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
  x <- fs::path_abs(report_path)
  x <- map(x, path_split) |> list_flatten()

  idx_reports <- map_int(x, \(x) which(x == "reports"))

  sboe_id <- map2_vec(x, idx_reports, \(x, i) x[i + 1])
  report_id <- map2_int(x, idx_reports, \(x, i) as.integer(strsplit(x[i + 3], "_")[[1]][1]))

  sboe_id <- fix_sboe_id_missing(sboe_id, report_id)

  tibble(sboe_id, report_id)
}

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
}