V8 (version 6.0.1)

V8: Run JavaScript in a V8 context

Description

The v8() function (formerly called new_context) creates a new V8 context. A context provides an execution environment that allows separate, unrelated, JavaScript code to run in a single instance of V8, like a tab in a browser.

Usage

v8(global = "global", console = TRUE, ...)
engine_info()

Arguments

global

character vector indicating name(s) of the global environment. Use NULL for no name.

console

expose console API (console.log, console.warn, console.error).

...

ignored parameters for past/future versions.

V8 Context Methods

V8:::generate_rd()

The ct$eval method evaluates a string of JavaScript code in the same way as eval in JavaScript. By default eval() returns a string with console output; but when the serialize parameter is set to TRUE it serializes the JavaScript return object to a JSON string or a raw buffer.

The ct$get, ct$assign and ct$call functions automatically convert arguments and return value between R and JavaScript (using JSON). To pass literal JavaScript arguments that should not be converted to JSON, wrap them in JS(), see examples.

If a call to ct$eval(),ct$get(), or ct$call() returns a JavaScript promise, you can set await = TRUE to wait for the promise to be resolved. It will then return the result of the promise, or an error in case the promise is rejected.

The ct$validate function is used to test if a piece of code is valid JavaScript syntax within the context, and always returns TRUE or FALSE.

In an interactive R session you can use ct$console() to switch to an interactive JavaScript console. Here you can use console.log to print objects, and there is some support for JS tab-completion. This is mostly for testing and debugging, it may not work perfectly in every IDE or R-frontend.

Data Interchange

JSON is used for data interchange between R and JavaScript. Therefore you can (and should) only exchange data types that have a sensible JSON representation. One exception is raw vectors which are converted to/from Uint8Array buffers, see below. All other arguments and objects are automatically converted according to the mapping described in Ooms (2014), and implemented by the jsonlite package in jsonlite::fromJSON() and jsonlite::toJSON().

As for version 3.0 of this R package, Raw vectors are converted to Uint8Array typed arrays, and vice versa. This makes it possible to efficiently copy large chunks binary data between R and JavaScript, which is useful for running wasm or emscripten.

Note about Linux and Legacy V8 engines

This R package can be compiled against modern (V8 version 6+) libv8 API, or the legacy libv8 API (V8 version 3.15 and below). You can check V8::engine_info() to see the version that is running. The legacy version does not support modern JS (ES6) or WASM, but it is still the default on older versions of Ubuntu and CentOS. The latest versions of all major Linux distributions now provide a modern version of V8. For Ubuntu 16.04 and 18.04 we provide backports of libv8 (via libnode-dev), see the readme for details.

Details

A V8 context cannot be saved or duplicated, but creating a new context and sourcing code is very cheap. You can run as many parallel v8 contexts as you want. R packages that use V8 can use a separate V8 context for each object or function call.

The name of the global object (i.e. global in node and window in browsers) can be set with the global argument. A context always have a global scope, even when no name is set. When a context is initiated with global = NULL, the global environment can be reached by evaluating this in the global scope, for example: ct$eval("Object.keys(this)").

References

A Mapping Between JSON Data and R Objects (Ooms, 2014): https://arxiv.org/abs/1403.2805

Examples

# Create a new context
ctx <- v8();

# Evaluate some code
ctx$eval("var foo = 123")
ctx$eval("var bar = 456")
ctx$eval("foo+bar")

# Functions and closures
ctx$eval("JSON.stringify({x:Math.random()})")
ctx$eval("(function(x){return x+1;})(123)")

# Objects (via JSON only)
ctx$assign("mydata", mtcars)
ctx$get("mydata")
outlist <- ctx$get("mydata", simplifyVector = FALSE)
outlist[1]

# Assign JavaScript
ctx$assign("foo", JS("function(x){return x*x}"))
ctx$assign("bar", JS("foo(9)"))
ctx$get("bar")

# Validate script without evaluating
ctx$validate("function foo(x){2*x}") #TRUE
ctx$validate("foo = function(x){2*x}") #TRUE
ctx$validate("function(x){2*x}") #FALSE

# Use a JavaScript library
ctx$source("https://cdnjs.cloudflare.com/ajax/libs/underscore.js/1.13.6/underscore-min.js")
ctx$call("_.filter", mtcars, JS("function(x){return x.mpg < 15}"))

# Example from underscore manual
ctx$eval("_.templateSettings = {interpolate: /\\{\\{(.+?)\\}\\}/g}")
ctx$eval("var template = _.template('Hello {{ name }}!')")
ctx$call("template", list(name = "Mustache"))

# Call anonymous function
ctx$call("function(x, y){return x * y}", 123, 3)

if (FALSE) {
#CoffeeScript
ct2 <- v8()
ct2$source("http://coffeescript.org/v1/browser-compiler/coffee-script.js")
jscode <- ct2$call("CoffeeScript.compile", "square = (x) -> x * x", list(bare = TRUE))
ct2$eval(jscode)
ct2$call("square", 9)

# Interactive console
ct3 <- v8()
ct3$console()
# //this is JavaScript
# var test = [1,2,3]
# JSON.stringify(test)
# exit
}