public async static Task <List <TagImg> > GetImgFromHtml(string html, bool isHotspot = false)
{
Task <List <TagImg> > task = Task.Run(() => {
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(html);
var imgList = doc.DocumentNode.SelectNodes("//img");
var w = 0;
var h = 0;
HtmlAgilityPack.HtmlAttribute tempAttribute = null;
List <TagImg> list = new List <TagImg>();
foreach (var item in imgList)
{
TagImg tagImg = new TagImg();
tempAttribute = item.Attributes["alt"];
tagImg.Alt = tempAttribute == null ? "":tempAttribute.Value;
tempAttribute = item.Attributes["src"];
tagImg.Src = tempAttribute == null ? "" : tempAttribute.Value;
tempAttribute = item.Attributes["h"];
if (tempAttribute != null)
{
int.TryParse(tempAttribute.Value, out h);
}
tempAttribute = item.Attributes["w"];
if (tempAttribute != null)
{
int.TryParse(tempAttribute.Value, out w);
}
//Search Detail
/*
* <a class="iusc" style="height:208px;width:333px" m="{"cid":"1jz2ZvDM","purl":"https://www.927tour.com/News_newsDetail_id_20180408195735146766.html","murl":"http://ynwgm.ynurl.cn/uploadfile/s10/2018/0408/20180408075500850.jpg","turl":"https://tse1-mm.cn.bing.net/th?id=OIP.1jz2ZvDMIyhtns4hK1ay-AHaFJ&pid=15.1","md5":"d63cf666f0cc23286d9ece212b56b2f8","shkey":"","t":"铁路、民航保障游客正常出游","mid":"1034F8C523DE0FCD1B8302CF3C0D52E2DA5E1CD3","desc":""}" onclick="sj_evt.fire('IFrame.Navigate', this.href); return false;" href="/images/search?view=detailV2&ccid=1jz2ZvDM&id=1034F8C523DE0FCD1B8302CF3C0D52E2DA5E1CD3&thid=OIP.1jz2ZvDMIyhtns4hK1ay-AHaFJ&mediaurl=http%3a%2f%2fynwgm.ynurl.cn%2fuploadfile%2fs10%2f2018%2f0408%2f20180408075500850.jpg&exph=407&expw=585&q=%e6%b8%85%e6%98%8e%e5%81%87%e6%9c%9f%e5%9b%bd%e5%86%85%e6%97%85%e6%b8%b8%e6%8e%a5%e5%be%85%e6%80%bb%e4%ba%ba%e6%95%b01.12%e4%ba%bf&simid=608053044385353052&selectedIndex=32&qft=+filterui%3aphoto-photo" h="ID=images.5601_7,5217.1">
* <div class="img_cont hoff">
* <img class="mimg" style="background-color:#c10a34;color:#c10a34" height="208" width="299" src="https://tse3-mm.cn.bing.net/th?id=OIP.1jz2ZvDMIyhtns4hK1ay-AHaFJ&w=299&h=208&c=7&o=5&pid=1.7" alt="清明假期国内旅游接待总人数1.12亿 的图像结果" />
* </div>
* </a>
*/
/*< a class="iusc" style="height:207px;width:276px" m="{"cid":"Ox2V7JRH","purl":"http://www.wall001.com/nature/under_sky/html/image8.html","murl":"http://wall001.com/nature/under_sky/mxxx01/[wall001.com]_sky_AP23070.jpg","turl":"https://tse2-mm.cn.bing.net/th?id=OIP.Ox2V7JRHXMInhT3_WlPpVgHaFj&pid=15.1","md5":"3b1d95ec94475cc227853dff5a53e956","shkey":"","t":"桌布天堂 --- 晴朗天空 - 藍天白云8","mid":"8A372FC995FECC38853858A07F4171C439B8FA58","desc":""}" onclick="sj_evt.fire('IFrame.Navigate', this.href); return false;" href="/images/search?view=detailV2&ccid=Ox2V7JRH&id=8A372FC995FECC38853858A07F4171C439B8FA58&thid=OIP.Ox2V7JRHXMInhT3_WlPpVgHaFj&mediaurl=http%3a%2f%2fwall001.com%2fnature%2funder_sky%2fmxxx01%2f%5bwall001.com%5d_sky_AP23070.jpg&exph=768&expw=1024&q=%e5%a4%a9%e7%a9%ba&simid=608010515721882861&selectedIndex=5&qft=+filterui%3aphoto-photo" h="ID=images.5601_7,5055.1"><div class="img_cont hoff"><img class="mimg" style="background-color:#1543b6;color:#1543b6" height="207" width="276" src="https://tse4-mm.cn.bing.net/th?id=OIP.Ox2V7JRHXMInhT3_WlPpVgHaFj&w=276&h=207&c=7&o=5&pid=1.7" alt="天空 的图像结果"></div></a>*/
Tuple <bool, string> extractResult = RegexUtil.ExtractBingImage(item.ParentNode.ParentNode.OuterHtml);
if (extractResult.Item1 == true || isHotspot == true)
{
tagImg.DetailUrl = extractResult.Item2;
tagImg.Width = w;
tagImg.Height = h;
list.Add(tagImg);
}
}
return(list);
});
return(await task);
}
public static string GetSrc(this TagImg tag)
{
var attr = tag.Attribute("src");
if (attr == null)
{
return(string.Empty);
}
return(attr.Value);
}
public static TagImg Main(string name)
{
var size = Urls.GetSize(Urls.MainSys(name));
var tag = new TagImg().Src(Urls.Main(name)).Alt("");
if (size.HasValue)
{
tag.Size(size.Value.Width, size.Value.Height);
}
return(tag);
}
private string GetCertView(TagImg certImg, decimal sigId, bool eng = false, bool vendor = false,
