public static string Submit(this HtmlAgilityPack.HtmlNode form, Encoding encoding)
{
if (form == null)
{
throw new ArgumentNullException(nameof(form));
}
if (encoding == null)
{
throw new ArgumentNullException(nameof(encoding));
}
string postDataStr = form == null ? string.Empty :
form.SelectNodes("//input").ToList()
.ToDictionary(
(input => input.GetAttributeValue("name", string.Empty)),
(input => HttpUtility.UrlEncode(input.GetAttributeValue("value", string.Empty)))
)
.SerializeData <string>();
byte[] responseData = HttpRequestUtil.Post(form?.GetAttributeValue("action", null),
encoding.GetBytes(postDataStr)
);
return(encoding.GetString(responseData));
}
private List <string> FindAllAppKeysId(string htmlContent, string keyName = null)
{
List <string> htmlValues = new List <string>();
HtmlAgilityPack.HtmlDocument htmlDocument = new HtmlAgilityPack.HtmlDocument();
htmlDocument.LoadHtml(htmlContent);
HtmlAgilityPack.HtmlNode node = htmlDocument.DocumentNode.SelectSingleNode("//div[contains(@class, 'ui key list')]");
HtmlAgilityPack.HtmlNodeCollection nodes = node.ChildNodes;
foreach (HtmlAgilityPack.HtmlNode keyNode in nodes)
{
if (keyNode.OuterHtml.Contains(keyName == null ? "AltinnStudioAppKey" : keyName))
{
// Returns the button node
HtmlAgilityPack.HtmlNode deleteButtonNode = keyNode.SelectSingleNode("./div/button");
string dataId = deleteButtonNode.GetAttributeValue("data-id", string.Empty);
htmlValues.Add(dataId);
}
}
return(htmlValues);
}
public static XRefDetails From(HtmlAgilityPack.HtmlNode node)
{
if (node.Name != "xref")
{
throw new NotSupportedException("Only xref node is supported!");
}
var xref = new XRefDetails();
var uid = node.GetAttributeValue("uid", null);
var rawHref = node.GetAttributeValue("href", null);
NameValueCollection queryString = null;
if (!string.IsNullOrEmpty(rawHref))
{
if (!string.IsNullOrEmpty(uid))
{
Logger.LogWarning($"Both href and uid attribute are defined for {node.OuterHtml}, use href instead of uid.");
}
string others;
var anchorIndex = rawHref.IndexOf("#");
if (anchorIndex == -1)
{
xref.Anchor = string.Empty;
others = rawHref;
}
else
{
xref.Anchor = rawHref.Substring(anchorIndex);
others = rawHref.Remove(anchorIndex);
}
var queryIndex = others.IndexOf("?");
if (queryIndex == -1)
{
xref.Uid = HttpUtility.UrlDecode(others);
}
else
{
xref.Uid = HttpUtility.UrlDecode(others.Remove(queryIndex));
queryString = HttpUtility.ParseQueryString(others.Substring(queryIndex));
}
}
else
{
xref.Uid = uid;
}
xref.InnerHtml = node.InnerHtml;
xref.DisplayProperty = node.GetAttributeValue("displayProperty", queryString?.Get("displayProperty") ?? XRefSpec.NameKey);
xref.AltProperty = node.GetAttributeValue("altProperty", queryString?.Get("altProperty") ?? "fullName");
xref.Text = node.GetAttributeValue("text", node.GetAttributeValue("name", StringHelper.HtmlEncode(queryString?.Get("text"))));
xref.Alt = node.GetAttributeValue("alt", node.GetAttributeValue("fullname", StringHelper.HtmlEncode(queryString?.Get("alt"))));
xref.Title = node.GetAttributeValue("title", queryString?.Get("title"));
xref.SourceFile = node.GetAttributeValue("sourceFile", null);
xref.SourceStartLineNumber = node.GetAttributeValue("sourceStartLineNumber", 0);
xref.SourceEndLineNumber = node.GetAttributeValue("sourceEndLineNumber", 0);
// Both `data-raw-html` and `data-raw-source` are html encoded. Use `data-raw-html` with higher priority.
