active phantomjs development has ended, in favor of using Chrome's new headless functionality (reference). Instead of using this library, consider using a go package that uses this new api such as chromedp.

This is a Go wrapper for the phantomjs command line program. It provides the full webpage API and has a strongly typed API. The wrapper provides an idiomatic Go interface while allowing you to communicate with the underlying WebKit and JavaScript engine in a seamless way.

First, install phantomjs on your machine. This can be done using your package manager (such as apt-get or brew). Then install this package using the Go toolchain:

$ go get -u github.com/benbjohnson/phantomjs

This wrapper works by communicating with a separate phantomjs process over HTTP. The process can take several seconds to start up and shut down so you should do that once and then share the process. There is a package-level variable called phantomjs.DefaultProcess that exists for this purpose.

package main

import (
	"github.com/benbjohnson/phantomjs"
)

func main() {
	// Start the process once.
	if err := phantomjs.DefaultProcess.Open(); err != nil {
		fmt.Println(err)
		os.Exit(1)
	}
	defer phantomjs.DefaultProcess.Close()

	// Do other stuff in your program.
	doStuff()
}

You can have multiple processes, however, you will need to change the port used for each one so they do not conflict. This library uses port 20202 by default.

The WebPage will be the primary object you work with in phantomjs. Typically you will create a web page from a Process and then either open a URL or you can set the content directly:

// Create a web page.
// IMPORTANT: Always make sure you close your pages!
page, err := p.CreateWebPage()
if err != nil {
	return err
}
defer page.Close()

// Open a URL.
if err := page.Open("https://google.com"); err != nil {
	return err
}

The HTTP API uses a reference map to track references between the Go library and the phantomjs process. Because of this, it is important to always Close() your web pages or else you can experience memory leaks.

You can synchronously execute JavaScript within the context of a web page by by using the Evaluate() function. This example below opens Hacker News, retrieves the text and URL from the first link, and prints it to the terminal.

// Open a URL.
if err := page.Open("https://news.ycombinator.com"); err != nil {
	return err
}

// Read first link.
info, err := page.Evaluate(`function() {
	var link = document.body.querySelector('.itemlist .title a');
	return { title: link.innerText, url: link.href };
}`)
if err != nil {
	return err
}

// Print title and URL.
link := info.(map[string]interface{})
fmt.Println("Hacker News Top Link:")
fmt.Println(link["title"])
fmt.Println(link["url"])
fmt.Println()

You can pass back any object from Evaluate() that can be marshaled over JSON.

Another common task with PhantomJS is to render a web page to an image. Once you have opened your web page, simply set the viewport size and call the Render() method:

// Open a URL.
if err := page.Open("https://news.ycombinator.com"); err != nil {
	return err
}

// Setup the viewport and render the results view.
if err := page.SetViewportSize(1024, 800); err != nil {
	return err
}
if err := page.Render("hackernews.png", "png", 100); err != nil {
	return err
}

You can also use the RenderBase64() to return a base64 encoded image to your program instead of writing the file to disk.