Implementation of list() that uses NSE to name components lazily

[jennybc]

Opening at @hadley's suggestion on twitter. Relevant to a triplicated question on stackoverflow. Overlaps with the store() proposal by @lionel- over in purrr.

If list components could get named at creation time, that would be handy.

thing_one <- head(iris, 3)
thing_two <- head(iris, 3)

## our current options
thing_list <- list(thing_one = thing_one, thing_two = thing_two)
thing_list <- mget(c("thing_one", "thing_two"))
str(thing_list, max.level = 1)

## List of 2
##  $ thing_one:'data.frame':   3 obs. of  5 variables:
##  $ thing_two:'data.frame':   3 obs. of  5 variables:

## names can by so handy in downstream operations, e.g.
plyr::ldply(thing_list, .id = "came_from")

##   came_from Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## 1 thing_one          5.1         3.5          1.4         0.2  setosa
## 2 thing_one          4.9         3.0          1.4         0.2  setosa
## 3 thing_one          4.7         3.2          1.3         0.2  setosa
## 4 thing_two          5.1         3.5          1.4         0.2  setosa
## 5 thing_two          4.9         3.0          1.4         0.2  setosa
## 6 thing_two          4.7         3.2          1.3         0.2  setosa

## wishful thinking
#thing_list <- list(thing_one, thing_two)
#thing_list <- magical_list_fxn("^thing_")

[lionel-]

Indeed, dplyr may well be a better fit than purrr for store() and a list method for mutate_().

[jimhester]

I think this does the trick?

auto_list <- function(...) {
  lapply(dplyr:::named_dots(...), eval)
}

a <- 5
auto_list(a, 1:10, c = 'hi')

# $a
# [1] 5
# 
# $`1:10`
#  [1]  1  2  3  4  5  6  7  8  9 10
# 
# $c
# [1] "hi"

auto_list(thing_one, thing_two)
# $thing_one
#   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
# 1          5.1         3.5          1.4         0.2  setosa
# 2          4.9         3.0          1.4         0.2  setosa
# 3          4.7         3.2          1.3         0.2  setosa
# 
# $thing_two
#   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
# 1          5.1         3.5          1.4         0.2  setosa
# 2          4.9         3.0          1.4         0.2  setosa
# 3          4.7         3.2          1.3         0.2  setosa

[hadley]

I think this should be as similar as possible to data_frame(), and hopefully we can make it so that

data_frame <- function(...) {
  as_data_frame(auto_list(...))
}

works. This would improve the design of data_frame and as_data_frame my moving the currently duplicated checks into one place.

But what's a good name? lst()?

[jimhester]

This produces the same functionality as my previous implementation and uses lazyeval style NSE.

lst <- function(...) {
  lst_(lazyeval::lazy_dots(...))
}

lst_ <- function(columns) {
  lazyeval::lazy_eval(lazyeval::auto_name(columns))
}

thing_one <- head(iris, 3)
thing_two <- head(iris, 3)

lst(thing_one, thing_two)
#> $thing_one
#>   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
#> 1          5.1         3.5          1.4         0.2  setosa
#> 2          4.9         3.0          1.4         0.2  setosa
#> 3          4.7         3.2          1.3         0.2  setosa
#> 
#> $thing_two
#>   Sepal.Length Sepal.Width Petal.Length Petal.Width Species
#> 1          5.1         3.5          1.4         0.2  setosa
#> 2          4.9         3.0          1.4         0.2  setosa
#> 3          4.7         3.2          1.3         0.2  setosa

Also seems to work fine as a replacement for data_frame() (with limited testing).

data_frame2 <- function(...) {
  as_data_frame(lst(...))
}

data_frame2(a=1:10, b=letters[1:10])
#> Source: local data frame [10 x 2]
#> 
#>     a b
#> 1   1 a
#> 2   2 b
#> 3   3 c
#> 4   4 d
#> 5   5 e
#> 6   6 f
#> 7   7 g
#> 8   8 h
#> 9   9 i
#> 10 10 j

data_frame2(a=1:10, b="a")
#> Error: Columns are not all same length

[hadley]

In midst of refactoring data_frame_ into two pieces.