// `data-raw-html` will be decoded then displayed, while `data-raw-source` will be displayed directly.
xref.RawSource = node.GetAttributeValue("data-raw-source", null);
var raw = node.GetAttributeValue("data-raw-html", null);
if (!string.IsNullOrEmpty(raw))
{
xref.Raw = StringHelper.HtmlDecode(raw);
}
else
{
xref.Raw = xref.RawSource;
}
xref.ThrowIfNotResolved = node.GetAttributeValue("data-throw-if-not-resolved", false);
return(xref);
}
static UpdateInfo GetUpdateInfo(HtmlAgilityPack.HtmlNode doc, string NewVersion = null)
{
UpdateInfo updateInfo = new UpdateInfo
{
verson = NewVersion,
files = new List <UpdateFileInfo>(),
desc = new List <string>()
};
//label-latest
HtmlAgilityPack.HtmlNode releaseMain = GetReleaseMain(doc);
HtmlAgilityPack.HtmlNode Pre = releaseMain.SelectSingleNode("//div[contains(@class,'flex-self-start')]");
HtmlAgilityPack.HtmlNode header = releaseMain.SelectSingleNode("//div[contains(@class,'release-header')]");
HtmlAgilityPack.HtmlNode body = releaseMain.SelectSingleNode("//div[contains(@class,'markdown-body')]");
if (Pre != null)
{
if (Pre.InnerText.Contains("Pre"))
{
updateInfo.pre = true;
}
}
if (header != null)
{
if (NewVersion != null)
{
HtmlAgilityPack.HtmlNode headers = header.SelectSingleNode("div[1]/div[1]");
if (headers != null)
{
updateInfo.title = headers.InnerText.Replace("\n", "").Trim();
}
}
else
{
HtmlAgilityPack.HtmlNode headers = header.SelectSingleNode("div[1]/div[1]/a[1]");
if (headers != null)
{
string _Newversion = headers.GetAttributeValue("href", null);
if (_Newversion != null)
{
updateInfo.verson = _Newversion.Substring(_Newversion.LastIndexOf("/") + 1).TrimStart('v');
}
updateInfo.title = headers.InnerText.Replace("\n", "").Trim();
}
}
HtmlAgilityPack.HtmlNode headersTime = header.SelectSingleNode("p[1]/relative-time[1]");
if (headersTime != null)
{
string time = headersTime.GetAttributeValue("datetime", null);//2020-07-07T07:58:07Z
updateInfo.time = time.ToDateTime(DateTime.Now);
}
}
if (body != null)
{
HtmlAgilityPack.HtmlNodeCollection nodes = body.SelectNodes("p");
if (nodes != null && nodes.Count > 0)
{
foreach (HtmlAgilityPack.HtmlNode item in nodes)
{
HtmlAgilityPack.HtmlNodeCollection Anodes = item.ChildNodes;
if (Anodes != null && Anodes.Count > 0)
{
foreach (HtmlAgilityPack.HtmlNode Aitem in Anodes)
{
switch (Aitem.Name)
{
case "a":
string url = Aitem.GetAttributeValue("href", null);
if (url != null)
{
updateInfo.files.Add(new UpdateFileInfo
{
name = Aitem.InnerText.Trim(),
url = url
});
}
break;
case "#text":
updateInfo.desc.Add(Aitem.InnerText.Replace("\n", "").Trim());
break;
default:
Debug.WriteLine(Aitem.Name);
break;
}
}
}
else
{
updateInfo.desc.Add(item.InnerText.Trim());
}
}
}
else
{
var reg = new System.Text.RegularExpressions.Regex(@"(?is)<a(?:(?!href=).)*href=(['""]?)(?<url>[^""\s>]*)\1[^>]*>(?<text>(?:(?!</?a\b).)*)</a>");
var mc = reg.Matches(body.InnerHtml);
foreach (System.Text.RegularExpressions.Match m in mc)
{
string url = m.Groups["url"].Value;
if (url.StartsWith("/files/"))
{
updateInfo.files.Add(new UpdateFileInfo
{
name = m.Groups["text"].Value,
url = url
});
}
}
}
//string title = header.InnerText.Trim();
}
if (updateInfo.files.Count == 0)
{
HtmlAgilityPack.HtmlNode filebody = releaseMain.SelectSingleNode("//div[contains(@class,'Box--condensed')]");
if (filebody != null)
{
HtmlAgilityPack.HtmlNodeCollection files = filebody.SelectNodes("div[1]/div");
if (files != null)
{
foreach (HtmlAgilityPack.HtmlNode file in files)
{
HtmlAgilityPack.HtmlNode Afiles = file.SelectSingleNode("a[1]");
if (Afiles != null)
{
string url = Afiles.GetAttributeValue("href", null);
if (url != null)
{
if (url.StartsWith("/"))
{
url = "https://github.com" + url;
}
//https://github.com/giant-app/LiveWallpaper/archive/v1.4.67.zip
updateInfo.files.Add(new UpdateFileInfo
{
