void CrawSite_CrawInfoEvent(CrawInfoCls CrawInfo)
{
dbModelDataContext db = new dbModelDataContext();
var iCheck = (from i in db.CRItemPrices
where i.SiteCode == CrawInfo.SiteCode && i.ItemSiteCode == CrawInfo.ItemSiteCode
select i).ToList();
if (iCheck.Count > 0)
{
db.CRItemPrices.DeleteAllOnSubmit(iCheck);
db.SubmitChanges();
}
CRItemPrice iNew = new CRItemPrice();
iNew.SiteCode = CrawInfo.SiteCode;
iNew.ItemSiteCode = CrawInfo.ItemSiteCode;
iNew.ItemBrand = CrawInfo.ItemBrand;
iNew.ItemSiteName = CrawInfo.ItemSiteName;
iNew.SiteItemGroup = CrawInfo.SiteItemGroup;
iNew.SitePrice = CrawInfo.SitePrice;
iNew.UrlCheck = CrawInfo.UrlCheck;
iNew.CreateDate = DateTime.Now;
iNew.CreateBy = "SoftWare";
db.CRItemPrices.InsertOnSubmit(iNew);
db.SubmitChanges();
cr_LogEvent(string.Format("Mã hàng: {0}, giá: {1}, Thương hiệu: {2}", CrawInfo.ItemSiteCode, CrawInfo.SitePrice, CrawInfo.ItemBrand));
}
public void RaiseCrawInfo(CrawInfoCls CrawInfo)
{
if (CrawInfoEvent != null)
{
CrawInfoEvent(CrawInfo);
}
}
public void LoadCrawSite(string url)
{
int CurrentItem = 0;
string baseUrl = string.Empty;
int CurrentPage = 1;
baseUrl = url + string.Format("?page={0}", CurrentPage.ToString());
HtmlAgilityPack.HtmlDocument document = LoadPage(baseUrl);
var listNodes = document.DocumentNode.SelectNodes("//ul[@class='cate ']/li");
if (listNodes == null)
{
return;
}
foreach (var ulItem in listNodes)
{
if (GlobalEnv.StopRuning)
{
break;
}
string ItemSiteName = ulItem.SelectSingleNode(ulItem.XPath + "//h3") != null?ulItem.SelectSingleNode(ulItem.XPath + "//h3").InnerText : string.Empty;
string SitePrice = ulItem.SelectSingleNode(ulItem.XPath + "/strong") != null?ulItem.SelectSingleNode(ulItem.XPath + "/strong").InnerText : string.Empty;
string linkItemCode = ulItem.SelectSingleNode(ulItem.XPath + "/div/a") != null?ulItem.SelectSingleNode(ulItem.XPath + "/div/a").Attributes["href"].Value : string.Empty;
string ItemSiteCode = linkItemCode.Substring(linkItemCode.LastIndexOf('/') + 1, linkItemCode.Length - linkItemCode.LastIndexOf('/') - 1);
string ItemBrand = ItemSiteCode.Substring(0, ItemSiteCode.IndexOf('-'));
SitePrice = SitePrice.Replace("₫", string.Empty);
SitePrice = SitePrice.Replace(".", string.Empty);
CrawInfoCls CrawInfo = new CrawInfoCls();
CrawInfo.SiteItemGroup = this.SiteItemGroup;
CrawInfo.ItemSiteCode = ItemSiteCode;
CrawInfo.SiteCode = this.SiteCode;
CrawInfo.ItemBrand = ItemBrand;
CrawInfo.SitePrice = Convert.ToDouble(SitePrice);
CrawInfo.ItemSiteName = ItemSiteName;
CrawInfo.UrlCheck = baseUrl;
RaiseCrawInfo(CrawInfo);
}
GlobalEnv.Waiting = false;
}
public void LoadCrawSite(string url)
{
string baseUrl = string.Empty;
int CurrentPage = 1;
baseUrl = url + string.Format("?p={0}", CurrentPage.ToString());
HtmlAgilityPack.HtmlDocument document = LoadPage(baseUrl);
var listNodes = document.DocumentNode.SelectNodes("//div[@class='category-products']/ol[@class='products-list']");
if (document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li") == null)
{
return;
}
var listPages = document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li").Count;
int MaxPage = 0;
for (int iIndex = 0; iIndex < listPages; iIndex++)
{
if (document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li")[iIndex] != null)
{
int checkPage = 0;
int.TryParse(document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li")[iIndex].InnerText, out checkPage);
if (checkPage > MaxPage)
{
MaxPage = checkPage;
}
}
}
string fisrtItemCode = string.Empty;
bool breakLoop = false;
while (CurrentPage < MaxPage + 2)
{
RaiseLog("Bắt đầu quét trang: " + baseUrl);
CurrentPage += 1;
baseUrl = url + string.Format("?p={0}", CurrentPage.ToString());
var itemNodes = document.DocumentNode.SelectNodes("//div[@class='category-products']/ol[@class='products-list']/li[@class='item']");
