eshost

eshost

Execute ECMAScript code uniformly across any ECMAScript host environment. See also eshost-cli for an easy way to use this library from the command line.

Using eshost, you can create an agent (eg. a web browser or a command-line ECMAScript host) and evaluate scripts within that agent. Code running within the agent has access to the eshost runtime API which enables code to evaluate scripts, create new realms, handle errors, and so forth all without worrying about the host-specific mechanisms for these capabilities are.

eshost consists of a wrapper around the various ways of executing a host and processing its output (called an Agent) and a runtime library for host-agnostic scripts to use.

Installation

npm install eshost

Supported Hosts

Host	Supported Platforms	Download	Notes
node	Any	https://nodejs.org
ch	Any	Download or build	Chakra console host.
d8	Any	Build from source	v8 console host. Errors are reported on stdout. Use $.getGlobal and $.setGlobal to get and set properties of global objects in other realms.
jsshell	Any	Download	SpiderMonkey console host.
jsc	Mac¹	Build from source²
nashorn	Any	Build from source
edge	Windows		Errors reported from Microsoft Edge are all of type Error. Requires Microsoft WebDriver in your path.
chrome	Any		Requires ChromeDriver in your path.
firefox	Any		Requires GeckoDriver in your path (possibly renamed to wires).
safari	Mac		Requires (SafariDriver browser extension)[https://github.com/SeleniumHQ/selenium/wiki/SafariDriver].

• 1: It is possible to build jsc on other platforms, but not supported.
• 2: Also available on your Mac system at /System/Library/Frameworks/JavaScriptCore.framework/Versions/A/Resources/jsc.

Examples

const esh = require('eshost');
const agent = esh.createAgent('d8', { hostPath: 'path/to/d8.exe' });
const result = await agent.evalScript(`
  print(1+1);
`);
console.log(result.stdout);

Documentation

eshost API

supportedHosts

An array of supported host types.

createAgent(type: string, options = {}): Runner

Gets an instance of a runner for a particular host type. See the table above for supported host types.

Options

• hostPath: Path to host to execute. For console hosts, this argument is required. For the specific browser runners, hostPath is optional and if omitted, the location for that browser will be detected automatically.
• hostArguments: Command line arguments used when invoking your host. Not supported for browser hosts. hostArguments is an array of strings as you might pass to Node's spawn API.

Agent API

initialize(): Promise

Initializes the host and returns a promise that is resolved once the host is initialized. Command line hosts have no initialization as a new process is started for each execution.

This is called for you if you use the createAgent factory.

evalScript(code, options = {}): Promise

Executes code in the host using the Script goal symbol. Returns a promise for a result object.

By default, a script will run in Eshost until the realm is destroyed. For most command-line hosts, this is done automatically when the script execution queues are empty. However, browsers will remain open waiting for more code to become available. Therefore, eshost will automatically append $.destroy() to the end of your scripts. This behavior is not correct if you are attempting to execute asynchronous code. In such cases, add async: true to the options.

Options:

• async: True if the test is expected to call $.destroy() on the root realm when it's finished. When false, $.destroy() is added for you.

stop(): Promise

Stops the currently executing script. For a console host, this simply kills the child process. For browser hosts, it will kill the current window and create a new one.

destroy(): Promise

Destroys the agent, closing any of its associated resources (eg. browser windows, child processes, etc.).

Result Object

An object with the following keys:

• stdout: anything printed to stdout (mostly what you print using print).
• stderr: anything printed to stderr
• error: if the script threw an error, it will be an error object. Else, it will be null.

The error object is similar to the error object you get in the host itself. Namely, it has the following keys:

• name: Error name (eg. SyntaxError, TypeError, etc.)
• message: Error message
• stack: An array of stack frames.

destroy(): Promise

Tears down the agent. For browsers, this will close the browser window.

Runtime Library

print(str)

Prints str to stdout.

$.global

A reference to the global object.

$.createRealm(options)

Creates a new realm, returning that realm's runtime library ($).

For example, creating two nested realms:

$sub = $.createRealm();
$subsub = $sub.createRealm();

You can also use a destroy callback that gets called when the code inside the realm calls $.destroy(). For example:

$sub = $.createRealm({
  destroy: function () {
    print('destroyed!')
  }
});

$sub.evalScript('$.destroy()'); // prints "destroyed!"

Options:

• globals: an object containing properties to add to the global object in the new realm.
• destroy: a callback that is called when the code executing in the realm destroys its realm (ie. by calling $.destroy()).

$.evalScript(code)

Creates a new script and evals code in that realm. If an error is thrown, it will be passed to the onError callback.

Scripts are different from eval in that lexical bindings go into the global lexical contour rather than being scoped to the eval.

$.destroy()

Destroys the realm. Note that in some hosts, $.destroy may not actually stop executing code in the realm or even destroy the realm.

$.getGlobal(name)

Gets a global property name.

$.setGlobal(name, value)

Sets a global property name to value.