#!/usr/bin/env bash
# =============================================================================
# evolution-daemon.sh — perpetual Push3 evolution loop
#
# Wraps the full per-run cycle so that a single command starts continuous
# evolution on a DigitalOcean (or similar) box with no manual intervention.
#
# Usage:
#   cd <repo-root>
#   BASE_RPC_URL=https://mainnet.base.org \
#   ./tools/push3-evolution/evolution-daemon.sh
#
# Per-run cycle:
#   1. git pull origin master          — sync latest code
#   2. git apply evolution.patch       — unbounded AW, gas limit override
#   3. Clean stale /tmp/tmp.* dirs     — prevent interference from killed runs
#   4. Run evolve.sh                   — full evolution pipeline
#   5. Results already in evolved/run_NNN/ (evolve.sh auto-increments)
#   6. Admission already done by evolve.sh (step 5 of its pipeline)
#   7. Write summary report            — best fitness, improvement, duration
#   8. Notify via openclaw             — SSH to main VPS
#   9. git apply --reverse             — revert evolution patches
#  10. Loop
#
# Configuration:
#   Load from tools/push3-evolution/evolution.conf (co-located with this script).
#   BASE_RPC_URL must be set in the environment or in evolution.conf.
#
# Signals:
#   SIGINT / SIGTERM — finish the current run cleanly, then exit.
# =============================================================================

set -euo pipefail

SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
REPO_ROOT="$(cd "$SCRIPT_DIR/../.." && pwd)"
CONF_FILE="$SCRIPT_DIR/evolution.conf"
PATCH_FILE="$SCRIPT_DIR/evolution.patch"
EVOLVE_SH="$SCRIPT_DIR/evolve.sh"

# =============================================================================
# Load config
# =============================================================================

if [ ! -f "$CONF_FILE" ]; then
  echo "[daemon] ERROR: config file not found: $CONF_FILE" >&2
  exit 2
fi

# Source the config so all variables are available.
# shellcheck source=evolution.conf
. "$CONF_FILE"

# Required: BASE_RPC_URL may come from the environment or from the conf file.
BASE_RPC_URL="${BASE_RPC_URL:-}"
if [ -z "$BASE_RPC_URL" ] && [ "${EVAL_MODE:-revm}" = "revm" ]; then
  echo "[daemon] ERROR: BASE_RPC_URL is not set. Set it in the environment or in $CONF_FILE" >&2
  exit 2
fi
export BASE_RPC_URL

# Resolve seed path (relative to repo root if not absolute).
SEED="${SEED:-tools/push3-evolution/seeds/optimizer_v3.push3}"
if [[ "$SEED" != /* ]]; then
  SEED="$REPO_ROOT/$SEED"
fi

# Optional defaults for variables the conf might not set.
EVAL_MODE="${EVAL_MODE:-revm}"
POPULATION="${POPULATION:-20}"
GENERATIONS="${GENERATIONS:-30}"
MUTATION_RATE="${MUTATION_RATE:-1}"
ELITES="${ELITES:-2}"
DIVERSE_SEEDS="${DIVERSE_SEEDS:-true}"
OPENCLAW_SSH_TARGET="${OPENCLAW_SSH_TARGET:-}"

# Output directory (relative to repo root so evolve.sh's auto-increment finds prior runs).
OUTPUT_DIR="$REPO_ROOT/evolved"

# =============================================================================
# Patch state tracking
# =============================================================================

PATCH_APPLIED=false

cleanup_patch() {
  if [ "$PATCH_APPLIED" = "true" ]; then
    echo "[daemon] Reverting evolution patches…" >&2
    (cd "$REPO_ROOT" && git apply --reverse "$PATCH_FILE") 2>/dev/null || true
    PATCH_APPLIED=false
  fi
}

# =============================================================================
# Signal handling — finish current run, then exit cleanly
# =============================================================================

STOP_REQUESTED=false

handle_signal() {
  echo "" >&2
  echo "[daemon] Stop requested — will exit after current run completes." >&2
  STOP_REQUESTED=true
}

trap handle_signal SIGINT SIGTERM
trap cleanup_patch EXIT

# =============================================================================
# Helpers
# =============================================================================

log() {
  echo "[daemon] $*" >&2
}

ts() {
  date -u '+%Y-%m-%dT%H:%M:%SZ'
}

notify() {
  local msg="$*"
  if [ -n "$OPENCLAW_SSH_TARGET" ]; then
    ssh "$OPENCLAW_SSH_TARGET" "openclaw system event '$msg'" 2>/dev/null || true
  fi
}