name = Afiles.InnerText.Trim(),
url = url
});
}
}
}
}
}
}
return(updateInfo);
}
private static void UpdateXref(HtmlAgilityPack.HtmlNode xref, Dictionary <string, XRefSpec> internalXRefMap, Dictionary <string, XRefSpec> externalXRefMap, Func <string, string> updater, string language)
{
var key = xref.GetAttributeValue("href", null);
// If name | fullName exists, use the one from xref because spec name is different from name for generic types
// e.g. return type: IEnumerable<T>, spec name should be IEnumerable
var name = xref.GetAttributeValue("name", null);
var fullName = xref.GetAttributeValue("fullName", null);
string displayName;
string href = null;
XRefSpec spec = null;
if (internalXRefMap.TryGetValue(key, out spec))
{
href = updater(spec.Href);
var hashtagIndex = href.IndexOf('#');
if (hashtagIndex == -1)
{
var htmlId = GetHtmlId(key);
// TODO: What if href is not html?
href = href + "#" + htmlId;
}
}
else if (externalXRefMap.TryGetValue(key, out spec) && !string.IsNullOrEmpty(spec.Href))
{
href = spec.Href;
}
// If href is not null, use name
if (href != null)
{
if (!string.IsNullOrEmpty(name))
{
displayName = name;
}
else
{
displayName = string.IsNullOrEmpty(fullName) ? key : fullName;
if (spec != null)
{
displayName = StringHelper.HtmlEncode(GetLanguageSpecificAttribute(spec, language, displayName, "name"));
}
}
var anchorNode = $"<a class=\"xref\" href=\"{href}\">{displayName}</a>";
xref.ParentNode.ReplaceChild(HtmlAgilityPack.HtmlNode.CreateNode(anchorNode), xref);
}
else
{
// If href is null, use fullName
if (!string.IsNullOrEmpty(fullName))
{
displayName = fullName;
}
else
{
displayName = string.IsNullOrEmpty(name) ? key : name;
if (spec != null)
{
displayName = StringHelper.HtmlEncode(GetLanguageSpecificAttribute(spec, language, displayName, "fullName", "name"));
}
}
var spanNode = $"<span class=\"xref\">{displayName}</span>";
xref.ParentNode.ReplaceChild(HtmlAgilityPack.HtmlNode.CreateNode(spanNode), xref);
}
}
private string[] ProcessNode(HtmlAgilityPack.HtmlNode htmlNode)
{
if (htmlNode == null || string.IsNullOrWhiteSpace(htmlNode.Name))
{
return(null);
}
string nodeName = htmlNode.Name.ToLower().Trim();
if (nodeName.StartsWith("#"))
{
return(null);
}
switch (nodeName)
{
case "a":
{
string href = htmlNode.GetAttributeValue("href", string.Empty);
if (!string.IsNullOrWhiteSpace(href))
{
string url = ProcessUrl(href, false);
if (!string.IsNullOrWhiteSpace(url))
{
return new string[] { url }
}
;
}
else if (_isLiveATC)
{
string onClick = htmlNode.GetAttributeValue("onClick", string.Empty);
if (!string.IsNullOrWhiteSpace(onClick) && onClick.ToLower().Contains("mydirectstream") && onClick.IndexOf('\'') > 0)
{
onClick = onClick.Substring(onClick.IndexOf('\'') + 1);
if (!string.IsNullOrWhiteSpace(onClick) && onClick.IndexOf('\'') > 0)
{
onClick = onClick.Substring(0, onClick.IndexOf('\''));
string url = ProcessUrl(string.Format("http://d.liveatc.net/{0}.m3u", onClick), false);
if (!string.IsNullOrWhiteSpace(url))
{
return new string[] { url }
}
;
}
}
}
break;
}
case "meta":
{
string metaType = htmlNode.GetAttributeValue("http-equiv", string.Empty);
if (string.Equals(metaType, "refresh", StringComparison.InvariantCultureIgnoreCase))
{
string metaContent = htmlNode.GetAttributeValue("content", string.Empty);
if (!string.IsNullOrWhiteSpace(metaContent))
{
int indUrl = metaContent.ToLower().IndexOf("url=");
if (indUrl >= 0)
{
metaContent = metaContent.Substring(indUrl + 4);
}
string url = ProcessUrl(metaContent, false);
if (!string.IsNullOrWhiteSpace(url))
{
return new string[] { url }
}
;
}
}
break;
}
case "object":
{
string classid = htmlNode.GetAttributeValue("classid", string.Empty);
if (string.Equals(classid, "CLSID:22d6f312-b0f6-11d0-94ab-0080c74c7e95", StringComparison.InvariantCultureIgnoreCase))
{
//this is a media player object...
