Archive social care extracts (#927)

* Set up `get_sandpit_extract_path`

* Update documentation

* Update sc `all` data paths

* Write sandpit extract if file does not exist

* Style code

---------

Co-authored-by: Jennit07 <[email protected]>

NAMESPACE

@@ -54,6 +54,7 @@ export(get_nsu_path)
 export(get_pop_path)
 export(get_practice_details_path)
 export(get_readcode_lookup_path)
+export(get_sandpit_extract_path)
 export(get_sc_at_episodes_path)
 export(get_sc_ch_episodes_path)
 export(get_sc_client_lookup_path)

R/get_sandpit_extract_path.R

@@ -0,0 +1,39 @@
+#' Sandpit Extract File Path
+#'
+#' @description Get the file path for sandpit extracts
+#'
+#' @param update The update month to use,
+#' defaults to [latest_update()]
+#'
+#' @param ... additional arguments passed to [get_file_path()]
+#'
+#' @return The path to the sandpit extracts as an [fs::path()]
+#' @export
+#' @family social care sandpit extract paths
+#' @seealso [get_file_path()] for the generic function.
+get_sandpit_extract_path <- function(type = c(
+                                       "at", "ch", "hc",
+                                       "sds", "client", "demographics"
+                                     ),
+                                     year = NULL,
+                                     update = latest_update(), ...) {
+  dir <- fs::path(get_slf_dir(), "Social_care", "Sandpit_Extracts")
+
+  file_name <- dplyr::case_match(
+    type,
+    "at" ~ "sandpit_at_extract",
+    "ch" ~ "sandpit_ch_extract",
+    "hc" ~ "sandpit_hc_extract",
+    "sds" ~ "sandpit_sds_extract",
+    "client" ~ "sandpit_sc_client_extract",
+    "demographics" ~ "sandpit_sc_demographics_extract"
+  )
+
+  if (type == "client") {
+    sandpit_extract_path <- fs::path(dir, stringr::str_glue("{file_name}_{year}_{update}.parquet"))
+  } else {
+    sandpit_extract_path <- fs::path(dir, stringr::str_glue("{file_name}_{update}.parquet"))
+  }
+
+  return(sandpit_extract_path)
+}

R/get_sc_episodes_path.R

@@ -13,7 +13,7 @@
 #' @seealso [get_file_path()] for the generic function.
 get_sc_ch_episodes_path <- function(update = latest_update(), ...) {
   sc_ch_episodes_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Social_care"),
+    directory = fs::path(get_slf_dir(), "Social_care", "processed_sc_all_care_home"),
     file_name = stringr::str_glue("all_ch_episodes_{update}.parquet"),
     ...
   )
@@ -33,7 +33,7 @@ get_sc_ch_episodes_path <- function(update = latest_update(), ...) {
 #' @seealso [get_file_path()] for the generic function.
 get_sc_at_episodes_path <- function(update = latest_update(), ...) {
   sc_at_episodes_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Social_care"),
+    directory = fs::path(get_slf_dir(), "Social_care", "processed_sc_all_alarms_telecare"),
     file_name = stringr::str_glue("all_at_episodes_{update}.parquet"),
     ...
   )
@@ -53,7 +53,7 @@ get_sc_at_episodes_path <- function(update = latest_update(), ...) {
 #' @seealso [get_file_path()] for the generic function.
 get_sc_hc_episodes_path <- function(update = latest_update(), ...) {
   sc_hc_episodes_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Social_care"),
+    directory = fs::path(get_slf_dir(), "Social_care", "processed_sc_all_home_care"),
     file_name = stringr::str_glue("all_hc_episodes_{update}.parquet"),
     ...
   )
@@ -73,7 +73,7 @@ get_sc_hc_episodes_path <- function(update = latest_update(), ...) {
 #' @seealso [get_file_path()] for the generic function.
 get_sc_sds_episodes_path <- function(update = latest_update(), ...) {
   sc_sds_episodes_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Social_care"),
+    directory = fs::path(get_slf_dir(), "Social_care", "processed_sc_all_sds"),
     file_name = stringr::str_glue("all_sds_episodes_{update}.parquet"),
     ...
   )

