Skip to content

Commit

Permalink
feat: #2125 separate exist_flag and missing_value functionality
Browse files Browse the repository at this point in the history
  • Loading branch information
Zelos Zhu committed Oct 12, 2023
1 parent f34ae89 commit 13ea0a6
Show file tree
Hide file tree
Showing 2 changed files with 75 additions and 36 deletions.
28 changes: 19 additions & 9 deletions R/derive_merged.R
View file
Original file line number Diff line number Diff line change
Expand Up @@ -385,18 +385,26 @@ derive_vars_merged <- function(dataset,
select(!!!by_vars_right, !!!replace_values_by_names(new_vars))
}

if (!is.null(exist_flag)) {
add_data <- mutate(
add_data,
!!exist_flag := TRUE
)
}

if (!is.null(missing_values)) {
exist_flag_var <- get_new_tmp_var(add_data, prefix = "tmp_exist_flag")
missing_values_var <- get_new_tmp_var(add_data, prefix = "tmp_missing_flag")
} else {
exist_flag_var <- exist_flag
missing_values_var <- missing_values
}

if (!is.null(exist_flag_var)) {
if (!is.null(missing_values_var)) {
add_data <- mutate(
add_data,
!!exist_flag_var := true_value
!!missing_values_var := TRUE
)
}


# check if there are any variables in both datasets which are not by vars
# in this case an error is issued to avoid renaming of varibles by left_join()
common_vars <-
Expand All @@ -420,20 +428,22 @@ derive_vars_merged <- function(dataset,
}
dataset <- left_join(dataset, add_data, by = vars2chr(by_vars))

if (!is.null(missing_values)) {
if (!is.null(missing_values_var)) {
update_missings <- map2(
syms(names(missing_values)),
missing_values,
~ expr(if_else(is.na(!!exist_flag_var), !!.y, !!.x))
~ expr(if_else(is.na(!!missing_values_var), !!.y, !!.x))
)
names(update_missings) <- names(missing_values)
dataset <- dataset %>%
mutate(!!!update_missings)
}
if (!is.null(exist_flag_var)) {

if (!is.null(exist_flag)) {
dataset <- dataset %>%
mutate(!!exist_flag_var := ifelse(is.na(!!exist_flag_var), false_value, !!exist_flag_var))
mutate(!!exist_flag := ifelse(is.na(!!exist_flag), false_value, true_value))
}

dataset %>%
remove_tmp_vars()
}
Expand Down
83 changes: 56 additions & 27 deletions tests/testthat/test-derive_merged.R
View file
Original file line number Diff line number Diff line change
Expand Up @@ -193,8 +193,36 @@ test_that("derive_vars_merged Test 7: expressions for new_vars and missing_value
)
})

## Test 8: use new variables in filter_add and order ----
test_that("derive_vars_merged Test 8: use new variables in filter_add and order", {

## Test 8: expressions for new_vars and missing_values and exist_flags ----
test_that("derive_vars_merged Test 8: expressions for new_vars and missing_values and exist_flags", {
actual <- derive_vars_merged(
adsl,
dataset_add = advs,
by_vars = exprs(USUBJID),
order = exprs(AVISIT),
new_vars = exprs(LASTVIS = str_to_upper(AVISIT)),
mode = "last",
missing_values = exprs(LASTVIS = "UNKNOWN"),
exist_flag = matched,
true_value = "Yes",
false_value = "No"
)

expected <- adsl %>%
mutate(LASTVIS = c("WEEK 2", "BASELINE", "WEEK 4", "UNKNOWN"),
matched = c("Yes", "Yes", "Yes", "No"))


expect_dfs_equal(
base = expected,
compare = actual,
keys = "USUBJID"
)
})

## Test 9: use new variables in filter_add and order ----
test_that("derive_vars_merged Test 9: use new variables in filter_add and order", {
expected <- tibble::tribble(
~USUBJID, ~TRTSDT, ~TRTSSEQ,
"ST42-1", "2020-12-14", 2,
Expand Down Expand Up @@ -232,8 +260,8 @@ test_that("derive_vars_merged Test 8: use new variables in filter_add and order"
)
})

## Test 9: warning if not unique w.r.t the by variables and the order ----
test_that("derive_vars_merged Test 9: warning if not unique w.r.t the by variables and the order", {
## Test 10: warning if not unique w.r.t the by variables and the order ----
test_that("derive_vars_merged Test 10: warning if not unique w.r.t the by variables and the order", {
expect_warning(
actual <- derive_vars_merged(advs,
dataset_add = adsl2,
Expand All @@ -246,8 +274,8 @@ test_that("derive_vars_merged Test 9: warning if not unique w.r.t the by variabl
)
})

## Test 10: error if not unique w.r.t the by variables and the order ----
test_that("derive_vars_merged Test 10: error if not unique w.r.t the by variables and the order", {
## Test 11: error if not unique w.r.t the by variables and the order ----
test_that("derive_vars_merged Test 11: error if not unique w.r.t the by variables and the order", {
expect_error(
actual <- derive_vars_merged(advs,
dataset_add = adsl2,
Expand All @@ -261,8 +289,8 @@ test_that("derive_vars_merged Test 10: error if not unique w.r.t the by variable
)
})

## Test 11: error if variables in missing_values but not in new_vars ----
test_that("derive_vars_merged Test 11: error if variables in missing_values but not in new_vars", {
## Test 12: error if variables in missing_values but not in new_vars ----
test_that("derive_vars_merged Test 12: error if variables in missing_values but not in new_vars", {
expect_error(
derive_vars_merged(
adsl,
Expand All @@ -279,8 +307,8 @@ test_that("derive_vars_merged Test 11: error if variables in missing_values but
})

