基于类型的自定义“ggplot2”扩展中的子集数据

Question

我正在开发一个适用于 data.frames 的 ggplot2 扩展程序，看起来有点像：

data <- data.frame(
  type = c("text", "text", "line", "line"),
  label = c("some label", "another one", NA, NA),
  x = c(0,10,2,4),
  y = c(0,10,3,7),
  xend = c(NA, NA, 8, 10),
  yend = c(NA, NA, 3, 4)
)

这意味着我们有 对象（行） 具有不同的type。现在我想根据类型对我的Geoms 和Stats 中的数据 进行子集化。

考虑以下示例（使用ggplot2 标准函数）：

library(ggplot2)
ggplot(data, aes(x, y)) + 
  geom_text(aes(label = label)) + 
  geom_segment(aes(xend = xend, yend = yend))

这会绘制人们期望的内容（text 作为文本，lines 作为段）。

现在我有自己的geom_text 版本，称为geom_var：

GeomVar <- ggproto("GeomVar", ggplot2::GeomText,
                   default_aes = aes(x = x, y = y, label = label, colour = "black", 
                                     size = 4, angle = 0, hjust = 0.5, vjust = 0.5, 
                                     alpha = NA, family = "Arial", fontface = 1, 
                                     lineheight = 1.2, length = 10)
)
geom_var <- function(mapping = aes(label = label), data = NULL, position = "identity", 
                     ..., parse = FALSE, nudge_x = 0, nudge_y = 0, check_overlap = FALSE, 
                     na.rm = FALSE, show.legend = NA, inherit.aes = TRUE) 
{
  ggplot2::layer(data = data, mapping = mapping, stat = StatVar, geom = GeomVar, 
                 position = position, show.legend = show.legend, inherit.aes = inherit.aes, 
                 params = list(parse = parse, check_overlap = check_overlap, 
                               na.rm = na.rm, ...))
}

StatVar <- ggproto("StatVar", ggplot2::Stat,
                   compute_group = function(data, scales, length = 5) {
                     data$label <- sapply(data$label, 
                                          function(x) {paste0(strwrap(x, width = length), 
                                                              collapse = "\n")})
                     data
                   }
)

stat_var <- function(mapping = NULL, data = NULL, geom = "var",
                     position = "identity", na.rm = FALSE, show.legend = NA,
                     inherit.aes = TRUE, ...) {
  ggplot2::layer(
    stat = StatVar, data = data, mapping = mapping, geom = geom,
    position = position, show.legend = show.legend, inherit.aes = inherit.aes,
    params = list(na.rm = na.rm, ...)
  )
}

当我使用我自己的版本时，情节看起来像：

ggplot(data, aes(x, y)) + 
  geom_var(aes(label = label)) + 
  geom_segment(aes(xend = xend, yend = yend))

---

TL;DR：

• 如何更改我的 GeomVar 和/或 StatVar 以使 NAs 不再被绘制？

• 或者：如何在 GeomVar 和 StatVar 函数中根据 type 对 data 进行子集化？

（我在data 基本上在GeomVar、geom_var、StatVar 和stat_var 函数中都尝试过data <- data[data$type == "text", ]。）

Answer 1

您可以在geom_var(aes(...)) 中将类型作为美学传递，并为您的StatVar 指定setup_data 函数以查看它：

# define setup_data in StatVar
StatVar <- ggproto("StatVar", 
                   ggplot2::Stat,
                   setup_data = function(data, params){
                     # print(data) # I like doing this while debugging code to see what data
                                   # actually looks like at this point
                     data <- data[data$type == "text", ]
                   },
                   compute_group = function(data, scales, length = 5) {
                     data$label <- sapply(data$label, 
                                          function(x) {paste0(strwrap(x, width = length), 
                                                              collapse = "\n")})
                     data
                   }
)

# include type as a required aesthetic mapping in GeomVar
GeomVar <- ggproto("GeomVar", ggplot2::GeomText,
                   required_aes = c("x", "y", "label", "type"),
                   default_aes = aes(x = x, y = y, label = label, colour = "black", 
                                     size = 4, angle = 0, hjust = 0.5, vjust = 0.5, 
                                     alpha = NA, family = "Arial", fontface = 1, 
                                     lineheight = 1.2, length = 10))

# map type in geom_var
ggplot(data, aes(x, y)) + 
  geom_var(aes(label = label, type = type)) + 
  geom_segment(aes(xend = xend, yend = yend))