// Define Method
public Object _openFile(INakoFuncCallInfo info)
{
string fileName = info.StackPopAsString();
string e = info.StackPopAsString();
// Exists?
if (!System.IO.File.Exists(fileName))
{
throw new NakoPluginRuntimeException("ファイル『" + fileName + "』は存在しません。");
}
// Load
//string src = File.ReadAllText(fileName);
string src;
if (e == "Auto")
{
src = StrUnit.LoadFromFileAutoEnc(fileName);
}
else
{
if (sjis.Contains(e))
{
src = File.ReadAllText(fileName, Encoding.GetEncoding(932));
}
else
{
src = File.ReadAllText(fileName, Encoding.GetEncoding(e));
}
}
return(src);
}
private System.Text.Encoding GetCode()
{
using (System.IO.FileStream fr = new FileStream(path, FileMode.Open, FileAccess.Read)){
byte[] bs = new byte[Math.Min(fr.Length, 100)];
int read = fr.Read(bs, 0, bs.Length);
if (read == 0)
{
return(Encoding.UTF8);
}
Encoding e = StrUnit.GetCode(bs);
return(e);
}
}
private string GetNextUrl(string text)
{
string str = string.Empty;
string[] strArray = text.Split(new string[] { "</a>" }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < strArray.Length; i++)
{
if (strArray[i].IndexOf("下一页") > -1)
{
return(StrUnit.MidStrEx(strArray[i], "href=\"", "\"").Trim());
}
}
return(str);
}
/// <summary>
/// ファイルからソースコードを読みこんでパースする
/// </summary>
/// <param name="filename"></param>
public void LoadFromFile(string filename)
{
string filename2 = this.RemoveBasePath(filename);
int i = files.IndexOf(filename2);
if (i >= 0)
{
// 二重で取り込みはしない
return;
}
files.Add(filename2);
string src = StrUnit.LoadFromFileAutoEnc(filename);
// Parse
this.ParseEx(src, filename);
}
public Object _dataGet(INakoFuncCallInfo info)
{
string s = info.StackPopAsString();
string e = info.StackPopAsString();
string ret;
WebClient client = new WebClient();
if (e == "Auto")
{
byte[] bytes = client.DownloadData(s);
ret = StrUnit.ToStringAutoEnc(bytes);
}
else
{
client.Encoding = Encoding.GetEncoding(e);
ret = client.DownloadString(s);
}
client.Dispose();
return(ret);
}
private string GetData1(string text)
{
string str = string.Empty;
innerHtml = string.Empty;
string source = string.Empty;
try
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(text);
HtmlNode node = null;
node = document.DocumentNode.SelectSingleNode("//ul[@id='hs-below-list-items']");
if (node == null)
{
base.updateTextBox("获取商品列表失败", true);
throw new Exception("获取商品列表失败");
}
foreach (string itemInfo in node.InnerHtml.Split(new string[] { "<li qrdata" }, StringSplitOptions.RemoveEmptyEntries))
{
if (itemInfo.IndexOf("class=\"list-item") != -1)
{
//source = StrUnit；
DataRow row = base.OutDataTable.NewRow();
row["Platform"] = "Ali Express";
row["Keyword"] = base.keywordInf.keyword;
row["ItemID"] = string.Empty;
//
string goodInfo = StrUnit.MidStrEx(itemInfo, "<a class=\"history-item product", ">");
row["Url"] = StrUnit.MidStrEx(goodInfo, "href=\"//", "\"");
row["Title"] = StrUnit.MidStrEx(goodInfo, "title=\"", "\"");
row["Price"] = StrUnit.MidStrEx(itemInfo, "itemprop=\"price\">", "</span>");
string storeInfo = StrUnit.MidStrEx(itemInfo, "class=\"store-name", "</div>");
row["StoreName"] = StrUnit.MidStrEx(storeInfo, "title=\"", "\"");
row["StoreUrl"] = StrUnit.MidStrEx(storeInfo, "href=\"//", "\"");
