沪商财富爬虫源码

var configs = {
    domains: ["www.hushangcaifu.com"],
    scanUrls: ["http://www.hushangcaifu.com/invest/index1.html"],
    contentUrlRegexes: ["http://www\\.hushangcaifu\\.com/invest/a\\d{4}\\.html"],
    helperUrlRegexes: ["http://www\\.hushangcaifu\\.com/invest/index\\d+\\.html"],
    fields: [
        {
            name: "title",
            selector: "//div[contains(@class,‘product-content-top-left-top‘)]/h3/text()",
            required: true 
        },
        {
            name: "user_name",
            selector: "//div[contains(@class,‘product-content-top-left-top‘)]/p/span/text()"
        },
        {
            name: "total_money",
            selector:"//div[contains(@class,‘product-content-top-left-middle‘)]/div[1]/h4/text()"
        },
        {
            name: "project_time",
            selector:"//div[contains(@class,‘product-content-top-left-middle‘)]/div[2]/h4/text()"
        },
        {
            name: "annual_return",
            selector:"//div[contains(@class,‘product-content-top-left-middle‘)]/div[3]/h4/text()"
        },
        {
            name: "return_method",
            selector:"//div[contains(@class,‘product-content-top-left-middle‘)]/div[4]/h4/text()"
        }
        
    ]
};

var crawler = new Crawler(configs);
crawler.start();