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

ROOT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
LOGS_DIR="${ROOT_DIR}/logs"
OLDER_THAN_DAYS="${OLDER_THAN_DAYS:-1}"
WARN_TOTAL_MB="${WARN_TOTAL_MB:-300}"
WARN_CANDIDATE_FILES="${WARN_CANDIDATE_FILES:-500}"

# Keep in sync with archive-logs.sh patterns.
PATTERN_PATHS=(
  "logs/e2e-automation/run_*.log"
  "logs/e2e-automation/report_*.md"
  "logs/prd-review/review_*.md"
  "logs/prd-review/claude_apply_*.md"
  "logs/prd-review/execution_report_*.md"
  "logs/prd-review/optimization_report_*.md"
)

log() { echo "[logs-health] $*"; }

to_mb() {
  local bytes="$1"
  awk -v b="${bytes}" 'BEGIN { printf "%.2f", b / 1024 / 1024 }'
}

if [[ ! -d "${LOGS_DIR}" ]]; then
  log "logs directory not found, skip."
  exit 0
fi

cutoff_epoch="$(date -d "${OLDER_THAN_DAYS} days ago" +%s)"
total_bytes="$(du -sb "${LOGS_DIR}" | awk '{print $1}')"
total_mb="$(to_mb "${total_bytes}")"

candidate_files=0
shopt -s nullglob
for pattern in "${PATTERN_PATHS[@]}"; do
  for abs in "${ROOT_DIR}"/${pattern}; do
    [[ -f "${abs}" ]] || continue
    mtime_epoch="$(stat -c %Y "${abs}")"
    if [[ "${mtime_epoch}" -lt "${cutoff_epoch}" ]]; then
      candidate_files=$((candidate_files + 1))
    fi
  done
done
shopt -u nullglob

log "total_size_mb=${total_mb} (threshold=${WARN_TOTAL_MB})"
log "archive_candidates_older_than_${OLDER_THAN_DAYS}d=${candidate_files} (threshold=${WARN_CANDIDATE_FILES})"

if awk -v a="${total_mb}" -v b="${WARN_TOTAL_MB}" 'BEGIN { exit !(a > b) }'; then
  log "WARN: logs directory is large. Consider: npm run logs:archive:apply"
fi

if [[ "${candidate_files}" -gt "${WARN_CANDIDATE_FILES}" ]]; then
  log "WARN: many archive candidates detected. Consider archiving historical logs."
fi

log "top 5 largest log files:"
find "${LOGS_DIR}" -type f -printf '%s %p\n' | sort -nr | sed -n '1,5p' | awk '
  {
    mb = $1 / 1024 / 1024
    $1 = ""
    sub(/^ /, "", $0)
    printf "  - %.2f MB %s\n", mb, $0
  }
'

exit 0