Use case: unit testing code that manipulates data frames obtained by querying a Postgres database. Auto-generate functions for creating in memory data frames that mirror the structure of the database, for easy mocking of the tables.

Note that for now this only supports Postgres databases with the following types: “integer”, “smallint”, “boolean”, “numeric”, “real”, “date”, “timestamp without time zone”, “timestamp with time zone”


  1. Create a db connection. To start a dockerised Postgres database containing the Northwind sample database using this image, run ./scripts/ Then you can use the following connection:
con <- DBI::dbConnect(RPostgres::Postgres(),
                        dbname = "northwind",
                        host = "localhost",
                        user = "northwind",
                        password = "northwind")
  1. Generate and load functions for creating fake db tables
db <- fakerbase::fb_generate(con, schema_name = "public")
  1. Use generated functions to create fake db tables
    region <- db$region(region_id = 123L, region_description = "Central America")
    'data.frame':   1 obs. of  2 variables:
     $ region_id  : int 123
     $ region_description: chr "Central America"
  1. To re-use generated functions without re-connecting to the database, use
db <- fakerbase::fb_load("northwind", schema_name = "public")


If running integration tests, first run ./scripts/ which starts a Postgres instance running in a docker container (using host networking.) To remove the test database, run ./scripts/