row["SellerID"] = StrUnit.MidStrEx(storeInfo, "store/", "\"");
base.OutDataTable.Rows.Add(row);
Thread.Sleep(100);
if (string.IsNullOrEmpty(row["Url"].ToString()))
{
throw new Exception("获取商品列表数据失败！ -1");
}
}
}
base.updateTextBox(base.keywordInf.keyword + " 第 " + (page) + " 页获取完成", true);
if (page >= base.keywordInf.endPage)
{
str = string.Empty;
return(str);
}
++page;
str = "http://www.aliexpress.com/wholesale?SearchText=" + base.keywordInf.keyword + "&page=" + page;
return(str);
}
catch (Exception e)
{
base.updateTextBox("GetData Err:" + e.Message, true);
}
return(str);
}
public object _post(INakoFuncCallInfo info)
{
string url = info.StackPopAsString();
object val = info.StackPop();//TODO:Array
string e = info.StackPopAsString();
string query = "";
if (val is string)
{
query = (string)val;
}
else if (val is NakoVariable)
{
List <string> qp = new List <string>();
NakoVarArray arr = (NakoVarArray)val;
foreach (NakoVariable kv in arr)
{
qp.Add(String.Format("{0}={1}", HttpUtility.UrlEncode(kv.key, Encoding.GetEncoding(e)), HttpUtility.UrlEncode((string)kv.Body, Encoding.GetEncoding(e))));
}
query = String.Join("&", qp.ToArray());
}
else
{
throw new Exception("variable is not NakoVariable or string");
}
byte[] queryBytes = Encoding.ASCII.GetBytes(query);
HttpWebRequest req = (HttpWebRequest)WebRequest.Create(url);
req.Method = "POST";
req.ContentType = "application/x-www-form-urlencoded";
using (Stream str = req.GetRequestStream())
{
str.Write(queryBytes, 0, queryBytes.Length);
}
WebResponse response = req.GetResponse();
string ret = "";
if (e == "Auto")
{
using (Stream sr = response.GetResponseStream()) {
byte[] buf = new byte[32768];
int read = 0;
using (MemoryStream ms = new MemoryStream()){
do
{
read = sr.Read(buf, 0, buf.Length);
if (read > 0)
{
ms.Write(buf, 0, read);
}
} while(read > 0);
if (ms.Length > 0)
{
ret = StrUnit.ToStringAutoEnc(ms.ToArray());
}
}
}
}
else
{
using (StreamReader sr = new StreamReader(response.GetResponseStream(), Encoding.GetEncoding(e))) {
ret = sr.ReadToEnd();
}
}
return(ret);
}
public string GetData(string ss)
{
string nextUrl = string.Empty;
string sourse = "";
try
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(ss);
HtmlNode node = document.DocumentNode.SelectSingleNode("//div[@class='nextpage']");
if (node.SelectSingleNode("//span[@class='current']") != null)
{
byte result = 0;
if (!byte.TryParse(node.SelectSingleNode("//span[@class='current']").InnerText.Trim(), out result))
{
base.updateTextBox(base.keywordInf.keyword + " 获取当前页面出错", true);
}
if (base.keywordInf.endPage < result)
{
throw new Exception("已到规定获取的页数，退出");
}
base.updateTextBox(base.keywordInf.keyword + " 正在获取第 " + result.ToString() + " 页", true);
}
if (node.InnerHtml.IndexOf("下一页") > -1)
{
nextUrl = this.GetNextUrl(node.InnerHtml);
}
node = document.DocumentNode.SelectSingleNode("//div[@class='s_product']");
if (node == null)
{
return(nextUrl);
}
string[] strArray = node.InnerHtml.Split(new string[] { "<div class=\"s_product_item\">" }, StringSplitOptions.RemoveEmptyEntries);
for (int i = 0; i < strArray.Length; i++)
{
if (BaseWorkth.DoStop)
{
return(nextUrl);
}
if (strArray[i].Contains("<div class=\"s_product_info\">"))
