jjgonecrypto / hounds

Release the hounds and smoke test any site. The beasts will follow links and stream any page errors to a pipe of your choice.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

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).

npm version GitHub version CircleCI

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

hounds-simple

Unleash them on the latest sites from HackerNews via node hackerNews

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

hounds-preauth

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

About

Release the hounds and smoke test any site. The beasts will follow links and stream any page errors to a pipe of your choice.

License:MIT License


Languages

Language:JavaScript 95.5%Language:HTML 4.5%