get_database_overview

Instructions

Implementation Reference

• src/jvlink_mcp_server/server.py:688-692 (handler)

  The MCP tool handler for 'get_database_overview'. Decorated with @mcp.tool(), it opens a DatabaseConnection and delegates to _get_data_snapshot (which is an alias for get_data_snapshot from sample_data_provider).

  @mcp.tool()
  def get_database_overview() -> dict:
      """データベース全体の概要を取得"""
      with DatabaseConnection() as db:
          return _get_data_snapshot(db)

• src/jvlink_mcp_server/database/sample_data_provider.py:198-248 (helper)

  The actual implementation function get_data_snapshot(). Iterates over 9 database tables (NL_RA, NL_SE, NL_UM, NL_KS, NL_CH, NL_HR, NL_O1, NL_RA_NAR, NL_SE_NAR) to count records and determine the data period from NL_SE. Returns table record counts, total_records, and data_period with earliest/latest dates.

  def get_data_snapshot(db_connection) -> Dict[str, Any]:
      """データベース全体のスナップショット情報を取得
  
      Args:
          db_connection: DatabaseConnectionインスタンス
  
      Returns:
          dict: {
              'tables': テーブルごとの概要情報,
              'total_records': 総レコード数,
              'data_period': データ期間
          }
      """
      results = {
          "tables": {},
          "total_records": 0,
      }
  
      # 各テーブルのレコード数を取得
      for table_name in ["NL_RA", "NL_SE", "NL_UM", "NL_KS", "NL_CH", "NL_HR", "NL_O1", "NL_RA_NAR", "NL_SE_NAR"]:
          try:
              count_sql = f"SELECT COUNT(*) as cnt FROM {table_name}"
              df = db_connection.execute_safe_query(count_sql)
              count = int(df.iloc[0]["cnt"]) if not df.empty else 0
              results["tables"][table_name] = {
                  "record_count": count,
                  "description": _get_table_description(table_name),
              }
              results["total_records"] += count
          except Exception:
              results["tables"][table_name] = {"record_count": 0, "error": "取得失敗"}
  
      # データ期間を取得（NL_SEから）
      try:
          period_sql = """
          SELECT
              MIN(Year || '-' || MonthDay) as earliest,
              MAX(Year || '-' || MonthDay) as latest
          FROM NL_SE
          WHERE KakuteiJyuni IS NOT NULL
          """
          df = db_connection.execute_safe_query(period_sql)
          if not df.empty:
              results["data_period"] = {
                  "earliest": df.iloc[0]["earliest"],
                  "latest": df.iloc[0]["latest"],
              }
      except Exception:
          results["data_period"] = {"error": "取得失敗"}
  
      return results

• src/jvlink_mcp_server/server.py:688-688 (registration)

  The @mcp.tool() decorator registers 'get_database_overview' as an MCP tool on the FastMCP server instance.

  @mcp.tool()

• src/jvlink_mcp_server/database/sample_data_provider.py:321-332 (helper)

  Helper function _get_table_description() used by get_data_snapshot to provide human-readable Japanese descriptions for each table.

  def _get_table_description(table_name: str) -> str:
      """テーブルの説明を取得"""
      descriptions = {
          "NL_RA": "レース情報テーブル",
          "NL_SE": "出馬表・レース結果テーブル",
          "NL_UM": "馬マスタテーブル",
          "NL_KS": "騎手マスタテーブル",
          "NL_CH": "調教師マスタテーブル",
          "NL_HR": "払戻テーブル",
          "NL_O1": "単勝複勝オッズテーブル",
      }
      return descriptions.get(table_name, "不明")

• src/jvlink_mcp_server/server.py:42-46 (helper)

  Import statement: get_data_snapshot is imported from sample_data_provider and aliased as _get_data_snapshot for use in the handler.

  from .database.sample_data_provider import (
      get_sample_data as _get_sample_data,
      get_column_value_examples as _get_column_value_examples,
      get_data_snapshot as _get_data_snapshot,
  )