#!/usr/bin/env bash
# dump_project.sh — собирает всё нужное для аудита проекта в один файл
# Запуск: sudo bash dump_project.sh
# Результат: /root/morozovka_dump_YYYY-MM-DD_HH-MM.txt

set -uo pipefail

PROJECT_DIR="/opt/morozovka-portal"
OUT="/root/morozovka_dump_$(date +%F_%H-%M).txt"
MAX_FILE_SIZE=$((512 * 1024))   # файлы больше 512 КБ пропускаем (обычно мусор)

# ----- исключения -----
EXCLUDES=(
  "*/venv/*"
  "*/__pycache__/*"
  "*.pyc"
  "*/.git/*"
  "*/node_modules/*"
  "*.log"
  "*.sqlite"
  "*.db"
  "google_credentials.json"     # СЕКРЕТ — не включаем
  "*.png" "*.jpg" "*.jpeg" "*.gif" "*.ico" "*.pdf" "*.zip" "*.tar" "*.gz" "*.woff" "*.woff2" "*.ttf"
)

should_skip() {
  local f="$1"
  for pat in "${EXCLUDES[@]}"; do
    # shellcheck disable=SC2053
    [[ "$f" == $pat ]] && return 0
  done
  return 1
}

section() {
  echo "" >> "$OUT"
  echo "================================================================================" >> "$OUT"
  echo "== $*" >> "$OUT"
  echo "================================================================================" >> "$OUT"
}

dump_file() {
  local f="$1"
  [[ -f "$f" ]] || { echo "[skip: not a file] $f" >> "$OUT"; return; }
  local size
  size=$(stat -c%s "$f" 2>/dev/null || echo 0)
  if (( size > MAX_FILE_SIZE )); then
    echo "" >> "$OUT"
    echo "---- FILE: $f  (SKIPPED, size=${size}B > ${MAX_FILE_SIZE}B) ----" >> "$OUT"
    return
  fi
  echo "" >> "$OUT"
  echo "---- FILE: $f  (size=${size}B) ----" >> "$OUT"
  cat "$f" >> "$OUT"
  echo "" >> "$OUT"
  echo "---- END: $f ----" >> "$OUT"
}

# ----- начинаем -----
: > "$OUT"
echo "DUMP GENERATED: $(date -Is)" >> "$OUT"
echo "HOSTNAME: $(hostname -f 2>/dev/null || hostname)" >> "$OUT"
echo "UPTIME: $(uptime -p 2>/dev/null || uptime)" >> "$OUT"

# 1) Дерево файлов проекта
section "TREE: $PROJECT_DIR"
if command -v tree >/dev/null 2>&1; then
  tree -a -I 'venv|__pycache__|.git|node_modules' -L 4 "$PROJECT_DIR" >> "$OUT" 2>&1
else
  find "$PROJECT_DIR" \
    \( -path '*/venv' -o -path '*/__pycache__' -o -path '*/.git' -o -path '*/node_modules' \) -prune -o \
    -type f -print | sort >> "$OUT"
fi

# 2) requirements.txt + версия Python
section "ENVIRONMENT"
echo "python: $(command -v python3) $(python3 --version 2>&1)" >> "$OUT"
if [[ -f "$PROJECT_DIR/requirements.txt" ]]; then
  echo "" >> "$OUT"
  echo "--- pip freeze (из venv) ---" >> "$OUT"
  if [[ -x "$PROJECT_DIR/venv/bin/pip" ]]; then
    "$PROJECT_DIR/venv/bin/pip" freeze >> "$OUT" 2>&1
  else
    cat "$PROJECT_DIR/requirements.txt" >> "$OUT"
  fi
fi

# 3) Все текстовые файлы проекта
section "PROJECT FILES ($PROJECT_DIR)"
while IFS= read -r -d '' f; do
  should_skip "$f" && continue
  # бинарные — пропускаем по mime
  if command -v file >/dev/null 2>&1; then
    case "$(file -b --mime-encoding "$f" 2>/dev/null)" in
      binary) echo "" >> "$OUT"; echo "---- BINARY SKIPPED: $f ----" >> "$OUT"; continue ;;
    esac
  fi
  dump_file "$f"
done < <(find "$PROJECT_DIR" \
          \( -path '*/venv' -o -path '*/__pycache__' -o -path '*/.git' -o -path '*/node_modules' \) -prune -o \
          -type f -print0 | sort -z)

