Shiny w/ ggplot produces erroneous legend: duplicate legend entries with facets - r

Question

I want to plot a facet plot with added geom_hline and show the line in the legend. However, when I add the line to the legend, all entries duplicate to the number of facets.
How can I avoid this behaviour?
Here is my MWE
library(shiny)
library(plotly)
library(ggplot2)
df <- mpg
# Define UI for application that draws a histogram
ui <- fluidPage(
mainPanel(
plotlyOutput('graph')
)
)
# Define server logic required to draw a histogram
server <- function(input, output) {
output$graph <- renderPlotly({
p <- ggplot2::mpg %>%
plot_ly %>%
ggplot() +
geom_point(aes(displ, hwy, color = class))
facet <- p + facet_wrap(~year)
facet + geom_hline(aes(yintercept = 20, linetype = 'hline20'), color = '#00b700') +
scale_linetype_manual(name = 'line', values = 1,
guide = guide_legend(aes = list(color = "#00b700")))
})
}
# Run the application
shinyApp(ui = ui, server = server)

Answer 1

The problem is not shiny nor ggplot2. The problem is the conversion of a ggplot2 object into a plotly object.
Below I've isolated the ggplot steps from the plotly step and completely removed shiny from the picture.
p <- ggplot2::mpg %>%
ggplot() +
geom_point(aes(x = displ, y = hwy, color = class))
facet <- p + facet_wrap(~year)
final <- facet + geom_hline(aes(yintercept = 20, linetype = 'hline20'), color = '#00b700') +
scale_linetype_manual(name = 'line', values = 1, guide = guide_legend(aes = list(color = "#00b700")))
print(final)
The above displays fine.
ggplotly(final)
This above has the same display errors as yours, without involving shiny.
Legend issues and other display issues are common when converting ggplot2 objects to plotly objects. Plotly and ggplot2 are completely independent plotting frameworks with their own syntax and graphics objects. Plotly provides conversion function methods but things still get lost in translation from one format to the other, because there is not a complete equivalency between them.
I'd recommend trying to implement your plot in native plotly syntax instead of trying to convert it from ggplot2.