I'd like to bind (row-wise) two dataframes of different column size, but enforce that the columns' formats of the first dataframe be kept.

Example:

d1 <- tibble(x = 1, y = "b") # before edit: d1 <- tibble(x = 1)
d2 <- tibble(x = c(2, "a"))

When employing, say, plyr's rbind.fill, the combined column will be of the character type. However, I'd like to obtain the (final) result

str(data.frame(x = c(1, 2, NA)))
#> 'data.frame':    3 obs. of  1 variable:
#>  $ x: num  1 2 NA

I suppose for this to work, I'd have to

• look up x's format in d1 and then
• force x in d2 to be of the same format, producing NA's along the way, and
• rbind thereafter.

The problem is, I cannot come up with an elegant solution to this little problem. Any help will be much appreciated!

Edited to account for data frames of different column size.

CodePudding user response：

Your approach is reasonable and can be done in a single line:

bind_rows(d1, map2_df(d1, d2, ~ `class<-`(.y, class(.x))))
#> # A tibble: 3 x 1
#>       x
#>   <dbl>
#> 1     1
#> 2     2
#> 3    NA

CodePudding user response：

In base R with match.fun:

common <- intersect(names(d1), names(d2))
d2[common] <- lapply(common, function(x) {
  match.fun(paste0("as.", class(d1[[x]])))(d2[[x]])
})

dplyr::bind_rows(d1, d2)

output

# A tibble: 3 × 1
      x
  <dbl>
1     1
2     2
3    NA