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

prep_candidate_listing <- function(
  path_candidate_listing_raw
) {

  candidate_listing_contest <-
    prep_open_dataset_db(fs::path_dir(path_candidate_listing_raw)) |>
    filter(name_on_ballot != "No Preference") |>
    tidyr::replace_na(list(
      first_name = "",
      middle_name = "",
      last_name = "",
      name_suffix_lbl = ""
    ))

  # candidate names ----
  # Extract candidate names, these will be primary keys for the candidates table
  candidate_names <-
    candidate_listing_contest |>
    dbplyr::window_order(last_name, first_name, middle_name) |>
    distinct(first_name, middle_name, last_name, name_suffix_lbl) |>
    mutate(candidate_id = row_number(), .before = 1)

  # candidate_name_on_ballot ----
  candidate_name_on_ballot <-
    extract_candidate_info(
      candidate_listing_contest,
      candidate_names,
      info_vars = c("name_on_ballot")
    ) |>
    collect()

  # candidate_address ----
  candidate_address <-
    candidate_listing_contest |>
    mutate(
      phone = coalesce(phone, office_phone, business_phone),
      street_address = toupper(street_address),
      street_address = REGEXP_REPLACE(street_address, " +", " ", "g"),
      street_address = trimws(street_address),
    ) |>
    select(-office_phone, -business_phone) |>
    extract_candidate_info(
      candidate_names,
      info_vars = c("street_address", "city", "state", "zip_code", "phone", "email")
    ) |>
    collect() |>
    rename(street = street_address)

  # candidate_party ----
  candidate_party <-
    candidate_listing_contest |>
    extract_candidate_info(
      candidate_names,
      info_vars = c("party_candidate")
    ) |>
    collect() |>
    mutate(party_candidate = forcats::fct_inorder(party_candidate))

  # Extract contests (remaining data in candidate_listing) ----
  cols_candidate_id <- intersect(colnames(candidate_names), colnames(candidate_listing_contest))
  cols_related <- setdiff(
    c(colnames(candidate_name_on_ballot), colnames(candidate_address), colnames(candidate_party)),
    "election_dt"
  )

  contests <-
    candidate_listing_contest |>
    select(election_dt:name_suffix_lbl) |>
    select(-any_of(cols_related)) |>
    left_join(candidate_names, by = cols_candidate_id) |>
    relocate(candidate_id, .before = first_name) |>
    collect()

  # Get current complete contact information ----
  candidate_contact_current <-
    candidate_listing_current_contact_info(candidate_address)

  # Join candidates into one big table ----
  candidates <-
    candidate_names |>
    collect() |>
    left_join(
      candidate_name_on_ballot |>
        slice_max(election_dt, by = candidate_id, n = 1) |>
        select(-election_dt),
      by = "candidate_id",
      relationship = "one-to-one"
    ) |>
    left_join(
      candidate_contact_current,
      by = "candidate_id",
      relationship = "one-to-one"
    ) |>
    left_join(
      candidate_party |>
        filter(!is.na(party_candidate)) |>
        group_by(candidate_id) |>
        slice_max(election_dt, n = 1) |>
        arrange(party_candidate) |>
        slice_head(n = 1) |>
        ungroup() |>
        select(-election_dt) |>
        rename(party_last = party_candidate),
      by = "candidate_id",
      relationship = "one-to-one"
    ) |>
    left_join(
      candidate_party |>
        filter(!is.na(party_candidate)) |>
        group_by(candidate_id) |>
        count(party_candidate) |>
        slice_max(n, n = 1) |>
        arrange(party_candidate) |>
        slice_head(n = 1) |>
        ungroup() |>
        select(-n) |>
        rename(party_most = party_candidate),
      by = "candidate_id",
      relationship = "one-to-one"
    ) |>
    left_join(
      contests |>
        group_by(candidate_id) |>
        distinct(candidate_id, election_dt, contest_name) |>
        summarize(
          contest_n = n(),
          contest_first = min(election_dt),
          contest_latest = max(election_dt)
        ),
      by = "candidate_id",
      relationship = "one-to-one"
    ) |>
    relocate(name_on_ballot, .before = first_name) |>
    relocate(starts_with("party"), .before = street) |>
    relocate(starts_with("contest"), .before = street) |>
    add_address_lookup(street = street, postal_code = zip_code)

  # Return list of tables
  list(
    elections = contests,
    candidates = candidates,
    candidate_name_on_ballot = candidate_name_on_ballot,
    candidate_contact = candidate_address,
    candidate_party = candidate_party
  )
}

extract_candidate_info <- function(
  candidate_listing_contest,
  candidate_names,
  info_vars
) {
  candidate_listing_contest |>
    select(first_name:name_suffix_lbl, election_dt, all_of(info_vars)) |>
    distinct() |>
    left_join(
      candidate_names,
      by = c("first_name", "middle_name", "last_name", "name_suffix_lbl")
    ) |>
    select(-first_name, -middle_name, -last_name, -name_suffix_lbl) |>
    relocate(candidate_id, .before = 1) |>
    distinct() |>
    arrange(candidate_id, election_dt, !!info_vars)
}

candidate_listing_current_contact_info <- function(candidate_address) {
  candidate_phone_current <-
    candidate_address |>
    filter(!is.na(phone)) |>
    slice_max(election_dt, by = candidate_id, n = 1, with_ties = FALSE) |>
    select(candidate_id, phone)

  candidate_email_current <-
    candidate_address |>
    filter(!is.na(email)) |>
    slice_max(election_dt, by = candidate_id, n = 1, with_ties = FALSE) |>
    select(candidate_id, email)

  candidate_address_current <-
    candidate_address |>
    group_by(candidate_id) |>
    slice_max(election_dt, n = 1) |>
    select(-election_dt) |>
    filter(
      n_distinct(street) == 1 | !grepl("PO BOX", street)
    ) |>
    slice_head(n = 1) |>
    select(candidate_id, street:zip_code) |>
    ungroup()

  candidate_address_current |>
    left_join(candidate_phone_current, by = "candidate_id", relationship = "one-to-one") |>
    left_join(candidate_email_current, by = "candidate_id", relationship = "one-to-one")
}