Allow yyyy-mm-dd formatted character input

.travis.yml

@@ -23,4 +23,4 @@ notifications:
     on_failure: change
 
 after_success:
-- if [[ "${{R_CODECOV}}" ]]; then Rscript -e 'covr::codecov()'; fi
+- if [[ "${R_CODECOV}" ]]; then Rscript -e 'covr::codecov()'; fi

NAMESPACE

@@ -27,6 +27,7 @@ S3method(group_names,default)
 S3method(group_names,incidence)
 S3method(incidence,Date)
 S3method(incidence,POSIXt)
+S3method(incidence,character)
 S3method(incidence,default)
 S3method(incidence,integer)
 S3method(incidence,numeric)

NEWS.md

@@ -3,26 +3,24 @@ incidence 1.5.4 unreleased
 
 ### BUG FIX
 
-* `incidence()` now returns an error when dates argument is character object.
+* `incidence()` now returns an error when supplied a character vector that is
+  not formatted as (yyyy-mm-dd).
   (See https://github.com/reconhub/incidence/issues/88)
 * `fit()` now returns correct coefficients when dates is POSIXt by converting to
   Date. (See https://github.com/reconhub/incidence/issues/91)
 * `plot.incidence()` now plots in UTC by default for POSIXt incidence objects.
   this prevents a bug where different time zones would cause a shift in the bars
   (See https://github.com/reconhub/incidence/issues/99).
 
-### NOTABLE CHANGES 
-
-* `incidence()` no longer accepts characters as input for dates, first_date, or
-  last_date arguments.
-
 ### MISC
 
 * A test that randomly failed on CRAN has been fixed.
   (See https://github.com/reconhub/incidence/issues/95).
 * Plotting tests have been updated for new version of vdiffr
   (See https://github.com/reconhub/incidence/issues/96).
 * POSIXct incidence are first passed through POSIXlt when initialized.
+* A more informative error message is generated for non ISO 8601 formatted 
+  `first_date` and `last_date` parameters.
 
 incidence 1.5.3 (2018-12-07)
 ============================

R/check_boundaries.R

@@ -14,15 +14,16 @@ check_boundaries <- function(dates, boundary = NULL, what = "first") {
     MINMAX <- if (what == "first") min else max
     boundary <- MINMAX(dates, na.rm = TRUE)
   }
-  if (is.character(boundary)) {
-    msg <- '%s_date is a character. Did you forget to convert to Date?'
-    stop(sprintf(msg, what), call. = FALSE)
+  msg <- "%s_date (%s) could not be converted to Date."
+  if (is.character(boundary) && !grepl("^[0-9]{4}-[01][0-9]-[0-3][0-9]$", boundary)) {
+    msg <- paste(msg, 'Dates must be in ISO 8601 standard format (yyyy-mm-dd).')
+    stop(sprintf(msg, what, boundary), call. = FALSE)
   }
   res <- try(check_dates(boundary), silent = TRUE)
   if (inherits(res, "try-error")) {
-    msg <- paste("%s_date could not be converted to Date. Accepted formats are:",
-                 "\n  Date, POSIXct, integer, numeric.")
-    stop(sprintf(msg, what), call. = FALSE)
+    msg <- paste(msg, "Accepted formats are:",
+                 "\n  Date, POSIXct, integer, numeric, character.")
+    stop(sprintf(msg, what, deparse(substitute(boundary))), call. = FALSE)
   }
   res
 }

R/check_dates.R

@@ -14,6 +14,10 @@ check_dates <- function(x, error_on_NA = FALSE, ...) {
     stop("dates is NULL", call. = FALSE)
   }
 
+  if (is.character(x)) {
+    x <- as.Date(x, ...)
+  }
+
   not_finite <- !is.finite(x)
   if (sum(not_finite) > 0) {
     x[not_finite] <- NA
@@ -55,7 +59,7 @@ check_dates <- function(x, error_on_NA = FALSE, ...) {
   }
 
 
-  formats <- c("Date", "POSIXct", "integer", "numeric")
+  formats <- c("Date", "POSIXct", "integer", "numeric", "character")
   msg <- paste0(
     "Input could not be converted to date. Accepted formats are:\n",
     paste(formats, collapse = ", "))

