old_flsystem/类库/Api.Framework/Tools/JDHelper.cs

using CsharpHttpHelper;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks;

namespace Api.Framework.Tools
{
    /// <summary>
    /// 京东解析助手
    /// </summary>
    public static class JDHelper
    {
        /// <summary>
        /// 查找宝贝Id的正则表达式
        /// </summary>
        private static string RegItemId = @"(?:item\.jd\.com/|item\.m\.jd\.com/product/|item\.jd\.hk/|jd(?:.*?)sku_id=|skuId=|re.jd.com/cps/item/|jd(?:.*?)[\?&]sku=|jd/dujia/item/|http(?:.*?)jd\.(?:.{0,15}?)|jd(?:.*?)wareId=|sku=|京东(?:[\w\W]*?)wareId=|wareId=)(?<itemId>\d{4,})";

        /// <summary>
        /// 通过字符串获取京东商品id
        /// </summary>
        /// <param name="str">包含京东链接的文本</param>
        /// <returns></returns>
        public static string GetJingdongItemId(string str)
        {
            try
            {
                if (str.ToLower().Contains("taobao") || str.ToLower().Contains("detail.tmall") || str.ToLower().Contains("<title><![CDATA[微信红包]]></title>")) return string.Empty;
                Match reg = null;
                if (!str.ToLower().Contains("sq.jd."))
                {
                    reg = Regex.Match(str, RegItemId, RegexOptions.IgnoreCase);
                    if (reg.Success)
                    {
                        return reg.Groups["itemId"].Value;//京东宝贝ID
                    }
                    else
                        reg = null;
                }

                if (reg == null)
                {
                    if (!str.ToLower().Contains("sq.jd."))
                    {
                        reg = Regex.Match(HttpHelper.URLDecode(str), RegItemId, RegexOptions.IgnoreCase);
                        if (reg.Success)
                        {
                            return reg.Groups["itemId"].Value;//京东宝贝ID
                        }
                    }

                    str = Regex.Replace(str, "u.jd", "https://u.jd", RegexOptions.IgnoreCase);
                    var _url = HttpExtend.RegexMatchUrl(str);
                    if (string.IsNullOrWhiteSpace(_url))
                        _url = HttpExtend.RegexMatch(str, @"""jumpUrl"":""(?<链接>http:[^""]+?)""");
                    if (string.IsNullOrWhiteSpace(_url))//没有前缀https：
                    {
                        _url = HttpExtend.RegexMatchUrl(str);
                    }
                    if (!string.IsNullOrWhiteSpace(_url))
                    {
                        var url = _url;
                        HttpHelper http = new HttpHelper();
                        var item = http.GetItem(url);
                        var result = http.GetHtml(item);
                        var html = result.Html;
                        //reg = Regex.Match(html, @"(?:window._itemOnly\s*?=\s*?\({\s*?""item"": {""skuId"":""(?<宝贝Id>\d+?)"",)");
                        if (str.Contains("taobao") || str.Contains("detail.tmall")) return string.Empty;
                        reg = Regex.Match(html, @"""skuId"":""(?<宝贝Id>\d+?)""");
                        if (reg.Success)
                        {
                            return reg.Groups["宝贝Id"].ToString();
                        }
                        else
                        {
                            reg = Regex.Match(html, @"hrl='(?<跳转A>.+?)';");
                            if (reg.Success)
                            {
                                var skip = reg.Groups["跳转A"].ToString();
                                item.URL = skip;
                                item.Referer = url;
                                item.Allowautoredirect = false;
                                result = http.GetHtml(item);
                                var header = result.Header;
                                var location = result.Header.Get("Location");
                                if (location != null)
                                {
                                    reg = Regex.Match(location, RegItemId, RegexOptions.IgnoreCase);
                                    if (reg.Success)
                                    {
                                        return reg.Groups["itemId"].Value;//京东宝贝ID
                                    }
                                }
                            }
                            else //http://3.cn/Vc0Avuc 这类链接
                            {
                                var location = result.ResponseUri;
                                if (!string.IsNullOrWhiteSpace(location))
                                {
                                    var url1 = location.Substring(location.IndexOf("appurl=") + 7);
                                    url1 = HttpHelper.URLDecode(url1);
                                    item.URL = url1;
                                    item.Referer = location;
                                    item.Allowautoredirect = false;
                                    result = http.GetHtml(item);
                                    reg = Regex.Match(result.Html, @"hrl='(?<hrl>.+?)';");
                                    if (reg.Success)
                                    {
                                        var hrl = reg.Groups["hrl"].Value;
                                        item.URL = hrl;
                                        item.Referer = hrl;
                                        item.Allowautoredirect = false;
                                        result = http.GetHtml(item);

                                        reg = Regex.Match(result.RedirectUrl, RegItemId, RegexOptions.IgnoreCase);
                                        if (reg.Success)
                                        {
                                            return reg.Groups["itemId"].Value;//京东宝贝ID
                                        }
                                    }
                                }

                            }
                        }

                        http = new HttpHelper();
                        item = new HttpItem()
                        {
                            URL = url,
                            Method = "GET",
                            Timeout = 100000,
                            ReadWriteTimeout = 30000,
                            IsToLower = false,
                            Cookie = "",
                            UserAgent = "Opera/9.27 (Windows NT 5.2; U; zh-cn)",
                            Accept = "text/html, application/xhtml+xml, */*",
                            ContentType = "text/html",
                            Referer = "",
                            Postdata = "",
                        };
                        result = http.GetHtml(item);
                        var _location = result.Header.Get("Location");
                        if (!string.IsNullOrWhiteSpace(_location))
                        {
                            reg = Regex.Match(_location, RegItemId, RegexOptions.IgnoreCase);
                            if (reg.Success)
                            {
                                return reg.Groups["itemId"].Value;//京东宝贝ID
                            }
                        }

                    }
                }
            }
            catch (Exception ex)
            {
                //throw ex;
            }
            return string.Empty;
        }

    }
}