fill in missing candidate address with committee's address

process/R/out-cover-committees.R

@@ -29,27 +29,30 @@ out_committees <- function(path_out_cover, path_out_report_list) {
   committee_name_address <-
     cover |>
     inner_join(
-      report_list |> select(report_id, doc_order),
+      report_list |> select(report_id, year, doc_order),
       by = "report_id"
     ) |>
     group_by(sboe_id) |>
+    slice_max(year, n = 1) |>
     slice_max(doc_order, n = 1) |>
+    slice_max(date_filed, n = 1, with_ties = FALSE) |>
     ungroup() |>
     select(sboe_id, committee_name, report_id, street_1, street_2, city, state, zip_code) |>
-    add_address_lookup(postal_code = zip_code)
+    add_address_lookup(postal_code = zip_code) |>
+    collect()
 
   committee_type <-
     cover |>
     group_by(sboe_id) |>
     count(committee_type, fund_type, fund_name) |>
-    slice_max(n, n = 1) |>
+    slice_max(n, n = 1, with_ties = FALSE) |>
     ungroup() |>
-    select(-n)
+    select(-n) |>
+    collect()
 
   committees <-
     committee_name_address |>
-    left_join(committee_type, by = "sboe_id") |>
-    collect()
+    left_join(committee_type, by = join_by(sboe_id), relationship = "one-to-one")
 
   arrow::write_parquet(committees, out)
 

process/R/prepare_candidates.R

@@ -1,17 +1,25 @@
 prepare_candidates_for_matching <- function(
   path_data_prep_officers,
   path_out_report_list,
+  path_out_committees,
   path_addresses = "data-out/addresses"
 ) {
   lg_info_target(lg_get_logger())
 
   report_list <- out_open_dataset_db(path_out_report_list)
   addresses <- out_open_dataset_db(path_addresses)
+  committees <- out_open_dataset_db(path_out_committees)
+  candidates_db <- out_open_dataset_db(path_data_prep_officers)
 
   candidates <-
-    out_open_dataset_db(path_data_prep_officers) |>
+    candidates_db |>
     semi_join(report_list, by = "report_id") |>
     filter(type == "Candidate") |>
+    rows_patch(
+      committees |> select(sboe_id, address = address_lookup),
+      by = "sboe_id",
+      unmatched = "ignore"
+    ) |>
     mutate(
       name_clean = toupper(name),
       name_clean = REGEXP_REPLACE(name_clean, " FOR .+$", "", "g"),

process/_targets.R

@@ -208,7 +208,7 @@ list(
   # Candidates --------------------------------------------------------------
   tar_target(
     candidates_for_matching,
-    prepare_candidates_for_matching(path_data_prep_officers, path_out_report_list)
+    prepare_candidates_for_matching(path_data_prep_officers, path_out_report_list, path_out_committees)
   ),
   tar_target(
     candidate_listing_for_matching,