pachong/spider/tmall_get_list_data.js

const cheerio = require('cheerio');

// 获取当前页数据
function tmallGetListData(htmlStr) {
    let goodsArr = [];

    let $ = cheerio.load(htmlStr);
    let list = $('#J_ShopSearchResult .J_TItems>div');
    for (var i = 0; i < list.length; i++) {
        const item = list[i];
        // 遇到分页直接停止，分页后面是推荐的
        if ($(item).hasClass('pagination')) {
            break;
        } else {
            $(item).find('.item').each((i, e) => {
                const title = $(e).find('.item-name').text();
                const img = $(e).find('.photo img').attr('src');
                const url = $(e).find('.item-name').attr('href');
                const id = url.replace(/(.+id=)(\d+)(&.+)/igm, '$2');
                const obj = {
                    title, img, url, id
                }
                goodsArr.push(obj)
            });
        }
    }
    console.log(`列表获取完成: 共${goodsArr.length}个数据`)
    return goodsArr;
}

exports.tmallGetListData = tmallGetListData;