static void Main(string[] args)
{
Console.WriteLine("Start Crawler...");
System.Threading.Thread.Sleep(2000);
Console.WriteLine("Running");
TinTrinhLibrary.WebClient client = new TinTrinhLibrary.WebClient();
string html = "";
for (int i = 1; i <= 4; i++)
{
html += client.Get("https://tiki.vn/bestsellers/sach-truyen-tieng-viet/c316?p=" + i, "https://tiki.vn/bestsellers/sach-truyen-tieng-viet/c316", "");
}
MatchCollection BookData = Regex.Matches(html, "data-price=\"(.*?)\" data-title=\"(.*?)\" data-brand=\"(.*?)\" data-category=\"(.*?)\">", RegexOptions.Singleline);
MatchCollection Review = Regex.Matches(html, "class=\"review\">((.*?))</p>", RegexOptions.Singleline);
MatchCollection Description = Regex.Matches(html, "description\">(.*?)<a", RegexOptions.Singleline);
StreamWriter File = new StreamWriter("DataSave.txt");
File.Flush();
int j = 0;
int a = 0;
foreach (Match Book in BookData)
{
File.Write("Giá: ");
File.WriteLine(Book.Groups[1].Value.Trim());
File.Write("Tên sách: ");
File.WriteLine(Book.Groups[2].Value.Trim());
File.Write("Tác Giả: ");
File.WriteLine(Book.Groups[3].Value.Trim());
File.Write("Đánh giá: ");
File.WriteLine(Review[j].Groups[1].Value.Trim());
File.Write("Miêu tả: ");
File.WriteLine(Description[a].Groups[1].Value.Trim());
File.Write("Thể loại: ");
File.WriteLine(Book.Groups[4].Value.Trim());
File.WriteLine("___________________________");
j++;
a++;
}
File.Close();
System.Threading.Thread.Sleep(5000);
Console.WriteLine("Completed. Let's check file DataSave.txt");
Console.ReadLine();
}
static void Main(string[] args)
{
Console.WriteLine("Start crawler...");
//GET : https://tiki.vn/
TinTrinhLibrary.WebClient client = new TinTrinhLibrary.WebClient();
//get book
for (int i = 1; i <= 4; i++)
{
string html = client.Get("https://tiki.vn/bestsellers/sach-truyen-tieng-viet/c316?p=" + i + "", "http://tiki.vn/", "");
//get book
MatchCollection book = Regex.Matches(html, "title=\"(.*?)\" data-brand", RegexOptions.Multiline);
foreach (Match title in book)
{
Console.WriteLine(title.Groups[1].Value.Replace("?", "").Trim());
}
//get author
MatchCollection authors = Regex.Matches(html, "author\">(.*?)</p>", RegexOptions.Singleline);
foreach (Match author in authors)
{
Console.WriteLine(author.Groups[1].Value.Replace("?", "").Trim());
}
//get price
MatchCollection priceList = Regex.Matches(html, "price-sale\">(.*?)<span", RegexOptions.Singleline);
foreach (Match price in priceList)
{
Console.WriteLine(price.Groups[1].Value.Replace("?", "").Trim());
}
//get comment
MatchCollection comments = Regex.Matches(html, "review\">(.*?)</p>", RegexOptions.Multiline);
foreach (Match review in comments)
{
Console.WriteLine(review.Groups[1].Value.Replace("?", "").Trim());
}
//get description
MatchCollection descriptions = Regex.Matches(html, "description\">(.*?)<a", RegexOptions.Singleline);
foreach (Match description in descriptions)
{
Console.WriteLine(description.Groups[1].Value.Replace("?", "").Trim());
}
}
System.Threading.Thread.Sleep(5000);
Console.WriteLine("OK");
Console.ReadLine();
}
static void Main(string[] args)
{
TinTrinhLibrary.WebClient client = new TinTrinhLibrary.WebClient();
int start = 0, dem;
String tieude, noidung, hinhanh;
String reg = "<a href=" + '"' + "(.*?)" + '"' + " itemprop=" + '"' + "url" + '"';
String imgs = "<img src=" + '"' + "(.*?)" + '"' + " alt=" + '"' + '"' + " itemprop=" + '"' + "thumbnailUrl" + '"' + "/>";
client.Post("http://dichvuketoantainha.net/login", "trankhanhtoan=b303baf471a2a712b2456ed41c822ee8&user_name=admin&user_pass=BUKT&login=Log+in", "http://dichvuketoantainha.net", "");
loop:
String html = client.Get("http://ketoanbanthoigian.com/kinh-nghiem-ke-toan.html?start=" + start, "http://ketoanbanthoigian.com", "");
start = start + 10;
List <String> imgurls = new List <String>();
if (start > 180)
{
goto finish;
}
MatchCollection rg = Regex.Matches(html, reg);
MatchCollection rgimg = Regex.Matches(html, imgs);
dem = -1;
foreach (Match rimg in rgimg)
{
dem++;
String img = "http://ketoanbanthoigian.com" + rimg.Groups[1].Value;
imgurls.Add(img);
}
dem = -1;
foreach (Match r in rg)
{
dem++;
String url = "http://ketoanbanthoigian.com" + r.Groups[1].Value;
String htmli = client.Get(url, "http://ketoanbanthoigian.com", "");
int i = htmli.IndexOf("<h2 itemprop=\"name\">");
int j = htmli.IndexOf("</h2>", i);
tieude = htmli.Substring(i + 20, j - i - 20).Trim();
int i1 = htmli.IndexOf("<div itemprop=\"articleBody\">");
