R excels at computing with dates, and times. Using typed representation for your data is highly recommended not only because of the functionality offered but also because of the added safety stemming from proper representation.
But there is a small nuisance cost in interactive work as well as in programming. Users must have told as.POSIXct() about a million times that the origin is (of course) the epoch. Do we really have to say it a million more times? Similarly, when parsing dates that are some form of YYYYMMDD format, do we really have to manually convert from integer or numeric or factor or ordered to character? Having one of several common separators and/or date / time month forms (YYYY-MM-DD, YYYY/MM/DD, YYYYMMDD, YYYY-mon-DD and so on, with or without times), do we really need a format string? Or could a smart converter function do this?
anytime() aims to be that general purpose converter returning a proper POSIXct (or Date) object no matter the input (provided it was somewhat parseable), relying on Boost date_time for the (efficient, performant) conversion. anydate() is an additional wrapper returning a Date object instead.
We show some simple examples on Date types.
(Note that in the first few examples, and for numeric conversion in this range we now only use anydate as anytime is consistent in computing seconds since epoch. If you want the behaviour of version older than 0.3.0, set oldHeuristic=TRUE, see help(anytime) for more.)
library(anytime) ## also caches TZ in local env
options(digits.secs=6) ## for fractional seconds below
## integer
anydate(20160101L + 0:2) ## older version used anytime for this too
[1] "2016-01-01 CST" "2016-01-02 CST" "2016-01-03 CST"
## numeric
anydate(20160101 + 0:2)
[1] "2016-01-01 CST" "2016-01-02 CST" "2016-01-03 CST"
## factor
anydate(as.factor(20160101 + 0:2))
[1] "2016-01-01 CST" "2016-01-02 CST" "2016-01-03 CST"
## ordered
anydate(as.ordered(20160101 + 0:2))
[1] "2016-01-01 CST" "2016-01-02 CST" "2016-01-03 CST"## Dates: Character
anydate(as.character(20160101 + 0:2))
[1] "2016-01-01 CST" "2016-01-02 CST" "2016-01-03 CST"
## Dates: alternate formats
anydate(c("20160101", "2016/01/02", "2016-01-03"))
[1] "2016-01-01 CST" "2016-01-02 CST" "2016-01-03 CST"## Datetime: ISO with/without fractional seconds
anytime(c("2016-01-01 10:11:12", "2016-01-01 10:11:12.345678"))
[1] "2016-01-01 10:11:12.000000 CST" "2016-01-01 10:11:12.345678 CST"
## Datetime: ISO alternate (?) with 'T' separator
anytime(c("20160101T101112", "20160101T101112.345678"))
[1] "2016-01-01 10:11:12.000000 CST" "2016-01-01 10:11:12.345678 CST"## ISO style
anytime(c("2016-Sep-01 10:11:12", "Sep/01/2016 10:11:12", "Sep-01-2016 10:11:12"))
[1] "2016-09-01 10:11:12 CDT" "2016-09-01 10:11:12 CDT" "2016-09-01 10:11:12 CDT"
## Datetime: Mixed format (cf https://stackoverflow.com/questions/39259184)
anytime(c("Thu Sep 01 10:11:12 2016", "Thu Sep 01 10:11:12.345678 2016"))
[1] "2016-09-01 10:11:12.000000 CDT" "2016-09-01 10:11:12.345678 CDT"This shows an important aspect. When not working localtime (by overriding to UTC) the changing difference UTC is correctly covered (which the underlying Boost Date_Time library does not by itself).
## Datetime: pre/post DST
anytime(c("2016-01-31 12:13:14", "2016-08-31 12:13:14"))
[1] "2016-01-31 12:13:14 CST" "2016-08-31 12:13:14 CDT"
anytime(c("2016-01-31 12:13:14", "2016-08-31 12:13:14"), tz="UTC") # important: catches change
[1] "2016-01-31 18:13:14 UTC" "2016-08-31 17:13:14 UTC"The heavy lifting is done by a combination of Boost lexical_cast to go from anything to string representation which is then parsed by Boost Date_Time. We use the BH package to access Boost, and rely on Rcpp for a seamless C++ interface to and from R.
Further, as the Boost Date_Time library cannot resolve timezones on the Windows platform (where timezone information is typically provided by R itself for its use), we offer a fallback of calling into R (via facilities from Rcpp); see the help for the useR argument for more details.
The package should work as expected.
Currently, just over ten different CRAN packages import this package. Among them are the following research-focused packages: - AGread by Hibbing et al which reads and transforms ActiGraph physical activity measures; - cqcr by Odell accesses ‘Care Quality Commission’ data from the health and adult social care regulator for England; - E4tools by Kleiman which reads data from Empatica wearable physiology monitors; - RDS by Handcock et al which is part of the “RDS Ananlyst” suite for analysing respondent-driven sampling data; - tsbox by Sax for format-agnostic time series data representation and conversions; - tsibble by Wang et al for temporal data in an explicit data- and model-oriented format.
See the NEWS.Rd file on CRAN or GitHub. In particular, version 0.3.0 corrects an overly optimistic heuristic for integer or numeric arguments and now behaves more like R itself. Specifically, epoch offsets are interpreted as seconds for datetime objects, and days for date objects. The prior behaviour can be restored with an option which also be be set globally, see the help page for details.
The package is now on CRAN and can be installed via a standard
As we rely on the tinytest package, the already-installed package can also be verified via
at any later point.
Any problems, bug reports, or features requests for the package can be submitted and handled most conveniently as Github issues in the repository.
Before submitting pull requests, it is frequently preferable to first discuss need and scope in such an issue ticket. See the file Contributing.md (in the Rcpp repo) for a brief discussion.
Dirk Eddelbuettel
GPL (>= 2)