37

When using map on a nested data_frame, I do not understand why the latter two version give an error, how should I use the dot (.)?

library(tidyverse)
# dummy data
df <- tibble(id = rep(1:10, each = 10), 
                 val = runif(100))
df <- nest(df, -id)

# works as expected
map(df$data, min)
df %>% .$data %>% map(., min)

# gives an error
df %>% map(.$data, min)
# Error: Don't know how to index with object of type list at level 1

df %>% map(data, min)
Axeman
  • 32,068
  • 8
  • 81
  • 94
johannes
  • 14,043
  • 5
  • 40
  • 51

1 Answers1

53

The problem isn't map, but rather how the %>% pipe deals with the .. Consider the following examples (remember that / is a two argument function in R):

Simple piping:

1 %>% `/`(2)

Is equivalent to `/`(1, 2) or 1 / 2 and gives 0.5.

It is also equivalent to 1 %>% `/`(., 2).

Simple . use:

1 %>% `/`(2, .)

Is equivalent to `/`(2, 1) or 2 / 1 and gives 2.

You can see that 1 is no longer used as the first argument, but only as the second.

Other . use:

This doesn't work however, when subsetting the .:

list(a = 1) %>% `/`(.$a, 2)

Error in `/`(., .$a, 2) : operator needs one or two arguments

We can see that . got injected twice, as the first argument and subsetted in the second argument. An expression like .$a is sometimes referred to as a nested function call (the $ function is used inside the / function, in this case).

We use braces to avoid first argument injection:

list(a = 1) %>% { `/`(.$a, 2) }

Gives 0.5 again.

Actual problem:

You are actually calling map(df, df$data, min), not map(df$data, min).

Solution:

Use braces:

df %>% { map(.$data, min) }

Also see the header Using the dot for secondary purposes in ?magrittr::`%>%` which reads:

In particular, if the placeholder is only used in a nested function call, lhs will also be placed as the first argument! The reason for this is that in most use-cases this produces the most readable code. For example, iris %>% subset(1:nrow(.) %% 2 == 0) is equivalent to iris %>% subset(., 1:nrow(.) %% 2 == 0) but slightly more compact. It is possible to overrule this behavior by enclosing the rhs in braces. For example, 1:10 %>% {c(min(.), max(.))} is equivalent to c(min(1:10), max(1:10)).

Axeman
  • 32,068
  • 8
  • 81
  • 94