xml-streamparser.js

// =================================================================
// 		EXAMPLE: Product XML-to-JSON Stream Parser
// =================================================================

const fs = require('fs')
const sax = require('sax')
const stream = require('stream')
const JSONStream = require('JSONStream')

// ==============================================
// 		Initialize
// -----------------------------------------

console.log('BEGIN...')

// ==============================================
// 		Products (Object/JSON) Stream
// -----------------------------------------

const productsStream = new stream.Readable({objectMode: true})

productsStream.on('error', function (err) {
	console.error('ERR', err)
})

// ==============================================
// 		XML Parser (SAX) Stream
// -----------------------------------------

const saxStream = sax.createStream(true, {})

let products = []
let product = undefined
let property = undefined

saxStream.on('error', function (err) {
	console.error('ERR', err)

	this._parser.error = null
	this._parser.resume()
})

saxStream.on('opentag', function (tag) {
	console.log('TAG:OPEN', tag.name)

	if (tag.name === 'catalog') {


	} else if (tag.name === 'product') {
		product = {}
		property = undefined
		return
	} else {
		property = tag.name
	}
})

saxStream.on('attribute', function (attribute) {
	console.log('ATTRIBUTE', attribute)
})

saxStream.on('text', function (text) {
	console.log('TEXT', text)

	if (property) {
		product[property] = product[property] || ''
		product[property] += text
	}
})

saxStream.on('closetag', function (name) {
	console.log('TAG:CLOSE', name)

	if (name === 'product') {
		Object.keys(product).forEach((key) => {
			product[key] = product[key].trim()
		})

		return productsStream.push(product)
	}
})

saxStream.on('end', function () {
	console.log('END')
})

// ==============================================
// 		Run
// -----------------------------------------

fs.createReadStream('/tmp/feed.xml')
	.pipe(saxStream)
	.pipe(process.stdout)
	// .pipe(fs.createWriteStream('feed-copy.xml'))

productsStream
	.pipe(JSONStream.stringify())
	.pipe(process.stdout)
	.pipe(fs.createWriteStream('products.json'))

// ==============================================
// 		Exit
// -----------------------------------------

process.stdout.on('drain', () => {
	console.log('EXIT')

	process.exit(1)
})