ipbcrawler/extractions/home.js

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

const { accessElementInformation, sendElementToAccessInformation } = require('./../utils/extractor-access-element-information')
const he = require('he')
const clear = require('./../utils/clear')(["\n", "\t"])

const accessSubCategoryInformation = accessElementInformation((element) => {
	
	return {
		title: he.decode(clear(element.find('a').text()))
	}
})
/**
 * [description]
 * @param  {[type]} (element [description]
 * @return {[type]}          [description]
 */
const accessCategoryInformation = accessElementInformation((element, $) => {
	/**
	 * [main description]
	 * @type {[type]}
	 */
	const main = $(element.find('.ipsDataItem_main'))

	return {
		/**
		 * [icon description]
		 * @type {[type]}
		 */
		icon: element.find('.ipsDataItem_icon img').attr('src'),

		/**
		 * [title description]
		 * @type {[type]}
		 */
		title: he.decode(clear(main.find('.ipsDataItem_title a').text())),

		/**
		 * [description description]
		 * @type {[type]}
		 */
		description: he.decode(clear(main.find('.ipsDataItem_meta').text())),

		/**
		 * [subCategories description]
		 * @type {[type]}
		 */
		subCategories: sendElementToAccessInformation({
			/**
			 * [elements description]
			 * @type {[type]}
			 */
		 	elements: main.find('.ipsDataItem_subList li'),

		 	/**
		 	 * [extractor description]
		 	 * @type {[type]}
		 	 */
		 	extractor: accessSubCategoryInformation($)
		})
	}
})


const accessRankInformation = accessElementInformation((element) => {
	/**
	 * Element rank element
	 * @type {[type]}
	 */
	const domElement = element.find('a')

	/**
	 * 
	 */
	return {
		name: he.decode(domElement.text()),
		withHTML: he.decode(domElement.html())
	}
})
/**
 * [description]
 * @param  {[type]} (element [description]
 * @return {[type]}          [description]
 */
const accessZoneInformation = accessElementInformation((element, $) => {
	return {
		/**
		 * [title description]
		 * @type {[type]}
		 */
		title: he.decode(element.find('h2.cForumTitle a').text()),

		/**
		 * [categories description]
		 * @type {[type]}
		 */
		categories: sendElementToAccessInformation({
			/**
			 * [elements description]
			 * @type {[type]}
			 */
			elements: element.find('ol li.ipsDataItem'),

			/**
			 * [extractor description]
			 * @type {[type]}
			 */
			extractor: accessCategoryInformation($)
		})
	}
})

module.exports = ($) => {
	return {
		zones: sendElementToAccessInformation({
			elements: $('.cForumRow.ipsBox'),
			extractor: accessZoneInformation($)
		}),


		/**
		 * [ranks description]
		 * @type {[type]}
		 */
		ranks: sendElementToAccessInformation({
			/**
			 * [elements description]
			 * @type {[type]}
			 */
			elements: $(
						$('.cWidgetContainer .ipsWidget_inner.ipsPad')
							.find('.ipsList_inline')
							.not('.ipsList_csv')
						).find('li'),
			/**
			 * [extractor description]
			 * @type {[type]}
			 */
			extractor: accessRankInformation($)
		})
	}
}