scripts/run_all_local.sh

#!/usr/bin/env bash
set -euo pipefail

ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
cd "$ROOT_DIR"
source "$ROOT_DIR/scripts/venv_utils.sh"

PROFILE="full"
SKIP_UI="false"
SKIP_TRAIN="false"
FRONTIER_READY="false"

for arg in "$@"; do
  case "$arg" in
    --full) PROFILE="full" ;;
    --quick) PROFILE="quick" ;;
    --skip-ui) SKIP_UI="true" ;;
    --skip-train) SKIP_TRAIN="true" ;;
    --frontier-ready) FRONTIER_READY="true" ;;
    *)
      echo "Unknown flag: $arg"
      exit 1
      ;;
  esac
done

mkdir -p outputs/logs outputs/reports checkpoints
LOG_DIR="outputs/logs"
ENV_LOG="$LOG_DIR/env.log"
API_LOG="$LOG_DIR/api.log"
UI_LOG="$LOG_DIR/ui.log"
PIDS_FILE="$LOG_DIR/pids.txt"
rm -f "$PIDS_FILE"

cleanup() {
  if [[ -f "$PIDS_FILE" ]]; then
    while IFS= read -r pid; do
      if [[ -n "$pid" ]] && kill -0 "$pid" >/dev/null 2>&1; then
        kill "$pid" >/dev/null 2>&1 || true
      fi
    done < "$PIDS_FILE"
  fi
}
trap cleanup EXIT

echo "[run_all_local] Installing python deps"
ensure_polyguard_venv
if ! sync_polyguard_requirements >/dev/null 2>&1; then
  echo "[run_all_local] venv requirements sync skipped (offline or restricted environment)"
fi
if ! venv_has_required_imports >/dev/null 2>&1; then
  echo "[run_all_local] Missing required Python dependencies in .venv. Run: bash scripts/bootstrap_venv.sh"
  exit 1
fi
activate_polyguard_path
export MPLCONFIGDIR="${MPLCONFIGDIR:-$ROOT_DIR/outputs/.mplconfig}"
mkdir -p "$MPLCONFIGDIR"

if [[ "$PROFILE" == "quick" ]]; then
  export POLYGUARD_RISK_DATASET_SIZE=180
  export POLYGUARD_DOSE_DATASET_SIZE=120
  export POLYGUARD_SUPERVISOR_EPISODES=3
  export POLYGUARD_PLANNER_EPISODES=4
  export POLYGUARD_DOSING_EPISODES=3
else
  export POLYGUARD_RISK_DATASET_SIZE=500
  export POLYGUARD_DOSE_DATASET_SIZE=300
  export POLYGUARD_SUPERVISOR_EPISODES=6
  export POLYGUARD_PLANNER_EPISODES=8
  export POLYGUARD_DOSING_EPISODES=6
fi

if [[ "${POLYGUARD_ENABLE_OLLAMA:-false}" == "true" ]] && command -v ollama >/dev/null 2>&1; then
  MODEL_NAME="${POLYGUARD_OLLAMA_MODEL:-qwen2.5:3b-instruct}"
  echo "[run_all_local] Ensuring Ollama model is available: $MODEL_NAME"
  ollama pull "$MODEL_NAME" >/dev/null 2>&1 || echo "[run_all_local] ollama pull skipped/failed; continuing"
else
  echo "[run_all_local] Ollama disabled/unavailable; using HF Transformers path"
fi

echo "[run_all_local] Building data assets"
"$POLYGUARD_PYTHON_BIN" scripts/build_synthetic_patients.py
"$POLYGUARD_PYTHON_BIN" scripts/ingest_open_drug_sources.py
"$POLYGUARD_PYTHON_BIN" scripts/build_drug_knowledge.py
"$POLYGUARD_PYTHON_BIN" scripts/build_retrieval_index.py
"$POLYGUARD_PYTHON_BIN" scripts/build_scenarios.py
"$POLYGUARD_PYTHON_BIN" scripts/bootstrap_data.py
"$POLYGUARD_PYTHON_BIN" scripts/build_training_corpus.py --profile "$([[ "$PROFILE" == "quick" ]] && echo small || echo massive)" --with-local --with-synthetic
if [[ "${POLYGUARD_ALLOW_WEB_FETCH:-false}" == "true" ]]; then
  "$POLYGUARD_PYTHON_BIN" scripts/crawl_labels_and_guidelines.py
  "$POLYGUARD_PYTHON_BIN" scripts/build_training_corpus.py --profile small --with-local --with-synthetic --enable-web-fallback
fi

echo "[run_all_local] Training predictive models"
"$POLYGUARD_PYTHON_BIN" scripts/train_graph_model.py
"$POLYGUARD_PYTHON_BIN" scripts/train_risk_model.py
"$POLYGUARD_PYTHON_BIN" scripts/train_dose_model.py

