R Shiny plotly::plotlyProxyInvoke Update on subplot - Updates all traces/subplots not target plot of a subplot

Question

Good afternoon,

I'm attempting to use the plotly::plotlyProxyInvoke() function to update the data in a single plot of a subplot. I'm having issues, however, as it appears the indices argument does not apply to the correct trace in a subplot. Each plot gets updated with the same data.

Plot Before Update Original Data

Plot After Update The second plot 'hp' was targets to have 'Camero z28' set to 0. The code is doing that, but copying the data for hp, and applying it to disp, and mpg.

REPEX of my problem*

library(shiny)
library(tidyverse)
library(plotly)

cars <- mtcars %>% rownames_to_column(var = "carname") %>% 
  arrange(carname) %>% 
  select(carname, cyl, mpg, hp, disp) %>% 
  unique() %>% 
  mutate(cyl = as.factor(cyl)) %>% 
  pivot_longer(cols = -c(carname, cyl))

ui <- fluidPage(
  plotlyOutput("myplot", height = '600px'),
  actionButton("update", "Update Data"), 
  numericInput('newValue', "New hp Value", 0), 
  selectInput(inputId = "carName", "Car Name", multiple = FALSE, selectize = TRUE, choices = cars$carname, selected = unique(cars$carname[1]))
)

server <- function(input, output, session) {

  output$myplot <- renderPlotly({

    pp <- cars %>%
      group_by(name) %>%
      do(p =
           plot_ly(data = .) %>%
           add_bars(
             x = ~carname,
             y = ~value
             # color = ~cyl
           ) %>%
           layout(showlegend = F, yaxis = list(title = ~unique(name)))
      ) %>%
    subplot(nrows = 3, titleY = TRUE, titleX = FALSE, shareX = FALSE, shareY = FALSE)  
  
    pp

  })
   
  observeEvent(input$update, {

    cars2 <- cars %>% 
      filter(name == "hp") %>% 
      rows_update(., tibble(carname = input$carName, value = input$newValue))

    proxy <- plotlyProxy("myplot", session = shiny::getDefaultReactiveDomain())

    proxy %>% plotlyProxyInvoke("update",
                                y = list(cars2$value),
                                x = list(cars2$carname),
                                #Defining the index here is not helping? 
                                as.integer(1) #either does this. 
                                # list(list(1)) #this doesn't work
    )

  })

}


shinyApp(ui, server)

Answer 1

Please check the modified plotlyProxyInvoke call:

library(shiny)
library(tidyverse)
library(plotly)

cars <- mtcars %>% rownames_to_column(var = "carname") %>% 
  arrange(carname) %>% 
  select(carname, cyl, mpg, hp, disp) %>% 
  unique() %>% 
  mutate(cyl = as.factor(cyl)) %>% 
  pivot_longer(cols = -c(carname, cyl))

ui <- fluidPage(
  plotlyOutput("myplot", height = '600px'),
  actionButton("update", "Update Data"), 
  numericInput('newValue', "New hp Value", 0), 
  selectInput(inputId = "carName", "Car Name", multiple = FALSE, selectize = TRUE, choices = unique(cars$carname), selected = cars$carname[1])
)

server <- function(input, output, session) {
  
  output$myplot <- renderPlotly({
    
    pp <- cars %>%
      group_by(name) %>%
      do(p =
           plot_ly(data = .) %>%
           add_bars(
             x = ~carname,
             y = ~value
             # color = ~cyl
           ) %>%
           layout(showlegend = F, yaxis = list(title = ~unique(name)))
      ) %>%
      subplot(nrows = 3, titleY = TRUE, titleX = FALSE, shareX = FALSE, shareY = FALSE)  
    
    pp
    
  })
  
  observeEvent(input$update, {
    cars2 <- cars %>% 
      filter(name == "hp") %>% 
      rows_update(., tibble(carname = input$carName, value = input$newValue))
    proxy <- plotlyProxy("myplot", session = shiny::getDefaultReactiveDomain())
    proxy %>% plotlyProxyInvoke("restyle", list(y = list(cars2$value), x = list(cars2$carname)), list(0L))
    # update = restyle + relayout # subplots share a single layout
    # proxy %>% plotlyProxyInvoke("update", list(y = list(cars2$value), x = list(cars2$carname)), list(0L))
  })
  
}


shinyApp(ui, server)