forked from molecule-ai/molecule-core
test(e2e): poll-mode + since_id cursor round-trip (#2339 PR 4)
End-to-end coverage for the canvas-chat unblocker. Exercises every moving part of the #2339 stack against a real platform instance: Phase 1 — register a workspace as delivery_mode=poll WITHOUT a URL; verify the response carries delivery_mode=poll. Phase 2 — invalid delivery_mode rejected with 400 (typo defense). Phase 3 — POST A2A to the poll-mode workspace; verify proxyA2ARequest short-circuits and returns 200 {status:queued, delivery_mode:poll, method:message/send} without ever resolving an agent URL. Phase 4 — verify the queued message appears in /activity?type=a2a_receive with the right method + payload (the polling agent reads from here). Phase 5 — since_id cursor returns ASC-ordered rows STRICTLY AFTER the cursor; the cursor row itself must NOT be replayed. Sends two follow-up messages and asserts ordering: rows[0] is the older new event, rows[-1] is the newer. Phase 6 — unknown / pruned cursor returns 410 Gone with an explanation. Phase 7 — cross-workspace cursor isolation: a UUID belonging to one workspace cannot be used to peek at another workspace's feed (returns 410, same as pruned, no info leak). Idempotent: per-run unique workspace ids (date+pid). Trap-based cleanup deletes the test rows on exit; no e2e_cleanup_all_workspaces call (see feedback_never_run_cluster_cleanup_tests_on_live_platform.md). Wired into .github/workflows/e2e-api.yml so it runs on every PR that touches workspace-server/, tests/e2e/, or the workflow file itself — same gate as the existing test_a2a_e2e + test_notify_attachments suites. Stacked on #2354 (PR 3: since_id cursor). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
Hongming Wang
parent
b5bde0399a
commit
a495b86a06
3
.github/workflows/e2e-api.yml
vendored
3
.github/workflows/e2e-api.yml
vendored
@ -169,6 +169,9 @@ jobs:
|
||||
- name: Run priority-runtimes E2E (claude-code + hermes — skips when keys absent)
|
||||
if: needs.detect-changes.outputs.api == 'true'
|
||||
run: bash tests/e2e/test_priority_runtimes_e2e.sh
|
||||
- name: Run poll-mode + since_id cursor E2E (#2339)
|
||||
if: needs.detect-changes.outputs.api == 'true'
|
||||
run: bash tests/e2e/test_poll_mode_e2e.sh
|
||||
- name: Dump platform log on failure
|
||||
if: failure() && needs.detect-changes.outputs.api == 'true'
|
||||
run: cat workspace-server/platform.log || true
|
||||
|
||||
297
tests/e2e/test_poll_mode_e2e.sh
Executable file
297
tests/e2e/test_poll_mode_e2e.sh
Executable file
@ -0,0 +1,297 @@
|
||||
#!/usr/bin/env bash
|
||||
# E2E for delivery_mode=poll + since_id cursor (#2339).
|
||||
#
|
||||
# Round-trip: register a workspace as poll-mode (no URL) → POST A2A to it →
|
||||
# verify the proxy short-circuits to {status:"queued"} → verify the message
|
||||
# appears in /activity → verify the since_id cursor returns ONLY new events
|
||||
# in ASC order → verify a stale cursor returns 410.
|
||||
#
|
||||
# Requires: platform running on localhost:8080 with migrations applied.
|
||||
# bash workspace-server/scripts/dev-start.sh
|
||||
# bash workspace-server/scripts/run-migrations.sh
|
||||
#
|
||||
# Idempotent: each run uses fresh per-script workspace ids so reruns don't
|
||||
# collide. Does NOT call e2e_cleanup_all_workspaces — see
|
||||
# `feedback_never_run_cluster_cleanup_tests_on_live_platform.md`.
|
||||
|
||||
set -euo pipefail
|
||||
|
||||
source "$(dirname "$0")/_lib.sh"
|
||||
|
||||
PASS=0
|
||||
FAIL=0
|
||||
TIMEOUT="${A2A_TIMEOUT:-30}"
|
||||
|
||||
# Per-run unique ids — same shape as test_2307_peer_visibility_staging.sh.
|
||||
# Date+pid keeps reruns isolated; trap cleans up the row at the end so
|
||||
# /workspaces lists don't accumulate test garbage on a long-lived dev DB.
|
||||
RUN_TAG="poll-e2e-$(date +%s)-$$"
|
||||
POLL_WS_ID="ws-${RUN_TAG}"
|
||||
CALLER_WS_ID="ws-caller-${RUN_TAG}"
|
||||
|
||||
cleanup() {
|
||||
local rc=$?
