public static List<LinkGrap> getLinkLists(string link, string _class)
{
// string status = "Nạp danh sách";
List<LinkGrap> List = new List<LinkGrap>();
HttpWebRequest wrq;
wrq = (HttpWebRequest)(WebRequest.Create(link));
string host = new Uri(link).Host;
wrq.Credentials = CredentialCache.DefaultCredentials;
wrq.Method = "GET";
wrq.UserAgent = "Mozilla/5.0 (Windows; U; Windows NT 6.1; vi; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3";
wrq.SendChunked = true;
HttpWebResponse wrp = (HttpWebResponse)wrq.GetResponse();
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.Load(wrp.GetResponseStream(), Encoding.UTF8);
HtmlNode _n1 = doc.DocumentNode.SelectSingleNode(@"//td[@class='"+_class+"']");
HtmlNode _n2 = _n1.SelectSingleNode(@"table");
HtmlNodeCollection _nl = _n2.SelectNodes("tr");
// status = "Đọc danh sách";
foreach (HtmlNode _n3 in _nl)
{
LinkGrap item = new LinkGrap();
HtmlNode _n11 = _n3.SelectSingleNode("td/table");
HtmlNode _n111 = _n11.ChildNodes[1].ChildNodes[1].ChildNodes[1];
item.Link = _n111.Attributes["href"].Value;
item.Title = _n3.InnerText;
List.Add(item);
}
//foreach (HtmlNode _n3 in _nl)
//{
// HtmlNode _n11 = _n3.SelectSingleNode("td/table");
// HtmlNode _n111 = _n11.ChildNodes[1].ChildNodes[1].ChildNodes[1];
// HtmlNode _n112 = _n11.ChildNodes[5].ChildNodes[1].ChildNodes[1];
// insertLinkDeledate _dele = new insertLinkDeledate(insertLink);
// _dele.BeginInvoke(string.Format("http://www.ttnn.com.vn{0}", _n111.Attributes["href"].Value), null, null);
//}
return List;
}
public static List<LinkGrap> GetLinkFromRss(string link)
{
List<LinkGrap> List = new List<LinkGrap>();
#region xử lý Link
HttpWebRequest wrq;
wrq = (HttpWebRequest)(WebRequest.Create(link));
string host = new Uri(link).Host;
wrq.Credentials = CredentialCache.DefaultCredentials;
wrq.Method = "GET";
wrq.UserAgent = "Mozilla/5.0 (Windows; U; Windows NT 6.1; vi; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3";
wrq.SendChunked = false;
if (link.IndexOf("zing.vn") != -1)
{
wrq.Referer = "http://mp3.zing.vn";
}
try
{
using (HttpWebResponse wrp = (HttpWebResponse)wrq.GetResponse())
{
XmlDocument doc = new XmlDocument();
try
{
doc.Load(wrp.GetResponseStream());
XmlNodeList _l = doc.SelectNodes("//item");
if (List != null)
{
foreach (XmlNode n in _l)
{
LinkGrap item = new LinkGrap();
if (n.SelectSingleNode("title") != null)
{
item.Title = n.SelectSingleNode("title").InnerText;
}
if (n.SelectSingleNode("description") != null)
{
item.Description = n.SelectSingleNode("description").InnerText;
}
if (n.SelectSingleNode("link | LINK") != null)
{
string itemlink = n.SelectSingleNode("link | LINK").InnerText.Trim();
if (itemlink.IndexOf("http://nld.com.vnhttp://worldcup") != -1) itemlink = itemlink.Substring("http://nld.com.vn".Length);
item.Link = itemlink;
}
List.Add(item);
}
}
}
catch (XmlException xmlex)
{
}
}
}
catch (WebException ex)
{
LinkGrap _obj2 = new LinkGrap();
_obj2.Title = "0";
}
#endregion
return List;
}
public LinkGrap(string link, bool extractLink)
{
LinkGrap Item = null;
string saveLocation = HostingEnvironment.MapPath("~/lib/up/");
//string uploadDir = @"D:\InetPub\tintucme\wwwroot\lib\up\";
//C:\inetpub\wwwroot\choNongNghiep\web\lib\up\rss
string uploadDir = @"C:\inetpub\wwwroot\choNongNghiep\web\lib\up\tintuc\rss\";
Item = (LinkGrap)HttpRuntime.Cache[string.Format(cacheKey, link)];
Item = null;
if (Item == null)
{
#region xử lý Link
HttpWebRequest wrq;
