hounds

Release the hounds and smoke test any site. Sanity check after a large refactor, or plug into your CI tool of choice.

The beasts will follow all links (internal or otherwise) and round up any uncaught page errors. As a seamless unit, they will scour the field for their quarry by spanning out, containing an area, and expanding (breadth-first search).

Uses nightmare to fire up an Electron webkit browser (with optional UI).

Usage

npm i hounds

const hounds = require('hounds')

const hunt = hounds.release({ url: 'http://localhost:8080' })
    .on('error', console.error)
    .on('end', process.exit)

const quarry = new Writable({
    objectMode: true,
    write: (chunk, enc, next) => {
        console.dir(chunk)
        next()
    }
})

hunt.pipe(quarry)

To use the built-in stdout writers, you can use hounds.writers.error and hounds.writers.url:

const hunt = hounds.release({
    url: 'http://localhost:8080',
    logTo: hounds.writers.url()
})

hunt.pipe(hounds.writers.error())

To show the Electron UI, with devTools open, and keep both alive, use the following options when releasing the hounds:

{
    url: '...',
    keepAlive: true,
    nightmare: {
        show: true, openDevTools: true
    }
}

Options

• url base URL to start from
• keepAlive don't end the stream or the nightmare session when complete (when combined with nightmare.show, allows you to interact with the browser when done).
• waitAfterLoadedFor The number of milliseconds to wait after each page is loaded before following the next link in the queue
• maxFollows The maximum number of links to follow and track (the default is Infinity)
• timeout The number of ms before ending the session. When the timeout is reached, the system will end on the next attempt to read from the queue (Note: this has no use when keepAlive is true). (No default)
• logTo An optional writable stream that all URLs attempting to be processed will be written to.
• urlFilter An optional predicate function, taking the current url as a parameter, and returning true or false as to whether or not to include it in the hunt. Second argument of domainFiltered is a bool stating whether or not the host matches (use it if you'd like to include that check in your filter)
• before and after callbacks receive nightmare instance and if defined, must return it (see examples/preAuth.js)
• screnshot A function that given the current url, returns the path string of the PNG to save
• consoleLevel A string corresponding to one of the console logging levels. The default is error. In this way, console events that are not Javascript errors can be reported.
• nightmare All nightmare 3.0.1 options are supported

Known Issues

• screenshot with outside domains causes nightmare to hang periodically (ref issue on nightmare)
• console.errors not currently handled
• 404s are not currently handled
• /index.html and / are not treated as the same URL, and are both processed
• Subdomains (including www) are treated as different hosts
• unpipe() won't stop the stream from finding results
• links which are hidden in the page are still detected, could use jQuery's approach as an optional workaround

Examples

npm i hounds
cd node_modules/hounds/examples
npm i

Then try out node . for a basic example based on the test fixtures

Unleash them on the latest sites from HackerNews via node hackerNews

Or use them against a local site with auth (see examples/preAuth.js)

Changelog

• 0.2.0 Supports a single url with a promise
• 0.3.0 Stream support (instead of promises)
  • 0.3.1 Migrated to nightmare - 3x faster than Webdriver/Phantom2 and option to open up devTools
  • 0.3.2 Stream usage cleanup
• 0.4.0 Support to keep the session alive via keepAlive
• 0.5.0 Handles console errors that occur after DOMContentLoaded (with configurable timeout waitAfterLoadedFor (ms)). Follows links now.
• 0.6.0 Prevent visiting the same link twice.
• 0.7.0 Allow max number of links to follow or timeout.
• 0.8.0 Support for logTo writable stream for URLs processed, and correct error emitting bugfix.
• 0.9.0 By default, only links within same hostname:port are considered. Override with predicate function urlFilter
• 0.10.0 urlFilter also receives result of domain check as second argument. Bug fix: no dupes anchors in the one page
• 1.0.0 before and after callbacks receive nightmare instance and if defined, must return it (see examples/preAuth.js)
• 1.1.0 Upgrading to nightmare 2.8.1
• 1.2.0 Upgrading to nightmare 2.10.0
• 1.2.1 Fixing tests
• 1.3.0 Exposing the writers
• 1.4.0 Support for screenshot
• 1.4.1 prettyjson to regular dependency
• 1.5.0 Bugfix: before is invoked during the initial run and the system waits until it completes