How to safely replace values in data.frame even when data.frame is empty?

Question

I'm having trouble writing safe code to manipulate data frames in R. The problem I'm having is that df[...] <- (...) is vulnerable to the data.frame having no rows (among other things!).

Example 1:

df <- data.frame(a = 1:2, b = c(NA, 5))
df[is.na(df$b), 'b'] <- 0

Does what I want - replaces NA with 0.

Example 2:

df.empty <- data.frame(a = character(), b = character())
df.empty[is.na(df.empty$b), 'b'] <- 0
df.empty[is.na(df.empty$b), 'b', drop = F] <- 0

Both attempts to mutate the data frame in Example 2 result in an error (either replacement has 1 row, data has 0; or 'unused argument (drop = F)).

I find that very annoying.

How do I universally mutate a column based on arbitrary (vector) criterion in a data.frame, without having to manually check for all possible conditions like NA, NULL, no rows, coercion to vector because of no rows, etc.?

Answer 1

Perhaps using dplyr?

This example, combining mutate() and ifelse(), seems to achieve what you want.

df.empty <- df.empty %>% 
    mutate(b = ifelse(is.na(b), 0, b))