locusbuilder-utility/server/modules/htmlparser/index.js

Locusbuilder Utility server for asset scraping and processing

/**
 * Module: HTML Parser
 * @author  Richard Abear
 * 
 * @description HTML parser recieves a url and returns a fully processed output depending on request
 */
const phantom = require('phantom');

// Disable jshint for async programming
// jshint ignore: start
class HtmlParser {
    constructor(opts) {
        this.id = opts.id;
    }

    async init() {
        console.log('Initializing phantomjs instance');
        try {
            this.instance = await phantom.create();
            this.page = await this.instance.createPage();
        } catch(e) {
            console.log(e);
        }
    }

    async openPage(page) {
        const status = await this.page.open(page);
        const content = await this.page.property('content');

        return content;
    }
}
module.exports = HtmlParser;