crosstab fails if 1st variable has spaces, a pipe, and multiple quoted arguments in its name

[sfirke]

Wildly obscure, but I found it when in the course of actual analysis I tried:

crosstab(responses_principals$`The candidate(s) applied directly to my school` %>% convert_to_top_2_agree(c("All", "Most")),
            responses_principals$region) 

and got:

Error in stats::setNames(., c(var_name, names(.)[-1])) :
'names' attribute [11] must be the same length as the vector [10]

Here is a reprex:

dat <- data.frame(
  `The candidate(s) applied directly to my school` = c("a", "b", "a", "b"),
  x = 1:4,
  check.names = FALSE,
  stringsAsFactors = FALSE
)

crosstab(dat$`The candidate(s) applied directly to my school` %>% na_if(c("a", "c", "d", "e")), dat$x)

[sfirke]

Looks like it's the line var_name <- deparse(substitute(vec1)) in crosstab(), which in this case returns a vector of length 2 to assign to var_name.

Adding this immediately after var_name is assigned seems to fix it, and be safe; I will write a test for it and then merge in:

  # an odd variable name can be deparsed into a vector of length >1, rare but breaks function, see issue #87
  if(length(var_name) > 1){ var_name <- paste(var_name, collapse = "") }

[sfirke]

It's the fault of the width.cutoff argument in deparse(): http://stackoverflow.com/q/41596098/4470365. I like the fix above as more robust than just increasing width.cutoff to a very high number. But maybe I should also increase it, if it would result in prettier results than the collapsed version?