R/incidence.R

@@ -8,12 +8,13 @@
 #' and 2, the second interval includes 3, 4 and 5 etc.
 #'
 #' @param dates A vector of dates, which can be provided as objects of the
-#' class: integer, numeric, Date, POSIXct. Note that decimal numbers will be
-#' floored with a warning.
+#' class: integer, numeric, Date, POSIXct, POSIXlt, and character. (See Note 
+#' about `numeric` and `character` formats) 
 #'
-#' @param interval An integer or character indicating the (fixed) size of the time interval
-#' used for computing the incidence; defaults to 1 day. This can also be a text string that corresponds to a valid date
-#' interval: day, week, month, quarter, or year. See Note.
+#' @param interval An integer or character indicating the (fixed) size of the
+#' time interval used for computing the incidence; defaults to 1 day. This can
+#' also be a text string that corresponds to a valid date interval: day, week,
+#' month, quarter, or year. (See Note)
 #'
 #' @param groups An optional factor defining groups of observations for which
 #' incidence should be computed separately.
@@ -54,7 +55,14 @@
 #' @details For details about the `incidence class`, see the dedicated
 #' vignette:\cr `vignette("incidence_class", package = "incidence")`
 #'
-#' @note If `interval` is a valid character (e.g. "week" or "month"), then
+#' @note \subsection{Input data (`dates`)}{
+#'  - **Decimal (numeric) dates**: will be truncated with a warning
+#'  - **Character dates** should be in the unambiguous `yyyy-mm-dd` (ISO 8601)
+#'   format. Any other format will trigger an error.
+#' }
+#' 
+#' \subsection{Interval specification (`interval`)}{
+#' If `interval` is a valid character (e.g. "week" or "month"), then
 #' the bin will start at the beginning of the interval just before the first
 #' observation by default. For example, if the first case was recorded on
 #' Wednesday, 2018-05-09:
@@ -74,6 +82,7 @@
 #' number of days they encompass and warnings will be generated when the first
 #' date falls outside of a calendar date that is easily represented across the
 #' interval.
+#' }
 #'
 #' @seealso
 #' The main other functions of the package include:
@@ -155,12 +164,7 @@ incidence <- function(dates, interval = 1L, ...) {
 #' @export
 #' @rdname incidence
 incidence.default <- function(dates, interval = 1L, ...) {
-  if (is.character(dates)) {
-    stop('Input is a character. Did you forget to convert to Date?')
-  }
-  check_dates(dates)
-  msg <- "Unknown date input; accepted formats are Date, POSIXct, integer, numeric."
-  stop(msg)
+  incidence(check_dates(dates), interval = interval, ...)
 }
 
 #' @export
@@ -200,6 +204,32 @@ incidence.Date <- function(dates, interval = 1L, standard = TRUE, groups = NULL,
 }
 
 
+#' @export
+#' @rdname incidence
+incidence.character <- function(dates, interval = 1L, standard = TRUE, groups = NULL,
+                           na_as_group = TRUE, first_date = NULL,
+                           last_date = NULL, ...) {
+  iso_std <- grepl("^[0-9]{4}-[01][0-9]-[0-3][0-9]$", trimws(dates))
+  iso_std[is.na(dates)] <- TRUE # prevent false alarms
+  if (!all(iso_std)) {
+    msg <- paste("Not all dates are in ISO 8601 standard format (yyyy-mm-dd).",
+                 "The first incorrect date is %s"
+    )
+    stop(sprintf(msg, dates[!iso_std][1]))
+  }
+  dots  <- check_dots(list(...), names(formals(incidence.Date)))
+  dates <- check_dates(dates)
+
+  ret <- incidence(as.Date(trimws(dates)),
+                   interval = interval,
+                   standard = standard,
+                   groups = groups,
+                   na_as_group = na_as_group,
+                   first_date = first_date,
+                   last_date = last_date,
+                   ...)
+  ret
+}
 ## The default incidence is designed for dates provided as integers, and a fixed
 ## time interval defaulting to 1. 'bins' are time intervals, identified by the
 ## left date, left-inclusive and right-exclusive, i.e. the time interval defined

tests/testthat/test-incidence.R

@@ -107,7 +107,7 @@ test_that("construction - Date input", {
   expect_message(x.i.trim  <- incidence(dat, first_date = 0),
                  "[0-9]+ observations outside of \\[0, [0-9]+\\] were removed."
   )
-  expect_message(x.d.trim  <- incidence(dat_dates, first_date = as.Date("2016-01-01")),
+  expect_message(x.d.trim  <- incidence(dat_dates, first_date = "2016-01-01"),
                  "[0-9]+ observations outside of \\[2016-01-01, [-0-9]{10}\\] were removed."
   )
   x.7       <- incidence(dat_dates, 7L, standard = FALSE)
@@ -201,6 +201,21 @@ test_that("construction - POSIXct input", {
   expect_is(x.pos$dates, "POSIXct")
 })
 
