public RegularTin(string url)
{
string _host = string.Empty;
_host = new Uri(url).Host.ToLower();
HtmlAgilityPack.HtmlDocument doc = loadDoc(url);
List<QuyTac> List = QuyTacDal.SelectByHost(_host).ToList();
WrapperClean(_host, url, doc);
#region Lấy Ten,Mota, Noi Dung
foreach (QuyTac item in List)
{
HtmlNode _c = doc.DocumentNode.SelectSingleNode(@item.Xpath);
switch (item.Loai)
{
case 0:// renew doc
if (_c != null)
{
doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(_c.InnerHtml);
}
break;
case 1:// Title
if (_c != null)
{
Ten = _c.InnerText;
}
else
{
_c = doc.DocumentNode.SelectNodes(@"//title | //TITLE")[0];
if (_c != null)
{
Ten = _c.InnerText;
}
}
break;
case 2:
if (_c != null)
{
MoTa = _c.InnerHtml;
}
else
{
_c = doc.DocumentNode.SelectNodes(@"//meta[@name='description'] | //meta[@name='DESCRIPTION'] | //meta[@name='Description']")[0];
if (_c != null)
{
MoTa = _c.InnerText;
}
}
break;
case 3:
if (item.Xoa)
{
if (_c != null)
{
_c.ParentNode.RemoveChild(_c, false);
}
}
else
{
if (_c != null)
{
NoiDung = _c.InnerHtml;
}
}
break;
default:
break;
}
}
#endregion
#region Images
if (!string.IsNullOrEmpty(NoiDung))
{
HtmlAgilityPack.HtmlDocument _doc1 = new HtmlAgilityPack.HtmlDocument();
_doc1.LoadHtml(NoiDung);
#region Keywords
using (LinkKeyword _linkKeyword = new LinkKeyword(_doc1.DocumentNode.InnerText))
{
if (_linkKeyword.ListKeyWord != null)
{
KeyWords = _linkKeyword.ListKeyWord;
}
}
#endregion
string domain = "http://" + _host;
string saveLocation = HostingEnvironment.MapPath("~/lib/up/");
//string uploadDir = @"D:\Work\linh\ktt_x1\web\lib\up\rss\";
string uploadDir = @"C:\inetpub\wwwroot\kttvn\web\lib\up\rss\";
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
{
var gimg = new ImageProcess(new Uri(src), src);
if (gimg.Width > 250 && gimg.Heigth > 200)
{
_list.Add(src);
break;
}
}
catch (Exception ex)
{
}
}
}
if (_list.Count == 0) return;
Anh = _list[0];
}
}
#endregion
}
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 RegularTin(string url, string uploadDir, List<ImageSize> images)
{
var host = new Uri(url).Host.ToLower();
var doc = loadDoc(url);
var list = QuyTacDal.SelectByHost(host).ToList();
WrapperClean(host, url, doc);
#region Lấy Ten,Mota, Noi Dung
foreach (var item in list)
{
var c = doc.DocumentNode.SelectSingleNode(@item.Xpath);
switch (item.Loai)
{
case 0:// renew doc
if (c != null)
{
doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(c.InnerHtml);
}
break;
case 1:// Title
if (c != null)
{
Ten = c.InnerText;
}
else
{
c = doc.DocumentNode.SelectNodes(@"//title | //TITLE")[0];
if (c != null)
{
Ten = c.InnerText;
}
}
break;
case 2:
if (c != null)
{
MoTa = c.InnerHtml;
}
else
{
c = doc.DocumentNode.SelectNodes(@"//meta[@name='description'] | //meta[@name='DESCRIPTION'] | //meta[@name='Description']")[0];
if (c != null)
{
MoTa = c.InnerText;
}
}
break;
case 3:
if (item.Xoa)
{
if (c != null)
{
c.ParentNode.RemoveChild(c, false);
}
}
else
{
if (c != null)
{
NoiDung = c.InnerHtml;
}
}
break;
default:
break;
}
}
#endregion
#region Images
