I am about to create plots in a dynamic way using R shiny and the ggplot2 library. The function I use for the plot creation is ggplot() and I need to display the plot on a tab inside of a tabset.

The position of the plot is fixed but the plot type, the columns for x and y, even the number of data columns and the astethic parameters of aes mapping such as colour, fill, etc. are stored and mapped in the variable “aes_mapping”. The plot type is provided as a character “plot_type”. The data is given as a data frame “data”.

In the end the plot has to be shown in the object “output[[plot_id]]” using the renderPlot function.

I tried to create the functions for aes mapping danamically by using do.call:

    output[[plot_id]] <- renderPlot({
      # convert ot data frame
      data_df <- as.data.frame(data)
      # create plot
      p <- ggplot(data_df) + do.call(get(plot_type), list(mapping = aes_mapping))
    })

This leads to an error:

Error in as.vector(x, "character") : 
  cannot coerce type 'environment' to vector of type 'character'

I tried to define an environment function but the error persists.

    output[[plot_id]] <- renderPlot({
      data_df <- as.data.frame(data)
      plot_function <- get(plot_type, mode = "function", envir = asNamespace("ggplot2"))
      p <- ggplot(data_df) + do.call(get(plot_type), list(mapping = aes_mapping))
    })

When I use browser() to stop just before the output[[plot_id]]… block and try the lines in the console then I can create the plot without any error and display it with print(p).

Then, I tried to find out more about the variables:

Browse[1]> class(data_df)
[1] "data.frame"

Browse[1]> class(plot_type)
[1] "character"

Browse[1]> class(aes_mapping)
[1] "uneval"

Browse[1]> print(aes_mapping)
Aesthetic mapping: 
* `alpha`  -> NULL
* `colour` -> NULL
* `fill`   -> NULL
* `group`  -> NULL
* `shape`  -> NULL
* `size`   -> NULL
* `x`      -> `msglvl_times_readable`
* `y`      -> `msglvl_seqNumber`

I tried only to use the mandatory requested parameters x and y:

Browse[1]> m <- list(aes_mapping$x, aes_mapping$y)
Browse[1]> class(m)
[1] "list"

Browse[1]> print(m)
[[1]]
<quosure>
expr: ^msglvl_times_readable
env:  0x0000019d7aa09328

[[2]]
<quosure>
expr: ^msglvl_seqNumber
env:  0x0000019d7aa09328

And I tried to find out more about the classes fo the aes string:

Browse[1]> plot_function <- get(plot_type, mode = "function", envir = asNamespace("ggplot2"))
Browse[1]> c <- plot_function(mapping = aes_mapping)
Browse[1]> class(c)
[1] "LayerInstance" "Layer"         "ggproto"       "gg"    

The error persists when I let the script run and it is valid and the plot is created when I do it by console command.

Does someone see what I am doing wrong? Is it possible to create a plot that dynamically when using a shiny app?

This should be a running code which is producing the error. But it isn’t. I can’t reproduce it in a smaller way than the original which is quite large. But the variables are created correctly:

library(shiny)
library(ggplot2)
library(data.table)

# define ui object
ui <- fluidPage(
  sidebarLayout(
    sidebarPanel(
      actionButton(inputId = "action", label = "Action")
    ), 
    mainPanel(
      tabsetPanel(
        id = "tabset1",
        tabPanel("plot_1",
                 plotOutput(outputId = "plot_1"))
      )
    )
  )
)

server <- function(input, output, session) {
  
  observeEvent(input$action,{
    browser()
    data <- data.table(
      msglvl_times_readable= 1:10,
      msglvl_seqNumber= rnorm(10),
      z_var = runif(10)
    )
    
    
    settings <- list(
      plot_1 = list(
        "geom_point",
        list("msglvl_times_readable",
             "msglvl_seqNumber"),
        list(NA, NA, NA, NA, NA, NA)
      )
    )
    
    mapping_param <- list(
      "geom_point",
      c("x", "y"),
      c("alpha", "colour", "fill", "group", "shape", "size")
    )
    
    plot_id <- "plot_1"
    
    #create plot
    # get plot type
    plot_type <- settings[[plot_id]][[1]]
    # get mpp
    mpp <- mapping_param[[2]]
    # get mppv
    mppv <- settings[[plot_id]][[2]]
    # get opp
    opp <- mapping_param[[3]]
    # get oppv
    oppv <- settings[[plot_id]][[3]]
    # create aes mapping
    aes_mapping_list <- setNames(c(mppv, oppv), c(mpp, opp))
    # create aes dynamically
    aes_mapping <- do.call(aes_string, aes_mapping_list)
    
    output[[plot_id]] <- renderPlot({
      data_df <- as.data.frame(data)
      plot_function <- get(plot_type, mode = "function", envir = asNamespace("ggplot2"))
      p <- ggplot(data_df) + do.call(get(plot_type), list(mapping = aes_mapping))
    })
  })
}

# start ui
shinyApp(ui, server)