# =============================================================================
# Pre-flight checks
# =============================================================================

[ -f "$EVOLVE_SH" ] || { log "ERROR: evolve.sh not found at $EVOLVE_SH"; exit 2; }
[ -x "$EVOLVE_SH" ] || chmod +x "$EVOLVE_SH"
[ -f "$SEED" ]      || { log "ERROR: seed file not found: $SEED"; exit 2; }

if [ -f "$PATCH_FILE" ] && [ -s "$PATCH_FILE" ]; then
  HAS_PATCH=true
else
  HAS_PATCH=false
  log "WARNING: patch file is empty or missing — no evolution-specific overrides will be applied"
fi

log "========================================================"
log "evolution-daemon.sh — $(ts)"
log "  Repo:         $REPO_ROOT"
log "  Seed:         $SEED"
log "  Config:       $CONF_FILE"
log "  Patch:        $PATCH_FILE (has_patch=$HAS_PATCH)"
log "  Eval mode:    $EVAL_MODE"
log "  Population:   $POPULATION"
log "  Generations:  $GENERATIONS"
log "  Mutation:     $MUTATION_RATE"
log "  Elites:       $ELITES"
log "  Diverse:      $DIVERSE_SEEDS"
log "  Output dir:   $OUTPUT_DIR"
log "  Notify via:   ${OPENCLAW_SSH_TARGET:-<disabled>}"
log "========================================================"

RUN_NUM=0

# =============================================================================
# Main loop
# =============================================================================

while true; do

  RUN_NUM=$((RUN_NUM + 1))
  RUN_START="$(date +%s)"

  log ""
  log "════════════════════════════════════════════════════"
  log "Run #${RUN_NUM} — $(ts)"
  log "════════════════════════════════════════════════════"

  # ── Step 1: Sync master ──────────────────────────────────────────────────────

  log "[1/7] Syncing master…"
  if (cd "$REPO_ROOT" && git pull origin master --ff-only 2>&1); then
    log "  git pull OK"
  else
    log "  WARNING: git pull failed — continuing with current tree"
  fi

  # ── Step 2: Apply evolution patches ─────────────────────────────────────────

  PATCH_APPLIED=false
  if [ "$HAS_PATCH" = "true" ]; then
    log "[2/7] Applying evolution patches…"
    if (cd "$REPO_ROOT" && git apply "$PATCH_FILE"); then
      PATCH_APPLIED=true
      log "  Patches applied OK"
    else
      log "  WARNING: patch failed to apply — running without evolution-specific overrides"
    fi
  else
    log "[2/7] No patch file — skipping"
  fi

  # ── Step 3: Clean stale tmpdirs ─────────────────────────────────────────────

  log "[3/7] Cleaning stale /tmp/tmp.* directories…"
  STALE_COUNT=0
  # Only remove directories older than 1 hour to avoid disturbing very recent runs.
  while IFS= read -r -d '' STALE_DIR; do
    rm -rf "$STALE_DIR"
    STALE_COUNT=$((STALE_COUNT + 1))
  done < <(find /tmp -maxdepth 1 -name 'tmp.*' -type d -mmin +60 -print0 2>/dev/null)
  log "  Removed $STALE_COUNT stale tmpdir(s)"

  # ── Step 4: Run evolve.sh ────────────────────────────────────────────────────

  log "[4/7] Starting evolve.sh…"

  DIVERSE_FLAG=""
  [ "$DIVERSE_SEEDS" = "true" ] && DIVERSE_FLAG="--diverse-seeds"

  EVOLVE_EC=0
  EVOLVE_OUT=""
  EVOLVE_OUT=$(
    EVAL_MODE="$EVAL_MODE" \
    BASE_RPC_URL="$BASE_RPC_URL" \
    bash "$EVOLVE_SH" \
      --seed "$SEED" \
      --population "$POPULATION" \
      --generations "$GENERATIONS" \
      --mutation-rate "$MUTATION_RATE" \
      --elites "$ELITES" \
      --output "$OUTPUT_DIR" \
      $DIVERSE_FLAG \
      2>&1
  ) || EVOLVE_EC=$?

