| @@ -19,15 +19,23 @@ out_cover <- function( | |||
| dirname(out) | |||
| } | |||
| out_committees <- function(path_out_cover) { | |||
| out_committees <- function(path_out_cover, path_out_report_list) { | |||
| out <- path("data-out", "committees", "committees.parquet") | |||
| dir_create(path_dir(out)) | |||
| cover <- out_open_dataset_db(path_out_cover) | |||
| report_list <- out_open_dataset_db(path_out_report_list) | |||
| committee_name_address <- | |||
| cover |> | |||
| distinct(sboe_id, committee_name, street_1, street_2, city, state, zip_code) |> | |||
| inner_join( | |||
| report_list |> select(report_id, doc_order), | |||
| by = "report_id" | |||
| ) |> | |||
| group_by(sboe_id) |> | |||
| slice_max(doc_order, n = 1) |> | |||
| ungroup() |> | |||
| select(sboe_id, committee_name, report_id, street_1, street_2, city, state, zip_code) |> | |||
| add_address_lookup(postal_code = zip_code) | |||
| committee_type <- | |||
| @@ -60,7 +60,11 @@ out_receipts <- function( | |||
| receipts <- | |||
| receipts |> | |||
| left_join(receipts_payer, by = cols_payer_common) |> | |||
| select(-one_of(cols_payer_common)) |> | |||
| select(-any_of(cols_payer_common)) |> | |||
| mutate( | |||
| is_donation = receipt_type_code %in% c("IND", "CPCM", "GEN", "PPTY", "OUTS", "NFPC"), | |||
| .after = receipt_type_code | |||
| ) |> | |||
| relocate(payer_id, .after = report_id) |> | |||
| collect() | |||
| @@ -17,10 +17,12 @@ prepare_report_list <- function(path_report_list) { | |||
| } | |||
| out_report_list <- function(report_list) { | |||
| out <- path("data-out", "report_list", "report_list.parquet") | |||
| out <- path("data-out", "reports", "reports.parquet") | |||
| dir_create(path_dir(out)) | |||
| arrow::write_parquet(report_list, out) | |||
| report_list |> | |||
| mutate(doc_order = as.integer(doc_name), .after = doc_name) |> | |||
| arrow::write_parquet(out) | |||
| dirname(out) | |||
| } | |||
| @@ -13,7 +13,8 @@ voter_statewide_download <- function(output_dir = here::here("../data-raw/voters | |||
| voter_statewide_convert_parquet <- function(path) { | |||
| path <- fs::path_norm(path) | |||
| path_out <- fs::path_ext_set(tolower(path), "parquet") | |||
| path_out <- path("data-out", "voters", "voters.parquet") | |||
| dir_create(path_dir(path_out)) | |||
| x <- readr::read_tsv(path, col_types = voter_statewide_spec()) | |||
| arrow::write_parquet(x, path_out) | |||
| @@ -139,7 +139,7 @@ list( | |||
| ## Voter registration records | |||
| tar_target(path_voters_txt, voter_statewide_download(), cue = tar_cue("never")), #<< invalidate to get latest | |||
| tar_target( | |||
| path_voters_parquet, | |||
| path_out_voters, | |||
| voter_statewide_convert_parquet(path_voters_txt), | |||
| cue = tar_cue("never"), | |||
| format = "file" | |||
| @@ -250,7 +250,7 @@ list( | |||
| tar_target( | |||
| path_out_committees, | |||
| out_committees(path_out_cover), | |||
| out_committees(path_out_cover, path_out_report_list), | |||
| format = "file" | |||
| ), | |||