Using an IF statement on a reactive variable inside an observe context - r

Question

I am trying to create a shiny flexdashboard within R Markdown that will display different tables based on user input. Some tables are formattable, others are just regular tables. The following code will do the rendering of the table if the tables in a list are all Formattables.
num <- reactive(as.integer(input$qualityDataNum))
renderFormattable(qualityData[[num()]])
But since some are not formattables, I want to check before doing the render. Instead of putting the tables in a list I created the following code to pick the table based on user input. It doesn't work, I get a warning: "Error in if: argument is of length zero".
num <- reactive(as.integer(input$qualityDataNum))
observe({
if (num() == 1) {
renderFormattable(qualityData1)
} else {
renderTable(qualityData2)
}
})
The complete code is below (since this is RMarkdown code it was reformated when
the three ticks were used for code marking. Sorry.):
title: "Dashboard Prototype"
output: flexdashboard::flex_dashboard
runtime: shiny
# allow sharing of dashboard
library(datasets)
library(flextable)
library(formattable)
library(dplyr)
options(stringsAsFactors = FALSE)
qData <- data.frame(Name = "AAA", Releases = 10, Coverage = 23.0)
qData <- bind_rows(qData, data.frame(Name = "BBB", Releases = 35, Coverage = 88.0))
#Using Formattable
data_formatter_dd <-
formattable::formatter("span", style = x ~ style( font.weight = "bold",
color = ifelse(x > 80.0 & x <= 100.0, "green", ifelse(x > 50.0 & x <= 80.0, "orange", "red"))))
qualityData1 <- formattable::formattable(qData, align = c("l", rep("r", ncol(qData) - 1)),
list('Name' = formattable::formatter("span", style = ~ style(color = "grey", font.weight = "bold")),
'Coverage' = data_formatter_dd))
#Using flextable
qualityData2 <- flextable(head(qData, col_keys = c("Name", "Releases", "Coverage")))
Sidebar {.sidebar}
# shiny inputs
selectInput("qualityDataNum", label = h3("Quality Number Set"), choice = list("1" = 1, "2" = 2),
selected = 1)
Quality Statistics
Quality
num <- reactive(as.integer(input$qualityDataNum))
observe({
req(num())
if (num() == 1) {
renderFormattable(qualityData1)
} else {
renderTable(qualityData2)
}
})

Answer 1

For Standalone Shiny Applications:
In order to achieve what you desire easily i would recommend to generate an output for each plottype and only populate it when the corresponding input is selected.
Down below you can find a complete example of this.
Generally speaking it is not advised to mix outputTypes (different render functions on the server side) into one output function (on the UI side)
# allow sharing of dashboard
library(datasets)
library(flextable)
library(formattable)
library(dplyr)
library(shiny)
options(stringsAsFactors = FALSE)
qData <- data.frame(Name = "AAA", Releases = 10, Coverage = 23.0)
qData <- bind_rows(qData, data.frame(Name = "BBB", Releases = 35, Coverage = 88.0))
data_formatter_dd <-
formattable::formatter("span", style = x ~ style( font.weight = "bold",
color = ifelse(x > 80.0 & x <= 100.0, "green", ifelse(x > 50.0 & x <= 80.0, "orange", "red"))))
qualityData1 <- formattable::formattable(qData, align = c("l", rep("r", ncol(qData) - 1)),
list('Name' = formattable::formatter("span", style = ~ style(color = "grey", font.weight = "bold")),
'Coverage' = data_formatter_dd))
qualityData2 <- flextable(head(qData, col_keys = c("Name", "Releases", "Coverage")))
ui <- fluidPage(
fluidRow(
selectInput("qualityDataNum", label = h3("Quality Number Set"), choice = list("1" = 1, "2" = 2),
selected = 1)
,
tableOutput("table2"),
formattableOutput("table1")
)
)
server <- function(input, output, session) {
output$table1 <- renderFormattable({
req(as.integer(input$qualityDataNum) == 1)
qualityData1
})
output$table2 <- renderTable({
req(as.integer(input$qualityDataNum) != 1)
qualityData2$body$dataset
})
}
shiny::shinyApp(ui,server)
EDIT:
For RMarkdown Code
I just rechecked your question and noticed that you were specifically asking how to handle the problem in an RMarkdown document and not in a standalone shiny Application. So here is an Update answer:
For RMarkdown it is enough that you update the code in your Quality Statistics section to the following:
renderFormattable({
req(as.integer(input$qualityDataNum) == 1)
qualityData1
})
renderTable({
req(as.integer(input$qualityDataNum) != 1)
qualityData2$body$dataset
})
Here you render the table according to the select input, it would be possible to render also multiple tables using this approach at the same time. This works since the req() functions checks if the requirement is fullfilled and only proceeds with the procession (aka the rendering of the plot) when it does. If it doesn't just an empty object is returned. Since we are using an reactive input inside the render function the expression get evaluate every time the Input value changes. If you do not want that consider wrapping it with an isolate() function, which tells the encapsulation function to not revaluate the function every time the encapsulate reactive Values/Inputs changes.