{
string str3 = strArray[i].ToString();
DataRow row = base.OutDataTable.NewRow();
row["Platform"] = "makepolo.com";
row["Url"] = StrUnit.MidStrEx(str3, "href=\"", "\"");
row["ItemId"] = StrUnit.MidStrEx(row["Url"].ToString(), "product-detail/", ".");
row["Keyword"] = base.keywordInf.keyword;
row["Title"] = StrUnit.MidStrEx(str3, ".html\">", "</a>").Trim();
sourse = StrUnit.MidStrEx(str3, "报价:<strong>", "</strong>");
if (sourse.Trim() == "")
{
sourse = StrUnit.MidStrEx(str3, "报价:", "</");
}
row["Price"] = sourse;
sourse = StrUnit.MidStrEx(str3, "class=\"s_product_contact plistk\">", "<span");
row["StoreName"] = StrUnit.MidStrEx(sourse, "target=\"_blank\">", "</a>").Trim();
row["StoreUrl"] = StrUnit.MidStrEx(sourse, "<a href=\"", "\"");
row["SellerId"] = StrUnit.MidStrEx(row["StoreUrl"].ToString(), "://", ".");
Thread.Sleep(50);
base.OutDataTable.Rows.Add(row);
}
}
}
catch (Exception exception)
{
if (exception.Message != "已到规定获取的页数，退出")
{
base.updateTextBox("GetData取信息异常：" + exception.Message, true);
}
}
return(nextUrl);
}
private string GetJsonData(string text)
{
string str = string.Empty;
innerHtml = string.Empty;
string html = string.Empty;
string sourse = string.Empty;
try
{
if (text.IndexOf("itemSearchList") == -1)
{
throw new Exception("未知页面内容");
}
html = text.Replace(@"\r\n", "").Replace("\\\"", "\"");
HtmlDocument document = new HtmlDocument();
document.LoadHtml(html);
HtmlNode node = null;
node = document.DocumentNode.SelectSingleNode("//div[@id='itemSearchList']");
if (node == null)
{
throw new Exception("获取商品列表数据失败");
}
foreach (string str2 in node.InnerHtml.Split(new string[] { "<div class=\"mod_search_pro\"" }, StringSplitOptions.RemoveEmptyEntries))
{
if (str2.IndexOf("<div class=\"proImg\"") != -1)
{
sourse = StrUnit.MidStrEx(str2, "<div class=\"proImg\"", "<p class=\"proPrice\">");
DataRow row = base.OutDataTable.NewRow();
row["Platform"] = "yhd.com";
row["Keyword"] = base.keywordInf.keyword;
row["ItemId"] = StrUnit.MidStrEx(sourse, "pmid=\"", "\"");
if (string.IsNullOrEmpty(row["ItemId"].ToString()))
{
throw new Exception("获取商品列表数据失败1");
}
sourse = StrUnit.MidStrEx(sourse, "href=\"", "\"");
sourse = sourse.StartsWith("//") ? ("http:" + sourse) : sourse;
row["Url"] = sourse;
if (str2.IndexOf("1号店自营") == -1)
{
sourse = StrUnit.MidStrEx(str2, "<p class=\"storeName", "onclick=");
row["StoreName"] = StrUnit.MidStrEx(sourse, "title=\"", "\"");
row["StoreUrl"] = StrUnit.MidStrEx(sourse, "href=\"", "\"");
row["SellerId"] = "m-" + StrUnit.MidStrEx(sourse, "merchantid=\"", "\"");
}
else
{
row["StoreName"] = "1号店自营";
row["StoreUrl"] = string.Empty;
row["SellerId"] = string.Empty;
}
sourse = StrUnit.MidStrEx(str2, "<p class=\"proName clearfix\">", "onclick=");
row["Title"] = StrUnit.MidStrEx(sourse, "title=\"", "\"");
if (string.IsNullOrEmpty(row["Url"].ToString()))
{
row["Url"] = StrUnit.MidStrEx(sourse, "href=\"", "\"");
}
row["Price"] = StrUnit.MidStrEx(str2, "yhdprice=\"", "\"");
base.OutDataTable.Rows.Add(row);
}
}
node = document.DocumentNode.SelectSingleNode("//div[@class='search_select_page']");
if (node != null)
{
innerHtml = node.InnerHtml;
sourse = StrUnit.MidStrEx(innerHtml, "class=\"next iconSearch\"", "</a>");