+test_that("construction - character input", {
+  dats <- Sys.Date() + sample(-100:100, 5)
+  datc <- as.character(dats)
+
+  i.date <- incidence(dats)
+  i.char <- incidence(datc)
+  i.chaw <- incidence(paste(datc, "   "))
+  expect_message(i.cham <- incidence(c(datc, NA, NA)), "2 missing observations were removed.")
+  expect_is(i.date, "incidence")
+  expect_identical(i.date, i.char)  
+  expect_identical(i.date, i.chaw)  
+  expect_identical(i.date, i.cham)  
+})
+
+
 test_that("corner cases", {
 
 
@@ -223,7 +238,7 @@ test_that("corner cases", {
                "The interval 'grind' is not valid. Please supply an integer.")
 
   expect_error(incidence(as.Date(Sys.Date()), last_date = "core"),
-               "last_date is a character. Did you forget to convert to Date?")
+               "last_date \\(core\\) could not be converted to Date. Dates must be in ISO 8601 standard format \\(yyyy-mm-dd\\)")
 
   expect_error(incidence(1, "week"),
                "The interval 'week' can only be used for Dates")
@@ -240,6 +255,26 @@ test_that("corner cases", {
   expect_warning(incidence(c(dat_dates, as.Date("1900-01-01"))),
                  "greater than 18262 days \\[1900-01-01 to"
   )
+
+  msg <- 'Not all dates are in ISO 8601 standard format \\(yyyy-mm-dd\\). The first incorrect date is'
+  expect_error(incidence('daldkadl'), paste(msg, "daldkadl"))
+
+  dats <- as.character(Sys.Date() + sample(-10:10, 5))
+  dats[3] <- "1Q84-04-15" 
+  expect_error(incidence(dats), paste(msg, "1Q84-04-15"))
+
+  dats[3] <- "2018-69-11"
+  expect_error(incidence(dats), paste(msg, "2018-69-11"))
+
+  dats[3] <- "01-01-11"
+  expect_error(incidence(dats), paste(msg, "01-01-11"))
+
+  dats[3] <- "01-Apr-11"
+  expect_error(incidence(dats), paste(msg, "01-Apr-11"))
+
+  msg <- paste0("Input could not be converted to date. Accepted formats are:\n",
+                "Date, POSIXct, integer, numeric, character")
+  expect_error(incidence(factor("2001-01-01")), msg)
 })
 
 test_that("incidence constructor can handle missing data", {
@@ -318,7 +353,7 @@ test_that("user-defined group levels are preserved", {
 test_that("Printing returns the object", {
 
 
-  x <- incidence(as.Date("2001-01-01"))
+  x <- incidence("2001-01-01")
   y <- incidence(1:2, groups = factor(1:2))
   z <- incidence(dat_dates, interval = 7)
   expect_equal_to_reference(capture.output(print(x)),
@@ -329,20 +364,3 @@ test_that("Printing returns the object", {
                             file = "rds/print3.rds")
 })
 
-test_that("incidence returns error if input not in accepted format", {
-
-  msg <- 'Input is a character. Did you forget to convert to Date?'
-  expect_error(incidence('daldkadl'), msg)
-  expect_error(incidence('2001-01-01'), msg)
-
-  msg <- paste0("Input could not be converted to date. Accepted formats are:\n",
-                "Date, POSIXct, integer, numeric")
-  expect_error(incidence(factor("2001-01-01")), msg)
-
-  msg <- 'first_date is a character. Did you forget to convert to Date?'
-  expect_error(incidence(as.Date('2016-02-29'), "year", first_date = '2016-02-29'), msg)
-
-  msg <- 'last_date is a character. Did you forget to convert to Date?'
-  expect_error(incidence(as.Date('2016-02-29'), "year", last_date = '2016-02-29'), msg)
-
-})