I am trying to use JohnCoene/marker package to highlight sections of text in a shiny app. My intend is to first generate the text using some server logic and display it with textOutput. However, I am struggeling with how to trigger the marker after the text appeared on the website. Putting it in the same observeEvent() does not work.
Here is my reprex
# remotes::install_github("johncoene/marker")
library(shiny)
library(marker)
ui <- fluidPage(
use_marker(),
actionButton("click", "click"),
textOutput("text_to_mark")
)
server <- function(input, output) {
observeEvent(input$click,
{
output$text <- renderText("My house is yellow")
})
# observeEvent() below does not work. This is just for illustration
observeEvent(input$text_to_mark,
{
marker <- marker$new("#text_to_mark.shiny-text-output.shiny-bound-output")
marker$
unmark()$ # unmark all before we mark
mark("My house")
})
}
# Run the application
shinyApp(ui = ui, server = server)
Created on 2019-10-10 by the reprex package (v0.3.0)
For illustration: I can get the marker to work, by adding a second button as in the code below, but I am look for a solution where it gets triggered when the text appears.
# remotes::install_github("johncoene/marker")
library(shiny)
library(marker)
ui <- fluidPage(
use_marker(),
actionButton("click", "click"),
textOutput("text_to_mark"),
actionButton("mark", "Mark!")
)
server <- function(input, output) {
observeEvent(input$click,
{
output$text_to_mark <- renderText("My house is yellow")
})
observeEvent(input$mark,
{
marker <- marker$new("#text_to_mark.shiny-text-output.shiny-bound-output")
marker$
unmark()$ # unmark all before we mark
mark("My house")
})
}
# Run the application
shinyApp(ui = ui, server = server)
Created on 2019-10-10 by the reprex package (v0.3.0)
You could listen on DOM changes with javascript: Is there a JavaScript / jQuery DOM change listener?.
When a change happens you can check if your target element has text:
hasText = document.getElementById("text_to_mark").innerHTML != ""
Note that i assume that your element has the id "text_to_mark".
The result you can "send to R" with
Shiny.onInputChange("hasText", hasText);
On the R side you will know if the element has text via listening on input$hasText.
So you can add:
observeEvent(input$hasText,{
...
})
The javascript you can add to your app with tags$script(jsCode) or use shinyjs.
A reproducible example:
library(shiny)
library(marker)
jsCode <- '
MutationObserver = window.MutationObserver || window.WebKitMutationObserver;
var observer = new MutationObserver(function(mutations, observer) {
console.log(mutations, observer);
hasText = document.getElementById("text_to_mark").innerHTML != ""
Shiny.onInputChange("hasText", hasText);
});
observer.observe(document, {
subtree: true,
attributes: true
});
'
ui <- fluidPage(
use_marker(),
tags$script(jsCode),
actionButton("click", "click"),
textOutput("text_to_mark"),
actionButton("mark", "Mark!")
)
server <- function(input, output) {
observeEvent(input$click, {
output$text_to_mark <- renderText("My house is yellow")
})
observeEvent(input$hasText,{
marker <- marker$new("#text_to_mark.shiny-text-output.shiny-bound-output")
marker$
unmark()$ # unmark all before we mark
mark("My house")
})
}
# Run the application
shinyApp(ui = ui, server = server)
Note that this only works on the first appearance of the text. If you also want to listen for changes of the text, one could send the text to R instead and check on the R side if the text was updated. Not sure if it is needed here.
Listening on DOM changes is one option, but your approach already shows that there is a pure shiny (non-custom-JS) solution, it only takes one click more, so the question is how to do it with only one click. I suggest using invalidateLater and wrap it in an if statement to prevent it from running over and over again like seen here.
The trick is to run your marker calls in an observe statement. Include the invalidateLater here and wrap it in an if condition with a counter which counts how many times the statement has been executed. Play around with the number of milliseconds and counts, in my case it works fine with if(isolate(val$cnt) < 1) and invalidateLater(1000). Don't forget to wrap your counter in an isolate otherwise it will get stuck in a loop.
Note also that input$click not only writes the text into a reactiveValue but also resets the counter val$cnt to 0 so that you can use the invalidateLater again on a new text. The same procedure will help you if you want to update your text using an observeEvent or the like. Just make sure to also reset the counter to 0 and the highlighting works on your new text part.