wrq = (HttpWebRequest)(WebRequest.Create(link));
string host = new Uri(link).Host;
wrq.Credentials = CredentialCache.DefaultCredentials;
wrq.Method = "GET";
wrq.UserAgent = "Mozilla/5.0 (Windows; U; Windows NT 6.1; vi; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3";
wrq.SendChunked = false;
if (link.IndexOf("zing.vn") != -1)
{
wrq.Referer = "http://mp3.zing.vn";
}
try
{
HttpWebResponse wrp = (HttpWebResponse)wrq.GetResponse();
HtmlDocument doc = new HtmlDocument();
contentType = wrp.ContentType;
if (contentType.ToLower().IndexOf("html") != -1)
{
doc.Load(wrp.GetResponseStream(), Encoding.UTF8);
string domain = "http://" + (new Uri(link)).Host;
#region Title
if (doc.DocumentNode.SelectNodes("//title | //TITLE") != null)
{
HtmlNode titleNode = doc.DocumentNode.SelectNodes("//title | //TITLE")[0];
Title = titleNode.InnerText;
}
#endregion
#region Desc
if (doc.DocumentNode.SelectNodes("//meta[@name='description'] | //meta[@name='DESCRIPTION']") != null)
{
HtmlNode titleNode = doc.DocumentNode.SelectNodes("//meta[@name='description'] | //meta[@name='DESCRIPTION']")[0];
Description = titleNode.Attributes["content"].Value;
}
#endregion
#region Content
ContentRawHtml = doc.DocumentNode.InnerHtml;
ContentRawText = doc.DocumentNode.InnerText;
Content = Wrapper(host, link, doc);
if (string.IsNullOrEmpty(Content)) return;
string contentTokeyword = Giga.Common.Lib._string.getHTML(Content);
if (!string.IsNullOrEmpty(contentTokeyword))
{
using (LinkKeyword _linkKeyword = new LinkKeyword(contentTokeyword))
{
if (_linkKeyword.ListKeyWord != null)
{
KeyWords = _linkKeyword.ListKeyWord;
}
}
}
//if (string.IsNullOrEmpty(contentTokeyword)) contentTokeyword = ContentRawText;
//List<LinkKeyword> _ListKeyword = new List<LinkKeyword>();
#endregion
#region Images
HtmlDocument _doc1 = new HtmlDocument();
_doc1.LoadHtml(Content);
if (_doc1.DocumentNode.SelectNodes("//img | //IMG") != null)
{
List<string> _list = new List<string>();
foreach (HtmlNode _img in _doc1.DocumentNode.SelectNodes("//img | //IMG"))
{
if (_img.Attributes["src"] != null)
{
string src = _img.Attributes["src"].Value;
if (src.ToLower().IndexOf("http://") != 0)
{
if (src.IndexOf("/") != 0) src = "/" + src;
src = domain + src;
}
try
{
ImageProcess gimg = new ImageProcess(new Uri(src), src);
if (gimg.Width > 250 && gimg.Heigth > 200)
{
#region ảnh cũ
//string _newid = Guid.NewGuid().ToString();
//string _ten = saveLocation + _newid;
//gimg.Save(_ten + gimg.Ext);
//gimg.Crop(320, 188);
//gimg.Save(_ten + "320x188" + gimg.Ext);
//gimg.Crop(150, 160);
//gimg.Save(_ten + "150x160" + gimg.Ext);
//gimg.Crop(101, 58);
//gimg.Save(_ten + "101x58" + gimg.Ext);
//gimg.Crop(62, 36);
//gimg.Save(_ten + "62x36" + gimg.Ext);
//_list.Add(_newid + gimg.Ext);
//HttpRuntime.Cache.Remove(src);
#endregion
string gimg_t = Guid.NewGuid().ToString().Replace("-", "");
string gimg_ten = gimg_t + gimg.Ext;
//saveLocation = Path.Combine(uploadDir, gimg_ten);
//gimg.Save(saveLocation);
_list.Add(gimg_ten);
string gimg_ten_430x300 = gimg_t + "430x300" + gimg.Ext;
gimg.Crop(430, 300);
saveLocation = Path.Combine(uploadDir, gimg_ten_430x300);
gimg.Save(saveLocation);
_list.Add(gimg_ten_430x300);
//string gimg_ten_100 = gimg_t + "200x150" + gimg.Ext;
//gimg.Crop(200, 150);
//saveLocation = Path.Combine(uploadDir, gimg_ten_100);
//gimg.Save(saveLocation);
//_list.Add(gimg_ten_100);
string gimg_ten_101_58 = gimg_t + "100x100" + gimg.Ext;
gimg.Crop(100, 100);