string mediaURL = string.Empty;
foreach (HtmlAgilityPack.HtmlNode subNode in htmlNode.ChildNodes)
{
string subNodeName = subNode.Name.ToLower().Trim();
if (string.Equals(subNodeName, "param", StringComparison.InvariantCultureIgnoreCase) && string.Equals(subNode.GetAttributeValue("name", string.Empty), "filename", StringComparison.InvariantCultureIgnoreCase))
{
string strFileName = subNode.GetAttributeValue("value", string.Empty);
if (!string.IsNullOrWhiteSpace(strFileName))
{
return(new string[] { strFileName });
}
}
}
}
break;
}
case "script":
{
if (string.IsNullOrWhiteSpace(htmlNode.GetAttributeValue("src", string.Empty)) && !string.IsNullOrWhiteSpace(htmlNode.InnerText))
{
//process script...
System.IO.StringReader rdr = new System.IO.StringReader(htmlNode.InnerText);
string inLine = string.Empty;
do
{
inLine = rdr.ReadLine();
if (!string.IsNullOrWhiteSpace(inLine))
{
inLine = inLine.ToLower();
int indx = inLine.IndexOf(".playlist");
if (indx >= 0)
{
indx += 9;
inLine = inLine.Substring(indx);
indx = inLine.IndexOf('\'');
if (indx >= 0)
{
inLine = inLine.Substring(indx + 1);
indx = inLine.IndexOf('\'');
if (indx >= 0)
{
inLine = GetFeedUrlFromValue(inLine.Substring(0, indx), true);
if (!string.IsNullOrWhiteSpace(inLine))
{
return(new string[] { inLine });
}
}
}
}
else
{
indx = inLine.IndexOf("ipadurl");
if (indx >= 0)
{
indx += 7;
inLine = inLine.Substring(indx);
indx = inLine.IndexOf("\"");
if (indx < 0)
{
indx = inLine.IndexOf("'");
}
if (indx >= 0)
{
inLine = inLine.Substring(indx + 1);
indx = inLine.LastIndexOf("\"");
if (indx < 0)
{
indx = inLine.LastIndexOf("'");
}
if (indx >= 0)
{
inLine = GetFeedUrlFromValue(inLine.Substring(0, indx), true);
if (!string.IsNullOrWhiteSpace(inLine))
{
return(new string[] { inLine });
}
}
}
}
else
{
indx = inLine.IndexOf(".m3u");
if (indx >= 0)
{
int indxStart = inLine.IndexOf('"');
int indxEnd = inLine.LastIndexOf('"');
if (indxEnd > 0)
{
inLine = inLine.Substring(0, indxEnd);
}
if (indxStart >= 0)
{
inLine = inLine.Substring(indxStart + 1);
}
if (_isBroadcastify && inLine.IndexOf('"') >= 0)
{
indxStart = inLine.IndexOf('"');
indxEnd = inLine.LastIndexOf('"');
string tmpStart = inLine.Substring(0, indxStart);
string tmpEnd = inLine.Substring(indxEnd + 1);
if (string.IsNullOrWhiteSpace(_broadcastifyFeedId))
{
_broadcastifyListPrefix = tmpStart;
_broadcastifyListSuffix = tmpEnd;
}
else
{
inLine = tmpStart + _broadcastifyFeedId + tmpEnd;
if (!inLine.StartsWith("/"))
{
inLine = _broadcastifyPrefix + "/" + inLine;
}
else
{
inLine = _broadcastifyPrefix + inLine;
}
if (!string.IsNullOrWhiteSpace(inLine))
{
return(new string[] { inLine });
}
}
}
}
}
}
}
}while (inLine != null);
}
break;
}
default:
{
if (htmlNode.HasChildNodes)
{
List <string> subLinks = new List <string>();
foreach (HtmlAgilityPack.HtmlNode subNode in htmlNode.ChildNodes)
{
string[] links = ProcessNode(subNode);
if (links != null)
{
foreach (string strLink in links)
{
if (!subLinks.Contains(strLink))
{
subLinks.Add(strLink);
}
}
}
}
PostProcessBroadcastifyFeeds(subLinks);
return(subLinks.ToArray());
}
break;
}
}
return(null);
}
