Bug fix: single-column data frame input for row_to_names() now works

[matanhakim]

Description

Before this PR and after merging #542, code like this would error:

library(janitor)
df <- data.frame(x = c("", "", "", "1"))

df |>
  row_to_names(1:3) |> 
  names()

this is because in the code:

new_names <- sapply(dat[row_number,], paste_skip_na, collapse = sep) %>% 
    stringr::str_replace_na()

the specific call dat[row_number,] turns the single-column data frame (i.e., a list of length 1) to a character vector (of length 3 in this case). This, in turn, makes sapply() to iterate over a wrong vector and therefore error.

Changing this to dat[row_number, , drop = FALSE] solves this.

A relevant test was added to prevent this in the future.

Finally, documentation for the changed behavior of the row_number argument in row_to_names() was added, and the NEWS file was edited to reflect changes in this PR and in #542.

Related Issue

Fixes #536 and follows up on #542.

Example

See above.

[codecov]

Codecov Report

Merging #545 (555eceb) into main (5a34f00) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##              main      #545   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           26        26           
  Lines         1182      1182           
=========================================
  Hits          1182      1182           

Impacted Files	Coverage Δ
R/row_to_names.R	100.00% <100.00%> (ø)

[billdenney]

This looks good. Thanks!