  # Always print evolve.sh output for visibility.
  printf '%s\n' "$EVOLVE_OUT" >&2

  if [ "$EVOLVE_EC" -ne 0 ]; then
    log "  WARNING: evolve.sh exited $EVOLVE_EC — results may be incomplete"
  else
    log "  evolve.sh completed OK"
  fi

  # ── Step 5: Locate the run directory just created ────────────────────────────

  # evolve.sh already saves to evolved/run_NNN/ and admits to seed pool.
  # Find the most recent run dir to extract summary data.
  LATEST_RUN_DIR=""
  LATEST_RUN_DIR=$(python3 - "$OUTPUT_DIR" <<'PYEOF' 2>/dev/null || true
import sys, os, re
base = sys.argv[1]
max_n = -1
best_dir = ''
if os.path.isdir(base):
    for name in os.listdir(base):
        m = re.fullmatch(r'run_(\d+)', name)
        if m and os.path.isdir(os.path.join(base, name)):
            n = int(m.group(1))
            if n > max_n:
                max_n = n
                best_dir = os.path.join(base, name)
print(best_dir)
PYEOF
)

  BEST_FITNESS=0
  BEST_RUN_DIR="${LATEST_RUN_DIR:-<unknown>}"

  if [ -n "$LATEST_RUN_DIR" ] && [ -d "$LATEST_RUN_DIR" ]; then
    # Extract best fitness from the run's generation JSONL files.
    BEST_FITNESS=$(python3 - "$LATEST_RUN_DIR" <<'PYEOF' 2>/dev/null || echo 0
import json, sys, os
run_dir = sys.argv[1]
best = 0
for fname in sorted(os.listdir(run_dir)):
    if not (fname.startswith('generation_') and fname.endswith('.jsonl')):
        continue
    with open(os.path.join(run_dir, fname)) as f:
        for line in f:
            try:
                d = json.loads(line)
                fitness = int(d.get('fitness', 0))
                if fitness > best:
                    best = fitness
            except (json.JSONDecodeError, ValueError, TypeError):
                pass
print(best)
PYEOF
)
    log "[5/7] Results: dir=$LATEST_RUN_DIR  best_fitness=$BEST_FITNESS"
  else
    log "[5/7] WARNING: could not locate run output directory"
  fi

  # ── Steps 6 (seed admission already done by evolve.sh) ──────────────────────
  # evolve.sh step 5 handles pool admission automatically.

  # ── Step 7: Write summary report ────────────────────────────────────────────

  RUN_END="$(date +%s)"
  DURATION=$(( RUN_END - RUN_START ))
  DURATION_FMT="$(printf '%02d:%02d:%02d' $((DURATION/3600)) $(( (DURATION%3600)/60 )) $((DURATION%60)))"

  if [ -n "$LATEST_RUN_DIR" ] && [ -d "$LATEST_RUN_DIR" ]; then
    SUMMARY_FILE="$LATEST_RUN_DIR/daemon-summary.txt"
    {
      echo "=== Evolution Daemon Run Summary ==="
      echo "Timestamp:     $(ts)"
      echo "Run dir:       $LATEST_RUN_DIR"
      echo "Daemon run #:  $RUN_NUM"
      echo "Duration:      $DURATION_FMT"
      echo "Best fitness:  $BEST_FITNESS"
      echo "Eval mode:     $EVAL_MODE"
      echo "Population:    $POPULATION"
      echo "Generations:   $GENERATIONS"
      echo "Diverse seeds: $DIVERSE_SEEDS"
      echo "Patch applied: $PATCH_APPLIED"
      echo "evolve.sh exit:$EVOLVE_EC"
    } > "$SUMMARY_FILE"
    log "[6/7] Summary written to $SUMMARY_FILE"
  fi

  # ── Step 8: Notify ──────────────────────────────────────────────────────────

  NOTIFY_MSG="evolution run #${RUN_NUM} complete — best_fitness=${BEST_FITNESS} duration=${DURATION_FMT} dir=$(basename "${LATEST_RUN_DIR:-unknown}")"
  log "[7/7] Notifying: $NOTIFY_MSG"
  notify "$NOTIFY_MSG"

  # ── Revert patches ──────────────────────────────────────────────────────────

  cleanup_patch

  # ── Check stop flag ─────────────────────────────────────────────────────────

  if [ "$STOP_REQUESTED" = "true" ]; then
    log ""
    log "Stop requested — daemon exiting after run #${RUN_NUM}."
    exit 0
  fi

  log "Run #${RUN_NUM} complete (${DURATION_FMT}). Starting next run…"
  log ""

done