constructive

{constructive} prints code that can be used to recreate R objects. In a sense it is similar to base::dput() or base::deparse() but {constructive} strives to use idiomatic constructors (factor for factors, as.Date() for dates, data.frame() for data frames etc), in order to get output readable by humans.

Some use cases are:

Installation

Install the last stable version from CRAN:

install.packages('constructive')

Or install the development version from cynkra R-universe:

install.packages('constructive', repos = c('https://cynkra.r-universe.dev', 'https://cloud.r-project.org'))

Or directly from github:

pak::pak("cynkra/constructive")

Comparison with dput()

A few examples compared to their dput() output.

library(constructive)

construct(head(iris, 2))
#> data.frame(
#>   Sepal.Length = c(5.1, 4.9),
#>   Sepal.Width = c(3.5, 3),
#>   Petal.Length = c(1.4, 1.4),
#>   Petal.Width = c(0.2, 0.2),
#>   Species = factor(c("setosa", "setosa"), levels = c("setosa", "versicolor", "virginica"))
#> )

dput(head(iris, 2))
#> structure(list(Sepal.Length = c(5.1, 4.9), Sepal.Width = c(3.5, 
#> 3), Petal.Length = c(1.4, 1.4), Petal.Width = c(0.2, 0.2), Species = structure(c(1L, 
#> 1L), levels = c("setosa", "versicolor", "virginica"), class = "factor")), row.names = 1:2, class = "data.frame")

construct(.leap.seconds)
#> as.POSIXct(
#>   c(
#>     "1972-07-01", "1973-01-01", "1974-01-01", "1975-01-01", "1976-01-01",
#>     "1977-01-01", "1978-01-01", "1979-01-01", "1980-01-01", "1981-07-01",
#>     "1982-07-01", "1983-07-01", "1985-07-01", "1988-01-01", "1990-01-01",
#>     "1991-01-01", "1992-07-01", "1993-07-01", "1994-07-01", "1996-01-01",
#>     "1997-07-01", "1999-01-01", "2006-01-01", "2009-01-01", "2012-07-01",
#>     "2015-07-01", "2017-01-01"
#>   ),
#>   tz = "GMT"
#> )

dput(.leap.seconds)
#> structure(c(78796800, 94694400, 126230400, 157766400, 189302400, 
#> 220924800, 252460800, 283996800, 315532800, 362793600, 394329600, 
#> 425865600, 489024000, 567993600, 631152000, 662688000, 709948800, 
#> 741484800, 773020800, 820454400, 867715200, 915148800, 1136073600, 
#> 1230768000, 1341100800, 1435708800, 1483228800), class = c("POSIXct", 
#> "POSIXt"), tzone = "GMT")

library(dplyr, warn.conflicts = FALSE)
grouped_band_members <- group_by(band_members, band)

dput(grouped_band_members)
#> structure(list(name = c("Mick", "John", "Paul"), band = c("Stones", 
#> "Beatles", "Beatles")), class = c("grouped_df", "tbl_df", "tbl", 
#> "data.frame"), row.names = c(NA, -3L), groups = structure(list(
#>     band = c("Beatles", "Stones"), .rows = structure(list(2:3, 
#>         1L), ptype = integer(0), class = c("vctrs_list_of", "vctrs_vctr", 
#>     "list"))), class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA, 
#> -2L), .drop = TRUE))

construct(grouped_band_members)
#> tibble::tibble(name = c("Mick", "John", "Paul"), band = c("Stones", "Beatles", "Beatles")) |>
#>   dplyr::group_by(band)

We can provide to the data argument a named list, an environment, a package where to look for data, or an unnamed list of such items, so we don’t print more than necessary, for instance improving the previous example:

construct(grouped_band_members, data = "dplyr")
#> band_members |>
#>   dplyr::group_by(band)

Customize the output using constructive options

Some objects can be constructed in several ways, for instance a tibble might be constructed using tibble::tibble() or using tibble::tribble().

The opts_*() family of functions provides ways to tweak the output code, namely setting the constructor itself or options used by the constructor

construct(band_members, opts_tbl_df("tribble"))
#> tibble::tribble(
#>   ~name,  ~band,
#>   "Mick", "Stones",
#>   "John", "Beatles",
#>   "Paul", "Beatles",
#> )
construct(band_members, opts_tbl_df("tribble", justify = "right"))
#> tibble::tribble(
#>    ~name,     ~band,
#>   "Mick",  "Stones",
#>   "John", "Beatles",
#>   "Paul", "Beatles",
#> )

r <- as.raw(c(0x68, 0x65, 0x6c, 0x6c, 0x6f))
construct(r)
#> as.raw(c(0x68, 0x65, 0x6c, 0x6c, 0x6f))
construct(r, opts_raw(representation = "decimal"))
#> as.raw(c(104, 101, 108, 108, 111))
construct(r, opts_raw("charToRaw"))
#> charToRaw("hello")

These functions have their own documentation page and are referenced in ?construct.

For every class that doesn’t refer to an internal type a “next” constructor is available, so we can conveniently explore objects using lower level constructors.

construct(band_members, opts_tbl_df("next"))
#> data.frame(name = c("Mick", "John", "Paul"), band = c("Stones", "Beatles", "Beatles")) |>
#>   structure(class = c("tbl_df", "tbl", "data.frame"))

construct(band_members, opts_tbl_df("next"), opts_data.frame("next"))
#> list(name = c("Mick", "John", "Paul"), band = c("Stones", "Beatles", "Beatles")) |>
#>   structure(class = c("tbl_df", "tbl", "data.frame"), row.names = c(NA, -3L))

Other functions

Note about environments and external pointers

Environments use reference semantics, they cannot be copied. An attempt to copy an environment would indeed yield a different environment and identical(env, copy) would be FALSE (read more about it in ?opts_environment).

In some case we can build code that points to a specific environment, for instance:

construct(globalenv())
#> .GlobalEnv
construct(environment(setNames))
#> asNamespace("stats")

When it’s not possible we use constructive::.env() function for this purpose.

e1 <- new.env(parent = .GlobalEnv)
e1$x <- 1
construct(e1)
#> constructive::.env("0x131515348", parents = "global")

constructive::.env() fetches the environment from its memory address. The parents argument doesn’t do anything, it provides as additional information the sequence of parents until we reach a special environment.

This strategy is convenient because it always works, but it’s not reproducible between sessions as the memory address is not stable. Moreover it doesn’t tell us anything about the environment’s content.

Depending on what compromise you’re ready to make, you might use different constructions in opts_environment(). For the case above, choosing "list2env" works well :

construct(e1, opts_environment("list2env"))
#> list2env(list(x = 1), parent = .GlobalEnv)

constructive::.xptr() is the counterpart of constructive::.env() to construct "externalptr" objects from a memory address.

Extending constructive

You can define your own constructors and methods!

For more information see vignette("User-defined-methods-and-constructors", package = "constructive")