adapt to reformat function in dunlin (#409)

adapt chevron preprocessing and tests to the new `reformat` method from dunlin. Sister PR to the `dunlin` PR: insightsengineering/dunlin#58 please merge at the same time. thank you for the review --------- Co-authored-by: benoit <[email protected]>
insightsengineering · Mar 8, 2023 · ded40c0 · ded40c0
1 parent 300fd35
commit ded40c0
Show file tree

Hide file tree

Showing 32 changed files with 153 additions and 159 deletions.
diff --git a/NAMESPACE b/NAMESPACE
@@ -240,7 +240,7 @@ import(rtables)
 import(scda.2022)
 import(tern)
 importFrom(checkmate,check_class)
-importFrom(dunlin,poly_pivot_wider)
+importFrom(dunlin,rule)
 importFrom(forcats,fct_relevel)
 importFrom(ggplot2,theme)
 importFrom(magrittr,"%>%")

diff --git a/R/aet01.R b/R/aet01.R
@@ -177,15 +177,15 @@ aet01_1_pre <- function(adam_db,
     ) %>%
     dm_update_zoomed()
 
-  missing_list <- list("<Missing>" = c("", NA))
+  missing_rule <- rule("<Missing>" = c("", NA))
 
   new_format <- list(
     adsl = list(
-      DCSREAS = missing_list
+      DCSREAS = missing_rule
     )
   )
 
-  db <- dunlin::apply_reformat(db, new_format)
+  db <- dunlin::reformat(db, new_format, na_last = TRUE)
 
   db
 }
@@ -482,15 +482,15 @@ aet01_2_pre <- function(adam_db) {
     ) %>%
     dm_update_zoomed()
 
-  missing_list <- list("<Missing>" = c("", NA))
+  missing_rule <- rule("<Missing>" = c("", NA))
 
   new_format <- list(
     adsl = list(
-      DCSREAS = missing_list
+      DCSREAS = missing_rule
     )
   )
 
-  db <- dunlin::apply_reformat(db, new_format)
+  db <- dunlin::reformat(db, new_format, na_last = TRUE)
 
   db
 }

diff --git a/R/aet03.R b/R/aet03.R
@@ -111,20 +111,18 @@ aet03_1_pre <- function(adam_db) {
 
   new_format <- list(
     adae = list(
-      AEBODSYS = list("No Coding available" = c("", NA, "<Missing>")),
-      AEDECOD = list("No Coding available" = c("", NA, "<Missing>")),
-      ASEV = list("<Missing>" = c("", NA, "<Missing>"))
+      AEBODSYS = rule("No Coding available" = c("", NA, "<Missing>")),
+      AEDECOD = rule("No Coding available" = c("", NA, "<Missing>")),
+      ASEV = rule("<Missing>" = c("", NA, "<Missing>"))
     )
   )
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adae") %>%
     filter(.data$ANL01FL == "Y") %>%
     filter(.data$ASEV != "<Missing>") %>%
-    mutate(ASEV = factor(.data$ASEV,
-      levels = setdiff(levels(.data$ASEV), "<Missing>")
-    )) %>%
+    mutate(ASEV = factor(.data$ASEV, levels = setdiff(levels(.data$ASEV), "<Missing>"))) %>%
     dm_update_zoomed()
 }
 

diff --git a/R/aet04.R b/R/aet04.R
@@ -130,13 +130,13 @@ aet04_1_pre <- function(adam_db) {
 
   new_format <- list(
     adae = list(
-      AEBODSYS = list("No Coding Available" = c("", NA, "<Missing>")),
-      AEDECOD = list("No Coding Available" = c("", NA, "<Missing>")),
-      ATOXGR = list("No Grading Available" = c("", NA, "<Missing>"))
+      AEBODSYS = rule("No Coding Available" = c("", NA, "<Missing>")),
+      AEDECOD = rule("No Coding Available" = c("", NA, "<Missing>")),
+      ATOXGR = rule("No Grading Available" = c("", NA, "<Missing>"))
     )
   )
 
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adae") %>%

diff --git a/R/cmt01a.R b/R/cmt01a.R
@@ -122,18 +122,18 @@ cmt01a_1_pre <- function(adam_db, medcat_var = "ATC2", medname_var = "CMDECOD")
     dm_update_zoomed()
 
   fmt_ls <- list(
-    medcat_var = list(
+    medcat_var = rule(
       "No Coding available" = c("", NA)
     ),
-    medname_var = list(
+    medname_var = rule(
       "No Coding available" = c("", NA)
     )
   )
 
   names(fmt_ls) <- c(medcat_var, medname_var)
   new_format <- list(adcm = fmt_ls)
 
-  dunlin::apply_reformat(adam_db, new_format)
+  dunlin::reformat(adam_db, new_format, na_last = TRUE)
 }
 
 #' @describeIn cmt01a_1 Postprocessing
@@ -255,18 +255,18 @@ cmt01a_2_pre <- function(adam_db, medcat_var = "ATC2", medname_var = "CMDECOD")
     dm_update_zoomed()
 
   fmt_ls <- list(
-    medcat_var = list(
+    medcat_var = rule(
       "No Coding available" = c("", NA)
     ),
-    medname_var = list(
+    medname_var = rule(
       "No Coding available" = c("", NA)
     )
   )
 
   names(fmt_ls) <- c(medcat_var, medname_var)
   new_format <- list(adcm = fmt_ls)
 
