How to retain the filtered (or different page) view of a datatable after editing in a Shiny app?

I am writing a Shiny app where I am rendering an editable datatable. In some instances I would like the user to filter the table or go to a different page and then make edits to the table. After this editing, I would like the current view of the table to remain while showing the edits in the table. The editing of a column also changes the values in another column, so both these edits need to be displayed while retaining the view.

This is the code I have come up with so far

library(shiny)
library(DT)
# create a sample data frame with 50 entries
df <- data.frame(
name = paste0("Person ", 1:50),
age = sample(20:60, 50, replace = TRUE),
salary = sample(2000:5000, 50, replace = TRUE),
stringsAsFactors = FALSE
)
# create a shiny app
ui <- fluidPage(
DTOutput("table")
)
server <- function(input, output, session) {
# create a reactiveValues object to store the data frame
rv <- reactiveValues()
rv$df <- df
# render the table
output$table <- renderDT({
datatable(
rv$df,
rownames = F,
editable = T,
filter = "top",
selection = "none",
extensions = c(
"ColReorder",
"Buttons"
),
options = list(
dom = "lBRrftpi",
autoWidth = F,
pageLength = 20,
scrollX = F,
ColReorder = T,
buttons = list(list(extend = 'csv', filename= 'data'), 'print', 'copy')
)
)
})
# update the table when edited
observeEvent(input$table_cell_edit, {
info <- input$table_cell_edit
i <- info$row
j <- info$col
v <- info$value
new_i <- i
new_j <- j + 1
new_val <- v * 100
new_info <- data.frame(
row = new_i,
col = new_j,
value = new_val
)
rv$df <<- editData(rv$df, info, rownames = F)
if(j == 1){
rv$df <<- editData(rv$df, new_info, rownames = F)
}
replaceData(proxy = dataTableProxy("table"), data = rv$df, resetPaging = FALSE)
})
}
# run the app
shinyApp(ui, server)

While the edits in the table and the dependant changes in the other column do work, the table view resets as soon as I edit a cell. Can someone please help me with code that retains the view after editing?

What you observe is occuring because the data object you use is reactive and you update it, and then the table is re-rendered when the update occurs. The cool thing with a proxy is that you don't need a reactive data object. Also, you can pass the proxy object to editData and then there's no need of replaceData.

library(shiny)
library(DT)

# create a sample data frame with 50 entries
df <- data.frame(
  name = paste0("Person ", 1:50),
  age = sample(20:60, 50, replace = TRUE),
  salary = sample(2000:5000, 50, replace = TRUE),
  stringsAsFactors = FALSE
)

# create a shiny app
ui <- fluidPage(
  DTOutput("table")
)

server <- function(input, output, session) {

  # render the table
  output$table <- renderDT({
    datatable(
      df,
      rownames = F,
      editable = T,
      filter = "top",
      selection = "none",
      extensions = c(
        "ColReorder",
        "Buttons"
      ),
      options = list(
        dom = "lBRrftpi",
        autoWidth = F,
        pageLength = 20,
        scrollX = F,
        ColReorder = T,
        buttons = list(list(extend = 'csv', filename= 'data'), 'print', 'copy')
      )
    )
  })
  
  proxy <- dataTableProxy("table")
  
  # update the table when edited
  observeEvent(input$table_cell_edit, {
    info <- input$table_cell_edit
    i <- info$row
    j <- info$col
    v <- info$value
    
    new_i <- i
    new_j <- j + 1
    new_val <- v * 100
    new_info <- data.frame(
      row = new_i,
      col = new_j,
      value = new_val
    )
    
    df <<- editData(df, info, proxy, rownames = F)
    
    if(j == 1){
      df <<- editData(df, new_info, proxy, rownames = F)
    }
    
  })
  
}