bool ru = false)
{
var view = H.div[
certImg.Style("margin:10px 0;"), H.br,
Url.Graduate().DownloadCert(sigId, eng, vendor, ru, "Скачать").Class("ui-button"),
ru&& !vendor ? H.div[H.h3["Поделитесь с друзьями!"],
Htmls.AddThis(Html)] : null
];
return(view.ToString());
}
public static TagImg lang(this TagImg tag, LangCode value)
{
tag.Lang = value; return(tag);
}
public static TagImg style(this TagImg tag, string value)
{
tag.Style = value; return(tag);
}
public static TagImg name(this TagImg tag, string value)
{
tag.Name = value; return(tag);
}
public static TagImg ismap(this TagImg tag, IsMap value)
{
tag.IsMap = value; return(tag);
}
public static TagImg width(this TagImg tag, Length value)
{
tag.Width = value; return(tag);
}
public static TagImg longdesc(this TagImg tag, string value)
{
tag.LongDesc = value; return(tag);
}
public static TagImg ondblclick(this TagImg tag, string value)
{
tag.OnDblClick = value; return(tag);
}
public static TagImg dir(this TagImg tag, Dir value)
{
tag.Dir = value; return(tag);
}
public static TagImg Size(this TagImg tag, object width, object height)
{
return(tag.Width(width).Height(height));
}
public string WebChecker(string url)
{
string http = "http://";
string https = "https://";
string lblResult = String.Empty;
string lblStatusDescription = String.Empty;
string lblCause = String.Empty;
string innerTextOfTitle = String.Empty;
string contentOfTagMeta = String.Empty;
string tagA = String.Empty;
string tagImg = String.Empty;
string innerTextOfH1 = String.Empty;
string all = String.Empty;
DateTime date = DateTime.Now;
Urls urllink = new Urls();
List <TagA> tagAList = new List <TagA>();
List <TagImg> tagImgList = new List <TagImg>();
List <TagH1> tagH1List = new List <TagH1>();
try
{
//------проверяем есть ли ссылка-------------
if (string.IsNullOrEmpty(url))
{
MessageBox.Show(@"Вы не вввели ссылку !");
return(null);
}
//---делаем проверку на наличие протокола "http://" в ссылке запроса
if (!(url.StartsWith(http) || url.StartsWith(https)))
{
url = url.Insert(0, http);
}
//------------------------------------------------------------------
// Создаём объект запроса
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(url);
// Получаем ответ с сервера, если запрашиваемый URL не действителен,
// переходим к блоку catch, иначе идем дальше.
HttpWebResponse response = (HttpWebResponse)request.GetResponse();
if (request.HaveResponse)
{
lblResult = (response.StatusCode == HttpStatusCode.OK) ? "Сайт доступен" : "Сайт не доступен";
lblStatusDescription = response.StatusDescription;
lblCause = ((int)response.StatusCode).ToString();
string respStream = new StreamReader(response.GetResponseStream(), Encoding.GetEncoding("utf-8")).ReadToEnd();
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
htmlDoc.LoadHtml(respStream);
//------ получаем значение тэга Title -------------
var nodeTitle = htmlDoc.DocumentNode.SelectNodes("//title");
if (nodeTitle != null)
{
innerTextOfTitle = nodeTitle["title"].InnerText;
}
//--------------------------------------------------
//------ получаем значение атрибута content тэга meta у которого
//------ значение атрибута name является description
var nodeMeta = htmlDoc.DocumentNode.SelectNodes("//meta");
if (nodeMeta != null)
{
foreach (var tag in nodeMeta)
{
if (tag.Attributes["name"] != null && tag.Attributes["name"].Value == "description")
{
contentOfTagMeta = tag.Attributes["content"].Value;
}
}
}
//---------------------------------------------------------------
//------ получаем значение атрибута href у тэга A -------------
var nodesA = htmlDoc.DocumentNode.SelectNodes("//a");
if (nodesA != null)
{
foreach (var tag in nodesA)
{
if (tag.Attributes["href"] != null)
{
var link = tag.Attributes["href"].Value;
tagA += link + "\n";
TagA insTagA = new TagA
{
UrlId = urllink.Id,
Href = link
};
tagAList.Add(insTagA);
}
}
}
//--------------------------------------------------------------
//------ получаем значение атрибута src у тэга img -------------
var nodesImg = htmlDoc.DocumentNode.SelectNodes("//img");