-  dunlin::apply_reformat(adam_db, new_format)
+  dunlin::reformat(adam_db, new_format, na_last = TRUE)
 }
 
 #' @describeIn cmt01a_2 Postprocessing
@@ -440,18 +440,18 @@ cmt01a_3_pre <- function(adam_db, medcat_var = "ATC2", medname_var = "CMDECOD")
     dm_update_zoomed()
 
   fmt_ls <- list(
-    medcat_var = list(
+    medcat_var = rule(
       "No Coding available" = c("", NA)
     ),
-    medname_var = list(
+    medname_var = rule(
       "No Coding available" = c("", NA)
     )
   )
 
   names(fmt_ls) <- c(medcat_var, medname_var)
   new_format <- list(adcm = fmt_ls)
 
-  dunlin::apply_reformat(adam_db, new_format)
+  dunlin::reformat(adam_db, new_format, na_last = TRUE)
 }
 
 #' @describeIn cmt01a_3 Preprocessing

diff --git a/R/cmt02_pt.R b/R/cmt02_pt.R
@@ -77,17 +77,17 @@ cmt02_pt_1_pre <- function(adam_db) {
     dm_update_zoomed()
 
   fmt_ls <- list(
-    CMDECOD = list(
+    CMDECOD = rule(
       "No Coding available" = c("", NA)
     ),
-    CMSEQ = list(
+    CMSEQ = rule(
       "<Missing>" = c("", NA)
     )
   )
 
   new_format <- list(adcm = fmt_ls)
 
-  dunlin::apply_reformat(adam_db, new_format)
+  dunlin::reformat(adam_db, new_format, na_last = TRUE)
 }
 
 #' @describeIn cmt02_pt_1 Postprocessing

diff --git a/R/dmt01.R b/R/dmt01.R
@@ -94,17 +94,6 @@ dmt01_1_lyt <- function(arm_var,
 dmt01_1_pre <- function(adam_db) {
   checkmate::assert_class(adam_db, "dm")
 
-  new_format <- list(
-    adsl = list(
-      SEX = list(
-        "Female" = "F",
-        "Male" = "M"
-      )
-    )
-  )
-
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
-
   adam_db <- dunlin::dm_explicit_na(adam_db)
 
   db <- adam_db %>%

diff --git a/R/dst01.R b/R/dst01.R
@@ -155,8 +155,8 @@ dst01_1_pre <- function(adam_db,
                         disc_reason_var = "DCSREAS") {
   new_format <- list(
     adsl = list(
-      list("<Missing>" = c("", NA)),
-      list(
+      rule("<Missing>" = c("", NA)),
+      rule(
         "COMPLETED" = "Completed",
         "DISCONTINUED" = "Discontinued",
         "ONGOING" = "Ongoing",
@@ -167,11 +167,14 @@ dst01_1_pre <- function(adam_db,
 
   names(new_format$adsl) <- c(disc_reason_var, status_var)
 
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adsl") %>%
     mutate(DOMAIN = "ADSL") %>%
+    mutate(
+      !!status_var := factor(.data[[status_var]], levels = c("COMPLETED", "DISCONTINUED", "ONGOING", "<Missing>"))
+    ) %>%
     dm_update_zoomed()
 }
 
@@ -380,8 +383,8 @@ dst01_2_pre <- function(adam_db,
 
   new_format <- list(
     adsl = list(
-      list("<Missing>" = c("", NA)),
-      list(
+      rule("<Missing>" = c("", NA)),
+      rule(
         "COMPLETED" = "Completed",
         "DISCONTINUED" = "Discontinued",
         "ONGOING" = "Ongoing",
@@ -392,10 +395,13 @@ dst01_2_pre <- function(adam_db,
 
   names(new_format$adsl) <- c(disc_reason_var, status_var)
 
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adsl") %>%
+    mutate(
+      !!status_var := factor(.data[[status_var]], levels = c("COMPLETED", "DISCONTINUED", "ONGOING", "<Missing>"))
+    ) %>%
     mutate(reasonGP = case_when(
       .data[[disc_reason_var]] %in% c("ADVERSE EVENT", "DEATH") ~ "Safety",
       (.data[[disc_reason_var]] != "<Missing>" &
@@ -606,14 +612,14 @@ dst01_3_pre <- function(adam_db,
 
   new_format <- list(
     adsl = list(
-      list("<Missing>" = c("", NA)),
-      list(
+      rule("<Missing>" = c("", NA)),
+      rule(
         "COMPLETED" = "Completed",
         "DISCONTINUED" = "Discontinued",
         "ONGOING" = "Ongoing",
         "<Missing>" = c("", NA)
       ),
-      list(
+      rule(
         "COMPLETED" = "Completed",
         "DISCONTINUED" = "Discontinued",
         "ONGOING" = "Ongoing",
@@ -624,10 +630,13 @@ dst01_3_pre <- function(adam_db,
 
   names(new_format$adsl) <- c(disc_reason_var, status_var, status_treatment_var)
 
