Interested in learning what code is being used inside a serverless R Shinylive application? If so, this is the repository for you! Here’s a summary of what you can find in the repository:

• 00a-convert-shinylive-to-shiny.R: An R script that obtains and converts an R Shinylive script into an R Shiny script.
• 01-downloaded-r-shinylive-app/app.json: The source code and files powering the serverless R Shinylive app.
• 02-converted-r-shinylive-app-to-r-shiny/app.R: The converted code to run the R Shinylive app as a regular R Shiny app.
• zz-patched-converted-r-shinylive-app-to-r-shiny/app.R: A patched version of the Shiny app that fixes a hiccup with DataTable rendering.
• zz-patched-r-shiny-to-r-shinylive-app/: The conversion back into an R Shinylive app of patched Shiny app.
• experiments/: Exploring various parts of a Shinylive app and exporting Shiny apps.

R Shinylive is a technology from Posit that combines Shiny, webR, and WebAssembly to enable Shiny applications to be run entirely in a web browser without the need for a compute server. In essence, an R Shinylive application does not require a Shiny server to run. Instead, it can run anywhere that can serve static HTML web page files.

For this exercise, we’re going to focus on being able to obtain the R Shinylive app source that is powering the webR binary R package repository dashboard.

R Shinylive app data is stored using [JavaScript Object Notation (JSON)] inside of an app.json file. For each file that is converted from a regular R Shiny app, there is an entry inside of a JSON array that contains a subfield with three key-value pairs:

1. "name": The name of the file that was converted.
   • app.R or two separate entries with ui.R or server.R.
2. "content": The escaped contents of the file that was created.
3. "type": The kind of data stored inside "content".
   • This will likely contain the value of "text".

For instance, if we have a single file Shiny app in app.R and we have an RStudio project called shinylive-to-shiny, then we would expect two entries in the app.json:

[
  {
    "name": "app.R",
    "content": "library(shiny)\nlibrary(dplyr)\n ... \nshinyApp(ui = ui, server = server)\n",
    "type": "text"
  },
  {
    "name": "shinylive-to-shiny.Rproj",
    "content": "Version: 1.0\n\nRestoreWorkspace: Default ...\n\nAutoAppendNewline: Yes\n",
    "type": "text"
  }
]

For a two-file Shiny app that uses ui.R and server.R alongside an RStudio project called shinylive-to-shiny, we would expect to have three different entries array stored in app.json:

[
  {
    "name": "server.R",
    "content": server <- function(input, output) { ... }\n",
    "type": "text"
  },
  {
    "name": "ui.R",
    "content": "ui <- fluidPage(\n  sidebarLayout( ... )\n  )\n)",
    "type": "text"
  },
  {
    "name": "shinylive-to-shiny.Rproj",
    "content": "Version: 1.0\n\nRestoreWorkspace: Default ...\n\nAutoAppendNewline: Yes\n",
    "type": "text"
  }
]

Given the above structure, we can use existing R packages and features to obtain the app.json.