Locate values outside base R's numeric grammar

Return the positions of present values that become missing when converted by as.numeric(as.character(x)). Use this diagnostic to find labels, unit suffixes, localized separators, or other text that base R cannot parse as a number.

Usage

which.nonnum(x)

Arguments

x

An atomic vector or factor. Factor levels are tested through their labels. The result has the same positional indexing as x.

Value

An integer vector of ascending one-based positions. Returns integer(0) when every non-missing value is accepted.

Details

Conversion warnings are suppressed. Original missing values, including a numeric NaN, are excluded from the result. A present string such as "NA" or "NaN" is reported because its conversion is missing while the original element is not. Finite numbers and Inf are accepted by base R.

This helper does not apply toNumeric() first. Consequently, "1,25" is reported even though toNumeric("1,25") returns 1.25. Values containing units or currency symbols are also reported; neither units nor locale are inferred.

See also

toNumeric() to normalize the package's supported separator formats before numeric conversion.

Examples

x <- c("2.5", "1,25", "12 kg", NA, "Inf")
which.nonnum(x)
#> [1] 2 3
x[which.nonnum(x)]
#> [1] "1,25"  "12 kg"