CODEWITHSUNDEEP
rtime
PaulHurleyuk
PaulHurleyuk

Reputation: 8159

Converting date times in POSIXct gives screwy result?

I'm doing some extracting of data from a database, and running the results through rehsape2. For some reason this is mangling the POSIXct datetime stamps into numeric. No Problem I think, you can just turn them back, except I'm an hour out.

Here's a minimal example

foo<-as.POSIXct("2011-04-04 14:18:58")
as.numeric(foo)     #gives 130192318
bar<-as.POSIXct(as.numeric(foo), 
                tz=Sys.timezone(),
                origin=as.POSIXct(
                  strptime("1970-01-01 00:00:00", "%Y-%m-%d %H:%M:%S", tz="UTC")))
as.numeric(bar)     #gives 130192318 identical !
foo     #Gives "2011-04-04 14:18:58 BST"
bar     #Gives "2011-04-04 13:18:58 UTC"

Obviously foo and bar are numerically identical, but R thinks foo needs to be displayed as BST and bar as UTC. How do I get both displayed as BST. This doesn't work either;

as.POSIXct(bar, tz="BST")   #still gives "2011-04-04 13:18:58 UTC"

Upvotes: 7

Views: 10819

Answers (2)

PaulHurleyuk
PaulHurleyuk

Reputation: 8159

Here's the kludge I've employed to get round this. By specifying the origin as the current time minus the numeric version of the current time, the timezones don't seem to get screwed up.

foo<-as.POSIXct("2011-04-04 14:18:58")
as.numeric(foo)     #gives 130192318
bar<-as.POSIXct(as.numeric(foo), tz="GB", origin=Sys.time()-as.numeric(Sys.time()))
as.numeric(bar)     #gives 130192318 identical !
foo     #Gives "2011-04-04 14:18:58 BST"
[1] "2011-04-04 14:18:58 BST"
bar     #Gives "2011-04-04 14:18:58 BST"
[1] "2011-04-04 14:18:58 BST

Upvotes: 2

Joshua Ulrich
Joshua Ulrich

Reputation: 176648

Here's what's going on. bar is created using as.POSIXct.numeric, which is defined as:

as.POSIXct.numeric
function (x, tz = "", origin, ...) 
{
    if (missing(origin)) 
        stop("'origin' must be supplied")
    as.POSIXct(origin, tz = tz, ...) + x
}
<environment: namespace:base>

You supply an origin that is a POSIXct object. That means the as.POSIXct call in as.POSIXct.numeric dispatches to as.POSIXct.default, which is defined as:

as.POSIXct.default
function (x, tz = "", ...) 
{
    if (inherits(x, "POSIXct")) 
        return(x)
    if (is.character(x) || is.factor(x)) 
        return(as.POSIXct(as.POSIXlt(x, tz, ...), tz, ...))
    if (is.logical(x) && all(is.na(x))) 
        return(.POSIXct(as.numeric(x)))
    stop(gettextf("do not know how to convert '%s' to class \"POSIXct\"", 
        deparse(substitute(x))))
}
<environment: namespace:base>

x is a POSIXct class object (the origin you supplied in your initial call), so it is simply returned and the tz= argument is ignored.

UPDATE:
Here's how you can convert foo back to POSIXct with the appropriate time zone.

(foo <- as.POSIXct("2011-04-04 14:18:58", tz="GB"))
# [1] "2011-04-04 14:18:58 BST"
.POSIXct(as.numeric(foo), tz="GB")
# [1] "2011-04-04 14:18:58 BST"

Upvotes: 13

Related Questions