带有gridSVG和ggplot2 v.0.9.0的交互式点标签

Question

我想以交互方式在ggplot中标记点，以便鼠标悬停在一个点上显示标签。

我正在尝试调整this question中给出的答案，以便它可以在最新版本的ggplot2中运行。受ggplot google组here的评论影响，我使用最新版本的geom-point-.r作为模板，在各个地方的gp参数中添加了“label”字段。然后我从kohske的答案中复制了剩下的代码。但它不起作用 - 结果svg中没有任何标签，我无法弄清楚原因。

我注意到point_grobs_labels中的所有内容都为空，当我查看grid.get(point_grob_names[1])$gp时，没有标签字段...

library(ggplot2)
library(gridSVG)
library(proto)
library(rjson)

geom_point2 <- function (mapping = NULL, data = NULL, stat = "identity", 
                         position = "identity",
                         na.rm = FALSE, ...) {
  ggplot2:::GeomPoint$new(mapping = mapping, data = data, stat = stat, 
                          position = position, 
                          na.rm = na.rm, ...)
}

GeomPoint2 <- proto(ggplot2:::Geom, {
  objname <- "point"

  draw_groups <- function(., ...) .$draw(...)
  draw <- function(., data, scales, coordinates, na.rm = FALSE, ...) {    
    data <- remove_missing(data, na.rm, 
                           c("x", "y", "size", "shape"), name = "geom_point")
    if (empty(data)) return(zeroGrob())

    with(coord_transform(coordinates, data, scales), 
         ggname(.$my_name(), pointsGrob(x, y, size=unit(size, "mm"), pch=shape, 
                                        gp=gpar(
                                          col=alpha(colour, alpha),
                                          fill = alpha(fill, alpha),  
                                          label = label, 
                                          fontsize = size * .pt)))
    )
  }

  draw_legend <- function(., data, ...) {
    data <- aesdefaults(data, .$default_aes(), list(...))

    with(data,
         pointsGrob(0.5, 0.5, size=unit(size, "mm"), pch=shape, 
                    gp=gpar(
                      col = alpha(colour, alpha), 
                      fill = alpha(fill, alpha), 
                      label = label,
                      fontsize = size * .pt)
         )
    )
  }

  default_stat <- function(.) StatIdentity
  required_aes <- c("x", "y")
  default_aes <- function(.) aes(shape=16, colour="black", size=2, 
                                 fill = NA, alpha = NA, label = NA)

})

p <- ggplot(mtcars, aes(mpg, wt, label = rownames(mtcars))) + geom_point2() + facet_wrap(~ gear)
print(p)

grob_names <- grid.ls(print = FALSE)$name
point_grob_names <- sort(grob_names[grepl("point", grob_names)])
point_grobs_labels <- lapply(point_grob_names, function(x) grid.get(x)$gp$label)  

jlabel <- toJSON(point_grobs_labels)

grid.text("value", 0.05, 0.05, just = c(0, 0), name = "text_place", gp = gpar(col = "red"))

script <- '
var txt = null;
function f() {
var id = this.id.match(/geom_point2.([0-9]+)\\.points.*\\.([0-9]+)$/);
txt.textContent = label[id[1]-1][id[2]-1];
}

window.addEventListener("load",function(){
var es = document.getElementsByTagName("circle");
for (i=0; i<es.length; ++i) es[i].addEventListener("mouseover", f, false);

txt = (document.getElementById("text_place").getElementsByTagName("tspan"))[0];

},false);
'

grid.script(script = script)
grid.script(script = paste("var label = ", jlabel))

gridToSVG()

Answer 1

试试这个：

library(ggplot2)
library(gridSVG)
library(proto)
library(rjson)
mtcars2 <- data.frame(mtcars, names = rownames(mtcars))

geom_point2 <- function (...) {
  GeomPoint2$new(...)
}

GeomPoint2 <- proto(ggplot2:::Geom, {
  objname <- "point"

  draw_groups <- function(., ...) .$draw(...)
  draw <- function(., data, scales, coordinates, na.rm = FALSE, ...) {    
    data <- remove_missing(data, na.rm, 
                           c("x", "y", "size", "shape"), name = "geom_point")
    if (empty(data)) return(zeroGrob())
    name <- paste(.$my_name(), data$PANEL[1], sep = ".")
    with(coord_transform(coordinates, data, scales), 
         ggname(name, pointsGrob(x, y, size=unit(size, "mm"), pch=shape, 
                                        gp=gpar(
                                          col=alpha(colour, alpha),
                                          fill = alpha(fill, alpha),  
                                          label = label, 
                                          fontsize = size * .pt)))
    )
  }

  draw_legend <- function(., data, ...) {
    data <- aesdefaults(data, .$default_aes(), list(...))

