use of bquote in ggplot2 custom labeller-CodePudding

Is it possible to properly use bquote in some instances of ggplot2 custom labellers? The following example does not produce the desired output (and does not show any warning/error):

library(ggplot2)
data(Orange)
Orange$tree2 <- paste0("T", Orange$Tree)

tree.lab <- c(
  T1 = bquote("Tree 1"[test]), 
  T2 = "Tree 2", 
  T3 = "Tree 3",  
  T4 = "Tree 4", 
  T5 = "Tree 5"
)

ggplot(Orange, aes(x = age, y = circumference))  
  facet_wrap(~tree2, labeller = labeller(tree2 = tree.lab))  
  geom_point()

The following old style code still works but is deprecated (ggplot2 version 3.3.5 warns "Warning message: The labeller API has been updated. Labellers taking variable and value arguments are now deprecated. See labellers documentation."):

vlabeller <- function (variable, value) {
  return(tree.lab[value])
}

ggplot(Orange, aes(x = age, y = circumference))  
  facet_wrap(~tree2, labeller = vlabeller)  
  geom_point()

What would be a proper way to do that nowadays? Even without bquote...

CodePudding user response：

Basically with your list, you would want to update the vlabeller to use the as_labeller function. The default=identity prevents the expressions from being coerced back into text.

vlabeller <- as_labeller(function (value) {
  tree.lab[value]
} , default=identity)

To avoid the bquote, you could with with an exprssion list. Here's a helper function that will make that easier to work with.

label_expressions <- function(values) {
  stopifnot(is.expression(values))
  as_labeller(function(x) {
    if (is.null(names(values))) {
      x <- seq_along(x)
      if (length(x)!=length(values)) warning(paste0("Number of labels(", 
        length(values), ") does not match number of values (", length(x), ")"))
    }
    as.list(values[x])
  }, default=identity)
}

Then you can draw your plot with

tree.lab <- expression(
  T1 = "Tree 1"[test], 
  T2 = "Tree 2", 
  T3 = "Tree 3",  
  T4 = "Tree 4", 
  T5 = "Tree 5"
)

ggplot(Orange, aes(x = age, y = circumference))  
  facet_wrap(~tree2, labeller = label_expressions(tree.lab))  
  geom_point()

Note that tree.lab is an expression object so all the values are not parsed and remain language objects.