ipbcrawler/extractions/post.js

Package to mine forum data using the Invision Power Board platform.

const { accessElementInformation, sendElementToAccessInformation, asyncAccessElementInformation } = require('./../utils/extractor-access-element-information')
const clear = require('./../utils/clear')(["[E]",])
const he = require('he')

/**
 * [description]
 * @param  {[type]} (element, $             [description]
 * @return {[type]}           [description]
 */
const accessPostInformation = accessElementInformation((element, $) => {
	/**
	 * [postElement description]
	 * @type {[type]}
	 */
	const postElement = element.find('.ipsType_normal.ipsType_richText.ipsContained')

	/**
	 * 
	 */
	$(postElement).find('span.ipsType_reset').remove()

	return {
		/**
		 * [post description]
		 * @type {[type]}
		 */
		post: he.decode(element.find('.ipsType_normal.ipsType_richText.ipsContained').html()),
		
		/**
		 * [author description]
		 * @type {[type]}
		 */
		author: element.find('h3 strong a').text()
	}
})

module.exports = ($) => {

	return {
		/**
		 * [title description]
		 * @type {[type]}
		 */
		title: $('h1.ipsType_pageTitle span.ipsType_break.ipsContained span').text(),

		/**
		 * [post description]
		 * @type {[type]}
		 */
		...accessPostInformation($)($('.cPost.ipsBox.ipsComment')[0])
	}
}