if (!string.IsNullOrEmpty(sourse))
{
str = StrUnit.MidStrEx(sourse, "url=\"", "\"") + StrUnit.MidStrEx(sourse, "href=\"", "\"");
}
sourse = StrUnit.MidStrEx(StrUnit.MidStrEx(innerHtml, "id=\"currentPageNum\"", ">"), "value=\"", "\"");
if (string.IsNullOrEmpty(sourse))
{
sourse = StrUnit.MidStrEx(innerHtml, "pageSearch(", ",");
}
int result = 0;
if (!int.TryParse(sourse, out result))
{
return(str);
}
base.updateTextBox(base.keywordInf.keyword + " 第 " + sourse + " 页获取完成", true);
if (result >= base.keywordInf.endPage)
{
str = string.Empty;
}
//base.updateTextBox(base.keywordInf.keyword + " 已到规定页数，结束", true);
}
return(str);
}
catch (Exception exception)
{
base.updateTextBox("GetJsonData Err:" + exception.Message, true);
}
return(str);
}
public override void Excute()
{
try
{
DateTime now = DateTime.Now; //mark of time
string url = string.Empty;
string refererUrl = string.Empty; //use refererURLinstead if url in makepolo
string text = string.Empty; //crawled content
string s = string.Empty;
int num = 0;
base.IniDataTable(); //InitDataTable from the base
base.http = new HttpClientHelper(0x4e20); //new a HttpClientHelper
for (int i = 0; i < base.keywordInfList.Count; i++)
{ //go iterations
base.keywordInf = base.keywordInfList[i]; //setup the keyword during this iteration
if (string.IsNullOrEmpty(base.keywordInf.keyword)) //skip null or empty keyword
{
continue;
}
base.updateTextBox(base.keywordInf.keyword + " 开始查询", true);
url = "http://www.yhd.com?forceId=6";
byte num3 = 0;
while (true) //open yhd
{
if (base.http.Get(url).IndexOf("<title>网上超市1号店，省力省钱省时间</title>") != -1) //open success
{
goto Label_0107;
}
num3 = (byte)(num3 + 1); //retry, max: 3 times
if (num3 >= 3)
{
break;
}
base.updateTextBox("一号店打开主页失败，重试：" + num3.ToString(), true);
}
num++;
continue;
Label_0107:
Thread.Sleep(0x7d0);
url = "http://search.yhd.com/c0-0/k" + HttpUtility.UrlEncode(StrUnit.UrlEncode(base.keywordInf.keyword, true).ToUpper()) + "/null/"; //form of url
text = base.http.Get(url, "http://www.yhd.com");
//Console.WriteLine(text);
refererUrl = url; //copy search url to referer url
if (text.Contains("id=\"plist")) // id = "plist
{
//do nothing?
}
if (text.Contains("抱歉~没有找到"))
{
num++; //count the search num
base.updateTextBox(base.keywordInf.keyword + " 没有找到相关商品", true);
Thread.Sleep(200);
}
else if (!text.Contains("1号店为您找到"))
{
num++;
base.updateTextBox(base.keywordInf.keyword + " 返回页面错误", true);
Log.WritrTemp(text, "未知返回页面");
}
else
{
HtmlDocument document = new HtmlDocument(); //html document object
document.LoadHtml(text); //load the content got
HtmlNode node = null; //html node init
node = document.DocumentNode.SelectSingleNode("//small[@class='result_count']"); //select 'result_count'
if (node == null)
{
num++;
base.updateTextBox(base.keywordInf.keyword + " 获取商品数量错误1", false);
Log.WriteLog(base.keywordInf.keyword + " 节点不存在：//small[@class='result_count']");
}
else
{
s = StrUnit.MidStrEx(node.InnerText, "共", "条");
int result = 0;
if (!int.TryParse(s, out result))
{
base.updateTextBox(base.keywordInf.keyword + " 获取商品数量错误2", true);
num++;
}
else
{
for (string str6 = this.GetData(text); !string.IsNullOrEmpty(str6); str6 = this.GetJsonData(text)) //?
