ark_taxonomy pin

data-raw/nucc_files.R

@@ -83,7 +83,7 @@ parse_nucc_csvs <- function(path) {
 
 purrr::walk(nucc_paths, parse_nucc_csvs)
 
-##################### ARCHIVE
+# ARCHIVE
 archive::archive_write_files(
   archive = here::here("data-raw/raw/nucc_taxonomy.tar.xz"),
   files = c(
@@ -92,6 +92,12 @@ archive::archive_write_files(
     )
   )
 
-fs::dir_delete(
-  glue::glue("{here::here()}/data-raw/raw/csvs")
-)
+# fs::dir_delete(glue::glue("{here::here()}/data-raw/raw/csvs"))
+
+archive::archive(file = here::here("data-raw/raw/nucc_taxonomy.tar.xz"))
+
+readr::read_csv(
+  file = archive::archive_read(archive = here::here("data-raw/raw/nucc_taxonomy.tar.xz"), file = 1L),
+  show_col_types = FALSE,
+  col_types = "c",
+  name_repair = janitor::make_clean_names)

data-raw/nucc_pins.R

@@ -2,29 +2,34 @@ source(here::here("data-raw", "pins_internal.R"))
 
 cleaned_csvs <- fs::dir_info(glue::glue("{here::here()}/data-raw/clean/csvs"))$path
 
-tx_100 <- readr::read_csv(
-  file = tools::file_path_sans_ext(cleaned_csvs[1]),
-  id = "filename",
-  show_col_types = FALSE,
-  col_types = "c",
-  name_repair = janitor::make_clean_names
-  ) |>
-  dplyr::reframe(
-    release = basename(filename),
-    version,
-    release_date,
-    code,
-    grouping = type,
-    classification,
-    specialization,
-    definition,
-    notes
-    )
+cols <- c(
+  "version",
+  "release_date",
+  "code",
+  "grouping",
+  "grouping" = "type",
+  "classification",
+  "specialization",
+  "definition",
+  "notes"
+)
+
+read_clean_csvs <- function(x) {
+  readr::read_csv(
+    file = tools::file_path_sans_ext(x),
+    show_col_types = FALSE,
+    col_types = "c",
+    name_repair = janitor::make_clean_names) |>
+    dplyr::select(
+      dplyr::any_of(cols))
+}
+
+ark_taxonomy <- purrr::map(cleaned_csvs[1:20], read_clean_csvs) |>
+  purrr::list_rbind()
 
 pin_update(
-  tx_100,
-  name = "tax_v_10.0",
-  title = "Taxonomy Codeset v10.0",
-  description = "Health Care Provider Taxonomy Code Set, Version 10.1, Released 2010-01-01",
-  force = TRUE
+  ark_taxonomy,
+  name = "ark_taxonomy",
+  title = "NUCC Taxonomy Archive",
+  description = "Health Care Provider Taxonomy Code Set Archive"
 )

inst/extdata/pins/_pins.yaml

@@ -1,4 +1,2 @@
-tax_v_10.0:
-- tax_v_10.0/20241122T094232Z-7a358/
-tax_v_10.1:
-- tax_v_10.1/20241122T093658Z-1ced0/
+ark_taxonomy:
+- ark_taxonomy/20241122T213627Z-edbae/

inst/extdata/pins/ark_taxonomy/20241122T213627Z-edbae/data.txt

@@ -0,0 +1,10 @@
+file: ark_taxonomy.qs
+file_size: 722364
+pin_hash: edbaea85d09bbffa
+type: qs
+title: NUCC Taxonomy Archive
+description: Health Care Provider Taxonomy Code Set Archive
+tags: ~
+urls: ~
+created: 20241122T213627Z
+api_version: 1