tryCatch() or exists() in R pipe chain error handling

I am currently experiencing a hiccup where I am trying to create a table of values from some upstream calculations. Typically, I've been assuming that there would be at least one value of 1 each time these data frames are created; but I've encountered an example where this wasn't the case, and ended up with a table that looks like this:

df <- data.frame(
  Experiment_Batch = c(rep("008_1", 83),
                       rep("008_6", 82),
                       rep("520_0", 2),
                       rep("944_10", 84),
                       rep("944_8", 85),
                       rep("944_9", 72)),
  Overall = rep(0, 408)
  )

This has caused the following downstream processing:

df %>% 
  dplyr::count(Experiment_Batch, Overall) %>%
  tidyr::spread(Overall, n, fill = 0) %>% 
  dplyr::select(Experiment_Batch, `1`)

to error out: Error in overscope_eval_next(overscope, expr) : object '1' not found.

I've tried using tryCatch() and exists(), but I can't quite seem to get these to work properly. Ideally, this would all be handled elegantly using pipe operators. I already know I could create an object and add in a few if-else statements as necessary to my workflow, but I'm wondering if there's a... "more interesting" way to take care of this, so I wouldn't have to break up the work.

Solution

You can use select helper functions, if you want to ignore the selection if the column doesn't exist; Here matches("^1$") will try to select the column whose name exactly matches 1, since the data frame doesn't have the column, it simply ignores the selection as matches returns integer(0):

library(tidyverse)
df %>% 
    count(Experiment_Batch, Overall) %>%
    spread(Overall, n, fill = 0) %>% 
    select(Experiment_Batch, matches("^1$"))

# A tibble: 6 x 1
#  Experiment_Batch
#*           <fctr>
#1            008_1
#2            008_6
#3            520_0
#4           944_10
#5            944_8
#6            944_9

matches returns integer(0) when non of the column names matches the pattern which gets ignored in select:

matches("^1$", vars = c("0", "experiment"))
# integer(0)

matches("^1$", vars = c("0", "experiment", "1"))
# [1] 3

If you need to customize the error catch:

library(tidyverse)
df %>% 
    count(Experiment_Batch, Overall) %>%
    spread(Overall, n, fill = 0) %>% 
    {
        tryCatch(
            select(., Experiment_Batch, `1`), 
            error=function(e) select(., Experiment_Batch)
        )
    }
    # replace the error with the customized function to handle the exception

# A tibble: 6 x 1
#  Experiment_Batch
#*           <fctr>
#1            008_1
#2            008_6
#3            520_0
#4           944_10
#5            944_8
6            944_9