Now that we have a series of consistency checks built into Lint():
|
if (!missing(linter)) { |
|
lintr_deprecated( |
|
old = "Using the `linter` argument of `Lint()`", |
|
version = "3.0.0", |
|
type = "" |
|
) |
|
} |
|
|
|
if (length(line) != 1L || !is.character(line)) { |
|
stop("`line` must be a string.") |
|
} |
|
max_col <- max(nchar(line) + 1L, 1L, na.rm = TRUE) |
|
if (!is_number(column_number) || column_number < 0L || column_number > max_col) { |
|
stop(sprintf( |
|
"`column_number` must be an integer between 0 and nchar(line) + 1 (%d). It was %s.", |
|
max_col, column_number |
|
)) |
|
} |
|
if (!is_number(line_number) || line_number < 1L) { |
|
stop(sprintf("`line_number` must be a positive integer. It was %s.", line_number)) |
|
} |
|
check_ranges(ranges, max_col) |
It may make sense to offer a vectorized interface for constructing lints so these consistency checks can be vectorized for efficiency, rather than running the checks lint-by-lint.
