pipeline_name: operations_doc_onboarding
doc: Onboard example datasets to be used in the operations doc app.
steps:
  - module_type: import.file_bundle
    step_id: download_example_data
  - module_type: create.table.from.file_bundle
    step_id: create_files_table
    input_links:
      file_bundle: download_example_data.file_bundle
  - module_type: file_bundle.pick.file
    step_id: pick_journal_nodes
    input_links:
      file_bundle: download_example_data.file_bundle
  - module_type: file_bundle.pick.file
    step_id: pick_journal_edges
    input_links:
      file_bundle: download_example_data.file_bundle
  - module_type: create.table.from.file
    step_id: create_nodes_table
    input_links:
      file: pick_journal_nodes.file
  - module_type: create.table.from.file
    step_id: create_edges_table
    input_links:
      file: pick_journal_edges.file
  - module_type: assemble.tables
    step_id: assemble_tables
    module_config:
      constants:
        table_name_1: "nodes"
        table_name_2: "edges"
    input_links:
      table_1: create_nodes_table.table
      table_2: create_edges_table.table
  - module_type: create.database.from.tables
    step_id: create_database_from_tables
    input_links:
      tables: assemble_tables.tables


input_aliases:
  download_example_data.source: "example_data_source"
  download_example_data.sub_path: "sub_folder"
  download_example_data.include_files: "include_files"
  download_example_data.exclude_files: "exclude_files"
  download_example_data.onboard_type: "onboard_type"
  download_example_data.exclude_dirs: "exclude_dirs"
  pick_journal_edges.path: "journal_edges_path"
  pick_journal_nodes.path: "journal_nodes_path"
  create_nodes_table.first_row_is_header: "nodes_first_row_is_header"
  create_edges_table.first_row_is_header: "edges_first_row_is_header"

output_aliases:
  download_example_data.file_bundle: "file_bundle"
  create_files_table.table: "files_table"
  pick_journal_nodes.file: "nodes_file"
  pick_journal_edges.file: "edges_file"
  create_nodes_table.table: "nodes_table"
  create_edges_table.table: "edges_table"
  assemble_tables.tables: "journals_tables"
  create_database_from_tables.database: "journals_database"