Error handling for dynamic UI in Shiny app

shiny
,
1

I am having trouble with a Shiny app allowing dynamic changes in the display for one input (selectInput) based on the value of a second input (radioButtons). To clarify, I wrote a simplified version of my app relying on the Palmer penguins data producing the same error. I would like users to select multiple species from the data set, using either the scientific name or common name (selected with a radio button), for graphical summary. As far as I can tell, it looks like the species names are correctly updated using observeEvent and updateSelectInput, but the subsequent data selection isn't happening properly for some reason. When I try to plot the data, I receive the following error: ! Column scientificName not found in .data.

I would appreciate any insights about the error and strategies to fix it!

library(bslib)
library(palmerpenguins)
library(shiny)
library(tidyverse)
data(penguins)
penguins <- penguins |>
    rename(commonName=species) |>
    mutate(scientificName=factor(c(rep("Pygoscelis adeliae",152),rep("Pygoscelis papua",124),rep("Pygoscelis antarcticus",68))))
ui <- page_sidebar(
    title = "Penguins Reprex",
    sidebar = sidebar(selectInput('species','Choose species',choices=unique(penguins$scientificName),
                                  selected=first(unique(penguins$scientificName)),multiple=TRUE),
                      radioButtons('name','Choose name display',choiceNames=c("Scientific name","Common name"),
                                   choiceValues=c("scientificName","commonName"))),
    plotOutput("plot",width="100%")
)
server <- function(input, output, session) {
    data <- reactive({
        penguins |>
            filter(input$name %in% input$species) |> 
            group_by(input$name,year) |>
            summarize(n = n())
    })
    output$plot <- renderPlot(
        ggplot(data(), aes(x=.data[[input$name]],y=n,fill=year)) +
            geom_bar(position="stack",stat="identity",width=0.7) +
            xlab("Species") + ylab("Observations") +
            theme_nice() + 
            theme(axis.text.x = element_text(angle=45,vjust=1,hjust=1)) +
            scale_fill_custom() + 
            guides(fill=guide_legend(title="Years"))
    )
    observeEvent(input$name,{ updateSelectInput(session,'species','Choose species',choices=unique(penguins[[input$name]]),
                                                selected=first(unique(penguins[[input$name]]))) },
                 ignoreInit=TRUE)
}
shinyApp(ui,server)
2

Your issue concerns metaprogramming with tidyverse:
i.e.

data <- reactive({
        penguins |>
            filter(input$name %in% input$species) |> 
            group_by(input$name,year) |>
            summarize(n = n())
    })

should be

  data <- reactive({
    penguins |>
      filter(!!sym(input$name) %in% input$species) |>
      group_by(!!sym(input$name),year) |>
      summarize(n = n())
  })

so that the result of input$name is interpreted by R as a symbol rather than as a character string

3

Thanks for the suggestion - it worked!

I also found that using .data[[input$name]] worked equally well.