guess_datetime: Guess Timestamp Format

In enricoschumann/datetimeutils: Utilities for Dates and Times

Guess Timestamp Format

Description

Tries to convert a character vector to POSIXct.

Usage

guess_datetime(s, date.only = FALSE, within = FALSE, tz = "",
               try.patterns = NULL)

Arguments

s	character
date.only	logical: try to guess dates only (if TRUE) or times as well (if FALSE)
within	logical: ignore surrounding text? Note that trailing text is always ignored, see as.Date.
tz	character: timezone to assume for times. Default is the current timezone. See argument tz in as.POSIXct
try.patterns	either NULL or a character vector. See Details and Examples.

Details

The function first coerces its argument to character. It then applies a list of patterns to each element of s. Let d be a numeric digit; then the rules are roughly those in the table below. (For the precise rules, see Examples below.)

original pattern	assumed format
dddd-dd-dd dd:dd:dd	%Y-%m-%d %H:%M:%S
dd/dd/dddd dd:dd:dd	%m/%d/%Y %H:%M:%S
dd.dd.dddd dd:dd:dd	%d.%m.%Y %H:%M:%S

The rules are followed in the given order; an element will be matched only once. If there is a match, strptime will be tried with the assumed format (when date.only is TRUE, as.Date will be tried). For elements that do not match any pattern or for which strptime fails, NA is returned.

Additional patterns can be specified as try.patterns. This must be a character vector with an even number of elements: the first of each pair of elements is used as the pattern in a regular expression; the second as the format string passed to strptime. See Examples.

Value

POSIXct

Warning

If you know the format of a timestamp, then do not use this function (use strptime instead). If you have no idea at all about the format of a timestamp, then do not use this function.

Author(s)

Enrico Schumann

See Also

strptime

Examples

s <- c("  1999-08-19     10:00:31   ",
       "   1999-08-19 10:00",
       "19.8.1999 10:00",
       "8/19/99      10:00:31",
       "8/19/1999 10:00:31",
       "19.8.1999 10:00:31")

guess_datetime(s)

## the actual rules
rules <- as.data.frame(matrix(datetimeutils:::.dt_patterns,
                              byrow = TRUE, ncol = 2),
                       stringsAsFactors = FALSE)
names(rules) <- c("pattern", "assumed_format")
rules



## ----------------------------------

## a function for finding old files by looking at the
## dates in filenames (e.g. in a backup directory)
old_files <- function(min.age = 365,   ## in days
                      path = ".",
                      recursive = FALSE,
                      full.names = FALSE) {

    files <- dir(path, recursive = recursive, full.names = full.names)
    dates <- guess_datetime(files, date.only = TRUE, within = TRUE)

    age <- as.numeric(Sys.Date() - dates)
    old <- age >= min.age

    files[ !is.na(old) & old ]
}


## ----------------------------------

## specifying additional formats

s <- c("19-08-99",
       "29-2-00")
guess_datetime(s, date.only = TRUE)
## NA NA
guess_datetime(s, date.only = TRUE,
               try.patterns = c("[0-9]+-[0-9]+-[0-9]+", "%d-%m-%y"))
## "1999-08-19" "2000-02-29"

enricoschumann/datetimeutils documentation built on June 12, 2025, 1:06 p.m.