saveLocation = Path.Combine(uploadDir, gimg_ten_101_58);
gimg.Save(saveLocation);
_list.Add(gimg_ten_101_58);
string gimg_ten_62_36 = gimg_t + "50x50" + gimg.Ext;
gimg.Crop(50, 50);
saveLocation = Path.Combine(uploadDir, gimg_ten_62_36);
gimg.Save(saveLocation);
_list.Add(gimg_ten_62_36);
break;
}
}
catch (Exception ex)
{
}
}
}
if (_list.Count == 0) return;
Images = _list;
}
#endregion
#region Link
//List<string> _inLink = new List<string>();
//List<string> _outLink = new List<string>();
//foreach (HtmlNode a in doc.DocumentNode.SelectNodes("//a|//A"))
//{
// string href = string.Empty;
// if (a.Attributes["href"] != null)
// {
// href = a.Attributes["href"].Value;
// }
// else if (a.Attributes["HREF"] != null)
// {
// href = a.Attributes["HREF"].Value;
// }
// if (!string.IsNullOrEmpty(href))
// {
// if (href.ToLower().IndexOf("javascrip") != 0 && href.ToLower().IndexOf("#") != 0)
// {
// if (href.ToLower().IndexOf("http://") == 0)//HTTPLink
// {
// Uri _href = new Uri(href);
// if (_href.Host.ToLower().IndexOf(host.ToLower()) != -1)
// {
// _inLink.Add(href);
// }
// else
// {
// _outLink.Add(href);
// }
// }
// else
// {
// if (href.ToLower().IndexOf("/") == -1) href = "/" + href;
// if (href.ToLower().IndexOf("../") == 0) href = href.Substring(href.LastIndexOf("../") + 3);
// href = "http://" + host + href;
// _inLink.Add(href);
// }
// }
// }
// InLink = _inLink;
// OutLink = _outLink;
//}
#endregion
}
else
{
if (contentType.IndexOf("image") != -1)
{
Title = link;
Description = link;
List<string> _list = new List<string>();
ImageProcess gimg = new ImageProcess(new Uri(link), link);
if (gimg.Width > 100 && gimg.Heigth > 75)
{
_list.Add(link);
}
Images = _list;
}
}
LinkGrap _obj = new LinkGrap();
_obj.Title = Title;
_obj.Description = Description;
_obj.Images = Images;
_obj.contentType = contentType;
_obj.InLink = InLink;
_obj.OutLink = OutLink;
_obj.contentType = Content;
_obj.ContentRawText = ContentRawText;
_obj.ContentRawHtml = ContentRawHtml;
HttpRuntime.Cache.Insert(string.Format(cacheKey, link), _obj);
Item = _obj;
}
catch (WebException ex)
{
LinkGrap _obj2 = new LinkGrap();
_obj2.Title = "0";
HttpRuntime.Cache.Remove(string.Format(cacheKey, link));
Item = _obj2;
}
#endregion
}
Title = Item.Title;
Description = Item.Description;
Images = Item.Images;
contentType = Item.contentType;
KeyWordsIEnum = Item.KeyWordsIEnum;
}
public LinkGrap(string link)
{
LinkGrap Item = null;
string saveLocation = HostingEnvironment.MapPath("~/lib/u/");
Item = (LinkGrap)HttpContext.Current.Cache[string.Format(cacheKey, link)];
if (Item == null)
{
#region xử lý Link
HttpWebRequest wrq;
wrq = (HttpWebRequest)(WebRequest.Create(link));
string host = new Uri(link).Host;
wrq.Credentials = CredentialCache.DefaultCredentials;
wrq.Method = "GET";
wrq.UserAgent = "Mozilla/5.0 (Windows; U; Windows NT 6.1; vi; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3";
wrq.SendChunked = true;
if (link.IndexOf("zing.vn") != -1)
{
wrq.Referer = "http://mp3.zing.vn";
}
try
{
HttpWebResponse wrp = (HttpWebResponse)wrq.GetResponse();
HtmlDocument doc = new HtmlDocument();
contentType = wrp.ContentType;
if (contentType.ToLower().IndexOf("html") != -1)
{
doc.Load(wrp.GetResponseStream(), Encoding.UTF8);
string domain = "http://" + (new Uri(link)).Host;
#region Title
if (doc.DocumentNode.SelectNodes("//title | //TITLE") != null)
{
HtmlNode titleNode = doc.DocumentNode.SelectNodes("//title | //TITLE")[0];
