grid.grab: Grab the current grid output

Description

Creates a gTree object from the current grid display list or from a scene generated by user-specified code.

Usage

grid.grab(warn = 2, wrap = FALSE, ...)
grid.grabExpr(expr, warn = 2, wrap = FALSE,
              width = 7, height = 7, device = offscreen, ...)

Arguments

expr

An expression to be evaluated. Typically, some calls to grid drawing functions.

warn

An integer specifying the amount of warnings to emit. 0 means no warnings, 1 means warn when it is certain that the grab will not faithfully represent the original scene. 2 means warn if there's any possibility that the grab will not faithfully represent the original scene.

wrap

A logical indicating how the output should be captured. If TRUE, each non-grob element on the display list is captured by wrapping it in a grob.

width, height

Size of the device used for temporary rendering.

device

A function that opens a graphics device for temporary rendering. By default this is an off-screen, in-memory device based on the pdf device, but this default device may not be satisfactory when using custom fonts.

…

arguments passed to gTree, for example, a name and/or class for the gTree that is created.

Value

A gTree object.

Details

There are four ways to capture grid output as a gTree.

There are two functions for capturing output: use grid.grab to capture an existing drawing and grid.grabExpr to capture the output from an expression (without drawing anything).

For each of these functions, the output can be captured in two ways. One way tries to be clever and make a gTree with a childrenvp slot containing all viewports on the display list (including those that are popped) and every grob on the display list as a child of the new gTree; each child has a vpPath in the vp slot so that it is drawn in the appropriate viewport. In other words, the gTree contains all elements on the display list, but in a slightly altered form.

The other way, wrap=TRUE, is to create a grob for every element on the display list (and make all of those grobs children of the gTree).

The first approach creates a more compact and elegant gTree, which is more flexible to work with, but is not guaranteed to faithfully replicate all possible grid output. The second approach is more brute force, and harder to work with, but should always faithfully replicate the original output.

Examples

Run this code

# NOT RUN {
pushViewport(viewport(w=.5, h=.5))
grid.rect()
grid.points(stats::runif(10), stats::runif(10))
popViewport()
grab <- grid.grab()
grid.newpage()
grid.draw(grab)
# }