int j1 = htmli.IndexOf("<div class=\"extranews_separator\"></div>", i1);
noidung = htmli.Substring(i1 + 28, j1 - i1 - 28).Trim();
Console.WriteLine(dem + 1);
hinhanh = imgurls.ElementAt(dem);
/*
* thuc hien upload vao website dichvuketoantainha.net
* */
//client.Post("http://dichvuketoantainha.net/admin/new_blog", "blog_name="+tieude+ "&blog_image="+hinhanh+ "&blog_cat_ids[]=4&blog_seo_title="+tieude+ "&blog_seo_keyword=&blog_seo_description=&new_blog=Create&blog_content="+noidung, "http://dichvuketoantainha.net/admin/", "");
System.Threading.Thread.Sleep(1000);
}
goto loop;
finish:
Console.WriteLine("finish!");
Console.ReadLine();
}
static void Main(string[] args)
{
Console.WriteLine("Scraping....");
TinTrinhLibrary.WebClient client = new TinTrinhLibrary.WebClient();
StreamWriter file = new StreamWriter(@"D:\study\ECDESI\products.txt");
for (int t = 1; t <= 4; t++) // get multiple page
{
string html_source = client.Get($"https://tiki.vn/bestsellers/sach-truyen-tieng-viet/c316?p={t}", "https://tiki.vn/", "");
int i = 0;
/// Get Product div
MatchCollection products = Regex.Matches(html_source, "class=\"infomation\">(.*?)fa-caret-right", RegexOptions.Singleline);
//end Get
foreach (Match product in products)
{
string string_product = product.Groups[1].Value.Trim();
/// Remove htlm tag
MatchCollection products_detail = Regex.Matches(string_product, "[^>]+?(?=<)", RegexOptions.Singleline); //
/// write data to products.txt
foreach (Match product_detail in products_detail)
{
if (product_detail.Value.Trim() != "")
{
i++;
/// Set category
if (i == 1)
{
string p = "Name: " + product_detail.Value.Trim();
file.WriteLine(p);
}
if (i == 2)
{
string p = "Author: " + product_detail.Value.Trim();
file.WriteLine(p);
}
if (i == 5)
{
string p = "Price: " + product_detail.Value.Trim();
file.WriteLine(p);
}
if (i == 8)
{
string p = "Decription: " + product_detail.Value.Trim();
file.WriteLine(p);
}
if (i == 9)
{
string p = "end!\n\n------------------------------------";
file.WriteLine(p);
}
file.Flush();
///end Set catgory
///
}
}
i = 0;
}
}
//end write
System.Threading.Thread.Sleep(5000);
Console.WriteLine("Ok");
Console.ReadLine();
}
static void Main(string[] args)
{
MatchCollection matches;
//File BestSellingBook.txt lưu trữ thông tin cần lấy từ website Tiki.vn
StreamWriter text = new StreamWriter("BestSellingBooks.txt");
Console.WriteLine("Start Crawler...");
//Số trang trong danh mục Best Selling Books trong Tiki.vn
int page = 1;
//Thứ tự sách
int number = 1;
TinTrinhLibrary.WebClient client = new TinTrinhLibrary.WebClient();
while (page <= 4)
{
List <string> TITLE = new List <string>();
List <string> BRAND = new List <string>();
List <string> PRICE = new List <string>();
List <string> REVIEW = new List <string>();
List <string> DESCRIPTION = new List <string>();
string html = client.Get("https://tiki.vn/bestsellers/sach-truyen-tieng-viet/c316?p= " + page, "https://tiki.vn/", "");
//Get Title
var title = "data-title=\"(.*?)\"";
Regex re_title = new Regex(title);
matches = re_title.Matches(html);
foreach (Match correct in matches)
{
TITLE.Add(correct.Groups[1].Value);
}
//Console.WriteLine(matches.Count);
//Get Brand
var brand = "data-brand=\"(.*?)\"";
Regex re_brand = new Regex(brand);
matches = re_brand.Matches(html);
foreach (Match correct in matches)
{
BRAND.Add(correct.Groups[1].Value);
}
//Console.WriteLine(matches.Count);
//Get Price
var price = "data-price=\"(.*?)\"";
Regex re_price = new Regex(price);
matches = re_price.Matches(html);
foreach (Match correct in matches)
{
PRICE.Add(correct.Groups[1].Value);
}
//Console.WriteLine(matches.Count);
//Get Review
var review = "<p class=\"review\">(.*?)</p>";
Regex re_review = new Regex(review);
matches = re_review.Matches(html);
foreach (Match correct in matches)
{
REVIEW.Add(correct.Groups[1].Value);
}
//Console.WriteLine(matches.Count);
// Get Description
var description = "<div class=\"description\">[\r\n]([^><]+)<";
Regex re_description = new Regex(description);
matches = re_description.Matches(html);
foreach (Match correct in matches)
{
DESCRIPTION.Add(correct.Groups[1].Value.Trim());
}
//Console.WriteLine(matches.Count);
for (int i = 0; i < TITLE.Count; i++)
{
text.WriteLine(number + ") " + "Tên sách: " + TITLE[i] + " | Tác giả: " + BRAND[i] + " | Số nhận xét: " + REVIEW[i] + " | Giá: " + PRICE[i] + "đ | Giới thiệu sách: " + DESCRIPTION[i]);
text.WriteLine();
number++;
}
page++;
}
text.Close();
}