Title = titleNode.InnerText;
}
#endregion
#region Desc
if (doc.DocumentNode.SelectNodes("//meta[@name='description'] | //meta[@name='DESCRIPTION']") != null)
{
HtmlNode titleNode = doc.DocumentNode.SelectNodes("//meta[@name='description'] | //meta[@name='DESCRIPTION']")[0];
Description = titleNode.Attributes["content"].Value;
}
#endregion
#region Content
ContentRawHtml = doc.DocumentNode.InnerHtml;
ContentRawText = doc.DocumentNode.InnerText;
Content = Wrapper(host, link, doc);
#endregion
#region Images
//if (doc.DocumentNode.SelectNodes("//img | //IMG") != null)
//{
// List<string> _list = new List<string>();
// foreach (HtmlNode _img in doc.DocumentNode.SelectNodes("//img | //IMG"))
// {
// if (_img.Attributes["src"] != null)
// {
// string src = _img.Attributes["src"].Value;
// if (src.ToLower().IndexOf("http://") != 0)
// {
// src = domain + src;
// }
// _list.Add(src);
// }
// }
// Images = _list;
//}
#endregion
#region Link
List<string> _inLink = new List<string>();
List<string> _outLink = new List<string>();
foreach (HtmlNode a in doc.DocumentNode.SelectNodes("//a|//A"))
{
string href = string.Empty;
if (a.Attributes["href"] != null)
{
href = a.Attributes["href"].Value;
}
else if (a.Attributes["HREF"] != null)
{
href = a.Attributes["HREF"].Value;
}
if (!string.IsNullOrEmpty(href))
{
if (href.ToLower().IndexOf("javascrip") != 0 && href.ToLower().IndexOf("#") != 0)
{
if (href.ToLower().IndexOf("http://") == 0)//HTTPLink
{
Uri _href = new Uri(href);
if (_href.Host.ToLower().IndexOf(host.ToLower()) != -1)
{
_inLink.Add(href);
}
else
{
_outLink.Add(href);
}
}
else
{
if (href.ToLower().IndexOf("/") == -1) href = "/" + href;
if (href.ToLower().IndexOf("../") == 0) href = href.Substring(href.LastIndexOf("../") + 3);
href = "http://" + host + href;
_inLink.Add(href);
}
}
}
InLink = _inLink;
OutLink = _outLink;
}
#endregion
}
#region Ảnh
//else
//{
// if (contentType.IndexOf("image") != -1)
// {
// Title = link;
// Description = link;
// List<string> _list = new List<string>();
// ImageProcess gimg = new ImageProcess(new Uri(link), link);
// if (gimg.Width > 100 && gimg.Heigth > 75)
// {
// _list.Add(link);
// }
// Images = _list;
// }
//}
#endregion
HttpContext.Current.Cache.Insert(string.Format(cacheKey, link), this);
}
catch (WebException ex)
{
LinkGrap _obj2 = new LinkGrap();
_obj2.Title = "0";
HttpContext.Current.Cache.Remove(string.Format(cacheKey, link));
}
#endregion
}
}
public static List<LinkGrap> GetRss(string link)
{
List<LinkGrap> List = new List<LinkGrap>();
#region xử lý Link
HttpWebRequest wrq;
wrq = (HttpWebRequest)(WebRequest.Create(link));
string host = new Uri(link).Host;
wrq.Credentials = CredentialCache.DefaultCredentials;
wrq.Method = "GET";
wrq.UserAgent = "Mozilla/5.0 (Windows; U; Windows NT 6.1; vi; rv:1.9.2.3) Gecko/20100401 Firefox/3.6.3";
wrq.SendChunked = true;
if (link.IndexOf("zing.vn") != -1)
{
wrq.Referer = "http://mp3.zing.vn";
}
try
{
using (HttpWebResponse wrp = (HttpWebResponse)wrq.GetResponse())
{
HtmlDocument doc = new HtmlDocument();
doc.Load(wrp.GetResponseStream());
foreach (HtmlNode a in doc.DocumentNode.SelectNodes(@"//a[contains(text(),'rss') and contains(@href,'rss')]"))
{
LinkGrap item = new LinkGrap();
item.Link = a.Attributes["href"].Value;
item.Title = a.InnerText;
List.Add(item);
}
}
}
catch (WebException ex)
{
LinkGrap _obj2 = new LinkGrap();
_obj2.Title = "0";
}
#endregion
return List;
}
