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

report_path_info <- function(report_path) {
  # data-raw/reports/{sboe_id}/all/{report_id}_{received_date}.txt
  report_path <- fs::path_abs(report_path)
  x <- fs::path_rel(report_path, here::here("..", "data-raw", "reports"))
  x <- map(x, path_split) |> list_flatten()

  tibble(
    sboe_id = map_vec(x, \(x) x[1]),
    report_id = map_int(x, \(x) as.integer(strsplit(x[3], "_")[[1]][1])),
  )
}

report_info_in_report_list <- function(files, report_list = tar_read(report_list)) {
  info <- report_path_info(files)
  info$path <- files
  info <- semi_join(info, report_list, by = c("sboe_id", "report_id"))
  info
}