|
||||
# Best-effort delete; non-fatal if the row was never created.
|
||||
curl -s -X DELETE "$BASE/workspaces/$POLL_WS_ID" >/dev/null || true
|
||||
curl -s -X DELETE "$BASE/workspaces/$CALLER_WS_ID" >/dev/null || true
|
||||
exit $rc
|
||||
}
|
||||
trap cleanup EXIT
|
||||
|
||||
check() {
|
||||
local desc="$1"
|
||||
local expected="$2"
|
||||
local actual="$3"
|
||||
if echo "$actual" | grep -qF -- "$expected"; then
|
||||
echo "PASS: $desc"
|
||||
PASS=$((PASS + 1))
|
||||
else
|
||||
echo "FAIL: $desc"
|
||||
echo " expected to contain: $expected"
|
||||
echo " got: $(echo "$actual" | head -10)"
|
||||
FAIL=$((FAIL + 1))
|
||||
fi
|
||||
}
|
||||
|
||||
check_eq() {
|
||||
local desc="$1"
|
||||
local expected="$2"
|
||||
local actual="$3"
|
||||
if [ "$actual" = "$expected" ]; then
|
||||
echo "PASS: $desc"
|
||||
PASS=$((PASS + 1))
|
||||
else
|
||||
echo "FAIL: $desc"
|
||||
echo " expected: $expected"
|
||||
echo " got: $actual"
|
||||
FAIL=$((FAIL + 1))
|
||||
fi
|
||||
}
|
||||
|
||||
check_not_contains() {
|
||||
local desc="$1"
|
||||
local unexpected="$2"
|
||||
local actual="$3"
|
||||
if echo "$actual" | grep -qF -- "$unexpected"; then
|
||||
echo "FAIL: $desc"
|
||||
echo " should NOT contain: $unexpected"
|
||||
FAIL=$((FAIL + 1))
|
||||
else
|
||||
echo "PASS: $desc"
|
||||
PASS=$((PASS + 1))
|
||||
fi
|
||||
}
|
||||
|
||||
echo "=== Poll-Mode + since_id Cursor E2E (#2339) ==="
|
||||
echo " base: $BASE"
|
||||
echo " poll workspace: $POLL_WS_ID"
|
||||
echo " caller workspace: $CALLER_WS_ID"
|
||||
echo ""
|
||||
|
||||
# ---------- Phase 1: register as poll-mode ----------
|
||||
echo "--- Phase 1: Register poll-mode workspace (no URL) ---"
|
||||
|
||||
# A poll-mode workspace registers WITHOUT a URL — that's the contract from
|
||||
# PR 1 (#2348). The agent_card is required; everything else is optional.
|
||||
REG_RESP=$(curl -s -X POST "$BASE/registry/register" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d "{
|
||||
\"id\": \"$POLL_WS_ID\",
|
||||
\"delivery_mode\": \"poll\",
|
||||
\"agent_card\": {\"name\": \"poll-mode-test\"}
|
||||
}")
|
||||
|
||||
check "register accepts poll mode without URL" '"status":"registered"' "$REG_RESP"
|
||||
check "register response echoes delivery_mode=poll" '"delivery_mode":"poll"' "$REG_RESP"
|
||||
|
||||
# Capture the auth token for subsequent /activity reads (Phase 30.1).
|
||||
POLL_TOKEN=$(echo "$REG_RESP" | e2e_extract_token || true)
|
||||
if [ -z "$POLL_TOKEN" ]; then
|
||||
echo "WARN: no auth_token in register response — token-required reads will fail"
|
||||
fi
|
||||
|
||||
# ---------- Phase 2: invalid mode rejected ----------
|
||||
echo ""
|
||||
echo "--- Phase 2: Invalid delivery_mode rejected ---"
|
||||
|
||||
INVALID_RESP=$(curl -s -w '\n%{http_code}' -X POST "$BASE/registry/register" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d "{
|
||||
\"id\": \"${POLL_WS_ID}-bad\",
|
||||
\"delivery_mode\": \"webhook\",
|
||||
\"agent_card\": {\"name\": \"bad\"}
|
||||
}")
|
||||
INVALID_CODE=$(printf '%s' "$INVALID_RESP" | tail -n1)
|
||||
INVALID_BODY=$(printf '%s' "$INVALID_RESP" | sed '$d')
|
||||
|
||||
check_eq "register rejects unknown delivery_mode (HTTP 400)" "400" "$INVALID_CODE"
|
||||
check "error mentions delivery_mode" "delivery_mode" "$INVALID_BODY"
|
||||
|
||||
# ---------- Phase 3: A2A short-circuits to {status:"queued"} ----------
|
||||
echo ""
|
||||
echo "--- Phase 3: A2A to poll-mode workspace short-circuits ---"
|
||||
|
||||
A2A_RESP=$(curl -s --max-time "$TIMEOUT" -X POST "$BASE/workspaces/$POLL_WS_ID/a2a" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"jsonrpc": "2.0",
|
||||
"id": "msg-1",
|
||||
"method": "message/send",
|
||||
"params": {
|
||||
"message": {
|
||||
"role": "user",
|
||||
"parts": [{"type": "text", "text": "hello-from-e2e-1"}]
|
||||
}
|
||||
}
|
||||
}')
|
||||
|
||||
check "poll-mode A2A returns queued status" '"status":"queued"' "$A2A_RESP"
|
||||
check "queued response echoes delivery_mode=poll" '"delivery_mode":"poll"' "$A2A_RESP"
|
||||
check "queued response echoes the JSON-RPC method" '"method":"message/send"' "$A2A_RESP"
|
||||
|
||||
# ---------- Phase 4: queued message appears in /activity ----------
|
||||
echo ""
|
||||
echo "--- Phase 4: Queued message visible via /activity ---"
