ggplot2: fixed color scale for known levels and subsets

Question

Creating multiple plots makes it necessary to have a common color scale. That's easy by using scale_color_manual. But it gets tricky if not all levels are present in every plot like with this example

doG = function(df) {
  df$g = factor(df$g, levels = c("red_dot", "green_dot", "blue_dot"))
  g = ggplot(df, aes(x = x, y = y, color = g)) + geom_point()
  g + scale_color_manual(values = c("red", "green", "blue"))
}
df = data.frame(x = 1:3, y = c(1,1,1), g = c("green_dot", "blue_dot", "red_dot"))
doG(df)

looks perfectly fine but then

doG(df[1:2,])

destroys all the illusions. Is there some smart way I can reuse the function doGfor subsets where not all factor levels are present?

Answer 1

Ok. I just came up with that

doG = function(df) {
  ls = c("red_dot", "green_dot", "blue_dot")
  ls.col = c("red", "green", "blue")
  df$g = factor(df$g, levels = ls)
  g = ggplot(df, aes(x = x, y = y, color = g)) + geom_point()
  g + scale_color_manual(values = ls.col[ls %in% df$g])
}