job-hoarder/lib/indeed/parser.js

Job board aggregator to pull in standardized job postings from company job pages

"use strict";
var __importDefault = (this && this.__importDefault) || function (mod) {
    return (mod && mod.__esModule) ? mod : { "default": mod };
};
Object.defineProperty(exports, "__esModule", { value: true });
var cheerio_1 = __importDefault(require("cheerio"));
var IndeedParser = /** @class */ (function () {
    function IndeedParser(host, excludeSponsored) {
        var _this = this;
        /**
         * Parse jobs from request result
         * @param {string} data String of jobs
         * @returns {array} List of parsed jobs
         */
        this.parseJobs = function (body) {
            if (!body)
                throw new Error('No jobs to parse');
            var $ = cheerio_1.default.load(body);
            var jobTable = $('#resultsCol');
            var jobs = jobTable.find('.result');
            var cont = true;
            // Filter out ads
            var filtered = _this.excludeSponsored
                ? jobs.filter(function (_, e) {
                    var job = $(e);
                    var isSponsored = job.find('.jobsearch-SerpJobCard-footer span:contains("sponsored")');
                    return isSponsored == null;
                })
                : jobs;
            // Create objects
            var jobObjects = filtered
                .map(function (i, e) {
                var _a;
                var job = $(e);
                var id = (_a = job.attr('id')) === null || _a === void 0 ? void 0 : _a.trim().split('_')[1];
                var title = job.find('.jobtitle').text().trim();
                var url = 'https://' + _this.host + job.find('.jobtitle').attr('href');
                var company = job.find('.company').text().trim() || null;
                var companyUrl = job.find('.company a[data-tn-element="companyName"]').attr('href');
                var jobLocation = job.find('.location').text().trim();
                // TODO: Make date field
                var datePosted = job.find('.date').text().trim();
                var salary = job.find('.salary.no-wrap').text().trim();
                return {
                    id: id,
                    title: title,
                    url: url,
                    company: company,
                    companyUrl: companyUrl,
                    jobLocation: jobLocation,
                    datePosted: datePosted,
                    salary: salary,
                };
            })
                .get();
            if (jobTable.children().hasClass('dupetext')) {
                // We haven't seen all the results but indeed says the rest are duplicates
                cont = false;
            }
            else if ($('.pagination > *:last-child').hasClass('np')) {
                // We have seen all the results
                cont = false;
            }
            return {
                continue: cont,
                jobs: jobObjects,
            };
        };
        /**
         * Parses job from request result
         * @param {string} data String of job result
         * @returns {object} Object of parsed job
         */
        this.parseJob = function (data) {
            // TODO: Implement parse job
            throw new Error('Function not implemented: Indeed.parseJob');
        };
        this.host = host;
        this.excludeSponsored = excludeSponsored;
    }
    return IndeedParser;
}());
exports.default = IndeedParser;