if (itemNodes != null)
{
foreach (var iNode in itemNodes)
{
if (GlobalEnv.StopRuning)
{
break;
}
string ItemSiteName = iNode.SelectSingleNode("div[@class='product-shop']/div/div/h2[@class='product-name']") != null?iNode.SelectSingleNode("div[@class='product-shop']/div/div/h2[@class='product-name']").InnerText.Trim() : string.Empty;
string ItemBrand = ItemSiteName.Split(' ').Count <string>() > 1 ? ItemSiteName.Split(' ')[ItemSiteName.Split(' ').Count <string>() - 2] : string.Empty;
string ItemSiteCode = ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower()) > 0 ? ItemSiteName.Substring(ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower()), ItemSiteName.Length - ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower())).Trim() : ItemSiteName.Substring(ItemSiteName.LastIndexOf(" "), ItemSiteName.Length - ItemSiteName.LastIndexOf(" "));
string SitePrice = iNode.SelectSingleNode("div[@class='product-shop']/div/div/div[@class='price-box']/p[@class='special-price']/span[@class='price']") != null?iNode.SelectSingleNode("div[@class='product-shop']/div/div/div[@class='price-box']/p[@class='special-price']/span[@class='price']").InnerText.Trim() : string.Empty;
if (string.IsNullOrEmpty(SitePrice))
{
SitePrice = iNode.SelectSingleNode("div[@class='product-shop']/div/div/div[@class='price-box']/span[@class='regular-price']/span[@class='price']") != null?iNode.SelectSingleNode("div[@class='product-shop']/div/div/div[@class='price-box']/span[@class='regular-price']/span[@class='price']").InnerText.Trim() : string.Empty;
}
SitePrice = SitePrice.Replace("₫", string.Empty);
SitePrice = SitePrice.Replace(".", string.Empty);
SitePrice = SitePrice.Trim();
if (string.IsNullOrEmpty(SitePrice))
{
continue;
}
if (!string.IsNullOrEmpty(ItemSiteCode) && ItemSiteCode == fisrtItemCode)
{
breakLoop = true;
break;
}
CrawInfoCls CrawInfo = new CrawInfoCls();
CrawInfo.SiteItemGroup = this.SiteItemGroup;
CrawInfo.ItemSiteCode = ItemSiteCode;
CrawInfo.SiteCode = this.SiteCode;
CrawInfo.ItemBrand = ItemBrand;
CrawInfo.SitePrice = Convert.ToDouble(SitePrice);
CrawInfo.ItemSiteName = ItemSiteName;
CrawInfo.UrlCheck = baseUrl;
//Gán mã số 1 để check
if (string.IsNullOrEmpty(fisrtItemCode))
{
fisrtItemCode = ItemSiteCode;
}
RaiseCrawInfo(CrawInfo);
}
}
//Xử lý chốt
document = LoadPage(baseUrl);
listNodes = document.DocumentNode.SelectNodes("//div[@class='category-products']/ol[@class='products-list']");
if (listNodes == null)
{
continue;
}
for (int iIndex = 0; iIndex < listPages; iIndex++)
{
if (document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li") == null)
{
break;
}
if (document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li")[iIndex] != null)
{
int checkPage = 0;
int.TryParse(document.DocumentNode.SelectNodes("//div[@class='category-products']/div[@class='toolbar']/div[@class='pager']/div[@class='pages']/ol/li")[iIndex].InnerText, out checkPage);
if (checkPage > MaxPage)
{
MaxPage = checkPage;
}
}
}
}
GlobalEnv.Waiting = false;
}
public void LoadCrawSite(string url)
{
string baseUrl = string.Empty;
int CurrentPage = 1;
if (url.LastIndexOf("/") < url.Length - 2)
{
baseUrl = url + string.Format("/?trang={0}", CurrentPage.ToString());
}
else
{
baseUrl = url + string.Format("?trang={0}", CurrentPage.ToString());
}
HtmlAgilityPack.HtmlDocument document = LoadPage(baseUrl);
var listNodes = document.DocumentNode.SelectNodes("//div[@class='pca-pl-l']");
string fisrtItemCode = string.Empty;
bool breakLoop = false;
while (listNodes != null && breakLoop == false)
{
CurrentPage += 1;
if (url.LastIndexOf('/') >= url.Length - 1)
{
baseUrl = url + string.Format("?trang={0}", CurrentPage.ToString());
}
else
{
baseUrl = url + string.Format("/?trang={0}", CurrentPage.ToString());
}
RaiseLog("Bắt đầu quét trang: " + baseUrl);
