Use glyph plots in Shiny

glyphOutput() places a plot in a Shiny app's UI and renderGlyph() draws it from the server. renderGlyph() takes a plot from glyph(), a layout from compose(), or a widget from render().

What the user selects in the plot is sent back to the server as input$<outputId>_selected, holding row numbers of the data frame given to glyph(). See the section below and vignette("shiny", package = "glyph").

shiny is not installed with glyph. Install it to use these functions.

Usage

glyphOutput(outputId, width = "100%", height = "400px")

renderGlyph(expr, env = parent.frame(), quoted = FALSE)

Arguments

outputId

Output variable to read the plot from.

width, height

Size of the plot, as a CSS size such as "100%" or "400px", or a number of pixels.

expr

An expression that returns a glyph_spec, a glyph_layout or a widget made by render(). It can also return NULL, which shows nothing and clears the selection.

env

The environment in which to evaluate expr.

quoted

Is expr a quoted expression (with quote())? This is useful if you want to save an expression in a variable.

Value

glyphOutput() returns the UI element for the plot. renderGlyph() returns a render function to assign to an element of output.

Selections

A selection is sent when the user ends a brush, clicks a mark with interact(click = "select"), or clicks a legend entry added with selection(type = "legend"). It is not sent while a brush is being dragged. Clearing a brush (with the "Clear selection" button, Escape, or a click on an empty part of the plot) sends NULL, as does clicking a selected mark again. So does redrawing the plot, which happens when renderGlyph() runs again or the output is resized.

For a single plot the value is an integer vector of row numbers, so data[input$plot_selected, ] gives the selected rows. Row numbers are positions in the data frame, not row names. They are correct when rows with missing values were dropped and when a mapping is computed (such as x = log(wt)). Selecting a box, violin, histogram bin or polygon selects every row it summarises. A mark drawn from its own data takes no part in the value. A brush that takes in no rows gives integer(0).

A legend selects the rows whose entries are still shown. With every entry shown there is no selection, and the value is NULL.

A facet() grid has one data frame, so its value is one vector: the rows selected in any of its panels.

A compose() layout gives a list with one entry per panel, in the order the panels were given, named after them when compose() was called with names. Each entry holds the rows selected in that panel, as row numbers of that panel's data, or NULL when nothing is selected there. A panel that is itself a layout gives a list of the same shape. When no panel has a selection the whole value is NULL.

output$plots <- renderGlyph(compose(cars = p1, iris = p2))
# input$plots_selected is NULL, or for example
# list(cars = c(3L, 18L), iris = NULL)

Examples

if (interactive() && requireNamespace("shiny", quietly = TRUE)) {
  library(shiny)

  ui <- fluidPage(
    glyphOutput("plot"),
    tableOutput("rows")
  )

  server <- function(input, output) {
    output$plot <- renderGlyph(
      glyph(mtcars, x = wt, y = mpg) |>
        mark_point() |>
        interact(brush = TRUE)
    )
    output$rows <- renderTable(mtcars[input$plot_selected, ], rownames = TRUE)
  }

  shinyApp(ui, server)
}