# 4) systemd-юниты и таймеры проекта
section "SYSTEMD UNITS"
for u in \
  /etc/systemd/system/morozovka-portal.service \
  /etc/systemd/system/morozovka-sync.service \
  /etc/systemd/system/morozovka-sync.timer \
  /etc/systemd/system/gitea.service \
  /etc/systemd/system/bridge-webui.service; do
  [[ -f "$u" ]] && dump_file "$u" || echo "[missing] $u" >> "$OUT"
done

echo "" >> "$OUT"
echo "--- systemctl list-units (morozovka*) ---" >> "$OUT"
systemctl list-units --type=service --type=timer --no-pager --all --no-legend 2>/dev/null \
  | grep -Ei 'morozovka|gitea|bridge' >> "$OUT" || true

echo "" >> "$OUT"
echo "--- systemctl status (кратко) ---" >> "$OUT"
for s in morozovka-portal morozovka-sync.timer gitea; do
  echo "" >> "$OUT"
  echo "### $s" >> "$OUT"
  systemctl is-enabled "$s" 2>&1 >> "$OUT" || true
  systemctl is-active  "$s" 2>&1 >> "$OUT" || true
done

# 5) Nginx-конфиги (сервер 3)
section "NGINX CONFIGS (server 3)"
for d in /etc/nginx/sites-available /etc/nginx/sites-enabled /etc/nginx/conf.d; do
  [[ -d "$d" ]] || continue
  while IFS= read -r -d '' f; do
    dump_file "$f"
  done < <(find "$d" -maxdepth 1 -type f -o -type l -print0 2>/dev/null | sort -z)
done

section "NGINX TEST OUTPUT"
nginx -t 2>&1 >> "$OUT" || true

# 6) PostgreSQL: список таблиц и схема (без данных)
section "POSTGRESQL SCHEMA (morozovka_portal)"
if command -v psql >/dev/null 2>&1 || [[ -x /usr/bin/sudo ]]; then
  sudo -u postgres psql -d morozovka_portal -c "\dt" >> "$OUT" 2>&1 || true
  sudo -u postgres psql -d morozovka_portal -c "\d+" >> "$OUT" 2>&1 || true
  echo "" >> "$OUT"
  echo "--- pg_dump --schema-only ---" >> "$OUT"
  sudo -u postgres pg_dump --schema-only --no-owner --no-privileges morozovka_portal >> "$OUT" 2>&1 || true
else
  echo "[psql not found]" >> "$OUT"
fi

# 7) Проверка, что портал отвечает
section "PORTAL HEALTH"
curl -sS -m 5 http://127.0.0.1:8000/healthz >> "$OUT" 2>&1 || echo "[healthz failed]" >> "$OUT"
echo "" >> "$OUT"
curl -sS -m 5 -o /dev/null -w "HTTP %{http_code}  time=%{time_total}s\n" http://127.0.0.1:8000/ >> "$OUT" 2>&1 || true

# 8) Последние логи (обрезанные)
section "RECENT LOGS (last 40 lines each)"
for u in morozovka-portal morozovka-sync; do
  echo "" >> "$OUT"
  echo "### journalctl -u $u -n 40 --no-pager" >> "$OUT"
  journalctl -u "$u" -n 40 --no-pager >> "$OUT" 2>&1 || true
done

# 9) Проверка VPN/маршрутов на всякий
section "NETWORK (server 3)"
ip -brief addr >> "$OUT" 2>&1 || true
echo "" >> "$OUT"
ip route >> "$OUT" 2>&1 || true

echo "" >> "$OUT"
echo "DUMP COMPLETE: $(date -Is)" >> "$OUT"
echo "SIZE: $(du -h "$OUT" | cut -f1)" >> "$OUT"

echo "Готово: $OUT"
echo "Размер: $(du -h "$OUT" | cut -f1)"