var itemNodes = document.DocumentNode.SelectNodes("//div[@class='pca-pl-l']/ul[@class='pl-item-ul']/li");
if (itemNodes == null)
{
break;
}
foreach (var iNode in itemNodes)
{
if (GlobalEnv.StopRuning)
{
break;
}
string ItemSiteName = iNode.SelectSingleNode("div/p[@class='pl-item-name']") != null?iNode.SelectSingleNode("div/p[@class='pl-item-name']").InnerText.Trim() : string.Empty;
string ItemBrand = iNode.SelectSingleNode("div/p[@class='pl-item-brand']").FirstChild != null?iNode.SelectSingleNode("div/p[@class='pl-item-brand']").FirstChild.InnerText.Trim() : string.Empty;
string ItemSiteCode = ItemSiteName.Substring(ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower()) + ItemBrand.Length, ItemSiteName.Length - ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower()) - ItemBrand.Length).Trim();
string iProductPricehtml = iNode.SelectSingleNode("div/div[@class='pl-item-price']/div[@class='pl-item-pbuy']/div[@class='draw-price']/div[@class='draw-price-content']") != null?iNode.SelectSingleNode("div/div[@class='pl-item-price']/div[@class='pl-item-pbuy']/div[@class='draw-price']/div[@class='draw-price-content']").InnerHtml : string.Empty;
string[] arrItemPrice = Regex.Split(iProductPricehtml, "</span>");
string[] realPrice = arrItemPrice.Where(a => a.Contains("drw-pri-thumb-view")).ToArray <string>();
string SitePrice = "";
string getPrice = "drw-pri-thumb-view-";
foreach (string iPrice in realPrice)
{
if (iPrice.IndexOf("drw-pri-thumb-stt-0") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-1") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-2") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-3") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-4") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-5") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-6") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
if (iPrice.IndexOf("drw-pri-thumb-stt-7") > 0)
{
SitePrice += !string.IsNullOrEmpty(iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1)) ? iPrice.Substring(iPrice.IndexOf(getPrice) + getPrice.Length, 1) : string.Empty;
}
}
if (string.IsNullOrEmpty(SitePrice))
{
continue;
}
if (!string.IsNullOrEmpty(ItemSiteCode) && ItemSiteCode == fisrtItemCode)
{
breakLoop = true;
break;
}
CrawInfoCls CrawInfo = new CrawInfoCls();
CrawInfo.ItemSiteCode = ItemSiteCode;
CrawInfo.SiteCode = this.SiteCode;
CrawInfo.ItemBrand = ItemBrand;
CrawInfo.SitePrice = Convert.ToDouble(SitePrice) * 1000;
CrawInfo.ItemSiteName = ItemSiteName;
CrawInfo.SiteItemGroup = this.SiteItemGroup;
CrawInfo.UrlCheck = baseUrl;
//Gán mã số 1 để check
if (string.IsNullOrEmpty(fisrtItemCode))
{
fisrtItemCode = ItemSiteCode;
}
RaiseCrawInfo(CrawInfo);
}
//Xử lý chốt
document = LoadPage(baseUrl);
listNodes = document.DocumentNode.SelectNodes("//div[@class='pca-pl-l']");
}
GlobalEnv.Waiting = false;
}
public void LoadCrawSite(string url)
{
string baseUrl = string.Empty;
int CurrentPage = 1;
baseUrl = url + string.Format("?&pageIndex={0}", CurrentPage.ToString());
HtmlAgilityPack.HtmlDocument document = LoadPage(baseUrl);
var listNodes = document.DocumentNode.SelectNodes("//div[@class='row category-child']/div[@class='col-md-3 col-sm-4 col-xs-6 product']");
string fisrtItemCode = string.Empty;
bool breakLoop = false;
while (listNodes != null && breakLoop == false)
{
RaiseLog("Bắt đầu quét trang: " + baseUrl);
CurrentPage += 1;
baseUrl = url + string.Format("?&pageIndex={0}", CurrentPage.ToString());
var itemNodes = document.DocumentNode.SelectNodes("//div[@class='col-md-3 col-sm-4 col-xs-6 product']/div[@class='product-info']");
foreach (var iNode in itemNodes)
{
if (GlobalEnv.StopRuning)
{
break;
}
string ItemSiteName = iNode.SelectSingleNode("h6/a") != null?iNode.SelectSingleNode("h6/a").InnerText.Trim() : string.Empty;
