Flexdashboard, rhandsontable: how to programmatically access user updated table?

Question

Not a Shiny programmer. Simple question. rhandsontable in Flexdashboard app. How to access a column updated by the user? Sample code:

---
title: "Test"
runtime: shiny
output: 
      flexdashboard::flex_dashboard:
      orientation: columns
      vertical_layout: fill
---

```{r setup, include=FALSE}
library(flexdashboard)
library(shiny)
require(dplyr)
require(tidyverse)
require(rhandsontable)

hour <- 1:24
required <- c(2, 2, 2, 2, 2, 2, 8, 8, 8, 8, 4, 4, 3, 3, 3, 3, 6, 6, 5, 5, 5, 5, 3, 3)
required <- as.integer(required)
on_duty <- as.integer(rep(0, 24))
start <- on_duty

df <- data.frame(hour, required, on_duty, start)

```
Inputs {.sidebar data-width=w}
-----------------------------------------------------------------------

```{r Inputs}

```

Column {data-width=200}
-----------------------------------------------------------------------

### Chart A

```{r}

global <- reactiveValues(df = df)

rHandsontableOutput("dftbl1")

    output$dftbl1 = renderRHandsontable({
    rhandsontable(global$df, selectCallback = TRUE, readOnly = FALSE)
    })

```

So the code renders the table. A user could update the table by editing the table cells. But then how to reference the updated table to pass table columns to a function called with an actionButton? The complex examples I've found are difficult to decipher. Appreciate any feedback. SteveM

Answer 1

you can use hot_to_r as in

modified_table <- reactive({
  hot_to_r(req(input$table_id)) ## req!
})

to get access to the current state of the table including modifications from the user. req is needed because hot_to_r can't deal with NULLs. table_id should be the output-id you use for the return value of renderRHandsontable.

output$table_id <- renderRHandsontable({ 
  rhandsontable(initial_table) ## need to call converter
})

The complex examles you are referring to (like this one, #64-81) allow a two-way connection of the tables in the sense that they can be updated both from the user and from the server. In this simple setup I outlined here however, modified_table is created with reactive so it can only be updated by the user.

I totally agree that this package could be made more user friendly by allowing NULL in hot_to_r and by automatically calling rhandsontable in renderRHandsontable if the return value is a data.frame but this is what you will have to work with.

Here is a full app demonstrating this setup

library(shiny)
library(rhandsontable)

ui <- fluidPage(
  rHandsontableOutput("table_id"),
  tableOutput("second_table")
)

server <- function(input, output, session) {
  initial_table <- head(iris)

  output$table_id <- renderRHandsontable({
    rhandsontable(initial_table) ## need to call converter
  })

  modified_table <- reactive({
    hot_to_r(req(input$table_id)) ## req!
  })

  output$second_table <- renderTable({
    modified_table()
  })
}

shinyApp(ui, server)

In order to access a specific column, you can use modified_table()$column_name inside a reactive context.