/// <summary>
/// Get product information
/// </summary>
/// <param name="productLink">Example: https://www.jomashop.com/tissot-watch-t0064071603300.html</param>
private Models.Product GetProductInformation(string productLink)
{
//Create product to save crawled data
Models.Product product = new Models.Product();
//Redirect to site by URL
//browser.Navigate().GoToUrl("https://www.jomashop.com/tissot-watch-t0064071603300.html");
browser.Navigate().GoToUrl(productLink);
//browser.Navigate().GoToUrl("https://dosi-in.com/tee-black-myson-mix-reflective/");
//Select elements by CSS Selector (easiest way)
//You can also select element by ID, Class, Name, XPath,...
//Get brand by CSS Attribute Selectors (https://www.w3schools.com/css/css_attribute_selectors.asp)
//var element = browser.FindElement(By.CssSelector("[itemprop=\"brand manufacturer\"]>a"));
//product.Attribute1Value = element.GetAttribute("innerHTML"); //OuterHTML will give full element HTML code
//Get price by CSS Selectors (https://www.w3schools.com/css/css_selectors.asp)
//element = browser.FindElement(By.CssSelector("#final-price")); //No timeout, wait until page loaded
//string finalPrice = element.GetAttribute("innerHTML");
//finalPrice = finalPrice.Replace("$", "");
//double finalPriceInVnd = Double.Parse(finalPrice) * 24300 * 1.1 + 350000;
// product.SalePrice = finalPriceInVnd;
List <IWebElement> elementList0 = new List <IWebElement>();
elementList0.AddRange(browser.FindElements(By.CssSelector(".bigEntrance.banner")));
if (elementList0.Count < 1)
{
List <IWebElement> elementList1 = new List <IWebElement>();
elementList1.AddRange(browser.FindElements(By.CssSelector(".product-short_description li:first-child")));
if (elementList1.Count > 0)
{
string SKU = browser.FindElement(By.CssSelector(".product-short_description li:first-child")).GetAttribute("innerHTML");
product.SKU = SKU.Replace("Mã sản phẩm: ", "");
}
string Name = browser.FindElement(By.CssSelector("h1.mainbox-title")).GetAttribute("innerHTML");
product.Name = Name;
double finalPriceInVnd = Double.Parse(browser.FindElement(By.CssSelector(".price-num")).GetAttribute("innerHTML")) * 1.2 + 50000;
product.SalePrice = finalPriceInVnd;
List <IWebElement> elementList2 = new List <IWebElement>();
elementList2.AddRange(browser.FindElements(By.CssSelector(".strike span:first-child")));
if (elementList2.Count > 0)
{
Double retailPriceInVnd = Double.Parse(browser.FindElement(By.CssSelector(".strike span:first-child")).GetAttribute("innerHTML")) * 1.2 + 50000;
product.RegularPrice = retailPriceInVnd;
}
List <IWebElement> elementList3 = new List <IWebElement>();
elementList3.AddRange(browser.FindElements(By.CssSelector("label.radio.float-left.dosi_get_change_option.dosi_option_size")));
if (elementList3.Count > 0)
{
List <string> sizes = new List <string>();
foreach (var size in browser.FindElements(By.CssSelector("label.radio.float-left.dosi_get_change_option.dosi_option_size")))
{
sizes.Add(size.GetAttribute("innerHTML").Trim());
}
product.Attribute1Value = String.Join(", ", sizes);
}
List <string> images = new List <string>();
foreach (var image in browser.FindElements(By.CssSelector("a.cm-image-previewer.cm-previewer.previewer img")))
{
images.Add(image.GetAttribute("src").Replace(".webp", ""));
}
product.Images = String.Join(", ", images);
//string Description = browser.FindElement(By.CssSelector("#myTabproductContent div")).GetAttribute("innerHTML");
//product.Description = Description.Trim();
string brand = browser.FindElement(By.CssSelector("a.product_company")).GetAttribute("innerHTML");
product.Attribute2Value = brand.Trim();
product.Categories = product.Categories + "Brand>" + brand.Trim();
// Có 6 PriceRange: > 2.000.000, 1.000.000 - 2.000.000, 500.000 - 1.000.000, 350.000 - 500.000, 150.000 - 350.000, 95.000 - 150.000
string PriceRange = "0 - 95.000";
if (finalPriceInVnd > 2000000)
{
PriceRange = ">2000000";
}
if (finalPriceInVnd >= 1000000 && finalPriceInVnd <= 2000000)
{
PriceRange = "1.000.000 - 2.000.000";
}
if (finalPriceInVnd >= 350000 && finalPriceInVnd <= 500000)
{
PriceRange = "350.000 - 500.000";