{
Thread.Sleep(0x7d0);
text = base.http.Get(str6, refererUrl);
}
TimeSpan span = (TimeSpan)(DateTime.Now - now);
base.updateTextBox(string.Concat(new object[] { base.keywordInf.keyword, " 获取完毕,耗时：", span.TotalSeconds, "秒" }), true);
}
}
}
}
base.updateTextBox(base.keywordInf.keyword + " 已到规定页数，结束", true);
base.updateTextBox("共 " + base.keywordInfList.Count.ToString() + " 件商品查询完毕，其中 " + num.ToString() + "件未检索到数据", true);
base.http.Free();
base.Stoped = true;
}
catch (Exception exception)
{
Log.WriteLog("YHDTh Excute Err:" + exception.Message + " ,Err Stack:" + exception.StackTrace);
}
}
//excution part
public override void Excute()
{
//TODO: implement the excute function
try
{
DateTime begin = DateTime.Now;
string url = string.Empty;
string refererUrl = string.Empty;
string text = string.Empty;
string s = string.Empty;
int num = 0; //count
base.IniDataTable();
base.http = new DoNet4.Utilities.HttpClientHelper(0x4e20);
for (int i = 0; i < base.keywordInfList.Count; i++)
{ //iterate the keyword list
base.keywordInf = base.keywordInfList[i];
if (string.IsNullOrEmpty(base.keywordInf.keyword))
{
continue;
}
base.updateTextBox(base.keywordInf.keyword + " 开始查询", true);
url = "http://1688.com";
byte retry = 0;
while (true)
{
if (base.http.Get(url).IndexOf("阿里巴巴1688.com - 全球领先的采购批发平台") != -1)
{
goto Label_search_taobao;
}
retry = (byte)(retry + 1);
if (retry >= 3) { break; }
base.updateTextBox("阿里巴巴主页打开失败，重试： " + retry.ToString(), true);
}
num++;
continue;
Label_search_taobao:
Thread.Sleep(0x7d0);
System.Text.Encoding gb2312 = System.Text.Encoding.GetEncoding("gb2312");
url = "http://s.1688.com/selloffer/offer_search.htm?keywords=" + HttpUtility.UrlEncode(base.keywordInf.keyword, gb2312).ToUpper();//(base.keywordInf.keyword).ToString();
text = base.http.Get(url);
refererUrl = url;
if (text.Contains("没找到与"))
{
num++;
base.updateTextBox(base.keywordInf.keyword + " 没有找到相关商品", true);
Thread.Sleep(200);
}
else if (text.Contains("淘宝会员（仅限会员名）请在此登录")) {
base.updateTextBox(base.keywordInf.keyword + " 阿里巴巴要求登录", true);
Thread.Sleep(200);
}
else
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(text);
HtmlNode node = null;
int result = 0;
node = document.DocumentNode.SelectSingleNode("//span[@class='sm-widget-offer']");
if (node == null)
{
base.updateTextBox(base.keywordInf.keyword + " node为空（null）!", true);
}
else
{
s = StrUnit.MidStrEx(node.InnerHtml, "<em>", "</em>");
if (!int.TryParse(s, out result))
{
base.updateTextBox(base.keywordInf.keyword + " 获取商品数量错误！ - 2", true);
num++;
}
}
s = StrUnit.MidStrEx(text, "共<em>", "</em>件");
if (!int.TryParse(s, out result))
{
base.updateTextBox(base.keywordInf.keyword + " 获取商品数量错误！ - 2", true);
num++;
}
else
{ //get the useful data, go into iterations
//for ()
base.updateTextBox("共检索到" + s + " 件相关商品， 开始爬取", true);
for (string str6 = this.GetData(text); !string.IsNullOrEmpty(str6); str6 = this.GetData1(text)) //?
