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

prepare_addresses_lookup_db <- function(
  addresses,
  path_address_db = "../data-prep/address_lookup.sqlite"
) {
  if (!fs::file_exists(path_address_db)) {
    prepare_addresses_create_db(path_address_db)
  }

  con <- DBI::dbConnect(RSQLite::SQLite(), path_address_db)
  withr::defer(DBI::dbDisconnect(con))
  db <- dplyr::tbl(con, "resolved")

  seen <- db |> dplyr::pull(address)

  if (length(seen)) {
    addresses <- setdiff(trimws(addresses), seen)
  }

  if (!length(addresses)) {
    cli::cli_alert_success("All addresses have been geocoded")
    return(path_address_db)
  }

  cli::cli_inform("Geocoding {length(addresses)} addresses")

  blocks <- seq(1, length(addresses), by = 5000L)
  if (blocks[length(blocks)] != length(addresses)) {
    blocks <- c(blocks, length(addresses) + 1L)
  }
  blocks <- blocks - 1L
  for (i in seq_along(blocks)[-1]) {
    start <- blocks[i - 1] + 1
    end <- blocks[i]

    resolved <-
      tidygeocoder::geo(
        addresses[start:end],
        method = "census",
        full_results = TRUE
      ) |>
      dplyr::select(-input_address)

    dplyr::rows_upsert(db, resolved, by = "address", in_place = TRUE, copy = TRUE)
  }

  return(path_address_db)
}

prepare_addresses_create_db <- function(path_address_db) {
  sql <- "CREATE TABLE `resolved` (
  `address` TEXT UNIQUE,
  `lat` REAL,
  `long` REAL,
  `id` INTEGER,
  `match_indicator` TEXT,
  `match_type` TEXT,
  `matched_address` TEXT,
  `tiger_line_id` INTEGER,
  `tiger_side` TEXT
)"

  con <- DBI::dbConnect(RSQLite::SQLite(), path_address_db)
  DBI::dbExecute(con, sql)
  DBI::dbDisconnect(con)

  invisible(path_address_db)
}