Server logic for the Figure Builder module

Powers the multi-panel Figure Builder module rendered by figureBuilderUI(). Users can add any VizModules plot module to a free-form A4 canvas, drag and resize each plot, filter each plot's data independently, label panels automatically, and export the whole figure as a single editable SVG (or bundle every plot's source data, HTML plot, and statistics into one .zip).

Usage

figureBuilderServer(id, data_list = NULL, module_registry = NULL)

Arguments

id

The ID for the Shiny module. Must match the id given to figureBuilderUI().

data_list

An optional named list of data frames that seed the dataset registry. If NULL (the default), the bundled example datasets (plus a sales_by_region summary suited to the pie plot) are used. At least one element is required. An element is either a data frame, or a named list of data frames for a module that needs companion tables (the ComplexHeatmap module's list(matrix = , column_annotations = )); in the latter case only the primary table is filtered and shown in the panel's table pane.

module_registry

An optional named list describing the plot modules to offer. If NULL (the default), all bundled VizModules modules are offered, each opening on the same example figure as in moduleGalleryApp(). Each entry is itself a list with components: label (character, shown in the picker), dataset (character, the dataset name its defaults were written for), inputs_ui, output_ui, and server_fn (the module's three functions), and defaults (a named list of input defaults applied only when dataset is the chosen dataset). defaults is passed to both inputs_ui and server_fn, so it can seed server-rendered controls such as the group color picker. An entry may also carry primary.table, naming which table of a multi-table dataset gets filtered (the first by default); its presence is also what marks the module as able to take a multi-table dataset at all, so modules without it are handed the primary table alone and any dataset stays usable with any module.

A module whose output is not a plotly graph can still contribute to the SVG figure export by attaching a vector_svg attribute to the reactive its server returns: a function(width, height, res) yielding an <svg> element drawn at that pixel size, which is spliced into the figure in place of the Plotly.toImage() result. draw_to_svg() builds one from any grid or base drawing; ComplexHeatmap_HeatmapServer() is the worked example. A panel whose module attaches nothing simply contributes no artwork.

The same renderers feed the source archive, which additionally wants a raster_png counterpart returning PNG bytes (see draw_to_png()). Either can be given as an attribute on the reactive, as here, or as a field on the summary list the reactive returns – the latter being order-independent, since the summary is rebuilt on every download. See create_source_download_handler().

Value

Invisibly returns NULL; called for its side effects (wiring up the Figure Builder module's reactive logic).

Details

Call this from your app's server with the same id you passed to figureBuilderUI(). Because it is a proper Shiny module, several Figure Builders can coexist on one page, each with its own namespace and canvas.

Author

Jared Andrews

Examples

library(VizModules)
if (interactive()) {
    ui <- fluidPage(figureBuilderUI("figure_builder"))
    server <- function(input, output, session) {
        figureBuilderServer("figure_builder")
    }
    shinyApp(ui, server)
}