Sources

Source nodes represent external tables in a database or data warehouse and make up the foundational layer on which other nodes are built upon.

AttributeDescriptionType
nameUnique name used by other nodes to select from this nodestring
descriptionA human readable description of the nodestring
display_nameA human readable name for the nodestring
modepublished or draft (see Node Mode)string
catalogThe name of the external catalogstring
schema_The name of the external schemastring
tableThe name of the external tablestring
columnsA map of the external table’s column names and typesmap

Creating Catalogs and Engines

Before creating source nodes, a DataJunction server must contain at least one catalog. A catalog represents a catalog in your external data warehouse and includes one or more engine definitions. All nodes that select from a source node, automatically inherit that source node’s catalog. Here’s an example of creating a duckdb engine and a warehouse catalog.

Create a Catalog 

curl -X POST http://localhost:8000/catalogs/ \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "name": "warehouse"
}'

dj.catalogs.create({"name": "warehouse"}).then(data => console.log(data))

Create an Engine 

curl -X 'POST' \
  'http://localhost:8000/engines/' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '{
  "name": "duckdb",
  "version": "0.7.1",
  "dialect": "spark"
}'

dj.engines.create({
    name: "duckdb",
    version: "0.7.1",
    dialect: "spark",
}).then(data => console.log(data))

Attaching Existing Engines and Catalogs

Once a catalog and engine defintion are created, the engine definition can be attached to the catalog. Here is an example of attaching the existing duckdb engine definition to the existing warehouse catalog.

curl -X 'POST' \
  'http://localhost:8000/catalogs/warehouse/engines/' \
  -H 'accept: application/json' \
  -H 'Content-Type: application/json' \
  -d '[{
  "name": "duckdb",
  "version": "0.7.1"
}]'

dj.catalogs.addEngine("warehouse", "duckdb", "0.7.1").then(data => console.log(data))

Registering Tables as Source Nodes

Assuming that you’ve configured a reflection service and query service to interface with your data warehouse setup, you can just register a table as a source node by telling DJ about the table. The reflection service will automatically add the column names and types to all source nodes and update them whenever the external table changes.

To read more about DJRS, see the Table Reflection page.

curl -X POST http://localhost:8000//register/table/warehouse/roads/repair_orders/

table = dj.register_table(
    catalog="warehouse",
    schema_="roads",
    table="repair_orders",
)

Create Source Node Manually 

curl -X POST http://localhost:8000/nodes/source/ \
-H 'Content-Type: application/json' \
-d '{
    "name": "default.repair_orders",
    "description": "Repair orders",
    "mode": "published",
    "catalog": "warehouse",
    "schema_": "roads",
    "table": "repair_orders",
    "columns": [
        {"name": "repair_order_id", "type": "int"},
        {"name": "municipality_id", "type": "string"},
        {"name": "hard_hat_id", "type": "int"},
        {"name": "order_date", "type": "timestamp"},
        {"name": "required_date", "type": "timestamp"},
        {"name": "dispatched_date", "type": "timestamp"},
        {"name": "dispatcher_id", "type": "int"}
    ]
}'

source = dj.create_source(
    name="default.repair_orders",
    description="Repair orders",
    catalog="warehouse",
    schema_="roads",
    table="repair_orders",
    columns=[
        {"name": "repair_order_id", "type": "int"},
        {"name": "municipality_id", "type": "string"},
        {"name": "hard_hat_id", "type": "int"},
        {"name": "order_date", "type": "timestamp"},
        {"name": "required_date", "type": "timestamp"},
        {"name": "dispatched_date", "type": "timestamp"},
        {"name": "dispatcher_id", "type": "int"},
    ],
)

dj.sources.create({
    name: "default.repair_orders",
    description: "Repair orders",
    mode: "published",
    catalog: "warehouse",
    schema_: "roads",
    table: "repair_orders",
    columns: [
        {name: "repair_order_id", type: "int"},
        {name: "municipality_id", type: "string"},
        {name: "hard_hat_id", type: "int"},
        {name: "order_date", type: "timestamp"},
        {name: "required_date", type: "timestamp"},
        {name: "dispatched_date", type: "timestamp"},
        {name: "dispatcher_id", type: "int"}
    ]
}).then(data => console.log(data))
← Namespaces
Transforms →