{
Thread.Sleep(0x7d0);
text = base.http.Get(str6, refererUrl);
}
TimeSpan span = (TimeSpan)(DateTime.Now - begin);
base.updateTextBox(string.Concat(new object[] { base.keywordInf.keyword, " 获取完毕,耗时：", span.TotalSeconds, "秒" }), true);
}
}
}
base.updateTextBox(base.keywordInf.keyword + " 已到达规定页数，结束", true);
base.updateTextBox("共 " + base.keywordInfList.Count.ToString() + " 件商品查询完毕，其中 " + num.ToString() + "件未检索到数据", true);
base.http.Free();
base.Stoped = true;
}
catch (Exception e)
{
//exception handling
Log.WriteLog("taobaoTh Excute Err:" + e.Message + " ,Err Stack:" + e.StackTrace);
}
}
private string GetData1(string text)
{
string str = string.Empty;
innerHtml = string.Empty; //innerHtml content
string source = string.Empty;
try
{
HtmlDocument document = new HtmlDocument();
document.LoadHtml(text);
HtmlNode node = null;
node = document.DocumentNode.SelectSingleNode("//ul[@id='sm-offer-list']");
if (node == null)
{
base.updateTextBox("获取商品列表失败", true);
throw new Exception("获取商品列表失败");
}
foreach (string str2 in node.InnerHtml.Split(new string[] { "<li t" }, StringSplitOptions.RemoveEmptyEntries))
{
if (str2.IndexOf("-rank") != -1)
{
source = StrUnit.MidStrEx(str2, "rank", "</li>");
DataRow row = base.OutDataTable.NewRow();
row["Platform"] = "1688.com";
row["Keyword"] = base.keywordInf.keyword;
row["ItemID"] = StrUnit.MidStrEx(source, "t-offer-id=\"", "\"");
row["SellerId"] = StrUnit.MidStrEx(source, "t-member-id=\"", "\"");
string goodInfo = string.Empty;
goodInfo = StrUnit.MidStrEx(source, "sm-offer-photo sw-dpl-offer-photo", "</div>");
row["Url"] = StrUnit.MidStrEx(goodInfo, "href=\"", "\"");
row["Title"] = StrUnit.MidStrEx(goodInfo, "title=\"", "\"");
string priceInfo = string.Empty;
priceInfo = StrUnit.MidStrEx(source, "<div class=\"s-widget-offershopwindowprice sm-offer-price sw-dpl-offer-price\">", "</div>");
row["Price"] = StrUnit.MidStrEx(priceInfo, "title=\"¥", "\"");
string companyInfo = string.Empty;
companyInfo = StrUnit.MidStrEx(source, "<div class=\"s-widget-offershopwindowcompanyinfo sm-offer-company sw-dpl-offer-company\">", "</div>");
row["StoreName"] = StrUnit.MidStrEx(companyInfo, "title=\"", "\"");
row["StoreUrl"] = StrUnit.MidStrEx(companyInfo, "href=\"", "\"");
base.OutDataTable.Rows.Add(row);
Thread.Sleep(50);
if (string.IsNullOrEmpty(row["ItemID"].ToString()))
{
throw new Exception("获取商品列表数据失败！ - 1");
}
}
}//end of info retrieving
/*
HtmlNode node2;
//node2 = new HtmlNode();
node2 = document.DocumentNode.SelectSingleNode("//div[@id='fui_widget_4']/span");
string pageSelectionInfo = StrUnit.MidStrEx(text, "<span class=\"fui-paging-list\">", "</span>");
string nextPage = string.Empty;
if (node2 != null)
{
//innerHtml = node.InnerHtml;
foreach (string pageInfo in node2.InnerHtml.Split(new string[] { "<a href=#" }, StringSplitOptions.RemoveEmptyEntries))
{
if (pageInfo.IndexOf("fui-next") != -1)
{
nextPage = StrUnit.MidStrEx(pageInfo, "data-page=\"", "\"");
break;
}
int resultPage = 0;
if (!int.TryParse(nextPage, out resultPage))
{
System.Text.Encoding gb2312 = System.Text.Encoding.GetEncoding("gb2312");
str = "https://s.1688.com/selloffer/offer_search.htm?keywords=" + HttpUtility.UrlEncode(base.keywordInf.keyword, gb2312).ToUpper() + "&beginPage=" + nextPage;
return str;
}
base.updateTextBox(base.keywordInf.keyword + " 第 " + (resultPage-1) + " 页获取完成", true);
if (resultPage > base.keywordInf.endPage)
{
str = string.Empty;
}
}//end of for each
//source = StrUnit.