# remotes::install_github("johncoene/marker")
library(shiny)
library(marker)
ui <- fluidPage(
use_marker(),
actionButton("click", "click"),
textOutput("text_to_mark")
)
server <- function(input, output) {
val <- reactiveValues(cnt = 0,
text = NULL)
observeEvent(input$click, {
val$text <- "My house is yellow"
val$cnt <- 0
})
observe({
if(isolate(val$cnt) < 1) {
invalidateLater(1000)
}
marker <- marker$new("#text_to_mark.shiny-text-output.shiny-bound-output")
marker$
unmark()$ # unmark all before we mark
mark("My house")
val$cnt = isolate(val$cnt) + 1
})
output$text_to_mark <-renderText({
val$text
})
}
# Run the application
shinyApp(ui = ui, server = server)
Related
I'm trying to render some UI elements in Shiny on the client side using Javascript. I want the user to click a button which reveals a hidden panel. Once the panel is revealed, it should show a plot. When the button is clicked the panel becomes visible but the plot doesn't render.
I know this behaviour is achievable using server-side functions like observeEvent but the goal here is to leverage the client directly via JS.
Below is a reprex. For ease, I've inlined the JS and CSS.
library(shiny)
# Inline js to toggle display of element when button is clicked
js <- '
$(document).ready(function() {
document.getElementById("reveal").addEventListener("click", event => {
var panel = document.getElementById("init-hidden-panel");
panel.style.display = "block";
});
})
'
ui <- fluidPage(
tags$script(HTML(js)),
actionButton("reveal", "Reveal"),
div(
id = "init-hidden-panel",
plotOutput("plot"),
style = "display: none; background-color: darkgray" # initialize element as hidden
)
)
server <- function(input, output) {
output$plot <- renderPlot({
hist(iris$Sepal.Length)
})
}
# Run the application
shinyApp(ui = ui, server = server)
We can tell that the JS is in fact toggling the display because the gray background appears.
Perhaps you can use this method:
ui <- fluidPage(
actionButton("reveal", "Reveal"),
plotOutput("plot")
)
server <- function(input, output) {
plot_data <- reactiveValues(trigger = 0)
observe({
req(input$reveal)
isolate(plot_data$trigger <- plot_data$trigger + 1)
})
output$plot <- renderPlot({
input$reveal
if (plot_data$trigger %% 2 == 0) {
return(NULL)
} else {
hist(iris$Sepal.Length)
}
})
}
# Run the application
shinyApp(ui = ui, server = server)
In running the below code, each click of the actionButton() is correctly counted and output to the UI (using reactive value y). I'm trying to do the same with selectInput() (counting each time a selection is made)(using reactive value x), but it doesn't completely correctly count because when first invoking the App, selecting the default choice of "Cyl" is not included in the count (I'd like it to), and clicking the same choice more than once isn't included in the count (I'd like all clicks counted). One workaround is to include "multiple = TRUE" in the selectInput(), but I'd like to see if there's a JS solution instead so I don't have to change the UI the way including "multiple = TRUE" does.
As an aside, using "multiple = TRUE" also corrects the weird subtraction of 1 from (x) in output$clickSelInput...; hopefully a JS solution does the same.
Code:
library(shiny)
ui = fluidPage(hr(),
selectInput("selInput",label=NULL,c("Cyl"="cyl","Trans"="am","Gears"="gear"),selected=NULL),
actionButton("addBtn","Add"), hr(),
textOutput("clickSelInput"),
textOutput("clickAddBtn"),
tableOutput("data")
)
server = function(input, output) {
x = reactiveVal(0)
y = reactiveVal(0)
output$data <- renderTable({mtcars[1:10, c("mpg", input$selInput), drop = FALSE]})
observeEvent(input$selInput,{x(x()+1)})
observeEvent(input$addBtn,{y(y()+1)})
output$clickSelInput <- renderText({paste('Select Input clicks =',x()-1)})
output$clickAddBtn <- renderText({paste('Add Button clicks =',y())})
}
shinyApp(ui, server)
Here is a working solution, taking SamR's suggestion into account and using that example by analogy. I had to make other changes to make it work. I commented changes below from OP.
library(shiny)
library(shinyjs) # added
ui = fluidPage(hr(),
useShinyjs(), # added, this line is easy to forget to add
uiOutput("selInput"), # added, pulls renderUI output from server into UI
actionButton("addBtn","Add"), hr(),
textOutput("clickSelInput"),
textOutput("clickAddBtn"),
tableOutput("data")
)
server = function(input, output) {
x = reactiveVal(0)
y = reactiveVal(0)
output$data <- renderTable({mtcars[1:10, c("mpg", input$selInput), drop = FALSE]})
observeEvent(input$addBtn,{y(y()+1)})
# moved select input into renderUI so it can interact with JS in the observe further down:
output$selInput <- renderUI(
selectInput("selInput",label=NULL,c("Cyl"="cyl","Trans"="am","Gears"="gear"),selected=NULL)
)
output$clickSelInput <- renderText({paste('Select Input clicks =',input$rnd)}) # changed
output$clickAddBtn <- renderText({paste('Add Button clicks =',y())})
# added js:
observe({
if(is.null(input$rnd)){
runjs("
var click = 0;
Shiny.onInputChange('rnd', click)
var selInput = document.getElementById('selInput')
selInput.onclick = function() {click += 1; Shiny.onInputChange('rnd', click)};
")
}
})
}
shinyApp(ui, server)
Summary: When I put a shinyTree (from the R package of the same name) in a dropdownButton (from the shinyWidgets package), it seems that the tree in question is not accessible from the input variable until the user has interacted with the tree in some way. Is there a way I can force the tree to load when the app is loaded at the start?
