Refactor chi deaths

R/get_slf_lookup_paths.R

@@ -112,12 +112,21 @@ get_combined_slf_deaths_lookup_path <- function(update = latest_update(), ...) {
 #' @export
 #' @family slf lookup file path
 #' @seealso [get_file_path()] for the generic function.
-get_slf_chi_deaths_path <- function(update = latest_update(), ...) {
-  slf_chi_deaths_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Deaths"),
-    file_name = stringr::str_glue("anon-chi_deaths_{update}.parquet"),
-    ...
-  )
+get_slf_chi_deaths_path <- function(update = latest_update(), BYOC_MODE = FALSE, ...) {
+  file_name <- stringr::str_glue("anon-chi_deaths_{update}.parquet")
+
+  if (BYOC_MODE) {
+    slf_chi_deaths_path <- file.path(
+      directory = denodo_output_path(), ## TO-DO: waiting to be finalised
+      file_name = file_name
+    )
+  } else {
+    slf_chi_deaths_path <- get_file_path(
+      directory = fs::path(get_slf_dir(), "Deaths"),
+      file_name = file_name,
+      ...
+    )
+  }
 
   return(slf_chi_deaths_path)
 }

R/process_it_chi_deaths.R

@@ -10,7 +10,11 @@
 #' @return the final data as a [tibble][tibble::tibble-package].
 #' @export
 #' @family process extracts
-process_it_chi_deaths <- function(data, write_to_disk = TRUE) {
+process_it_chi_deaths <- function(data,
+                                  write_to_disk = TRUE,
+                                  BYOC_MODE = FALSE,
+                                  run_id = NA,
+                                  run_date_time = NA) {
   log_slf_event(stage = "process", status = "start", type = "it_chi_deaths", year = "all")
 
   it_chi_deaths_clean <- data %>%
@@ -21,16 +25,28 @@ process_it_chi_deaths <- function(data, write_to_disk = TRUE) {
     dplyr::distinct(.data$anon_chi, .keep_all = TRUE) %>%
     # remove death_date_nrs as this is the nrs weekly unvalidated data and we should not use this.
     # the boxi nrs death date is more reliable as this is provided monthly and is validated.
-    dplyr::select(.data$anon_chi, .data$death_date_chi) %>%
     dplyr::mutate(
-      death_date_chi = lubridate::ymd(.data$death_date_chi)
+      death_date_chi = lubridate::ymd(.data$death_date_chi),
+      run_id = run_id,
+      run_date_time = run_date_time
+    ) %>%
+    dplyr::select(
+      "run_id",
+      "run_date_time",
+      "anon_chi",
+      "death_date_chi"
     )
 
   if (write_to_disk) {
-    it_chi_deaths_clean %>%
-      write_file(get_slf_chi_deaths_path(check_mode = "write"),
-        group_id = 3206 # hscdiip owner
-      )
+    write_file(
+      it_chi_deaths_clean,
+      get_slf_chi_deaths_path(
+        BYOC_MODE,
+        check_mode = "write"
+      ),
+      BYOC_MODE = BYOC_MODE,
+      group_id = 3206 # hscdiip owner
+    )
   }
 
   log_slf_event(stage = "process", status = "complete", type = "it_chi_deaths", year = "all")

R/read_it_chi_deaths.R

@@ -6,19 +6,31 @@
 #' @return the final data as a [tibble][tibble::tibble-package].
 #' @export
 #' @family process extracts
-read_it_chi_deaths <- function(file_path = get_it_deaths_path()) {
+read_it_chi_deaths <- function(denodo_connect = get_denodo_connection(BYOC_MODE = BYOC_MODE),
+                               file_path = get_it_deaths_path(), ## CHECK + Is this needed?
+                               BYOC_MODE ## Is this needed?
+) {
   log_slf_event(stage = "read", status = "start", type = "it_chi_deaths", year = "all")
 
-  it_chi_deaths <- read_file(file_path) %>%
+  on.exit(try(DBI::dbDisconnect(denodo_connect), silent = TRUE), add = TRUE)
+
+  it_chi_deaths <- dplyr::tbl(
+    denodo_connect,
+    dbplyr::in_schema("sdl", "sdl_XXX")
+  ) %>% ## PLACEHOLDER
     dplyr::select(
-      anon_chi = "anon_chi",
-      death_date_nrs = "PATIENT DoD DATE (NRS)",
-      death_date_chi = "PATIENT DoD DATE (CHI)"
+      anon_chi = "patient_upi", ## CHECK
+      death_date_nrs = "patient_dod_nrs", ## CHECK
+      death_date_chi = "patient_dod_chi" ## CHECK
     ) %>%
     dplyr::mutate(
       death_date_nrs = lubridate::dmy(.data$death_date_nrs),
       death_date_chi = lubridate::dmy(.data$death_date_chi)
-    )
+    ) %>%
+    dplyr::collect() %>%
+    slfhelper::get_anon_chi("anon_chi") ## CHECK
+
+  log_slf_event(stage = "read", status = "complete", type = "it_chi_deaths", year = "all")
 
   log_slf_event(stage = "read", status = "complete", type = "it_chi_deaths", year = "all")
 

SDL_process/run_sdl.r

@@ -162,7 +162,7 @@ year <- "1920"
 # Build BYOC Output File Paths
 byoc_output_files <- get_byoc_output_files(
   year = year,
-  types = c("homelessness", "maternity") # using homelessness for test purpose. When development is complete, we change to "types = "byoc_input_files""
+  types = c("homelessness", "maternity", "it_chi_deaths") # using homelessness for test purpose. When development is complete, we change to "types = "byoc_input_files""
 ) # can always use any other type for testing also
 
 ## targets ----
@@ -207,4 +207,17 @@ maternity <- read_extract_maternity(
     run_date_time = run_date_time
   )
 
+# Logger is included within each function so is excluded here?
+it_chi_deaths <- read_it_chi_deaths(
+  denodo_connect = get_denodo_connection(BYOC_MODE = BYOC_MODE),
+  file_path = get_it_deaths_path(),
+  BYOC_MODE = BYOC_MODE
+) %>%
+  process_it_chi_deaths(
+    write_to_disk = TRUE,
+    BYOC_MODE = BYOC_MODE,
+    run_id = run_id,
+    run_date_time = run_date_time
+  )
+
 logger::log_info("Run SDL ended.")

_targets.R

@@ -258,9 +258,15 @@ list(
   ),
   # IT deaths-----------------------------------------------------------------
   # READ - IT CHI deaths------
-  tar_file_read(it_chi_deaths_extract,
-    command = get_it_deaths_path(),
-    read = read_it_chi_deaths(!!.x)
+  tar_target(
+    # Target name
+    it_chi_deaths_extract,
+    # Function
+    read_it_chi_deaths(
+      denodo_connect = get_denodo_connection(BYOC_MODE = BYOC_MODE),
+      file_path = get_it_deaths_path(),
+      BYOC_MODE = BYOC_MODE
+    )
   ),
   # PROCESS - IT CHI deaths------
   tar_target(
@@ -269,7 +275,10 @@ list(
     # Function
     process_it_chi_deaths(
       data = it_chi_deaths_extract,
-      write_to_disk = write_to_disk
+      write_to_disk = write_to_disk,
+      BYOC_MODE = BYOC_MODE,
+      run_id = run_id,
+      run_date_time = run_date_time
     ),
     priority = 0.9
   ),