if (string.IsNullOrEmpty(NoiDung)) return;
var doc1 = new HtmlAgilityPack.HtmlDocument();
doc1.LoadHtml(NoiDung);
#region Keywords
using (var linkKeyword = new LinkKeyword(doc1.DocumentNode.InnerText))
{
if (linkKeyword.ListKeyWord != null)
{
KeyWords = linkKeyword.ListKeyWord;
}
}
#endregion
var domain = "http://" + host;
var saveLocation = HostingEnvironment.MapPath("~/lib/up/");
List<string> _list = new List<string>();
if (doc1.DocumentNode.SelectNodes("//img | //IMG") != null)
{
foreach (var img in doc1.DocumentNode.SelectNodes("//img | //IMG"))
{
if (img.Attributes["src"] != null)
{
string src = img.Attributes["src"].Value;
if (src.ToLower().IndexOf("http://", System.StringComparison.Ordinal) != 0)
{
if (src.IndexOf("/", System.StringComparison.Ordinal) != 0) src = "/" + src;
src = domain + src;
}
try
{
var gimg = new ImageProcess(new Uri(src), src);
if (gimg.Width > 250 && gimg.Heigth > 200)
{
var imgTen = Guid.NewGuid().ToString().Replace("-", "");
foreach (var imageSize in images)
{
gimg.Crop(imageSize.Width, imageSize.Height);
gimg.Save(Path.Combine(uploadDir,
string.Format("{0}{1}.{2}", imgTen,
imageSize.DefaultImage
? ""
: string.Format("{0}x{1}",
imageSize.Width,
imageSize.Height),
gimg.Ext)));
}
break;
}
}
catch (Exception ex)
{
}
}
}
if (_list.Count == 0) return;
Anh = _list[0];
}
#endregion
}
public Dictionary<string, LinkKeyword> GetDic(Dictionary<string, LinkKeyword> inputDic
, Dictionary<int, string> IndexDic
, Dictionary<string, LinkKeyword> DuplicateDic)
{
Dictionary<string, LinkKeyword> Dic8 = new Dictionary<string, LinkKeyword>();
if (inputDic.Count > 0)
{
foreach (LinkKeyword item in new List<LinkKeyword>(inputDic.Values))
{
if (item.count > 1)
{
int KeySize = item.loai;
foreach (int _index in item.index)
{
int newIndex = _index + KeySize;
if (newIndex < IndexDic.Count)
{
if (DuplicateDic.ContainsKey(IndexDic[newIndex]))
{
string newKey = item.key + " " + IndexDic[newIndex];
if (Dic8.ContainsKey(newKey))
{
LinkKeyword _Key = Dic8[newKey];
List<int> List = _Key.index;
List.Add(_index);
Dic8[newKey] = new LinkKeyword(_Key.key, _Key.count + 1, List, KeySize + 1, false);
}
else
{
List<int> List = new List<int>();
List.Add(_index);
Dic8.Add(newKey, new LinkKeyword(newKey, 1, List, KeySize + 1, false));
}
//if (newIndex + 1 < IndexDic.Count)
//{
// if (!DuplicateDic.ContainsKey(IndexDic[newIndex + 1]))
// {
// }
//}
}
}
}
}
}
Dictionary<string, LinkKeyword> DicNone = GetDic(Dic8, IndexDic, DuplicateDic);
foreach (KeyValuePair<string, LinkKeyword> item in DicNone)
{
if (item.Value.loai > 1 && item.Value.count > 1)
{
string itemChild = item.Value.key.Substring(0, item.Value.key.LastIndexOf(" "));
if (Dic8.ContainsKey(itemChild))
{
LinkKeyword removeKey = Dic8[itemChild];
removeKey.kill = true;
Dic8[itemChild] = removeKey;
}
}
if (Dic8.ContainsKey(item.Key))
{
LinkKeyword _Key = Dic8[item.Key];
Dic8[item.Key] = new LinkKeyword(_Key.key, _Key.count + 1, _Key.index, item.Value.loai, _Key.kill);
}
else
{
Dic8.Add(item.Key, item.Value);
}
}
}
return Dic8;
}
