RShiny: Refer to data from UI in server - r

Question

Example Case: I have a function in my global.R called get_data which returns a list of many items. The reason I don't just put the data in global is so the data can automatically refresh after a certain amount of time
ui.R
my_data <- uiOutput("data") # Doesn't work
### Some more generic manipulation before final use
# The output of my_data will look like the following below.
my_data <- list()
my_data$first_entry <- c("a", "b", "d")
my_data$second_entry <- c("x", "y", "z") # and so on
shinyUI(navbarPage(theme=shinytheme("flatly"),
'App Name',
tabPanel('Title',
# Sidebar with a slider input for number of bins
sidebarLayout(
sidebarPanel(
width=3,
# new box
checkboxGroupButtons(
'name',
'label:',
choices = sort(my_data$first_value),
status = 'primary',
selected = sort(my_data$first_value)[1],
size = 'xs'
# inline = TRUE
))
server.R
shinyServer(function(input, output, session) {
data <- reactive({
invalidateLater(100000,session)
get_data()
})
output$data <- renderUI({
data()
})
})
Two questions:
Is there any way of referencing my_data correctly?
If my function get_data is simply reading a (large) csv which is updated systematically. Is there a better way of doing it than I am currently doing it?

Answer 1

I think you're wondering how to define possible choices= for something within the UI element, when the data is both (1) undefined at the start, and (2) changing periodically. The answer to that is to define it "empty" and update it as the new data is found.
library(shiny)
library(shinyWidgets)
get_data <- function() as.list(mtcars[sample(nrow(mtcars), size=3), sample(ncol(mtcars), size=3)])
logg <- function(...) message(paste0("[", format(Sys.time()), "] ", ...))
shinyApp(
ui = fluidPage(
title = "Hello",
checkboxGroupButtons(inputId = "cb", label = "label:", choices = c("unk"), selected = NULL,
status = "primary", size = "xs"),
br(),
textOutput("txt"),
br(),
textAreaInput("txtarea", NULL, rows = 4)
),
server = function(input, output, session) {
data <- reactive({
logg("in 'data'")
invalidateLater(3000, session)
get_data()
})
observe({
logg("in 'observe'")
req(length(data()) > 0)
updateCheckboxGroupButtons(session = session, inputId = "cb", choices = names(data()))
updateTextAreaInput(session, "txtarea", value = paste(capture.output(str(data())), collapse = "\n"))
})
output$txt <- renderPrint({
logg("in 'txt'")
req(length(data()) > 0)
str(data())
})
}
)
Notice that the definition of checkboxGroupButtons starts with no real choices. I'd prefer to start it empty, but unlike selectInput and similar functions, it does not like starting with an empty vector. It is quickly (nearly-immediately) changed, so I do not see "unk" in the interface.
I demoed two options for "displaying" the data in its raw form: as an output "txt", and as an updatable input "txtarea". I like the latter because it deals well with fixed-width, but it requires an update* function (which is really not a big deal).