Upload all the images of a page to s3

upload-images.js

/* 

# First, install node and:

	sudo npm install jsdom
	sudo npm install aws2js
	sudo npm install request

*/

var aws = {
	key: 'AWS KEY',
	secret: 'AWS SECRET',
	bucket: 'AWS BUCKET'
}
var base = 'http://www.your-domain.com'
var page = '/path/to/page'

var fs = require('fs')
var jsdom = require('jsdom')
var s3 = require('aws2js').load('s3', aws.key, aws.secret)
var request = require('request')

s3.setBucket(aws.bucket)

jsdom.env({
	html: base + page,
	scripts: ['http://code.jquery.com/jquery-1.5.min.js'],
	done: function(errors, window) {
		var $ = window.$;
		var urls = []
		$('img').each(function() {
			var url = this.src
			if (url[0] == '/') {
				url = base + url
			}
			urls.push(url)
		})
		
		function next() {
			if (!urls.length) {
				console.log("DONE")
				process.exit()
			}
			var url = urls.pop()
			var path = url.replace(/.*\.com\//, '')
			var name = encodeURIComponent(url)
			
			request({ url:url, encoding:null }, function(err, res, buf) {
				var extension = url.match(/.*\.(\w+)$/)
				console.log("HERE", path, buf.length, extension[1])
				s3.putBuffer(path, buf, 'public-read', { 'Content-Type':'image/'+extension[1] }, function(err, res) {
					if (err) { throw err }
					next()
				})
			})
		}
		next()
	}
})