MCP Trino Server

Instructions

Implementation Reference

• src/server.py:404-436 (handler)

  MCP tool handler for 'show_files'. Decorated with @mcp.tool(), validates inputs using pydantic Field schemas, and delegates to client.show_files().

  @mcp.tool(description="Show Iceberg table data files")
  def show_files(
      catalog: str = Field(description="catalog name "),
      schema_name: str = Field(description="schema name "),
      table: str = Field(description="The name of the table"),
  ) -> str:
      """Show Iceberg table data files in current snapshot.
  
      The files table contains:
      - content: Type of content (0=DATA, 1=POSITION_DELETES, 2=EQUALITY_DELETES)
      - file_path: Data file location
      - file_format: Format of the data file
      - record_count: Number of records in file
      - file_size_in_bytes: File size
      - column_sizes: Column ID to size mapping
      - value_counts: Column ID to value count mapping
      - null_value_counts: Column ID to null count mapping
      - nan_value_counts: Column ID to NaN count mapping
      - lower_bounds: Column ID to lower bound mapping
      - upper_bounds: Column ID to upper bound mapping
      - key_metadata: Encryption key metadata
      - split_offsets: Recommended split locations
      - equality_ids: Field IDs for equality deletes
  
      Args:
          catalog: catalog name
          schema_name: schema name
          table: The name of the table
  
      Returns:
          str: JSON-formatted table files info
      """
      return client.show_files(catalog, schema_name, table)

• src/trino_client.py:508-541 (helper)

  Core implementation of show_files in TrinoClient class. Constructs SQL query to query the Iceberg $files metadata table and returns JSON-formatted results.

  def show_files(self, table: str, catalog: str, schema: str) -> str:
      """Show Iceberg table data files in current snapshot.
  
      The files table contains:
      - content: INTEGER - Type of content (0=DATA, 1=POSITION_DELETES, 2=EQUALITY_DELETES)
      - file_path: VARCHAR - Data file location
      - file_format: VARCHAR - Format of the data file
      - record_count: BIGINT - Number of records in file
      - file_size_in_bytes: BIGINT - File size
      - column_sizes: map(INTEGER, BIGINT) - Column ID to size mapping
      - value_counts: map(INTEGER, BIGINT) - Column ID to value count mapping
      - null_value_counts: map(INTEGER, BIGINT) - Column ID to null count mapping
      - nan_value_counts: map(INTEGER, BIGINT) - Column ID to NaN count mapping
      - lower_bounds: map(INTEGER, VARCHAR) - Column ID to lower bound mapping
      - upper_bounds: map(INTEGER, VARCHAR) - Column ID to upper bound mapping
      - key_metadata: VARBINARY - Encryption key metadata
      - split_offsets: array(BIGINT) - Recommended split locations
      - equality_ids: array(INTEGER) - Field IDs for equality deletes
  
      Args:
          table: The name of the table
          catalog: Optional catalog name (defaults to configured catalog)
          schema: Optional schema name (defaults to configured schema)
  
      Returns:
          str: JSON-formatted string containing table files info
      """
      catalog = catalog or self.config.catalog
      schema = schema or self.config.schema
      if not catalog or not schema:
          raise CatalogSchemaError
      table_identifier = f"{catalog}.{schema}.{table}$files"
      query = 'SELECT * FROM "{}"'
      return self.execute_query(query.format(table_identifier))