Brewery2 pipelines – revival or former "forking forks" [1], now using operation kernel and virtual data objects. In this example: just denormalizing two tables into one dimension with field selection. Note that target table does not contain detail keys used for join. References: [1] http://blog.databrewery.org/posts/forking-forks-with-higher-ord…

gistfile1.txt

from brewery2 import Pipeline, open_store

stores = {
            "source": open_store("sql", "postgres://localhost/crm", schema="app"),
            "target": open_store("sql", connectable=source_store.connectable, schema="cubes")
        }

p = Pipeline(stores=stores)
p.source("source", "crm_contact")
p.field_filter(keep=["id",
                      "contact_type",
                      "first_name",
                      "last_name",
                      "display_name",
                      "sort_name",
                      "is_deleted"])

p2 = Pipeline(stores=stores)
p2.source("source", "crm_address")
p2.field_filter(keep=["contact_id",
                        "is_primary",
                        "city",
                        "postal_code" ])
p.join_details(p2.result, "id", "contact_id")
p.create("target", "dim_contact", replace=True)