Building Interactive Terminal Dashboards in R

A terminal dashboard keeps data and controls together inside a terminal. termr supplies widgets, layouts and events so application code can work with R data rather than terminal escape sequences. Start with the stable, in-memory table pattern from the verified data table recipe.

Build a small data explorer

Install termr, then save the following chunk as table-app.R. Only termr and base R are required.

library(termr)

items <- data.frame(
  item = c("Tea", "Coffee", "Apple"),
  units = c(5L, 8L, 3L),
  stringsAsFactors = FALSE
)

items_table <- data_table(
  items,
  id = "items",
  zebra = TRUE,
  header_sort = TRUE
)

application <- app(
  vertical(
    label("Inventory"),
    items_table,
    label("Use Up/Down to select a row. Click a header or press S to sort.", id = "selection")
  ),
  on("datatable.row_selected", "#items", function(event, app) {
    row <- event$data$value
    app$query_one("#selection")$update(
      sprintf("Selected: %s (%d units)", row$item, row$units)
    )
  }),
  bind("q", "quit")
)

vertical() reserves space for the title, table and selection label. Use horizontal() to arrange panels beside one another. Display widgets such as metric() and sparkline() can add summaries; see the widget guide.

Navigate and sort

Run in a real terminal:

Rscript -e "source('table-app.R'); termr::run(application)"

Up/Down move the row cursor. Click a header or press S to sort by the active column; repeated presses cycle ascending, descending and off. Press q to quit. The selection event provides the row as a named list in event$data$value; event$data$row is the original row number.

Check the interaction headlessly

After sourcing the app in a fresh session, run:

verify_table <- function() {
  pilot <- test_app(application, width = 60, height = 8)
  on.exit(pilot$stop(), add = TRUE)
  pilot$press("down")
  stopifnot(identical(pilot$query_one("#selection")$text,
                      "Selected: Coffee (8 units)"))
  pilot$press("s")
  stopifnot(identical(items_table$sort_state$column, "item"))
  stopifnot(grepl("Apple", pilot$screen_text()[[3]], fixed = TRUE))
}
verify_table()

The build executes these checks. The table formats and paints its visible viewport, but this example still holds the entire data frame in memory. Lazy table sources, database adapters and SQL editors are separate, experimental APIs. Database queries and page fetches are synchronous.

Extend an existing dashboard

For a larger, synthetic live dashboard, the repository includes terminal-dashboard.R. Run termr::run_example("terminal-dashboard") in a terminal. It combines metrics, sparklines, progress bars, a service table and an event log with pause and theme bindings. Its synthetic data is a demonstration, not a production monitoring backend.

See the data_table() reference, data tools guide, SQL workflows, original table recipe and testing tutorial for the next steps.