|
||||
|
||||
# The activity_logs INSERT runs in a goroutine — give it a moment.
|
||||
sleep 1
|
||||
|
||||
# Use bearer token if we got one; some platforms require it on /activity.
|
||||
ACTIVITY_AUTH=()
|
||||
[ -n "${POLL_TOKEN:-}" ] && ACTIVITY_AUTH=(-H "Authorization: Bearer $POLL_TOKEN")
|
||||
|
||||
ACT_RESP=$(curl -s --max-time "$TIMEOUT" "${ACTIVITY_AUTH[@]}" \
|
||||
"$BASE/workspaces/$POLL_WS_ID/activity?type=a2a_receive&limit=10")
|
||||
|
||||
check "activity feed has the queued message text" "hello-from-e2e-1" "$ACT_RESP"
|
||||
check "activity_type is a2a_receive" '"activity_type":"a2a_receive"' "$ACT_RESP"
|
||||
check "method preserved on the activity row" '"method":"message/send"' "$ACT_RESP"
|
||||
|
||||
# Pull the most-recent activity_id for use as a cursor.
|
||||
FIRST_ACTIVITY_ID=$(echo "$ACT_RESP" | python3 -c "
|
||||
import json, sys
|
||||
rows = json.load(sys.stdin)
|
||||
if not rows:
|
||||
print('')
|
||||
else:
|
||||
# Default ordering is DESC (newest-first) when no since_id is set.
|
||||
print(rows[0]['id'])
|
||||
")
|
||||
|
||||
if [ -z "$FIRST_ACTIVITY_ID" ]; then
|
||||
echo "FAIL: could not extract activity_id from /activity response"
|
||||
FAIL=$((FAIL + 1))
|
||||
exit 1
|
||||
fi
|
||||
echo " cursor candidate: $FIRST_ACTIVITY_ID"
|
||||
|
||||
# ---------- Phase 5: since_id returns only events strictly after ----------
|
||||
echo ""
|
||||
echo "--- Phase 5: since_id cursor returns ASC, strictly-after ---"
|
||||
|
||||
# Send a SECOND A2A message; it must appear in the cursor-filtered feed,
|
||||
# the FIRST message must NOT (cursor is strictly-after).
|
||||
A2A_RESP2=$(curl -s --max-time "$TIMEOUT" -X POST "$BASE/workspaces/$POLL_WS_ID/a2a" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"jsonrpc": "2.0",
|
||||
"id": "msg-2",
|
||||
"method": "message/send",
|
||||
"params": {
|
||||
"message": {
|
||||
"role": "user",
|
||||
"parts": [{"type": "text", "text": "hello-from-e2e-2"}]
|
||||
}
|
||||
}
|
||||
}')
|
||||
check "second A2A also queues" '"status":"queued"' "$A2A_RESP2"
|
||||
|
||||
sleep 1
|
||||
|
||||
CURSOR_RESP=$(curl -s --max-time "$TIMEOUT" "${ACTIVITY_AUTH[@]}" \
|
||||
"$BASE/workspaces/$POLL_WS_ID/activity?type=a2a_receive&since_id=$FIRST_ACTIVITY_ID&limit=10")
|
||||
|
||||
check "since_id feed includes the new message" "hello-from-e2e-2" "$CURSOR_RESP"
|
||||
check_not_contains "since_id feed excludes the cursor row itself" "hello-from-e2e-1" "$CURSOR_RESP"