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adsl") %>%
+    mutate(
+      !!status_var := factor(.data[[status_var]], levels = c("COMPLETED", "DISCONTINUED", "ONGOING", "<Missing>"))
+    ) %>%
     mutate(reasonGP = case_when(
       .data[[disc_reason_var]] %in% c("ADVERSE EVENT", "DEATH") ~ "Safety",
       .data[[disc_reason_var]] == "<Missing>" ~ "<Missing>",

diff --git a/R/dtht01.R b/R/dtht01.R
@@ -148,32 +148,25 @@ dtht01_1_opt_lyt <- function(arm_var,
 dtht01_1_pre <- function(adam_db) {
   checkmate::assert_class(adam_db, "dm")
 
-  adam_db <- adam_db %>%
-    dm_zoom_to("adsl") %>%
-    mutate(DTHFL = as.factor(.data$DTHFL)) %>%
-    dm_update_zoomed()
-
-  # Reorder factors to have "OTHER" last.
-  dthcat <- as.factor(adam_db$adsl$DTHCAT)
-  dthcat_lvl <- levels(dthcat)
-  dthcat_lvl <- setdiff(dthcat_lvl, c("OTHER", "", NA))
-  dthcat_lvl <- c(dthcat_lvl, "OTHER")
-
-  dthcat_lvl_order <- as.list(setNames(dthcat_lvl, dthcat_lvl))
-  na_lvl <- list("<Missing>" = c("", NA))
+  missing_rule <- rule("<Missing>" = c("", NA))
 
   new_formats <- list(
     adsl = list(
-      DTHCAT = c(
-        dthcat_lvl_order,
-        na_lvl
-      ),
-      DTHCAUS = na_lvl,
-      LDDTHGR1 = na_lvl
+      DTHCAT = missing_rule,
+      DTHCAUS = missing_rule,
+      LDDTHGR1 = missing_rule
     )
   )
 
-  dunlin::apply_reformat(adam_db, new_formats)
+  # Reorder factors to have "OTHER" last.
+  adam_db <- adam_db %>%
+    dm_zoom_to("adsl") %>%
+    mutate(DTHFL = as.factor(.data$DTHFL)) %>%
+    mutate(DTHCAT = as.factor(DTHCAT)) %>%
+    mutate(DTHCAT = factor(DTHCAT, levels = c(setdiff(levels(DTHCAT), "OTHER"), "OTHER"))) %>%
+    dm_update_zoomed()
+
+  dunlin::reformat(adam_db, new_formats, na_last = TRUE)
 }
 
 

diff --git a/R/lbt04.R b/R/lbt04.R
@@ -81,11 +81,11 @@ lbt04_1_pre <- function(adam_db, req_tables = c("adsl", "adlb"), arm_var = "ACTA
 
   new_format <- list(
     adlb = list(
-      ANRIND = list("<Missing>" = c("", NA, "<Missing>"))
+      ANRIND = rule("<Missing>" = c("", NA, "<Missing>"))
     )
   )
 
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adlb") %>%

diff --git a/R/lbt05.R b/R/lbt05.R
@@ -103,23 +103,23 @@ lbt05_1_pre <- function(adam_db, arm_var = "ACTARM") {
       .data$PARCAT2 == "LS",
       !is.na(.data$AVAL)
     ) %>%
-    dm_update_zoomed() %>%
-    dm_zoom_to("adlb") %>%
     mutate(abn_dir = factor(case_when(
       ANRIND == "LOW LOW" ~ "Low",
       ANRIND == "HIGH HIGH" ~ "High",
       TRUE ~ ""
     ), levels = c("Low", "High"))) %>%
     dm_update_zoomed()
 
+  missing_rule <- rule("<Missing>" = c("", NA, "<Missing>", "No Coding Available"))
+
   new_format <- list(
     adlb = list(
-      AVALCAT1 = list("<Missing>" = c("", NA, "<Missing>", "No Coding Available")),
-      abn_dir = list("<Missing>" = c("", NA, "<Missing>", "No Coding Available"))
+      AVALCAT1 = missing_rule,
+      abn_dir = missing_rule
     )
   )
 
-  db <- dunlin::apply_reformat(db, new_format)
+  db <- dunlin::reformat(db, new_format, na_last = TRUE)
 }
 
 #' @describeIn lbt05_1 Checks

diff --git a/R/lbt07.R b/R/lbt07.R
@@ -112,11 +112,11 @@ lbt07_1_pre <- function(adam_db) {
 
   new_format <- list(
     adlb = list(
-      ATOXGR = list("<Missing>" = c("", NA, "<Missing>", "No Coding available"))
+      ATOXGR = rule("<Missing>" = c("", NA, "<Missing>", "No Coding available"))
     )
   )
 
-  adam_db <- dunlin::apply_reformat(adam_db, new_format)
+  adam_db <- dunlin::reformat(adam_db, new_format, na_last = TRUE)
 
   adam_db %>%
     dm_zoom_to("adlb") %>%