Shiny creating a reactive legend - r

Question

I am following a solution given in R shiny Aesthetics must be either length 1 or the same as the data (8): y for that annoying problem which I have happily fixed.
The next issue I want to solve is that I want my plot to have a reactive legend - I only want the legend to display what's actually chosen and on the plot
I also want to set the colours of the lines to the ones I want. And finally I want to make sure the legend is always in the order that I specify
Here is a reproducible example (the commented out code is my attempt at solving my own issues.
As you can see, the commented out section is how I've tried to get the legend and colours I want:
library(shiny)
library(tidyverse)
library(reshape2)
library(scales)
time <- seq(-9, 60, 1)
var1 <- rnorm(70, 35, 2)
var2 <- rnorm(70, 50, 2)
var3 <- rnorm(70, 24, 2)
var4 <- rnorm(70, 17, 2)
data <- data.frame(time = time,
var1 = var1,
var2 = var2,
var3 = var3,
var4 = var4)
datamelt <- melt(data, "time")
p <- ggplot(datamelt, aes(x = time, y = value, color = variable)) +
# scale_color_manual(values = c(
# 'first' = 'red',
# 'second' = 'blue',
# 'third' = 'green',
# 'fourth' = 'orange'
# ),
# breaks = c("first", "second", "third", "fourth")) +
# labs(color = 'Legend') +
theme_classic() +
theme(axis.ticks = element_blank()) +
labs(title = 'it means nothing',
subtitle = 'these are made up data') +
theme(plot.title = element_text(hjust = 0.5),
plot.subtitle = element_text(hjust = 0.5)) +
scale_x_continuous(name ="a y variable", breaks = seq(-9, 60, 1)) +
scale_y_continuous(name = "yep an x variable",
breaks = seq(0, 60, 5), labels = comma) + geom_blank()
ui <- fluidPage(
titlePanel("trying to make this work"),
sidebarLayout(
sidebarPanel(
checkboxGroupInput("whichone", "Choose something:",
choiceNames = c("first",
"second",
"third",
"fourth"),
choiceValues = c("var1",
"var2",
"var3",
"var4"))
),
###the plot
mainPanel(
plotOutput("plot")
)
)
)
server <- function(input, output) {
output$plot <- renderPlot({
data_filtered <- datamelt %>% filter(variable %in% input$whichone)
p + geom_line(data = data_filtered)
})
}
shinyApp(ui, server)

Answer 1

The problem occurs because ggplot uses all the factor levels, which stay in place when you filter. So you need to drop these levels first.
Secondly, you are generating the plot statically with all the levels. So you need to update the data in teh plto as well to let ggplot know which levels to show in the legend. Putting that together you can use the following:
server <- function(input, output) {
output$plot <- renderPlot({
## 1. drop unused levels from teh filtered database
data_filtered <- datamelt %>% filter(variable %in% input$whichone) %>%
droplevels()
## 2. tell ggplot to update the data
p %+% data_filtered + geom_line()
})
}
Note. This approach has the nasty (?) side effect, that if you do not select any data to show you see only an empty canvas. This can be also remedied, but would require some change in your code logic (basically moving the construction of the ggplot inside the renderPlot)
Screenshots