Add internal function to check and convert Excel times. Closes #10. CC …

…#30.
mhpob · Jun 21, 2024 · f13f360 · f13f360
1 parent 66560cf
commit f13f360
Show file tree

Hide file tree

Showing 4 changed files with 91 additions and 32 deletions.
diff --git a/R/utilities_make.R b/R/utilities_make.R
@@ -5,7 +5,7 @@
 #' @param files Character. File paths of files to be unzipped or written to a directory
 #' @param temp_dir Character. File path of temporary directory
 #' @param detection_file Character. File path of detections.
-#'
+#' @param date_time Character or numeric. Date-time to convert.
 #'
 #' @name utilities-make
 #' @keywords internal
@@ -15,20 +15,20 @@ clean_otn_deployment <- function(deployment) {
     # Find which sheet has deployment data. If none are explicitly labeled, assume
     #   it's sheet 1
     sheet_id <- grep("dep", readxl::excel_sheets(deployment),
-      ignore.case = T, value = T
+                     ignore.case = T, value = T
     )
     if (length(sheet_id) == 0) {
       sheet_id <- 1
     }
 
     # Check for header: If the first row has no columns, it likely contains it.
     if (ncol(readxl::read_excel(deployment,
-      sheet = sheet_id,
-      range = "A1"
+                                sheet = sheet_id,
+                                range = "A1"
     )) == 0) {
       deployment <- readxl::read_excel(deployment,
-        sheet = sheet_id,
-        skip = 3
+                                       sheet = sheet_id,
+                                       skip = 3
       )
     } else {
       deployment <- readxl::read_excel(deployment, sheet = sheet_id)
@@ -38,12 +38,12 @@ clean_otn_deployment <- function(deployment) {
     check_head <- read.csv(deployment, nrows = 1, check.names = FALSE)
     if (ncol(check_head) > length(unique(names(check_head)))) {
       deployment <- read.csv(deployment,
-        skip = 3,
-        na.strings = c("NA", "")
+                             skip = 3,
+                             na.strings = c("NA", "")
       )
     } else {
       deployment <- read.csv(deployment,
-        na.strings = c("NA", "")
+                             na.strings = c("NA", "")
       )
     }
   } else {
@@ -59,28 +59,14 @@ clean_otn_deployment <- function(deployment) {
   deployment <- deployment[!is.na(deployment$deploy_date_time), ]
   deployment <- deployment[!deployment$recovered %in% c("l", "failed", NA), ]
 
-  deployment$deploy_date_time <- as.POSIXct(
-    deployment$deploy_date_time,
-    tz = "UTC",
-    tryFormats = c(
-      "%Y-%m-%dT%H:%M:%S",
-      "%Y-%m-%d %H:%M:%S"
-    )
-  )
-  deployment$recover_date_time <- as.POSIXct(
-    deployment$recover_date_time,
-    tz = "UTC",
-    tryFormats = c(
-      "%Y-%m-%dT%H:%M:%S",
-      "%Y-%m-%d %H:%M:%S"
-    )
-  )
+  deployment$deploy_date_time <- convert_times(deployment$deploy_date_time)
+  deployment$recover_date_time <- convert_times(deployment$recover_date_time)
 
   deployment <- deployment[!is.na(deployment$deploy_date_time) &
-    !is.na(deployment$recover_date_time), ]
+                             !is.na(deployment$recover_date_time), ]
   deployment$receiver <- paste(deployment$ins_model_no,
-    deployment$ins_serial_no,
-    sep = "-"
+                               deployment$ins_serial_no,
+                               sep = "-"
   )
   deployment$stationname <- deployment$station_no
 
@@ -98,6 +84,34 @@ clean_otn_deployment <- function(deployment) {
 }
 
 
+#' @rdname utilities-make
+#' @keywords internal
+convert_times <- function(date_time){
+
+  check_times <- function(x){
+    # check if Excel format
+    #   Assumes that it starts with 5 numbers
+    if (grepl('^\\d{5}', x)) {
+      as.POSIXct(
+        as.numeric(x) * (60 * 60 * 24),
+        tz = "UTC",
+        origin = '1899-12-30'
+      )
+    } else {
+      as.POSIXct(
+        x,
+        tz = "UTC",
+        tryFormats = c(
+          "%Y-%m-%dT%H:%M:%S",
+          "%Y-%m-%d %H:%M:%S"
+        )
+      )
+    }
+  }
+
+  do.call("c", lapply(date_time, check_times))
+}
+
 #' @rdname utilities-make
 #' @keywords internal
 provided_file_unzip <- function(files, temp_dir) {
@@ -109,8 +123,8 @@ provided_file_unzip <- function(files, temp_dir) {
     to_unzip,
     function(.) {
       unzip(.,
-        exdir = temp_dir,
-        setTimes = FALSE
+            exdir = temp_dir,
+            setTimes = FALSE
       )
     }
   )
@@ -144,7 +158,7 @@ write_to_tempdir <- function(type, files, temp_dir) {
   ##  Write file to temporary directory
   filepath <- file.path(temp_dir, paste0(type, ".csv"))
   write.csv(files, filepath,
-    row.names = F
+            row.names = F
   )
 
 

diff --git a/man/utilities-make.Rd b/man/utilities-make.Rd
diff --git a/tests/testthat/test-clean_otn_deployment.R b/tests/testthat/test-clean_otn_deployment.R
@@ -190,7 +190,8 @@ test_that("Correct names when no internal transmitter columns", {
   expect_named(
     clean_otn_deployment(deployment_sheet1),
     c(
-      "stationname", "receiver", "deploy_date_time", "deploy_lat", "deploy_long",
+      "stationname", "receiver", "internal_transmitter",
+      "deploy_date_time", "deploy_lat", "deploy_long",
       "recover_date_time"
     )
   )

diff --git a/tests/testthat/test-convert_times.R b/tests/testthat/test-convert_times.R
@@ -0,0 +1,39 @@
+test_datetime <- as.POSIXct("2020-04-09T11:04:59", tz = "UTC",
+                            format = "%Y-%m-%dT%H:%M:%S")
+
+test_that("converts ISO 8601", {
+  expect_equal(
+    convert_times("2020-04-09T11:04:59"),
+    test_datetime
+  )
+})
+
+
+
+
+test_that("converts space-separated date-times", {
+  expect_equal(
+    convert_times("2020-04-09 11:04:59"),
+    test_datetime
+  )
+})
+
+
+
+
+test_that("converts Excel numeric", {
+  expect_equal(
+    convert_times(43930.461793981),
+    test_datetime
+  )
+})
+
+
+
+
+test_that("converts Excel character", {
+  expect_equal(
+    convert_times("43930.461793981"),
+    test_datetime
+  )
+})