# derive_var_merged_exist_flag ----
## Test 11: merge existence flag ----
test_that("derive_var_merged_exist_flag Test 11: merge existence flag", {
## Test 13: merge existence flag ----
test_that("derive_var_merged_exist_flag Test 13: merge existence flag", {
actual <- derive_var_merged_exist_flag(
adsl,
dataset_add = advs,
Expand All @@ -300,8 +328,8 @@ test_that("derive_var_merged_exist_flag Test 11: merge existence flag", {
)
})

## Test 12: by_vars with rename ----
test_that("derive_var_merged_exist_flag Test 12: by_vars with rename", {
## Test 14: by_vars with rename ----
test_that("derive_var_merged_exist_flag Test 14: by_vars with rename", {
actual <- derive_var_merged_exist_flag(
adsl,
dataset_add = advs1,
Expand All @@ -322,8 +350,8 @@ test_that("derive_var_merged_exist_flag Test 12: by_vars with rename", {
})

# derive_vars_merged_lookup ----
## Test 13: merge lookup table ----
test_that("derive_vars_merged_lookup Test 13: merge lookup table", {
## Test 15: merge lookup table ----
test_that("derive_vars_merged_lookup Test 15: merge lookup table", {
param_lookup <- tibble::tribble(
~VSTESTCD, ~VSTEST, ~PARAMCD, ~DESCRIPTION,
"WEIGHT", "Weight", "WEIGHT", "Weight (kg)",
Expand Down Expand Up @@ -360,8 +388,8 @@ test_that("derive_vars_merged_lookup Test 13: merge lookup table", {


## the lookup table
## Test 14: all by_vars have records in the lookup table ----
test_that("derive_vars_merged_lookup Test 14: all by_vars have records in the lookup table", {
## Test 16: all by_vars have records in the lookup table ----
test_that("derive_vars_merged_lookup Test 16: all by_vars have records in the lookup table", {
param_lookup <- tibble::tribble(
~VSTESTCD, ~VSTEST, ~PARAMCD, ~DESCRIPTION,
"WEIGHT", "Weight", "WEIGHT", "Weight (kg)",
Expand Down Expand Up @@ -396,8 +424,8 @@ test_that("derive_vars_merged_lookup Test 14: all by_vars have records in the l
)
})

## Test 15: by_vars with rename ----
test_that("derive_vars_merged_lookup Test 15: by_vars with rename", {
## Test 17: by_vars with rename ----
test_that("derive_vars_merged_lookup Test 17: by_vars with rename", {
param_lookup <- tibble::tribble(
~TESTCD, ~VSTEST, ~PARAMCD, ~DESCRIPTION,
"WEIGHT", "Weight", "WEIGHT", "Weight (kg)",
Expand Down Expand Up @@ -433,8 +461,8 @@ test_that("derive_vars_merged_lookup Test 15: by_vars with rename", {


# get_not_mapped ----
## Test 16: not all by_vars have records in the lookup table ----
test_that("get_not_mapped Test 16: not all by_vars have records in the lookup table", {
## Test 18: not all by_vars have records in the lookup table ----
test_that("get_not_mapped Test 18: not all by_vars have records in the lookup table", {
param_lookup <- tibble::tribble(
~VSTESTCD, ~VSTEST, ~PARAMCD, ~DESCRIPTION,
"WEIGHT", "Weight", "WEIGHT", "Weight (kg)",
Expand Down Expand Up @@ -472,8 +500,8 @@ test_that("get_not_mapped Test 16: not all by_vars have records in the lookup ta
})

# derive_var_merged_summary ----
## Test 17: dataset == dataset_add, no filter ----
test_that("derive_var_merged_summary Test 17: dataset == dataset_add, no filter", {
## Test 19: dataset == dataset_add, no filter ----
test_that("derive_var_merged_summary Test 19: dataset == dataset_add, no filter", {
expected <- tibble::tribble(
~AVISIT, ~ASEQ, ~AVAL, ~MEANVIS,
"WEEK 1", 1, 10, 10,
Expand All @@ -499,8 +527,8 @@ test_that("derive_var_merged_summary Test 17: dataset == dataset_add, no filter"
)
})

## Test 18: dataset != dataset_add, filter ----
test_that("derive_var_merged_summary Test 18: dataset != dataset_add, filter", {
## Test 20: dataset != dataset_add, filter ----
test_that("derive_var_merged_summary Test 20: dataset != dataset_add, filter", {
expected <- tibble::tribble(
~USUBJID, ~MEANPBL,
"1", 13.5,
Expand Down Expand Up @@ -531,8 +559,8 @@ test_that("derive_var_merged_summary Test 18: dataset != dataset_add, filter", {
)
})

## Test 19: by_vars with rename ----
test_that("derive_var_merged_summary Test 19: by_vars with rename", {
## Test 21: by_vars with rename ----
test_that("derive_var_merged_summary Test 21: by_vars with rename", {
expected <- tibble::tribble(
~AVISIT, ~ASEQ, ~AVAL, ~MEANVIS,
"WEEK 1", 1, 10, 10,
Expand All @@ -559,7 +587,8 @@ test_that("derive_var_merged_summary Test 19: by_vars with rename", {
)
})

test_that("derive_var_merged_summary Test 19: deprecation warning", {
## Test 22: deprecation warning ----
test_that("derive_var_merged_summary Test 22: deprecation warning", {
expected <- tibble::tribble(
~AVISIT, ~ASEQ, ~AVAL, ~MEANVIS,
"WEEK 1", 1, 10, 10,
Expand Down

0 comments on commit 13ea0a6

Please sign in to comment.