Hiho,
i think you have to add an as.character like this:
sapply(sequence.to_roman, function(x) as.character(if(grepl("^[1-9]\\d*$",x)) as.roman(x) else "Some symbols"))
> sapply(sequence.to_roman, function(x) as.character(if(grepl("^[1-9]\\d*$",x)) as.roman(x) else "Some symbols"))
1 2 V1 df 3
"I" "II" "Some symbols" "Some symbols" "III"
looks like the shortening of the output in sapply would convert the class roman back to its numeric value by default. So converting all outputs to char first prevents this.
try:
lapply(sequence.to_roman, function(x) if(grepl("^[1-9]\\d*$",x)) as.roman(x) else "Some symbols")
> lapply(sequence.to_roman, function(x) if(grepl("^[1-9]\\d*$",x)) as.roman(x) else "Some symbols")
[[1]]
[1] I
[[2]]
[1] II
[[3]]
[1] "Some symbols"
[[4]]
[1] "Some symbols"
[[5]]
[1] III
this iw what we want, but:
unlist(lapply(sequence.to_roman, function(x) if(grepl("^[1-9]\\d*$",x)) as.roman(x) else "Some symbols"))
> unlist(lapply(sequence.to_roman, function(x) if(grepl("^[1-9]\\d*$",x)) as.roman(x) else "Some symbols"))
[1] "1" "2" "Some symbols" "Some symbols" "3"
also gives the recodet forms.
for a maybe more visible description what causes the problem:
> as.roman("3")
[1] III
> as.character(as.roman("3"))
[1] "III"
> c(as.roman("3"), "test")
[1] "3" "test"
> c(as.character(as.roman("3")), "test")
[1] "III" "test"