R/get_sc_lookup_paths.R

@@ -14,7 +14,7 @@
 #' @seealso [get_file_path()] for the generic function.
 get_sc_demog_lookup_path <- function(update = latest_update(), ...) {
   sc_demog_lookup_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Social_care"),
+    directory = fs::path(get_slf_dir(), "Social_care", "processed_sc_demographic_lookup"),
     file_name = stringr::str_glue("sc_demographics_lookup_{update}.parquet"),
     ...
   )
@@ -39,7 +39,7 @@ get_sc_demog_lookup_path <- function(update = latest_update(), ...) {
 #' @seealso [get_file_path()] for the generic function.
 get_sc_client_lookup_path <- function(year, update = latest_update(), ...) {
   sc_client_lookup_path <- get_file_path(
-    directory = fs::path(get_slf_dir(), "Social_care"),
+    directory = fs::path(get_slf_dir(), "Social_care", "processed_sc_client_lookup"),
     file_name = stringr::str_glue("sc_client_lookup_{year}_{update}.parquet"),
     ...
   )

R/read_lookup_sc_client.R

@@ -77,5 +77,12 @@ read_lookup_sc_client <- function(fyyear,
     ) %>%
     dplyr::collect()
 
+  if (!fs::file_exists(get_sandpit_extract_path(type = "client", year = fyyear))) {
+    client_data %>%
+      write_file(get_sandpit_extract_path(type = "client", year = fyyear))
+  } else {
+    client_data <- client_data
+  }
+
   return(client_data)
 }

R/read_lookup_sc_demographics.R

@@ -23,7 +23,16 @@ read_lookup_sc_demographics <- function(sc_connection = phs_db_connection(dsn =
       "submitted_postcode",
       "chi_gender_code"
     ) %>%
-    dplyr::collect() %>%
+    dplyr::collect()
+
+  if (!fs::file_exists(get_sandpit_extract_path(type = "demographics"))) {
+    sc_demog %>%
+      write_file(get_sandpit_extract_path(type = "demographics"))
+  } else {
+    sc_demog <- sc_demog
+  }
+
+  sc_demog <- sc_demog %>%
     dplyr::mutate(
       dplyr::across(c(
         "latest_record_flag",

R/read_sc_all_alarms_telecare.R

@@ -26,7 +26,16 @@ read_sc_all_alarms_telecare <- function(sc_dvprod_connection = phs_db_connection
       "service_start_date_after_period_end_date"
     ) %>%
     dplyr::collect() %>%
-    dplyr::distinct() %>%
+    dplyr::distinct()
+
+  if (!fs::file_exists(get_sandpit_extract_path(type = "at"))) {
+    at_full_data %>%
+      write_file(get_sandpit_extract_path(type = "at"))
+  } else {
+    at_full_data <- at_full_data
+  }
+
+  at_full_data <- at_full_data %>%
     dplyr::mutate(
       period_start_date = dplyr::if_else(
         .data$period == "2017",

R/read_sc_all_care_home.R

@@ -28,7 +28,16 @@ read_sc_all_care_home <- function(sc_dvprod_connection = phs_db_connection(dsn =
       "age"
     ) %>%
     dplyr::collect() %>%
-    dplyr::distinct() %>%
+    dplyr::distinct()
+
+  if (!fs::file_exists(get_sandpit_extract_path(type = "ch"))) {
+    ch_data %>%
+      write_file(get_sandpit_extract_path(type = "ch"))
+  } else {
+    ch_data <- ch_data
+  }
+
+  ch_data <- ch_data %>%
     # Correct FY 2017
     dplyr::mutate(period = dplyr::if_else(
       .data$period == "2017",

R/read_sc_all_home_care.R

@@ -45,7 +45,16 @@ read_sc_all_home_care <- function(sc_dvprod_connection = phs_db_connection(dsn =
     )) %>%
     # drop rows start date after end date
     dplyr::collect() %>%
-    dplyr::distinct() %>%
+    dplyr::distinct()
+
+  if (!fs::file_exists(get_sandpit_extract_path(type = "hc"))) {
+    home_care_data %>%
+      write_file(get_sandpit_extract_path(type = "hc"))
+  } else {
+    home_care_data <- home_care_data
+  }
+
+  home_care_data <- home_care_data %>%
     dplyr::mutate(dplyr::across(c(
       "sending_location",
       "financial_year",

R/read_sc_all_sds.R

@@ -26,7 +26,16 @@ read_sc_all_sds <- function(sc_dvprod_connection = phs_db_connection(dsn = "DVPR
       "sds_start_date_after_period_end_date" # get removed
     ) %>%
     dplyr::collect() %>%
-    dplyr::distinct() %>%
+    dplyr::distinct()
+
+  if (!fs::file_exists(get_sandpit_extract_path(type = "sds"))) {
+    sds_full_data %>%
+      write_file(get_sandpit_extract_path(type = "sds"))
+  } else {
+    sds_full_data <- sds_full_data
+  }
+
+  sds_full_data <- sds_full_data %>%
     dplyr::mutate(dplyr::across(c(
       "sending_location",
       "sds_option_1",