Column names reach model formulas, where text such as
"x+I(file.create('f'))" would run as code. Each name is trimmed, every
character outside A-Z a-z 0-9 . _ becomes ., the reserved ... and
..1, ..2, ... get an X prefix, and make.names() with
unique = TRUE fixes the rest, so every result matches
^[A-Za-z.][A-Za-z0-9._]*$ and parses as one symbol.
Value
data with syntactic, unique names. The attribute "renamed"
is a character vector of new names named by the original names, for
the changed columns only.
Examples
d <- safe_column_names(data.frame(`a b` = 1, `x+I(q())` = 2, check.names = FALSE))
attr(d, "renamed")
#> a b x+I(q())
#> "a.b" "x.I.q..."