CDM MCP Server

""" BERDL environment configuration for datalake-mcp-server. This module provides Pydantic-based settings validation for the MCP server, compatible with the spark_notebook environment. """ import logging from functools import lru_cache from pydantic import AnyHttpUrl, AnyUrl, Field from pydantic_settings import BaseSettings # Configure logging logger = logging.getLogger(__name__) class BERDLSettings(BaseSettings): """ BERDL environment configuration using Pydantic Settings. This configuration is for the shared MCP server that dynamically connects to user-specific Spark Connect servers based on authenticated requests. Most fields have defaults since user-specific values come from their individual notebook environments. """ # Core authentication (set dynamically per request) KBASE_AUTH_TOKEN: str = Field( default="", description="KBase auth token (set per request)" ) USER: str = Field( default="", description="KBase username (set dynamically per request)" ) # MinIO configuration (used for constructing S3 paths, actual creds from user's Spark) MINIO_ENDPOINT_URL: str = Field( default="minio:9002", description="MinIO endpoint (hostname:port)" ) MINIO_ACCESS_KEY: str = Field( default="", description="MinIO access key (from user's Spark)" ) MINIO_SECRET_KEY: str = Field( default="", description="MinIO secret key (from user's Spark)" ) MINIO_SECURE: bool = Field( default=False, description="Use secure connection (True/False)" ) # Redis configuration BERDL_REDIS_HOST: str = Field(default="redis", description="Redis host name") BERDL_REDIS_PORT: int = Field(default=6379, description="Redis port") # Spark configuration SPARK_HOME: str = Field( default="/usr/local/spark", description="Spark installation directory" ) SPARK_MASTER_URL: AnyUrl | None = Field( default=None, description="Spark Master URL (spark://host:port) - not used in Connect mode", ) SPARK_CONNECT_URL: AnyUrl = Field( default=AnyUrl("sc://localhost:15002"), description=( "Spark Connect URL - CONSTRUCTED DYNAMICALLY per user. " "In Kubernetes: sc://jupyter-{username}.jupyterhub-{env}:15002 (cross-namespace DNS). " "Override pattern with SPARK_CONNECT_URL_TEMPLATE env var for docker-compose. " "Set K8S_ENVIRONMENT (dev/prod/stage) to control namespace." ), ) # Hive configuration BERDL_HIVE_METASTORE_URI: AnyUrl = Field( default=AnyUrl("thrift://hive-metastore:9083"), description="Hive Metastore Thrift endpoint", ) # Profile-specific Spark configuration (defaults for session creation) SPARK_WORKER_COUNT: int = Field( default=1, description="Number of Spark workers from profile" ) SPARK_WORKER_CORES: int = Field( default=1, description="Cores per Spark worker from profile" ) SPARK_WORKER_MEMORY: str = Field( default="2GiB", pattern=r"^\d+[kmgKMGT]i?[bB]?$", description="Memory per Spark worker from profile", ) SPARK_MASTER_CORES: int = Field( default=1, description="Cores for Spark master from profile" ) SPARK_MASTER_MEMORY: str = Field( default="1GiB", pattern=r"^\d+[kmgKMGT]i?[bB]?$", description="Memory for Spark master from profile", ) # Data Governance API configuration GOVERNANCE_API_URL: AnyHttpUrl = Field( default=AnyHttpUrl("http://minio-manager-service:8000"), description="Data governance API endpoint", ) # Optional: Pod IP for legacy mode (not used in MCP server) BERDL_POD_IP: str | None = Field( default=None, description="Pod IP for legacy Spark mode" ) @lru_cache(maxsize=1) def get_settings() -> BERDLSettings: """ Get cached BERDLSettings instance. Only creates the object once. Returns: BERDLSettings: Cached settings instance Raises: ValidationError: If environment variables are missing or invalid """ return BERDLSettings()