automatically name unnamed pluckers in hoist

[mgirlich]

Often one wants to name the columns produced by hoist() the same as the element in the list. So to make hoist() less verbose and more comfortable for interactive use I propose automatically naming the column as follows for unnamed arguments:

• if the last part of the pluck specification is a character use that
• if the last part of the pluck specification is a number paste the number to the last character part of the spec, e.g. list("film", 1) would produce the column name film_1.

[hadley]

This is an interesting idea. What would you do if the pluck spec is a number? Or the number comes first?

[mgirlich]

maybe there should be a short and a long name option, e.g. something like

pluck spec	short name	long name
list("people", "name")	name	people_name
list("people", "name", 1)	name_1	people_name_1
list(1)	..1	..1
list(1, 2)	..1..2	..1..2
list(1, "people", "name")	..1_name	..1_people_name

as the numbers in a spec will probably be small one might also use words for the, i.e. list(1, "name") gives first_name

[hadley]

That's starting to feel a bit too magical too me — I think there'd have to be one principle applied consistently.

[mgirlich]

Looking at it again I agree that it looks quite magical.

One easy principle would be to simply use paste(pluck_spec, collapse = "_") for the long name version. Unfortunately, this generates non-syntactic names for leading numbers. But this could be resolved by adding a names_repair argument, so the user can choose how to handle this.

For short names one would use basically the same strategy but keep all trailing numbers, i.e.
list("people", "name", 1, 2) would give name_1_2.

Do you like that better?

[jennybc]

The case that seems like a no-brainer and a great place to start is when you pluck a string. That covers so much of the pain IMO. This is the case where I most feel like hoist() should be able to figure this out.

That being said, I feel like we discussed this once in slack @hadley, and you objected? There is some potential for muddying the waters re: our quoting conventions.

library(tidyverse)
#> Warning: package 'tibble' was built under R version 3.6.2

df <- tibble(
  character = c("Toothless", "Dory"),
  metadata = list(
    list(
      species = "dragon",
      color = "black",
      films = c(
        "How to Train Your Dragon",
        "How to Train Your Dragon 2",
        "How to Train Your Dragon: The Hidden World"
      )
    ),
    list(
      species = "blue tang",
      color = "blue",
      films = c("Finding Nemo", "Finding Dory")
    )
  )
)

df %>% 
  hoist(metadata, species = "species", color = "color")
#> # A tibble: 2 x 4
#>   character species   color metadata        
#>   <chr>     <chr>     <chr> <list>          
#> 1 Toothless dragon    black <named list [1]>
#> 2 Dory      blue tang blue  <named list [1]>

^{Created on 2020-04-03 by the reprex package (v0.3.0.9001)}

[hadley]

Ok, let's start with just naming single string pluckers.