tidyview

The goal of tidyview is to provide an extensible replacement for utils::View().

Installation

You can install the released version of tidyview from CRAN with:

install.packages("tidyview")

Example

By default, view() forwards to utils::View(). This isn't useful for rendering an rmarkdown document, so the first code shown here will be how to turn it off.

library(tidyview)
suppress_view()

Now we're safe to use view():

library(tidyverse)
#> ── Attaching packages ─────────────────────────────────────────────────── tidyverse 1.2.1 ──
#> ✔ ggplot2 2.2.1          ✔ purrr   0.2.5     
#> ✔ tibble  1.4.2.9002     ✔ dplyr   0.7.6     
#> ✔ tidyr   0.8.1          ✔ stringr 1.3.1     
#> ✔ readr   1.1.1          ✔ forcats 0.3.0
#> ── Conflicts ────────────────────────────────────────────────────── tidyverse_conflicts() ──
#> ✖ dplyr::filter() masks stats::filter()
#> ✖ dplyr::lag()    masks stats::lag()

view(mtcars)
#> Suppressed viewing of mtcars

mtcars %>%
  view() %>%
  nrow()
#> Suppressed viewing of .
#> [1] 32

Note the pipe-friendlyness -- the input is passed through, invisibly, and can be processed later on. In an interactive session, the code above would have opened two data viewers, via the default_view_handler() function:

default_view_handler
#> function(x, title) {
#>   View <- get("View", envir = as.environment("package:utils"))
#>   View(x, title)
#> }
#> <bytecode: 0x56424bdec790>
#> <environment: namespace:tidyview>

Custom view handlers can be made available through factories. The factory is a design pattern, see the Wikipedia article if you're not familiar with it.

1. A factory is registered via register_view_handler_factory().

2. Each time view() is called, all registered factories are consulted.

3. The first factory that returns a valid handler, i.e. a function similar to the default_view_handler() seen above, wins. The handler is called with the object.

4. If no factory takes responsibility, the default view handler is used.

The factory must be a pure function without (user-visible) side effects!

We register a view handler factory that outputs the title and the dimensions of the object for 2D objects, and does nothing for all other objects.

my_view_handler_factory <- function(x) {
  if (length(dim(x)) != 2) return (NULL)
  
  my_view_handler
}

my_view_handler <- function(x, title) {
  cat("Title: ", title, "\n", sep = "")
  cat("Dimensions: ", paste(dim(x), collapse = " x "), "\n", sep = "")
}

register_view_handler_factory(my_view_handler_factory)

Now, when we view() a data frame, we get console output:

view(mtcars)
#> Title: mtcars
#> Dimensions: 32 x 11

mtcars %>%
  view() %>%
  nrow()
#> Title: .
#> Dimensions: 32 x 11
#> [1] 32

mtcars %>%
  view("my title")
#> Title: my title
#> Dimensions: 32 x 11

(The dot . in the title is created by the pipe. If you need to distinguish views in a pipe-based workflow, use the title argument to view().)

Viewing a vector still doesn't do anything, this is how we designed our view handler factory:

view(1:10)
#> Suppressed viewing of 1:10

The implementation of suppress_view() shouldn't be surprising:

suppress_view
#> function() {
#>   register_view_handler_factory(void_view_handler_factory)
#> }
#> <bytecode: 0x56424a2c1020>
#> <environment: namespace:tidyview>

Factories are consulted in reverse registration order, calling suppress_view() again moves the void handler factory to the top.

suppress_view()
view(mtcars)
#> Suppressed viewing of mtcars