scripts/check.sh

#!/usr/bin/env bash
# scripts/check.sh — one-command health report for the AWS deployment.
#
# Written for an operator working from a phone terminal: five separate
# commands is five chances to mistype on a small keyboard, so this is the
# single "how is my bot doing?" answer. Read-only — it never restarts,
# edits or trades anything.
#
#   bash scripts/check.sh
#
# Nothing here prints a secret: the status endpoint carries no keys, the
# token is read into a variable for the curl call and never echoed, and the
# log tail goes through the bot's own redaction filter (modules/logger.py)
# before it is ever written to disk.

cd "$(dirname "$0")/.." || exit 1
SERVICE=garden-angel-terminal

echo "════════ SERVICE ════════"
STATE=$(systemctl is-active "$SERVICE" 2>/dev/null)
echo "state: $STATE"
if [ "$STATE" != "active" ]; then
    echo "⚠️  NOT RUNNING — recent failures:"
    sudo journalctl -u "$SERVICE" -n 15 --no-pager 2>/dev/null | tail -15
fi
systemctl show "$SERVICE" -p NRestarts 2>/dev/null
echo "uptime: $(systemctl show "$SERVICE" -p ActiveEnterTimestamp --value 2>/dev/null)"

echo
echo "════════ BOT STATUS ════════"
PORT=$(grep -m1 '^BOT_API_PORT=' .env 2>/dev/null | cut -d= -f2-)
PORT=${PORT:-8081}
curl -s --max-time 5 "localhost:${PORT}/status" \
  | python3 -c 'import json,sys
try:
    d = json.load(sys.stdin)
except Exception:
    print("  (no response — API not up yet)"); raise SystemExit
for k, v in d.items():
    print(f"  {k:20} {v}")' 2>/dev/null || echo "  (API not responding)"

echo
echo "════════ TRADE JOURNAL ════════"
if [ -f data/trade_journal.csv ]; then
    venv/bin/python scripts/journal.py 2>/dev/null | sed 's/^/  /'
else
    echo "  no journal yet — created on the first scan cycle"
fi

echo
echo "════════ RECENT ACTIVITY ════════"
LOGF=$(grep -m1 '^LOG_FILE=' .env 2>/dev/null | cut -d= -f2-)
if [ -n "$LOGF" ] && [ -f "$LOGF" ]; then
    grep -iE "signal|executed|scan|profit|error|warning" "$LOGF" 2>/dev/null \
      | tail -12 | cut -c1-160 | sed 's/^/  /'
else
    sudo journalctl -u "$SERVICE" -n 12 --no-pager 2>/dev/null | cut -c1-160 | sed 's/^/  /'
fi

echo
echo "════════ RESOURCES ════════"
echo "  disk: $(df -h . | awk 'NR==2 {print $4" free of "$2}')"
echo "  mem:  $(free -h | awk 'NR==2 {print $7" available of "$2}')"
echo
echo "Deeper checks:  bash scripts/check.sh --speed"

if [ "$1" = "--speed" ]; then
    echo
    echo "════════ SPEEDTEST ════════"
    set -a; . ./.env 2>/dev/null; set +a
    venv/bin/python scripts/speedtest.py
fi

if [ "$1" = "--tests" ]; then
    echo
    echo "════════ REGRESSION TESTS ════════"

    # v1.70 — A FAILING TEST NOW FAILS THIS SCRIPT.
    #
    # These three lines used to be `... | tail -2` and `... | grep -E`.
    # In a pipeline the shell reports the exit status of the LAST command,
    # so `grep` matching its own pattern returned 0 and the red ❌ it had
    # just printed was thrown away. Every run of `check.sh --tests` exited
    # 0 no matter what — the tests could print failure but could not BE a
    # failure, which is the one thing a regression suite exists to do.
    #
    # This is the same class of bug scripts/test_go_live.sh was written to
    # catch in go_live.sh (printing ✅ after a warning), living in the
    # runner that was supposed to be catching it.
    #
    # Output is still filtered — the point of the filtering was a readable
    # summary on a phone, and that is worth keeping — but now via a temp
    # file so the test's own status survives, and every failure is
    # re-printed IN FULL rather than reduced to one line.
    _test_failures=0
    _run_test() {
        local script="$1" filter="$2" out
        out="$(mktemp)"
        if venv/bin/python "$script" >"$out" 2>&1; then
            eval "$filter" <"$out"
        else
            _test_failures=$((_test_failures + 1))
            echo
            echo "❌ FAILED: $script"
            sed 's/^/   /' "$out"
            echo
        fi
        rm -f "$out"
    }

    _run_test scripts/test_blackout.py       "tail -2"
    _run_test scripts/test_pipeline.py       "grep -E '✅ all|❌' || true"
    _run_test scripts/test_signal_quality.py "grep -E '✅ all|❌' || true"
    _run_test scripts/test_hotpath.py        "grep -E '✅ all|❌' || true"
    _run_test scripts/test_ws_state.py       "grep -E '✅ all|❌' || true"
    _run_test scripts/test_feed_client.py    "grep -E '✅ all|❌' || true"
    _run_test scripts/test_truth.py          "grep -E '✅ all|❌' || true"
    _run_test scripts/test_agent.py          "grep -E '✅ all|❌' || true"
    _run_test scripts/test_scripts.py        "grep -E '✅ all|❌' || true"
    _run_test scripts/test_limiter.py        "grep -E '✅ all|❌' || true"
    _run_test scripts/test_ai_model.py       "grep -E '✅ all|❌' || true"

    echo
    if [ "$_test_failures" -gt 0 ]; then
        echo "❌ $_test_failures test script(s) FAILED — see full output above"
        exit 1
    fi
    echo "✅ all test scripts passed"
fi

if [ "$1" = "--net" ]; then
    echo
    bash scripts/latency_audit.sh
fi