if (nodesImg != null)
{
foreach (var tag in nodesImg)
{
if (tag.Attributes["src"] != null)
{
var src = tag.Attributes["src"].Value;
tagImg += src + "\n";
TagImg tagImgIns = new TagImg
{
UrlId = urllink.Id,
Src = src
};
tagImgList.Add(tagImgIns);
}
}
}
//---------------------------------------------------------------
//------ получаем значение атрибута src у тэга img -------------
var nodesH1 = htmlDoc.DocumentNode.SelectNodes("//h1");
if (nodesH1 != null)
{
foreach (var tag in nodesH1)
{
innerTextOfH1 += tag.InnerText + "\n";
int index = nodesH1[tag];
TagH1 tagH1Ins = new TagH1
{
UrlId = urllink.Id,
H1Text = tag.InnerText
};
tagH1List.Add(tagH1Ins);
}
}
//---------------------------------------------------------------
StringBuilder newContent = new StringBuilder();
newContent.AppendLine("URL: " + url +
"\nДата проверки: " + date +
"\nСтатус: " + lblResult +
"\nОписание статуса: " + lblStatusDescription +
"\nКод статуса: " + lblCause +
"\nTitle: " + innerTextOfTitle +
"\nContent: " + contentOfTagMeta + "\n" +
"\nТэги A: \n" + tagA + "\n" +
"\nТэги Img: \n" + tagImg +
"\nТэги H1: \n" + innerTextOfH1 +
"\n");
all = newContent.ToString();
urllink.Url = url;
urllink.Title = innerTextOfTitle;
urllink.MetaContent = contentOfTagMeta;
urllink.Status = lblResult;
urllink.StatusDescription = lblStatusDescription;
urllink.StatusCode = lblCause;
urllink.DateOfParsing = date;
using (var db = new ParsingResultsEntities())
{
db.Urls.Add(urllink);
db.TagH1.AddRange(tagH1List);
db.TagA.AddRange(tagAList);
db.TagImg.AddRange(tagImgList);
db.SaveChanges();
}
}
response.Close();
request.Abort();
return(all); //возвращаем полученные данные в интерфейс приложения
}
catch (WebException ex)
{
lblResult = @"Сайт не доступен";
lblStatusDescription = ex.Message;
//---делаем проверку является ли код ошибки HttpStatusCode или WebExceptionStatus
if (((HttpWebResponse)ex.Response) != null)
{
lblCause = ((int)(((HttpWebResponse)ex.Response).StatusCode)).ToString();
}
else
{
lblCause = ((int)(ex.Status)).ToString();
}
//-------------------------------------------------------------------------------
StringBuilder newContent = new StringBuilder();
newContent.AppendLine("URL: " + url +
"\nДата проверки: " + date +
"\nСтатус: " + lblResult +
"\nОписание статуса: " + lblStatusDescription +
"\nКод статуса: " + lblCause + "\n");
all = newContent.ToString();
using (var db = new ParsingResultsEntities())
{
db.Urls.Add(urllink);
db.SaveChanges();
}
return(all);
}
}
public static TagImg src(this TagImg tag, string value)
{
tag.Src = value; return(tag);
}
public static TagImg onmousedown(this TagImg tag, string value)
{
tag.OnMouseDown = value; return(tag);
}
public static TagImg alt(this TagImg tag, string value)
{
tag.Alt = value; return(tag);
}
public static TagImg onmouseup(this TagImg tag, string value)
{
tag.OnMouseUp = value; return(tag);
}
public static TagImg height(this TagImg tag, Length value)
{
tag.Height = value; return(tag);
}
public static TagImg onmousemove(this TagImg tag, string value)
{
tag.OnMouseMove = value; return(tag);
}
public static TagImg usemap(this TagImg tag, string value)
{
tag.UseMap = value; return(tag);
}
public static TagImg onmouseout(this TagImg tag, string value)
{
tag.OnMouseOut = value; return(tag);
}
public static TagImg id(this TagImg tag, string value)
{
tag.Id = value; return(tag);
}
public static TagImg onkeypress(this TagImg tag, string value)
{
tag.OnKeyPress = value; return(tag);
}
public static TagImg @class(this TagImg tag, string value)
{
tag.Class = value; return(tag);
}
public static TagImg onkeydown(this TagImg tag, string value)
{
tag.OnKeyDown = value; return(tag);
}
public static TagImg title(this TagImg tag, string value)
{
tag.Title = value; return(tag);
}
public static TagImg onkeyup(this TagImg tag, string value)
{
tag.OnKeyUp = value; return(tag);
}
public static TagImg xmllang(this TagImg tag, string value)
{
tag.XmlLang = value; return(tag);
}
public static TagImg FloatLeft(this TagImg tag)
{
return(tag.Class("float_left"));
}