}
if (finalPriceInVnd >= 150000 && finalPriceInVnd <= 350000)
{
PriceRange = "150.000 - 350.000";
}
if (finalPriceInVnd >= 95000 && finalPriceInVnd <= 150000)
{
PriceRange = "95.000 - 150.000";
}
product.Attribute3Value = PriceRange;
//string shortDes = browser.FindElement(By.CssSelector(".cm-disable-empty-files.cm-processed-form ul.product-short_description li")).GetAttribute("innerHTML");
//product.ShortDescription = shortDes.Replace("DOSI", "VENUS").Trim();
//----------------------------------------------
}
return(product);
}
/// <summary>
/// Get product information
/// </summary>
/// <param name="productLink">Example: https://www.jomashop.com/tissot-watch-t0064071603300.html</param>
private Models.Product GetProductInformation(string productLink)
{
//Create product to save crawled data
Models.Product product = new Models.Product();
//Redirect to site by URL
//browser.Navigate().GoToUrl("https://www.jomashop.com/tissot-watch-t0064071603300.html");
browser.Navigate().GoToUrl(productLink);
//Sử dụng try catch để over lỗi
//Lấy loại (Type) Dựa trên breadcumb
try
{
var elementtype = browser.FindElement(By.CssSelector("[class=\"mc-brea\"]"));
string breadcumb = elementtype.GetAttribute("innerHTML");
string temp = Regex.Match(breadcumb, "<ul>.*?</li>", RegexOptions.Singleline).Value;
breadcumb = breadcumb.Replace(temp, "").Trim(); //Replace Trang chủ
temp = Regex.Match(breadcumb, "<li>.*?</li>", RegexOptions.Singleline).Value;
breadcumb = breadcumb.Replace(temp, "").Trim(); //Replace Máy đổi trả
temp = Regex.Match(breadcumb, "<li>.*?</li>", RegexOptions.Singleline).Value;
product.Type = Regex.Match(temp, "<li><a.*?>(.*?)</a></li>", RegexOptions.Singleline).Groups[1].Value; //Lấy giá trị Type
breadcumb = breadcumb.Replace(temp, "").Trim(); //Replace Type
//Lấy Brand dựa trên link trên (Brand Attribute 1)
product.Attribute1Value = Regex.Match(breadcumb, "<li><a.*?>(.*?)</a></li>", RegexOptions.Singleline).Groups[1].Value;//Lấy giá trị Brand
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nType & Brand went wrong."); }
//Lấy SKU (SKU)
try
{
var element0 = browser.FindElement(By.CssSelector("[class=\"mc-ctname\"]"));
var SKU = element0.GetAttribute("innerHTML");
SKU = Regex.Match(SKU, "<span>(.*?)</span>", RegexOptions.Singleline).Groups[1].Value;
product.SKU = Regex.Replace(SKU, "\\W", "").Trim();
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nSKU went wrong."); }
//Lấy tên (Name)
try
{
var element1 = browser.FindElement(By.CssSelector("[class=\"mc-ctname\"]"));
var name = element1.GetAttribute("innerHTML");
product.Name = Regex.Replace(name, "<span>.*?</span>", "").Trim();
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nName went wrong."); }
//Lấy màu (Attribute 2)
try
{
var element2 = browser.FindElement(By.CssSelector("[class=\"mc-ctclo\"]"));
var color = element2.GetAttribute("innerHTML");
product.Attribute2Value = Regex.Replace(color, "<i.*?</i>", "").Trim();
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nColor went wrong."); }
//Lấy giá hiện tại (Regular Price)
try
{
var element3 = browser.FindElement(By.CssSelector("[class=\"mc-ctpri1\"]"));
var price1 = element3.GetAttribute("innerHTML");
price1 = Regex.Replace(price1, ".*?<span>", "").Trim();
price1 = Regex.Replace(price1, "đ</span>.*", "", RegexOptions.Singleline).Trim();
price1 = price1.Replace(".", "");
product.RegularPrice = Double.Parse(price1);
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nPrice went wrong."); }
//Lấy giá máy mới (Attribute 3)
try
{
var element4 = browser.FindElement(By.CssSelector("[class=\"mc-ctpri2\"]"));
product.Attribute3Value = element4.GetAttribute("innerHTML");
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nNewPrice went wrong."); }
//Tiết kiệm (Attribute 4)
try
{
var element5 = browser.FindElement(By.CssSelector("[class=\"mc-ctpri3\"]"));
var price3 = element5.GetAttribute("innerHTML");
product.Attribute4Value = Regex.Replace(price3, "<p>|</p>", "").Trim();
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nSavings went wrong."); }
//Tình trạng (Attribute 5)
try
{
