Use internal checking function to reduce duplication

R/cfr_rolling.R

@@ -63,32 +63,7 @@ cfr_rolling <- function(data,
     " the course of the outbreak."
   )
   # input checking
-  checkmate::assert_data_frame(
-    data,
-    min.rows = 1, min.cols = 3
-  )
-  # check that input `<data.frame>` has columns date, cases, and deaths
-  checkmate::assert_names(
-    colnames(data),
-    must.include = c("date", "cases", "deaths")
-  )
-  # check for any NAs among data
-  checkmate::assert_data_frame(
-    data[, c("date", "cases", "deaths")],
-    types = c("Date", "integerish"),
-    any.missing = FALSE
-  )
-  # check that data$date is a date column
-  checkmate::assert_date(data$date, any.missing = FALSE, all.missing = FALSE)
-  # check for excessive missing date and throw an error
-  # also check delay_density
-  stopifnot(
-    "Input data must have sequential dates with none missing or duplicated" =
-      identical(unique(diff(data$date)), 1) # use numeric 1, not integer
-    # this solution works when df$date is `Date`
-    # this may need more thought for dates that are integers, POSIXct,
-    # or other units; consider the units package
-  )
+  .check_input_data(data)
   checkmate::assert_count(poisson_threshold, positive = TRUE)
 
   # NOTE: delay_density is checked in estimate_outcomes() if passed and not NULL
@@ -97,8 +72,6 @@ cfr_rolling <- function(data,
   cumulative_cases <- cumsum(data$cases)
   cumulative_deaths <- cumsum(data$deaths)
 
-  # Check cumulative sums for count type
-  checkmate::assert_integerish(cumulative_cases, lower = 0)
   # use assert_number to set upper limit at total_cases
   checkmate::assert_integerish(
     cumulative_deaths,

R/cfr_static.R

@@ -124,33 +124,7 @@
 cfr_static <- function(data,
                        delay_density = NULL,
                        poisson_threshold = 100) {
-  # input checking
-  checkmate::assert_data_frame(
-    data,
-    min.rows = 1, min.cols = 3
-  )
-  # check that input `<data.frame>` has columns date, cases, and deaths
-  checkmate::assert_names(
-    colnames(data),
-    must.include = c("date", "cases", "deaths")
-  )
-  # check for any NAs among data
-  checkmate::assert_data_frame(
-    data[, c("date", "cases", "deaths")],
-    types = c("Date", "integerish"),
-    any.missing = FALSE
-  )
-  # check that data$date is a date column
-  checkmate::assert_date(data$date, any.missing = FALSE, all.missing = FALSE)
-
-  # check for excessive missing date and throw an error
-  stopifnot(
-    "Input data must have sequential dates with none missing or duplicated" =
-      identical(unique(diff(data$date)), 1) # use numeric 1, not integer
-    # this solution works when df$date is `Date`
-    # this may need more thought for dates that are integers, POSIXct,
-    # or other units; consider the units package
-  )
+  .check_input_data(data)
   checkmate::assert_count(poisson_threshold, positive = TRUE)
 
   # NOTE: delay_density is checked in estimate_outcomes() if passed and not NULL
@@ -160,8 +134,6 @@ cfr_static <- function(data,
   total_cases <- sum(data$cases, na.rm = TRUE)
   total_deaths <- sum(data$deaths, na.rm = TRUE)
 
-  # Add input checking for total cases and deaths
-  checkmate::assert_count(total_cases)
   # use assert_number to set upper limit at total_cases
   checkmate::assert_number(total_deaths, upper = total_cases, lower = 0)
 

R/cfr_time_varying.R

@@ -100,26 +100,10 @@ cfr_time_varying <- function(data,
   # input checking
   # zero count allowed to include all data
   checkmate::assert_count(burn_in)
-
-  # expect rows more than burn in value
-  checkmate::assert_data_frame(data, min.cols = 3, min.rows = burn_in + 1)
-  # check that input `<data.frame>` has columns date, cases, and deaths
-  checkmate::assert_names(
-    colnames(data),
-    must.include = c("date", "cases", "deaths")
-  )
-  # check for any NAs among data
-  checkmate::assert_data_frame(
-    data[, c("date", "cases", "deaths")],
-    any.missing = FALSE
-  )
-  # check that data$date is a date column
-  checkmate::assert_date(data$date, any.missing = FALSE, all.missing = FALSE)
   checkmate::assert_count(smoothing_window, null.ok = TRUE)
+  .check_input_data(data)
 
   stopifnot(
-    "Input data must have sequential dates with none missing or duplicated" =
-      identical(unique(diff(data$date)), 1), # use numeric 1, not integer
     "`smoothing_window` must be an odd number greater than 0" =
       (smoothing_window %% 2 != 0),
     "`delay_density` must be a function with a single required argument,

R/check_input_data.R

@@ -0,0 +1,43 @@
+.check_input_data <- function(data) {
+  coll <- checkmate::makeAssertCollection()
+
+  checkmate::assert_data_frame(
+    data,
+    min.rows = 1, min.cols = 3,
+    add = coll
+  )
+  # check that input `<data.frame>` has columns date, cases, and deaths
+  checkmate::assert_names(
+    colnames(data),
+    must.include = c("date", "cases", "deaths"),
+    add = coll
+  )
+  # check for any NAs among data
+  checkmate::assert_data_frame(
+    data[, c("date", "cases", "deaths")],
+    types = c("Date", "integerish"),
+    any.missing = FALSE,
+    add = coll
+  )
+  # check that data$date is a date column
+  checkmate::assert_date(
+    data$date,
+    any.missing = FALSE,
+    all.missing = FALSE,
+    add = coll
+  )
+
+  # Check count types
+  checkmate::assert_integerish(data$cases, lower = 0, add = coll)
+  checkmate::assert_integerish(data$deaths, lower = 0, add = coll)
+
+  stopifnot(
+    "Input data must have sequential dates with none missing or duplicated" =
+      identical(unique(diff(data$date)), 1) # use numeric 1, not integer
+    # this solution works when df$date is `Date`
+    # this may need more thought for dates that are integers, POSIXct,
+    # or other units; consider the units package
+  )
+
+  checkmate::reportAssertions(coll)
+}