Require a "T" before hour-only times; require calendar months and day…

…s not to include "00"

R/iso8601_regexp.R

@@ -66,15 +66,15 @@ pattern_ISO8601_calendar_datetime <- function(truncated=0, ...) {
   stopifnot(truncated >= 0 & truncated <= 6)
   # Min truncated is 2 because 3 will be handled below with the addition of the
   # "T"
-  pattern_time <- pattern_ISO8601_time(truncated=min(truncated, 2), ...)
+  pattern_time <- pattern_ISO8601_time(truncated=min(truncated, 3), require_T=TRUE, ...)
   pattern_time_aug <-
     pattern_ISO8601_truncated_helper(
-      paste0("T", pattern_time),
+      pattern_time,
       allow_truncation=truncated >= 3
     )
   pattern_ISO8601_calendar_date(
     truncated=max(truncated - 3, 0),
-    time_pattern=pattern_time_aug,
+    time_pattern=pattern_time,
     ...
   )
 }
@@ -114,9 +114,9 @@ pattern_ISO8601_calendar_date <- function(truncated=0, allow_before_year_1583=FA
     } else {
       "(158[3-9]|159[0-9]|1[6-9][0-9]{2}|[2-9][0-9]{3})"
     }
-  pattern_month <- "-(0[0-9]|1[0-2])"
+  pattern_month <- "-(0[1-9]|1[0-2])"
   # Not confirming that the date is valid for the month
-  pattern_day_number <- "[012][0-9]|3[01]"
+  pattern_day_number <- "0[1-9]|[12][0-9]|3[01]"
   pattern_day <-
     if (nchar(time_pattern) == 0) {
       sprintf("-(%s)", pattern_day_number)
@@ -159,10 +159,12 @@ pattern_ISO8601_calendar_date <- function(truncated=0, allow_before_year_1583=FA
 #' @param pattern_decimal_mark What regular expression patterns should be
 #'   allowed for the decimal mark?  Comma and period are both allowed in the
 #'   standard and by default.
-#' @param allow_leap_second Should leap seconds (61 seconds in a minute) be
-#'   allowed?
 #' @param timezone Should the timezone be optional (NA, default), required
 #'   (TRUE), or omitted (FALSE)?
+#' @param allow_leap_second Should leap seconds (61 seconds in a minute) be
+#'   allowed?
+#' @param require_T Require the "T" at the beginning of the time?  TRUE, FALSE,
+#'   or NA (optional)
 #' @param ... Passed to \code{pattern_ISO8601_timezone()}
 #' @family ISO8601 patterns
 #' @export
@@ -171,7 +173,8 @@ pattern_ISO8601_time <- function(truncated=0,
                                  allow_fractional_minutes=FALSE,
                                  allow_fractional_seconds=TRUE,
                                  fractional_digits=Inf, pattern_decimal_mark=c("\\.", ","),
-                                 timezone=NA, allow_leap_second=FALSE, ...) {
+                                 timezone=NA, allow_leap_second=FALSE, require_T=NA,
+                                 ...) {
   stopifnot(is.numeric(truncated))
   stopifnot(length(truncated) == 1)
   stopifnot(!is.na(truncated))
@@ -202,8 +205,19 @@ pattern_ISO8601_time <- function(truncated=0,
   stopifnot(is.logical(allow_leap_second))
   stopifnot(length(allow_leap_second) == 1)
   stopifnot(!is.na(allow_leap_second))
+
+  stopifnot(is.logical(require_T))
+  stopifnot(length(require_T) == 1)
 
-  pattern_hour <- "([01][0-9]|2[0-3])"
+  pattern_hour_number <- "([01][0-9]|2[0-3])"
+  pattern_hour_t <- paste0("T", pattern_hour_number)
+  if (is.na(require_T)) {
+    pattern_hour <- paste0("T?", pattern_hour_number)
+  } else if (require_T) {
+    pattern_hour <- pattern_hour_t
+  } else {
+    pattern_hour <- pattern_hour_number
+  }
   pattern_minute <- ":([0-5][0-9])"
   partial_pattern_second_whole <-
     if (allow_leap_second) {
@@ -236,12 +250,20 @@ pattern_ISO8601_time <- function(truncated=0,
 
   # Put it together
   pattern_second_aug <- pattern_ISO8601_truncated_helper(pattern_second, allow_truncation=truncated >= 1)
-  pattern_minute_aug <-
-    pattern_ISO8601_truncated_helper(
-      paste0(pattern_minute, pattern_second_aug),
-      allow_truncation=truncated >= 2
-    )
-  pattern_hour_aug <-
+  # To support the requirement of "T", pattern_minute_aug is not truncated by
+  # itself.
+  pattern_minute_aug <- paste0(pattern_minute, pattern_second_aug)
+  # T is required at the beginning for hour-only times
+  if (truncated >= 2) {
+    pattern_hour_aug <-
+      sprintf(
+        "(?:%s|%s)",
+        pattern_hour_t,
+        paste0(pattern_hour, pattern_minute_aug)
+      )
+  } else {
+    pattern_hour_aug <- paste0(pattern_hour, pattern_minute_aug)
+  }
   if (!isFALSE(timezone)) {
     pattern_timezone <-
       pattern_ISO8601_truncated_helper(
@@ -250,7 +272,7 @@ pattern_ISO8601_time <- function(truncated=0,
       )
     ret <-
       pattern_ISO8601_truncated_helper(
-        paste0(pattern_hour, pattern_minute_aug, pattern_timezone),
+        paste0(pattern_hour_aug, pattern_timezone),
         allow_truncation=truncated >= 3
       )
   } else {