|
||||
|
||||
# Verify ASC ordering: in a fresh cursor window with two new events the
|
||||
# array's first element must be the OLDER one (the test only sends one
|
||||
# event after the cursor, so this case is trivially "exactly one row";
|
||||
# the next sub-phase strengthens this with a second event).
|
||||
A2A_RESP3=$(curl -s --max-time "$TIMEOUT" -X POST "$BASE/workspaces/$POLL_WS_ID/a2a" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d '{
|
||||
"jsonrpc": "2.0",
|
||||
"id": "msg-3",
|
||||
"method": "message/send",
|
||||
"params": {
|
||||
"message": {
|
||||
"role": "user",
|
||||
"parts": [{"type": "text", "text": "hello-from-e2e-3"}]
|
||||
}
|
||||
}
|
||||
}')
|
||||
check "third A2A queues" '"status":"queued"' "$A2A_RESP3"
|
||||
|
||||
sleep 1
|
||||
|
||||
ASC_RESP=$(curl -s --max-time "$TIMEOUT" "${ACTIVITY_AUTH[@]}" \
|
||||
"$BASE/workspaces/$POLL_WS_ID/activity?type=a2a_receive&since_id=$FIRST_ACTIVITY_ID&limit=10")
|
||||
|
||||
# rows[0] should be msg-2 (older), rows[-1] should be msg-3 (newer) — that's
|
||||
# ASC. If the server still defaulted to DESC, rows[0] would be msg-3.
|
||||
ASC_FIRST=$(echo "$ASC_RESP" | python3 -c "
|
||||
import json, sys
|
||||
rows = json.load(sys.stdin)
|
||||
def text_of(r):
|
||||
body = r.get('request_body') or {}
|
||||
parts = (body.get('params') or {}).get('message', {}).get('parts') or []
|
||||
return ''.join(p.get('text','') for p in parts if p.get('type')=='text')
|
||||
if len(rows) < 2:
|
||||
print('NEED2_GOT_'+str(len(rows)))
|
||||
else:
|
||||
print(text_of(rows[0]) + '|' + text_of(rows[-1]))
|
||||
")
|
||||
check_eq "since_id feed orders ASC (oldest-new first, newest-new last)" \
|
||||
"hello-from-e2e-2|hello-from-e2e-3" "$ASC_FIRST"
|
||||
|
||||
# ---------- Phase 6: stale cursor returns 410 ----------
|
||||
echo ""
|
||||
echo "--- Phase 6: Stale / unknown cursor returns 410 ---"
|
||||
|
||||
GONE_RESP=$(curl -s -w '\n%{http_code}' --max-time "$TIMEOUT" "${ACTIVITY_AUTH[@]}" \
|
||||
"$BASE/workspaces/$POLL_WS_ID/activity?since_id=00000000-0000-0000-0000-000000000000")
|
||||
GONE_CODE=$(printf '%s' "$GONE_RESP" | tail -n1)
|
||||
GONE_BODY=$(printf '%s' "$GONE_RESP" | sed '$d')
|
||||
|
||||
check_eq "unknown since_id returns HTTP 410 Gone" "410" "$GONE_CODE"
|
||||
check "410 body explains how to recover" "since_id" "$GONE_BODY"
|
||||
|
||||
# ---------- Phase 7: cross-workspace cursor isolation ----------
|
||||
echo ""
|
||||
echo "--- Phase 7: Cross-workspace cursor isolation ---"
|
||||
|
||||
# Register a SECOND poll-mode workspace and try to read its activity
|
||||
# feed using a cursor from the FIRST workspace. Must 410 — the cursor
|
||||
# is workspace-scoped to prevent UUID-guessing peeks.
|
||||
REG2=$(curl -s -X POST "$BASE/registry/register" \
|
||||
-H "Content-Type: application/json" \
|
||||
-d "{
|
||||
\"id\": \"$CALLER_WS_ID\",
|
||||
\"delivery_mode\": \"poll\",
|
||||
\"agent_card\": {\"name\": \"poll-cross-test\"}
|
||||
}")
|
||||
check "second poll-mode workspace registers" '"status":"registered"' "$REG2"
|
||||
CALLER_TOKEN=$(echo "$REG2" | e2e_extract_token || true)
|
||||
CROSS_AUTH=()
|
||||
[ -n "${CALLER_TOKEN:-}" ] && CROSS_AUTH=(-H "Authorization: Bearer $CALLER_TOKEN")
|
||||
|
||||
CROSS_RESP=$(curl -s -w '\n%{http_code}' --max-time "$TIMEOUT" "${CROSS_AUTH[@]}" \
|
||||
"$BASE/workspaces/$CALLER_WS_ID/activity?since_id=$FIRST_ACTIVITY_ID")
|
||||
CROSS_CODE=$(printf '%s' "$CROSS_RESP" | tail -n1)
|
||||
check_eq "cross-workspace cursor blocked with 410 (no info leak)" "410" "$CROSS_CODE"
|
||||
|
||||
# ---------- Results ----------
|
||||
echo ""
|
||||
echo "=== Results: $PASS passed, $FAIL failed ==="
|
||||
[ "$FAIL" -eq 0 ]
|
||||
Loading…
Reference in New Issue
Block a user