    with(data,
         pointsGrob(0.5, 0.5, size=unit(size, "mm"), pch=shape, 
                    gp=gpar(
                      col = alpha(colour, alpha), 
                      fill = alpha(fill, alpha), 
                      label = label,
                      fontsize = size * .pt)
         )
    )
  }

  default_stat <- function(.) StatIdentity
  required_aes <- c("x", "y")
  default_aes <- function(.) aes(shape=16, colour="black", size=2, 
                                 fill = NA, alpha = NA, label = NA)

})

p <- ggplot(mtcars2, aes(mpg, wt, label = names)) + geom_point2() +facet_wrap(~ gear)
print(p)

grob_names <- grid.ls(print = FALSE)$name
point_grob_names <- sort(grob_names[grepl("point", grob_names)])
point_grobs_labels <- lapply(point_grob_names, function(x) grid.get(x)$gp$label)

jlabel <- toJSON(point_grobs_labels)

grid.text("value", 0.05, 0.05, just = c(0, 0), name = "text_place", gp = gpar(col = "red"))

script <- '
var txt = null;
function f() {
    var id = this.id.match(/geom_point.([0-9]+)\\.points.*\\.([0-9]+)$/);
    txt.textContent = label[id[1]-1][id[2]-1];
}

window.addEventListener("load",function(){
    var es = document.getElementsByTagName("circle");
    for (i=0; i<es.length; ++i) es[i].addEventListener("mouseover", f, false);

    txt = (document.getElementById("text_place").getElementsByTagName("tspan"))[0];

},false);
'
grid.script(script = paste("var label = ", jlabel))
grid.script(script = script)

gridToSVG()

没有太大的变化，但我不得不添加

mtcars2 <- data.frame(mtcars, names = rownames(mtcars))

然后

p <- ggplot(mtcars, aes(mpg, wt, label = rownames(mtcars)))
     + geom_point2() + facet_wrap(~ gear)

也改为

p <- ggplot(mtcars2, aes(mpg, wt, label = names)) 
     + geom_point2() +facet_wrap(~ gear)

因为我们有rownames(mtcars)

rownames(mtcars)
 [1] "Mazda RX4"           "Mazda RX4 Wag"       "Datsun 710"          "Hornet 4 Drive"     
 [5] "Hornet Sportabout"   "Valiant"             "Duster 360"          "Merc 240D"          
 [9] "Merc 230"            "Merc 280"            "Merc 280C"           "Merc 450SE"
.....

然后标签（我们设法通过其他修改获得）保持不变，即不由gears重新排列，仅由它分割：

point_grobs_labels
[[1]]
 [1] "Mazda RX4"          "Mazda RX4 Wag"      "Datsun 710"         "Hornet 4 Drive"    
 [5] "Hornet Sportabout"  "Valiant"            "Duster 360"         "Merc 240D"         
 [9] "Merc 230"           "Merc 280"           "Merc 280C"          "Merc 450SE"        
[13] "Merc 450SL"         "Merc 450SLC"        "Cadillac Fleetwood"
[[2]]
....

但将这些标签名称作为列可以解决问题。

point_grobs_labels 
[[1]]
 [1] "Hornet 4 Drive"      "Hornet Sportabout"   "Valiant"             "Duster 360"         
 [5] "Merc 450SE"          "Merc 450SL"          "Merc 450SLC"         "Cadillac Fleetwood" 
 [9] "Lincoln Continental" "Chrysler Imperial"   "Toyota Corona"       "Dodge Challenger"   
[13] "AMC Javelin"         "Camaro Z28"          "Pontiac Firebird"   

[[2]]
....

Answer 2

感谢Tracy提出了一个很好的问题，感谢Julius提供了非常有用的答案。

要让我在Chrome和Safari中使用Julius的javascript，我必须将this.id替换为this.correspondingUseElement.id。这是有道理的，因为单个<circle> SVG元素没有每个geom_point的id，我们想要的id附加到<use>元素。

即使这在我的Firefox中也不起作用，所以我将其更改为将事件监听器附加到<use>元素本身。请注意，如果SVG更复杂，它可能除了geom_points之外还有<use>，因此我添加了if仅将事件附加到geom_point.XX <use>元素。这适用于Chrome，Safari和Firefox：

window.addEventListener("load",function(){
  var es = document.getElementsByTagName("use"); 
  for (i=0; i<es.length; ++i) {
    if(es[i].id.search(/geom_point.([0-9]+)\.points.*\.([0-9]+)$/) >= 0) es[i].addEventListener("mouseover", f, false);
  }
  txt = (document.getElementById("text_place").getElementsByTagName("tspan"))[0];
},false);

（所有其他代码与Julius相同）

Answer 3

我们通过检测生成的.svg中的颜色属性并使用css检测鼠标悬停来解决这个问题。结果显示在本演示的步骤4,5,6中：

Showing svg highlighting using css

这是我的第一个stackoverflow响应 - 希望我得到正确的礼仪

Answer 4

这是一个一般性的答案。我只是在这里帮助您使剧情互动。您可以尝试-

library(ggplot2)
library(plotly)

# Plot how you would normally code for ggplot2
p <- ggplot(data,... 'add your variables and subsequent plots')
ggplotly(p)

享受！