Here is a minimal (almost) example:
library(shiny)
library(shinyWidgets)
library(shinyTree)
library(magrittr)
render.as.text <- function(t) {
if (is.null(t)) {
'argument is NULL'
} else {
q <- get_selected(t)
if (length(q) == 0) {
'none selected'
} else {
paste(q, sep=', ', collapse=', ')
}
}
}
ui <- fluidPage(
dropdownButton(shinyTree('tree', checkbox=TRUE), circle=FALSE, inputId='dd')
)
server <- function(input, output, session) {
output$tree <- renderTree({ list(OptionA = structure(list(OptionAA='', OptionAB=''), stopened=TRUE), OptionB='') })
updateActionButton(session, 'dd', label='initial label')
observe({
updateActionButton(session, 'dd', label=render.as.text(input$tree))
}) %>% bindEvent(input$dd_state)
}
shinyApp(ui = ui, server = server)
When I run this app, the button begins with the text 'initial label' (as expected), then when I click on it the first time the text changes to 'argument is NULL' (showing that the variable input$tree has a NULL value), but the tree displays correctly in the dropdown. If I then click the button again to close the dropdown, the text 'argument is NULL' remains - it remains until I actually select one or more of the tree's checkboxes after which (and after clicking the button again) it will display either a comma-separated list of selected options or 'none selected' as expected.
I would very much like to be able to refer to input$tree beginning with the first time the user takes any action in the app - ideally I am looking for some command to insert in either the server or ui code (or both) which will mean that the first time the user clicks the button its text will change directly to 'none selected' because input$tree is no longer NULL. Is this possible?
Would just changing the default of your render.as.text function work for your purpose?
library(shiny)
library(shinyWidgets)
library(shinyTree)
library(magrittr)
render.as.text <- function(t) {
if (is.null(t)) {
'none selected'
} else {
q <- get_selected(t)
if (length(q) == 0) {
'none selected'
} else {
paste(q, sep=', ', collapse=', ')
}
}
}
ui <- fluidPage(
dropdownButton(shinyTree('tree', checkbox=TRUE), circle=FALSE, inputId='dd')
)
server <- function(input, output, session) {
output$tree <- renderTree({ list(OptionA = structure(list(OptionAA='', OptionAB=''), stopened=TRUE), OptionB='') })
updateActionButton(session, 'dd', label='initial label')
observe({
updateActionButton(session, 'dd', label=render.as.text(input$tree))
}) %>% bindEvent(input$dd_state)
}
shinyApp(ui = ui, server = server)
I have a problem with this application, containing a javascript file for the user-interface part, which enables more tabs. However, the server part doesn't work when we have included the javascript file. I have a simple reactivity regarding the mtcars dataset here to showcase the problem. when I disable the "includeScript("script.js")", by merely put the # in front of it, the app works, so the problem is connected to this part. So my question would be, How can I fix this issue and also keep the javascript part in the shiny app.
Grateful for all your help.
Edit:
Now I have replaced includeScript("script.js") with tags$head(tags$script(src="./script.js")), and it appears to work, but with an extremely slow reactivity, I have to wait almost 1-2 min before seeing something. Any suggestion, or do you also experience this ?