return str;
}*/
base.updateTextBox(base.keywordInf.keyword + " 第 " + (page) + " 页获取完成", true);
if (page >= base.keywordInf.endPage)
{
str = string.Empty;
return str;
}
++page;
System.Text.Encoding gb2312 = System.Text.Encoding.GetEncoding("gb2312");
str = "https://s.1688.com/selloffer/offer_search.htm?keywords=" + HttpUtility.UrlEncode(base.keywordInf.keyword, gb2312).ToUpper() + "&beginPage=" + page;
return str;
}
catch (Exception e)
{
base.updateTextBox("GetData Err:" + e.Message, true);
}
return str;
}
//Get data
private string GetData(string text)
{
string str = string.Empty;
innerHtml = string.Empty;
string source = string.Empty;
try
{
text = StrUnit.MidStrEx(text, "\"normalList\":[", "</html>");
foreach (string itemInfo in text.Split(new string[] { "{\"productId\":" }, StringSplitOptions.RemoveEmptyEntries))
{
if (itemInfo.IndexOf("productId_f") != -1)
{
string info = "\"productID\":" + itemInfo;
DataRow row = base.OutDataTable.NewRow();
row["Platform"] = "alibaba.com";
row["Keyword"] = base.keywordInf.keyword;
row["ItemID"] = StrUnit.MidStrEx(info, "\"productID\":\"", "\",");
row["SellerID"] = StrUnit.MidStrEx(info, "supplierId\":\"", "\"");
row["Url"] = StrUnit.MidStrEx(info, "productHref\":\"//", "\"");
row["Title"] = StrUnit.MidStrEx(info, "productPuretitle\":\"", "\"");
row["Price"] = StrUnit.MidStrEx(info, "\"price\":\"US $", "\"");
row["StoreName"] = StrUnit.MidStrEx(info, "\"supplierName\":\"", "\"");
row["StoreUrl"] = StrUnit.MidStrEx(info, "supplierHref\":\"", "\"");
base.OutDataTable.Rows.Add(row);
Thread.Sleep(100);
if (string.IsNullOrEmpty(row["ItemID"].ToString()))
{
throw new Exception("获取商品列表数据失败！ - 1");
}
}
else
{
throw new Exception("Json data 获取错误");
}
}//end of foreach
base.updateTextBox(base.keywordInf.keyword + " 第 " + (page) + " 页获取完成", true);
if (page >= base.keywordInf.endPage)
{
str = string.Empty;
return(str);
}
++page;
str = "http://www.alibaba.com/products/F0/" + base.keywordInf.keyword + "/" + page + ".html";
return(str);
/*
* HtmlDocument document = new HtmlDocument();
* document.LoadHtml(text);
* HtmlNode node = null;
* node = document.DocumentNode.SelectSingleNode("//div[@data-content='abox-ProductNormalList']");
*
* if(node == null)
* {
* base.updateTextBox("获取商品列表失败！",true);
* throw new Exception("获取商品列表失败");
* }
*
* foreach (string itemInfo in node.InnerHtml.Split(new string[] { "<div class=\"m-product-item"},StringSplitOptions.None))
* {
* if (itemInfo.IndexOf("data-role=\"")!=-1)
* {
*
* }
* }//end of foreach
*
* base.updateTextBox(base.keywordInf.keyword + " 第 " + (page) + " 页获取完成", true);
* if (page >= base.keywordInf.endPage)
* {
* str = string.Empty;
* return str;
* }
++page;
* str = "http://www.alibaba.com/products/F0/"+base.keywordInf.keyword+"/"+page+".html";
* return str;
*/
}
catch (Exception e)
{
base.updateTextBox("GetData Err:" + e.Message, true);
}
return(str);
}