if [[ "$SKIP_TRAIN" != "true" ]]; then
  echo "[run_all_local] Running training and evaluation"
  "$POLYGUARD_PYTHON_BIN" scripts/generate_sft_data.py
  "$POLYGUARD_PYTHON_BIN" scripts/train_sft.py
  "$POLYGUARD_PYTHON_BIN" scripts/train_grpo_supervisor.py
  "$POLYGUARD_PYTHON_BIN" scripts/train_grpo_planner.py
  "$POLYGUARD_PYTHON_BIN" scripts/train_grpo_dosing.py
  "$POLYGUARD_PYTHON_BIN" scripts/train_grpo_trl.py --max-steps "$([[ "$PROFILE" == "quick" ]] && echo 3 || echo 12)"
  if [[ -d "checkpoints/sft_adapter" ]]; then
    "$POLYGUARD_PYTHON_BIN" scripts/merge_adapters_safe.py --adapter-dir checkpoints/sft_adapter --output-dir checkpoints/merged
  else
    echo "[run_all_local] sft_adapter not found; skipping adapter merge and using inference fallback checks"
  fi
  "$POLYGUARD_PYTHON_BIN" scripts/test_inference_postsave.py --samples "$([[ "$PROFILE" == "quick" ]] && echo 1 || echo 3)"
  "$POLYGUARD_PYTHON_BIN" scripts/evaluate_policy_ablations.py --episodes "$([[ "$PROFILE" == "quick" ]] && echo 3 || echo 8)"
  "$POLYGUARD_PYTHON_BIN" scripts/evaluate_baselines.py
  "$POLYGUARD_PYTHON_BIN" scripts/evaluate_all.py
  "$POLYGUARD_PYTHON_BIN" scripts/evaluate_compare_runs.py --baseline outputs/reports/baselines.json --candidate outputs/reports/benchmark_report.json --output outputs/reports/improvement_report.json
  "$POLYGUARD_PYTHON_BIN" scripts/benchmark_inference.py --provider transformers --model "${POLYGUARD_HF_MODEL:-Qwen/Qwen2.5-0.5B-Instruct}" --runs "$([[ "$PROFILE" == "quick" ]] && echo 2 || echo 5)"
  "$POLYGUARD_PYTHON_BIN" scripts/run_robustness_suite.py
  "$POLYGUARD_PYTHON_BIN" scripts/acceptance_gate.py
fi

if [[ "$FRONTIER_READY" == "true" ]]; then
  "$POLYGUARD_PYTHON_BIN" - <<'PY'
import json
from pathlib import Path
payload = {
    "frontier_models": ["qwen2.5:7b-instruct", "qwen2.5:14b-instruct"],
    "deployment_mode": "hf_or_vllm_ready",
    "notes": "Baseline complete; ready for larger model sweep."
}
out = Path("outputs/reports/frontier_ready.json")
out.parent.mkdir(parents=True, exist_ok=True)
out.write_text(json.dumps(payload, ensure_ascii=True, indent=2), encoding="utf-8")
print("frontier_ready_manifest_written")
PY
fi

echo "[run_all_local] Starting env service"
"$POLYGUARD_PYTHON_BIN" -m app.env.fastapi_app >"$ENV_LOG" 2>&1 &
echo "$!" >> "$PIDS_FILE"
sleep 2

echo "[run_all_local] Starting API service"
"$POLYGUARD_PYTHON_BIN" -m app.api >"$API_LOG" 2>&1 &
echo "$!" >> "$PIDS_FILE"
sleep 2

echo "[run_all_local] Health checks"
SERVICES_UP="false"
if curl -fsS http://127.0.0.1:8100/health >/dev/null 2>&1 && curl -fsS http://127.0.0.1:8200/health >/dev/null 2>&1; then
  SERVICES_UP="true"
  curl -fsS http://127.0.0.1:8100/metadata >/dev/null 2>&1 || true
  echo "[run_all_local] service health checks passed"
else
  echo "[run_all_local] network health checks unavailable, using in-process API/env smoke fallback"
  if ! "$POLYGUARD_PYTHON_BIN" -m pytest tests/test_api.py tests/test_remote_env.py >/dev/null 2>&1; then
    echo "[run_all_local] fallback smoke failed"
    exit 1
  fi
fi

if [[ "$SKIP_UI" != "true" ]]; then
  if command -v npm >/dev/null 2>&1; then
    echo "[run_all_local] Starting UI"
    (
      cd app/ui/frontend
      npm install >/dev/null
      npm run dev
    ) >"$UI_LOG" 2>&1 &
    echo "$!" >> "$PIDS_FILE"
    sleep 3
  else
    echo "[run_all_local] npm not found, UI skipped."
  fi
fi

if [[ "$SERVICES_UP" == "true" ]]; then
  bash scripts/smoke_test_all.sh >/dev/null 2>&1
fi

echo "[run_all_local] Completed profile=$PROFILE skip_ui=$SKIP_UI skip_train=$SKIP_TRAIN frontier_ready=$FRONTIER_READY"
echo "[run_all_local] Logs in $LOG_DIR"
if [[ "$SERVICES_UP" == "true" ]]; then
  echo "[run_all_local] Services are running. Press Ctrl-C to stop."
  wait
else
  echo "[run_all_local] Completed in fallback mode (services could not bind in this environment)."
fi