nc-campaign-finance/process/R/open_dataset.R

prep_open_dataset <- function(path_prep, partitioning = "sboe_id", ...) {
  cf_open_dataset(path_prep, partitioning = partitioning, ..., dir = "data-prep")
}

out_open_dataset <- function(path_out, partitioning = "sboe_id", ...) {
  cf_open_dataset(path_out, partitioning = partitioning, ..., dir = "data-out")
}

cf_open_dataset <- function(path, partitioning = "sboe_id", ..., dir = "data-prep") {
  path <- resolve_path_up_2(path, dir = dir)

  if (length(fs::dir_ls(path, type = "dir")) == 0) {
    partitioning <- NULL
  }

  arrow::open_dataset(path, partitioning = partitioning, ...)
}

prep_open_dataset_db <- function(table, ..., path_prep = table) {
  cf_open_dataset_db(table, ..., path = path_prep, dir = "data-prep")
}

out_open_dataset_db <- function(table, ..., path_out = table) {
  cf_open_dataset_db(table, ..., path = path_out, dir = "data-out")
}

cf_open_dataset_db <- function(table, ..., path = table, dir = "data-prep") {
  pq <- cf_open_dataset(path, ..., dir = dir)

  table <- fs::path_file(table)

  con <- duckdb_global_con()
  duckdb::duckdb_register_arrow(con, table, pq)
  dplyr::tbl(con, table)
}

prep_open_address_db <- function(
  path_db = "address_lookup.sqlite"
) {
  path_db <- resolve_path_up_2(path_db)

  con <- if (!is.null(.globals$con_address)) {
    .globals$con_address
  } else {
    .globals$con_address <- DBI::dbConnect(RSQLite::SQLite(), path_db)
  }

  tbl(con, "resolved")
}

# Utils ----
resolve_path_up_2 <- function(path, dir = "data-prep") {
  if (fs::file_exists(path)) {
    return(path)
  }

  path_here <- here::here(dir, path)
  path_wd <- fs::path(dir, path)
  path_up <- fs::path("..", dir, path)
  path_up2 <- fs::path("..", "..", dir, path)

  if (fs::file_exists(path_wd)) return(path_wd)
  if (fs::file_exists(path_here)) return(path_here)
  if (fs::file_exists(path_up)) return(path_up)
  if (fs::file_exists(path_up2)) return(path_up2)

   stop("File not found: ", path)
}