using renderUI to update a data set instead of conditional panel - r

Question

I am writing a shiny app to realize the following effects:
Whenever I choose variable included by categoryname, the web will generate the slider which provides a divider. It divides the selected variable into 2 groups and form a new column containing the group name added to the original data set.
People here helped me solve the problem using conditional panel in this question but now I use renderUI combined with shinyjs cause conditional panel doesn't work in my large project.
I am stuck by a tiny bug (seems):
Error in data.frame: arguments imply differing number of rows: 32, 0
The following is my code, how to change it so that the function works?
library(shiny)
library(shinyjs)
library(stringr)
categoryname = c("mpg_group", "disp_group")
MT_EG = mtcars[,1:5]
# Define UI for application that draws a histogram
ui <- fluidPage(
useShinyjs(),
# Application title
titlePanel("Mtcars Data"),
# Sidebar with a slider input for number of bins
sidebarLayout(
sidebarPanel(
selectInput(inputId = "arm",
label = "ARM VARIABLE",
choices = c("mpg_group", "cyl", "disp_group", "hp", "drat"),
selected = "cyl"),
# conditionalPanel(
# #condition = "categoryname.includes(input.arm)",
# condition = "input.arm == 'disp_group' | input.arm == 'mpg_group'",
#
# #sliderInput("divider", "divide slider", 1, 100, 20)
# optionalSliderInputValMinMax("divider", "divide slider", c(50,0,100), ticks = FALSE)
# )
uiOutput("divider")
),
# Show a plot of the generated distribution
mainPanel(
uiOutput("data")
)
)
)
# Define server logic required to draw a histogram
server <- function(input, output, session) {
output$divider <- renderUI({
if (input$arm %in% categoryname){
show("divider")
}
else{
hide("divider")
}
sliderInput("divider", "divide slider", 0, 100, 50)
})
observeEvent(
input$arm,
observe(
{
if (input$arm %in% categoryname){
#browser()
# start over and remove the former column if exists
MT_EG = MT_EG[, !(colnames(MT_EG) %in% input$arm)]
id_arm_var <- input$arm
id_arm <- unlist(str_split(id_arm_var,'_'))[1]
# change the range of the slider
val <- input$divider
mx = max(MT_EG[[id_arm]])
mn = min(MT_EG[[id_arm]])
updateSliderInput(session, inputId = "divider", min=floor(mn/2),max = mx + 4,step = 1,value = input$divider)
# generate a new column and bind
divi <- data.frame(id_arm_var = MT_EG[[id_arm]]>input$divider)
divi$id_arm_var[divi$id_arm_var==TRUE] <- paste0(id_arm_var, " Larger")
divi$id_arm_var[divi$id_arm_var==FALSE] <- paste0(id_arm_var, " Smaller")
colnames(divi) <- id_arm_var
MT_EG <- cbind(MT_EG,divi)
}
output$data=renderTable(MT_EG)
}
)
)
}
# Run the application
shinyApp(ui = ui, server = server)
I simply use mtcars dataset so that all of you can get access to

Answer 1

observeEvent and renderUI both depend on "or triggered by" input$arm. What happened is observeEvent ask for input$divider before renderUI rendered probably in the UI and input$divider becomes available for observeEvent which as I mentioned above input$divider ends up with value NULL.
To solve this problem just add req(input$divider) before MT_EG = MT_EG[, !(colnames(MT_EG) %in% input$arm)]. Also change output$divider to output$dividerUI since Shiny doesn't allow input and output with the same id.
See ?shiny::req for more details.