var element6 = browser.FindElement(By.CssSelector("[class=\"mc-ctttm\"]"));
string status = element6.GetAttribute("innerHTML");
string stttemp = Regex.Match(status, "<li>(.*?)</li>", RegexOptions.Singleline).Groups[1].Value;
product.Attribute5Value = stttemp;
status = status.Replace(stttemp, "");
//Phụ kiện (Attribute 6)
stttemp = Regex.Match(status, "<li></li>.*?<li>(.*?)</li>", RegexOptions.Singleline).Groups[1].Value;
product.Attribute6Value = stttemp;
status = status.Replace(stttemp, "");
//Bảo hành (Attribute 7)
stttemp = Regex.Match(status, "<li></li>.*?<li></li>.*?<li>(.*?)</li>", RegexOptions.Singleline).Groups[1].Value;
product.Attribute7Value = stttemp;
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nStatus, Acess & Grua went wrong."); }
//Thông số kĩ thuật (Description)
try
{
var element7 = browser.FindElement(By.CssSelector("[class=\"modal-body tskt-popct\"]"));
string infor = element7.GetAttribute("innerHTML");
infor.Trim();
infor = infor.Replace("@", "&");
product.Description = infor;
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nDecript went wrong."); }
//Hình ảnh (Images)
try
{
var element8 = browser.FindElement(By.CssSelector("[class=\"slick-list draggable\"]"));
string img = element8.GetAttribute("innerHTML");
string imgLinks = "";
string pattern = "https://.*?\"";
Regex r = new Regex(pattern);
foreach (Match m in r.Matches(img))
{
imgLinks += m.Value + ", ";
//Hiển thị kết quả
}
//loại bỏ cdn. mới truy cập được
imgLinks = imgLinks.Replace("cdn.", "").Replace("\"", "");
imgLinks = imgLinks.TrimEnd(',');
product.Images = imgLinks;
}
catch (Exception e)
{ Console.WriteLine(productLink + "\nImage went wrong."); }
//lấy từng địa chỉ cho từng biến (2 địa chỉ)
//string imageLink1 = Regex.Match(img, "src=\"(.*?)\"", RegexOptions.Singleline).Groups[1].Value;
//imageLink1 = imageLink1.Replace("cdn.", "");
//img = img.Replace("src=\"" + imageLink1 + "\"", "");
//string imageLink2 = Regex.Match(img, "src=\"(.*?)\"", RegexOptions.Singleline).Groups[1].Value;
//imageLink2 = imageLink2.Replace("cdn.", "");
//Lấy Địa chỉ cửa hàng (Địa chỉ -> Attribute8, link ->Attribute9)
try
{
var element9 = browser.FindElement(By.CssSelector("[class=\"mc-ctlocit\"]"));
string location = element9.GetAttribute("innerHTML");
var address = Regex.Match(location, "(.*?)<a", RegexOptions.Singleline).Groups[1].Value;
product.Attribute8Value = address;
string locLink = Regex.Match(location, "href=\"(.*?)\"", RegexOptions.Singleline).Groups[1].Value;
product.Attribute9Value = locLink;
}
catch (Exception e)
{ Console.WriteLine(productLink + "\n Address went wrong."); }
/* //Select elements by CSS Selector (easiest way)
* //You can also select element by ID, Class, Name, XPath,...
* //Get brand by CSS Attribute Selectors (https://www.w3schools.com/css/css_attribute_selectors.asp)
* var element = browser.FindElement(By.CssSelector("[itemprop=\"brand manufacturer\"]>a"));
* product.Attribute1Value = element.GetAttribute("innerHTML"); //OuterHTML will give full element HTML code
*
* //Get price by CSS Selectors (https://www.w3schools.com/css/css_selectors.asp)
* element = browser.FindElement(By.CssSelector("#final-price")); //No timeout, wait until page loaded
* string finalPrice = element.GetAttribute("innerHTML");
* finalPrice = finalPrice.Replace("$", "");
* double finalPriceInVnd = Double.Parse(finalPrice) * 24300 * 1.1 + 350000;
* product.SalePrice = finalPriceInVnd;
*/
//----------------------------------------------
return(product);
}
/// <summary>
/// Get product information
/// </summary>
/// <param name="productLink">Example
private Models.Product GetProductInformation(string productLink)
{
//Create product to save crawled data
Models.Product product = new Models.Product();
//Redirect to site by URL
//browser.Navigate().GoToUrl()
System.Threading.Thread.Sleep(new Random().Next(2) * 1000); //Sleep random from 1-5 seconds
WebDriverWait wait = new WebDriverWait(this.browser, TimeSpan.FromSeconds(120));
browser.Navigate().GoToUrl(productLink);
//Select elements by CSS Selector (easiest way)
//You can also select element by ID, Class, Name, XPath,...