library(shiny)
library(shinythemes)
library(shinymanager)
library(dplyr)
script.js
$(document).ready(function(){
$('.dropdown').on('click', function(e){
$(this).toggleClass('open');
e.stopPropagation();
e.preventDefault();
});
$('[data-toggle=tab]').on('click', function(e){
let dv = ($(this).attr('data-value'));
//Set active element in tabcontents
$('.tab-pane').removeClass('active');
$('.tab-pane[data-value="' + dv + '"]').addClass('active');
//Set active element in navbar
$('a[data-toggle=tab]').parent().removeClass('active');
$('a[data-value="' + dv + '"]').parent().addClass("active");
//Close the dropdowns
$('.dropdown').removeClass('open');
e.stopPropagation();
e.preventDefault();
});
});
Credentials
credentials <- data.frame(
user = c("Jhon", "Erik"), # mandatory
password = c("1", "1"), # mandatory
start = c("2022-02-14"), # optinal (all others)
expire = c(NA, "2022-12-31"),
admin = c(TRUE, TRUE),
comment = "Model Performance application",
stringsAsFactors = FALSE
)
Ui
ui <- fluidPage(
includeScript("script.js"),
navbarPage("Shiny",
collapsible = TRUE,
theme = shinytheme('yeti'),
tabPanel("Information" ,icon = icon("info"),
tags$h2("Information about the current user"),
verbatimTextOutput("auth_output")
),
tabPanel("Simulation 1",
tags$h2("Simulation"),
tags$hr(),
selectInput("vars", "Variables", names(mtcars), multiple = T),
tableOutput("data")
),
tabPanel("Upload",icon = icon("upload"),
tags$h2("Upload datasets"),
tags$hr(),
),
tabPanel("Simulation 2",
tags$h2("Simulation"),
tags$hr()
),
navbarMenu("Statistical outputs",
tabPanel("One"
),
tabPanel("Two"
),
tabPanel("Three"
),
tabPanel("Four"
),
tabPanel("Report"
),
navbarMenu("More",
tabPanel("Statistical", icon = icon("info")
),
tabPanel("Info",
icon = icon("info-circle")
),
tabPanel("Subpart 4", "Subpart 4"),
tabPanel("Subpart 5", "Subpart 5")
)
)
)
)
Wrap your UI with secure_app
ui <- secure_app(ui)
Server
server <- function(input, output, session) {
# call the server part
# check_credentials returns a function to authenticate users
res_auth <- secure_server(
check_credentials = check_credentials(credentials)
)
output$auth_output <- renderPrint({
reactiveValuesToList(res_auth)
})
output$data <-renderTable({
req(input$vars)
mtcars %>% select(all_of(input$vars))
})
}
shiny::shinyApp(ui, server)
Update:
Most important in addition to 1. and 2. from the first answer. The app works as desired only if split in ui and server part!
It seems that the server part is not working but after clicking on Statistical outputs the table appears!
First answer:
Put your script.js into a www folder. This should be in the same folder where your app is.
Change includeScript("script.js"), in ui part with tags$head(tags$script(src="script.js")),
I am builing a RShiny app where I need to create vanilla/plain html buttons and give basic functionality to them using JavaScript. In my original app I have a htmlOutput (or uiOutput) element containing these buttons (they are generated dynamically based on user input). Unfortunately JavaScript is not working properly inside this htmlOutput element and I cant figure out why. Please see my minimal reproducible example (app.R):
library(shiny)
library(shinyjs)
# define ui
ui <- fluidPage(
useShinyjs(),
tags$head(
tags$script(
HTML(
"window.onload = function(){
var coll = document.getElementsByClassName('testclass');
var i;
console.log(coll);
for (i = 0; i < coll.length; i++) {
coll[i].addEventListener('click', function() {
alert('clicked');
});
};
};
")
)
),
mainPanel(
# normal button (working)
tags$button(
type="button",
class="testclass",
"Click to alert (button inside main panel)"
),
# html output button (not working)
htmlOutput("html_out")
)
)
# define server
server <- function(input, output) {
# generate html output button (problematic with JS)
output$html_out <- renderUI({
tags$button(
type="button",
class="testclass",
"Click to alert (button inside htmlOutput)"
)
})
}
# run app
shinyApp(ui = ui, server = server)
The tags$button() element is working without problem if it is added statically into the main panel. But if the same tags$button() element is added via the htmlOutput it is not working with the JavaScript code. Is there any explanation or workaround for that?
The only difference in the html output code is that the htmlOutput element is wrapped inside a div with class = "shiny-html-output shiny-bound-output". And I know that I usually shouls use actionButton() but in my case this is not possible.
Thanks for your help!
The problem is that the initial JS in head is run when the app starts but the second button isn't available immediately. you can add the JS code directly to the HTML code
library(shiny)
library(shinyjs)
# define ui
ui <- fluidPage(
useShinyjs(),
tags$head(
tags$script(
HTML(
"window.onload = function(){
var coll = document.getElementsByClassName('testclass');
var i;
console.log(coll);
for (i = 0; i < coll.length; i++) {
coll[i].addEventListener('click', function() {
alert('clicked');
});
};
};
")
)
),
mainPanel(
# normal button (working)
tags$button(
type="button",
class="testclass",
"Click to alert (button inside main panel)"
),
# html output button (not working)
htmlOutput("html_out")
)
)
# define server
server <- function(input, output) {
# generate html output button (problematic with JS)
output$html_out <- renderUI({
tags$button(
type="button",
class="testclass",
# ADD JS HERE
onclick = "alert('clicked');",
"Click to alert (button inside htmlOutput)"
)
})
}
# run app
shinyApp(ui = ui, server = server)