Convertit une chaîne en date, le format “jj.mm.aaaa”

Pourrait être utile pour quelqu'un. J'ai trouvé cette fonction dans le tutoriel "Manipulation de la date et l'heure dans la R" par Cole Beck. La fonction identifie le format de vos données.

# FUNCTION guessDateFormat @x vector of character dates/datetimes @returnDates return
# actual dates rather than format convert character datetime to POSIXlt datetime, or 
# at least guess the format such that you could convert to datetime
guessDateFormat <- function(x, returnDates = FALSE, tzone = "") {
x1 <- x
# replace blanks with NA and remove
x1[x1 == ""] <- NA
x1 <- x1[!is.na(x1)]
if (length(x1) == 0)
return(NA)
# if it's already a time variable, set it to character
if ("POSIXt" %in% class(x1[1])) {
x1 <- as.character(x1)
}
dateTimes <- do.call(rbind, strsplit(x1, " "))
for (i in ncol(dateTimes)) {
dateTimes[dateTimes[, i] == "NA"] <- NA
}
# assume the time part can be found with a colon
timePart <- which(apply(dateTimes, MARGIN = 2, FUN = function(i) {
any(grepl(":", i))
}))
# everything not in the timePart should be in the datePart
datePart <- setdiff(seq(ncol(dateTimes)), timePart)
# should have 0 or 1 timeParts and exactly one dateParts
if (length(timePart) > 1 || length(datePart) != 1)
stop("cannot parse your time variable")
timeFormat <- NA
if (length(timePart)) {
# find maximum number of colons in the timePart column
ncolons <- max(nchar(gsub("[^:]", "", na.omit(dateTimes[, timePart]))))
if (ncolons == 1) {
timeFormat <- "%H:%M"
} else if (ncolons == 2) {
timeFormat <- "%H:%M:%S"
} else stop("timePart should have 1 or 2 colons")
}
# remove all non-numeric values
dates <- gsub("[^0-9]", "", na.omit(dateTimes[, datePart]))
# sep is any non-numeric value found, hopefully /or -
sep <- unique(na.omit(substr(gsub("[0-9]", "", dateTimes[, datePart]), 1, 1)))
if (length(sep) > 1)
stop("too many seperators in datePart")
# maximum number of characters found in the date part
dlen <- max(nchar(dates))
dateFormat <- NA
# when six, expect the century to be omitted
if (dlen == 6) {
if (sum(is.na(as.Date(dates, format = "%y%m%d"))) == 0) {
dateFormat <- paste("%y", "%m", "%d", sep = sep)
} else if (sum(is.na(as.Date(dates, format = "%m%d%y"))) == 0) {
dateFormat <- paste("%m", "%d", "%y", sep = sep)
} else stop("datePart format [six characters] is inconsistent")
}else if (dlen == 8) {
if (sum(is.na(as.Date(dates, format = "%Y%m%d"))) == 0) {
dateFormat <- paste("%Y", "%m", "%d", sep = sep)
} else if (sum(is.na(as.Date(dates, format = "%m%d%Y"))) == 0) {
dateFormat <- paste("%m", "%d", "%Y", sep = sep)
} else stop("datePart format [eight characters] is inconsistent")
} else {
stop(sprintf("datePart has unusual length: %s", dlen))
}
if (is.na(timeFormat)) {
format <- dateFormat
} else if (timePart == 1) {
format <- paste(timeFormat, dateFormat)
} else if (timePart == 2) {
format <- paste(dateFormat, timeFormat)
} else stop("cannot parse your time variable")
if (returnDates)
return(as.POSIXlt(x, format = format, tz = tzone))
format
}
# generate some dates
mydates <- format(as.POSIXct(sample(31536000, 20), origin = "2011-01-01", tz = "UTC"), "%m.%d.%Y %H:%M")
mydates
## [1] "02/07/2011 06:51" "11/21/2011 17:03" "09/17/2011 22:42" "02/16/2011 13:45"
## [5] "12/14/2011 19:11" "09/08/2011 09:22" "12/06/2011 14:06" "02/02/2011 11:00"
## [9] "03/27/2011 06:12" "01/05/2011 15:09" "04/15/2011 04:17" "10/20/2011 14:20"
## [13] "11/13/2011 21:46" "02/26/2011 03:24" "12/29/2011 11:02" "03/17/2011 02:24"
## [17] "02/27/2011 13:51" "06/27/2011 08:36" "03/14/2011 10:54" "01/28/2011 14:14"
guessDateFormat(mydates)

[1] "%m.%d.%Y %H:%M"

OriginalL'auteur Tymo