Dealing with tibbles?

[nickhaf]

@weirichs :

I ran into the following issue when working on eatRecode:

recodeList_2 <- readxl::read_xlsx(here::here("tests", "testthat", "helper_recodeDB_2.xlsx"))
recoded_df2 <- useRecodeList(df, varName = "country", new_varName = "country_r", 
                             recodeList = recodeList_2)

gave the error message:

Fehler in eatTools::recodeLookup(df[, varName], lookup = recodeList) :
Old values in lookup table are not unique.

The useRecodeList() function uses recodeLookup internally.

When loading the data like this it works just fine:

recodeList_2 <- as.data.frame(readxl::read_xlsx(here::here("tests", "testthat", "helper_recodeDB_2.xlsx")))

I pinned the error down to this line in recodeLookup():

if (length(unique(lookup[, 1])) < nrow(lookup)) {

This extracts a column in a tibble: my_tibble[, 1]
... and a vector in a data.frame: my_data.frame[, 1]

So in the first case the length is 1, and in the second 6 (in my example).

One fix might be using the double brackets in this case:

if (length(unique(lookup[[1])) < nrow(lookup)) {

But let me know what you think!

[weirichs]

Thanks for your tip! However, I think length(unique(lookup[[1])) might fail if lookup is given as a matrix. I think I would replace the first line of the function

if(!is.data.frame(lookup) && !is.matrix(lookup)) {stop("The lookup table must be a data.frame or matrix.")}

with

lookup <- makeDataFrame(lookup)

[beckerbenj]

Is there a reason this is still open? I think this was fixed in commit 77dbab7?

[weirichs]

Right. Issue can be closed.