string ItemBrand = iNode.SelectSingleNode("div/img[@alt]").Attributes["alt"].Value != null?iNode.SelectSingleNode("div/img[@alt]").Attributes["alt"].Value.Trim() : string.Empty;
string ItemSiteCode = ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower()) > 0 ? ItemSiteName.Substring(ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower()), ItemSiteName.Length - ItemSiteName.ToLower().LastIndexOf(ItemBrand.ToLower())).Trim() : ItemSiteName.Substring(ItemSiteName.LastIndexOf(" "), ItemSiteName.Length - ItemSiteName.LastIndexOf(" "));
string SitePrice = iNode.SelectSingleNode("div[@class='priceInfo']/span[@class='price']") != null?iNode.SelectSingleNode("div[@class='priceInfo']/span[@class='price']").InnerText.Trim() : string.Empty;
SitePrice = SitePrice.Replace("₫", string.Empty);
SitePrice = SitePrice.Replace(".", string.Empty);
if (string.IsNullOrEmpty(SitePrice))
{
continue;
}
if (!string.IsNullOrEmpty(ItemSiteCode) && ItemSiteCode == fisrtItemCode)
{
breakLoop = true;
break;
}
CrawInfoCls CrawInfo = new CrawInfoCls();
CrawInfo.SiteItemGroup = this.SiteItemGroup;
CrawInfo.ItemSiteCode = ItemSiteCode;
CrawInfo.SiteCode = this.SiteCode;
CrawInfo.ItemBrand = ItemBrand;
CrawInfo.SitePrice = Convert.ToDouble(SitePrice);
CrawInfo.ItemSiteName = ItemSiteName;
CrawInfo.UrlCheck = baseUrl;
//Gán mã số 1 để check
if (string.IsNullOrEmpty(fisrtItemCode))
{
fisrtItemCode = ItemSiteCode;
}
RaiseCrawInfo(CrawInfo);
}
//Xử lý chốt
document = LoadPage(baseUrl);
listNodes = document.DocumentNode.SelectNodes("//div[@class='row category-child']/div[@class='col-md-3 col-sm-4 col-xs-6 product']");
}
GlobalEnv.Waiting = false;
}
public void LoadCrawSite(string url)
{
string baseUrl = url;
HtmlAgilityPack.HtmlDocument document = LoadPage(baseUrl);
if (document.DocumentNode.SelectSingleNode("//html/body/div/div/script") == null)
{
return;
}
var htmlScript = document.DocumentNode.SelectSingleNode("//html/body/div/div/script").InnerText.Trim();
string jsScrip = htmlScript.Substring(htmlScript.IndexOf("products"), htmlScript.LastIndexOf("]") + 1 - htmlScript.IndexOf("products"));
string[] arrJson = jsScrip.Split('}');
foreach (var iStr in arrJson)
{
if (GlobalEnv.StopRuning)
{
break;
}
string iCheck = iStr.Substring(iStr.IndexOf('{') + 1, iStr.Length - iStr.IndexOf('{') - 1).Trim();
JObject jObject = null;
try
{
jObject = JObject.Parse("{" + iCheck + "}");
}
catch
{
continue;
}
string ItemSiteName = jObject["productname"].ToString();
string SitePrice = jObject["price"].ToString();
if (SitePrice.IndexOf("class=price>") > 0)
{
SitePrice = SitePrice.Substring(SitePrice.IndexOf("class=price>") + "class=price>".Length, SitePrice.Length - SitePrice.IndexOf("class=price>") - "class=price>".Length);
SitePrice = SitePrice.Substring(0, SitePrice.IndexOf("</span>"));
SitePrice = SitePrice.Replace(",", string.Empty);
SitePrice = SitePrice.Replace(".", string.Empty);
SitePrice = SitePrice.Trim();
}
else
{
SitePrice = "0";
}
string ItemBrand = jObject["brand"].ToString();
string categorie = jObject["categorie"].ToString();
string link = jObject["link"].ToString();
string id = jObject["id"].ToString();
string ItemSiteCode = "";
if (link.ToLower().IndexOf(ItemBrand.ToLower()) > 0 && link.ToLower().IndexOf(id.ToLower()) > 0)
{
ItemSiteCode = link.Substring(link.ToLower().IndexOf(ItemBrand.ToLower()) + ItemBrand.Length, link.ToLower().IndexOf(id.ToLower()) - link.ToLower().IndexOf(ItemBrand.ToLower()) - ItemBrand.Length).Replace("-", string.Empty);
}
CrawInfoCls CrawInfo = new CrawInfoCls();
CrawInfo.SiteItemGroup = this.SiteItemGroup;
CrawInfo.ItemSiteCode = ItemSiteCode;
CrawInfo.SiteCode = this.SiteCode;
CrawInfo.ItemBrand = ItemBrand;
CrawInfo.SitePrice = Convert.ToDouble(SitePrice);
CrawInfo.ItemSiteName = ItemSiteName;
CrawInfo.UrlCheck = baseUrl;
RaiseCrawInfo(CrawInfo);
}
GlobalEnv.Waiting = false;
}
