nc-campaign-finance/process/R/out-expenses.R

out_expenses_payee <- function(
  path_data_prep_expenditures,
  path_out_report_list
) {
  lg_info_target(lg_get_logger())

  out <- path("data-out", "expenses_payee", "expenses_payee.parquet")
  dir_create(path_dir(out))

  report_list <- out_open_dataset_db(path_out_report_list)

  expenses <-
    prep_open_dataset_db(path_data_prep_expenditures) |>
    semi_join(report_list, by = "report_id")

  expenses_payee <-
    expenses |>
    distinct(
      org_name,
      is_org,
      is_us,
      profession,
      employers_name,
      street_1,
      street_2,
      city,
      state,
      full_zip,
      country_name
    ) |>
    add_address_lookup(postal_code = full_zip, name = "address_lookup") |>
    mutate(payee_id = row_number(), .before = 1)

  arrow::write_parquet(collect(expenses_payee), out)

  dirname(out)
}

out_expenses <- function(
  path_data_prep_expenditures,
  path_out_expenses_payee,
  path_out_report_list
) {
  lg_info_target(lg_get_logger())

  out <- path("data-out", "expenses", "expenses.parquet")
  dir_create(path_dir(out))

  report_list <- out_open_dataset_db(path_out_report_list)
  expenses_payee <-
    out_open_dataset_db(path_out_expenses_payee) |>
    select(-address_lookup)

  expenses <-
    prep_open_dataset_db(path_data_prep_expenditures) |>
    semi_join(report_list, by = "report_id") |>
    select(-name_sort)

  # Replace payee info with payee_id
  cols_payee_common <- intersect(colnames(expenses_payee), colnames(expenses))

  expenses <-
    expenses |>
    left_join(expenses_payee, by = cols_payee_common) |>
    select(-one_of(cols_payee_common)) |>
    relocate(payee_id, .after = report_id) |>
    collect()

  arrow::write_parquet(expenses, out)

  dirname(out)
}