ONLY_TESTS=""
|
|
|
|
# Required as of MDK 0.9.x. `validate_relay_url` accepts `wss://`
|
|
# unconditionally but `ws://` only for a loopback host AND only behind this
|
|
# explicit opt-in; without it wnd refuses the harness relay with "invalid relay
|
|
# URL" and exits before creating its socket. 127.0.0.2 is inside 127.0.0.0/8 and
|
|
# already passes MDK's own loopback test, so the env var is the gate, not the
|
|
# address. Exported once here so `wn` and `wnd` both inherit it — `wn` runs the
|
|
# same validation on any relay argument.
|
|
export WN_ALLOW_LOOPBACK_RELAYS=1
|
|
# Same shape for blob stores: wn refuses a loopback Blossom endpoint unless it
|
|
# is told this is a dev/test run. The harness's blob store is loopback by
|
|
# design — nothing in a test run may leave the machine.
|
|
export WN_ALLOW_LOOPBACK_BLOB_ENDPOINTS=1
|
|
|
|
A_NPUB=""
|
|
A_HEX=""
|
|
B_NPUB=""
|
|
B_HEX=""
|
|
C_NPUB=""
|
|
C_HEX=""
|
|
|
|
while [[ $# -gt 0 ]]; do
|
|
case "$1" in
|
|
--port) RELAY_PORT="$2"; RELAY_URL="ws://$RELAY_HOST:$RELAY_PORT"; shift ;;
|
|
--host) RELAY_HOST="$2"; RELAY_URL="ws://$RELAY_HOST:$RELAY_PORT"; shift ;;
|
|
--no-build) NO_BUILD=1 ;;
|
|
--reuse-state) RESET_STATE=0 ;;
|
|
--tests) ONLY_TESTS="$2"; shift ;;
|
|
-h|--help)
|
|
sed -n '3,14p' "${BASH_SOURCE[0]}" | sed 's/^# \?//'
|
|
exit 0 ;;
|
|
*) printf 'unknown flag: %s\n' "$1" >&2; exit 2 ;;
|
|
esac
|
|
shift
|
|
done
|
|
|
|
if [[ $RESET_STATE -eq 1 && -d "$STATE_DIR" ]]; then
|
|
# Keep the relay checkout + its build (minutes to rebuild) and the log and
|
|
# results history; drop everything that holds protocol state.
|
|
#
|
|
# run.env counts as protocol state: it is where tests hand each other group
|
|
# ids. Leaving it behind a wipe leaves ids naming groups nobody is in any
|
|
# more, and a later `--tests` subset that consumes without re-creating then
|
|
# fails on "not a member" for a group id from a previous run.
|
|
rm -rf "$STATE_DIR/.amy" "$B_DIR" "$C_DIR" "$RELAY_DATA" "$STATE_DIR/run.env"
|
|
fi
|
|
|
|
mkdir -p "$STATE_DIR" "$LOG_DIR" "$B_DIR/logs" "$C_DIR/logs"
|
|
: >"$LOG_FILE"
|
|
: >"$RESULTS_FILE"
|
|
|
|
# Reuse colours / logging / dump_daemon_diagnostics from the interactive harness.
|
|
# shellcheck source=../lib.sh
|
|
source "$TESTS_DIR/lib.sh"
|
|
|
|
# shellcheck source=setup.sh
|
|
source "$SCRIPT_DIR/setup.sh"
|
|
# shellcheck source=../headless/helpers.sh
|
|
source "$TESTS_DIR/headless/helpers.sh"
|
|
# shellcheck source=tests-create.sh
|
|
source "$SCRIPT_DIR/tests-create.sh"
|
|
# shellcheck source=tests-manage.sh
|
|
source "$SCRIPT_DIR/tests-manage.sh"
|
|
# shellcheck source=tests-extras.sh
|
|
source "$SCRIPT_DIR/tests-extras.sh"
|
|
# shellcheck source=tests-media.sh
|
|
source "$SCRIPT_DIR/tests-media.sh"
|
|
|
|
# Make sure Ctrl+C / SIGTERM / SIGHUP all run the full cleanup path —
|
|
# otherwise wnd is nohup'd and keeps running after the script dies,
|
|
# and the next run's `ss` check then complains the port is in use.
|
|
# Trap INT/TERM/HUP forces `exit`, which triggers the EXIT handler once.
|
|
cleanup() {
|
|
local rc=$?
|
|
trap - EXIT INT TERM HUP
|
|
stop_daemons
|
|
stop_quic_broker
|
|
stop_blossom
|
|
stop_local_relay
|
|
print_summary
|
|
exit "$rc"
|
|
}
|
|
trap cleanup EXIT
|
|
trap 'exit 130' INT
|
|
trap 'exit 143' TERM
|
|
trap 'exit 129' HUP
|
|
|
|
banner "Marmot headless interop harness ($RUN_TS)"
|
|
preflight
|
|
start_local_relay
|
|
start_quic_broker || true
|
|
start_blossom || true
|
|
start_daemon B "$B_DIR" "$B_SOCKET"
|
|
start_daemon C "$C_DIR" "$C_SOCKET"
|
|
ensure_identity_a
|
|
ensure_identity B
|
|
ensure_identity C
|
|
configure_relays
|
|
|
|
ALL_TESTS=(
|
|
test_01_keypackage_discovery
|
|
test_02_a_creates_group
|
|
test_03_b_creates_group
|
|
test_04_three_member_group
|
|
test_05_b_adds_a_existing
|
|
test_06_member_removal
|
|
test_07_metadata_rename
|
|
test_08_admin_promote_demote
|
|
test_17_group_image_commit
|
|
test_09_reply_react_unreact
|
|
test_10_concurrent_commits
|
|
test_11_leave_group
|
|
test_12_offline_catchup
|
|
test_13_keypackage_rotation
|
|
test_14_wn_removes_a
|
|
test_15_wn_member_leaves
|
|
test_16_wn_keypackage_rotation
|
|
test_18_agent_stream_amy_publishes
|
|
test_19_agent_stream_wn_publishes
|
|
test_20_avatar_url_amy_to_wn
|
|
test_21_avatar_url_wn_to_amy
|
|
test_22_message_edit_amy_to_wn
|
|
test_23_deletion_amy_to_wn
|
|
test_24_media_v2_amy_to_wn
|
|
test_25_media_v2_wn_to_amy
|
|
)
|
|
|
|
# --tests runs a subset in the order given. Most tests read state a previous
|
|
# one saved (GROUP_02, GROUP_05, …), so a subset that skips a producer will
|
|
# report `skip`, not a false failure.
|
|
if [[ -n "$ONLY_TESTS" ]]; then
|
|
read -r -a ALL_TESTS <<<"$ONLY_TESTS"
|
|
fi
|
|
|
|
for t in "${ALL_TESTS[@]}"; do
|
|
if ! declare -F "$t" >/dev/null; then
|
|
fail_msg "unknown test: $t"
|
|
continue
|
|
fi
|
|
"$t"
|
|
done
|