//Select elements by CSS Selector (easiest way)
//You can also select element by ID, Class, Name, XPath,...
//Get brand by CSS Attribute Selectors (https://www.w3schools.com/css/css_attribute_selectors.asp)
wait.Until((x) =>
{
return(((IJavaScriptExecutor)this.browser).ExecuteScript("return document.readyState").Equals("complete"));
});
//Get SKU
bool test;
if (test = (verify(browser, "[itemprop = \"sku\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[itemprop = \"sku\"]"));
product.SKU = element.GetAttribute("innerHTML");
}
//Get Name
if (test = (verify(browser, "[itemprop =\"url\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[itemprop =\"url\"]"));
product.Name = element.GetAttribute("innerHTML");
}
//Get Regular price
if (test = (verify(browser, "[class=\"ngachngang\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[class=\"ngachngang\"]"));
var RegularPrice = element.GetAttribute("innerHTML");
RegularPrice = RegularPrice.Replace(" VNĐ", ""); // Remove Currency
double finalRegularPrice = Double.Parse(RegularPrice); // Parse to Double
product.RegularPrice = finalRegularPrice;
}
//Get Sale Price
if (test = (verify(browser, "[class=\"price_sale\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[class=\"price_sale\"]"));
var SalePrice = element.GetAttribute("innerHTML");
if (SalePrice != "Liên hệ")
{
SalePrice = SalePrice.Replace(" VNĐ", ""); // Remove Currency
double finalSalePrice = Double.Parse(SalePrice); // Parse to Double
product.SalePrice = finalSalePrice;
}
else
{
SalePrice = SalePrice.Replace("Liên hệ", "9999999"); // Remove Currency
double finalSalePrice = Double.Parse(SalePrice); // Parse to Double
product.SalePrice = finalSalePrice;
}
}
//Get Description
if (test = (verify(browser, "[class=\"view-content\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[class=\"view-content\"]"));
var Description = element.GetAttribute("innerText");
product.Description = Description.Replace("tại BinhMinhDigital", "").Replace("/s", "").Replace("\r", "").Replace("\n", "");
}
//Get Image
//if (test = (verify(browser, "[itemprop=\"image\"]") == true))
//{
// var element = browser.FindElement(By.CssSelector("[itemprop=\"image\"]"));
// product.Images = element.GetAttribute("src");
//}
// RegularExpression
//var eelement = browser.FindElement(By.XPath("//picture"));
//var ex = eelement.GetAttribute("innerHTML");
//product.Images = Regex.Match(ex.ToString(), @"img src(.*?)alt").Value.Replace("img src=\"", "").Replace("\" alt", "");
//Get Brand
if (test = (verify(browser, "[itemprop=\"brand\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[itemprop=\"brand\"]"));
product.Attribute1Value = element.GetAttribute("innerHTML");
}
//Get AdvanceInfor
if (test = (verify(browser, "[class=\"view-content\"]") == true))
{
var element = browser.FindElement(By.CssSelector("[class=\"product-recap\"]"));
var AdvanceInfor = element.GetAttribute("innerText");
product.Attribute2Value = AdvanceInfor = AdvanceInfor.Replace("TÍNH NĂNG NỔI BẬT", "").Replace("/s", "").Replace("\r", "").Replace("\n", "");
}
//Attribute Price
if (product.SalePrice <= 1000000)
{
product.Attribute3Value = "<= 1 000 000";
}
;
if (product.SalePrice > 1000000 && product.SalePrice <= 5000000)
{
product.Attribute3Value = "1 000 000 - 5 000 000";
}
;
if (product.SalePrice > 5000000 && product.SalePrice <= 20000000)
{
product.Attribute3Value = "5 000 000 - 20 000 000";
}
;
if (product.SalePrice > 2000000)
{
product.Attribute3Value = "> 20 000 000";
}
;
//Get category - brand, thuộc tính category, brand > thuộc tính category
product.Categories = "Chân Camera" + ">" + product.Attribute1Value;
//Get Images Galleries
string list = "";
string htmlpage = browser.PageSource;
var ListImages = Regex.Matches(htmlpage, @"data-standard(.*?)jpg", RegexOptions.Singleline);
foreach (var course in ListImages)
{
string Gallery = Regex.Match(course.ToString(), @"data-standard(.*?)jpg").Value.Replace("data-standard=\"/", "https://binhminhdigital.com/");
list += Gallery + ",";
}
product.Images = list.TrimEnd(',');
//----------------------------------------------
return(product);
}