public void Convert(HtmlAgilityPack.HtmlNode node, System.Xml.XmlWriter writer)
{
var classOfParagraph = node.GetAttributeValue("class", null);
if (string.IsNullOrEmpty(classOfParagraph))
{
if (!string.IsNullOrEmpty(node.InnerText.Trim()))
{
using (writer.StartElement("p"))
{
writer.WriteAttributeString("content-type", "2.2 Story Text");
var innerContent = new InnerTextNode();
innerContent.Convert(node, writer);
}
}
}
else if (classOfParagraph.Equals("article-paragraph", StringComparison.CurrentCultureIgnoreCase))
{
using (writer.StartElement("p"))
{
writer.WriteAttributeString("content-type", "2.2 Story Text");
var innerContent = new InnerTextNode();
innerContent.Convert(node, writer);
}
}
else if (classOfParagraph.Equals("nomargin", StringComparison.CurrentCultureIgnoreCase))
{
using (writer.StartElement("p"))
{
writer.WriteAttributeString("content-type", "2.25 Story Text - No Spacing");
var innerContent = new InnerTextNode();
innerContent.Convert(node, writer);
}
}
else if (classOfParagraph.Equals("exhibitnumber", StringComparison.CurrentCultureIgnoreCase))
{
// exhibit numbers are handled by the image/table.
return;
}
else if (classOfParagraph.Equals("exhibittitle", StringComparison.CurrentCultureIgnoreCase))
{
// exhibit titles are handled by the image/table.
return;
}
else if (classOfParagraph.Equals("source", StringComparison.CurrentCultureIgnoreCase))
{
// sources are handled by the image/table.
return;
}
else if (classOfParagraph.Equals("question", StringComparison.CurrentCultureIgnoreCase))
{
var interviewQuestion = new InterviewQuestionNode();
interviewQuestion.Convert(node, writer);
}
else if (classOfParagraph.Equals("answer", StringComparison.CurrentCultureIgnoreCase))
{
var interviewAnswer = new InterviewAnswerNode();
interviewAnswer.Convert(node, writer);
}
else if (classOfParagraph.Equals("CompanyName", StringComparison.CurrentCultureIgnoreCase))
{
var companyName = new CompanyNameNode();
companyName.Convert(node, writer);
}
}
public List <SubtitleWebList> Search(string search)
{
List <SubtitleWebList> zimu1s = new List <SubtitleWebList>();
List <HttpLib.Val> _conmand = new List <HttpLib.Val> {
new HttpLib.Val("q", search)
};
List <HttpLib.Val> _header = new List <HttpLib.Val> {
new HttpLib.Val("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.9"),
new HttpLib.Val("Accept-Encoding", "gzip, deflate"),
new HttpLib.Val("Accept-Language", "zh-CN,zh;q=0.9,en;q=0.8,en-GB;q=0.7,en-US;q=0.6"),
new HttpLib.Val("User-Agent", "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/83.0.4103.61 Safari/537.36 Edg/83.0.478.37"),
};
string html = HttpLib.Http.Get(urlbase + "/search").header(_header).data(_conmand).redirect(true).request();
if (!string.IsNullOrEmpty(html))
{
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(html);
HtmlAgilityPack.HtmlNode htmlNode = doc.DocumentNode;
var sd = htmlNode.SelectNodes("//div[@class='box clearfix']/div[@class='item prel clearfix']");
if (sd != null)
{
foreach (HtmlAgilityPack.HtmlNode item in sd)
{
try
{
HtmlAgilityPack.HtmlNode main1 = item.SelectSingleNode("div[1]/a");
HtmlAgilityPack.HtmlNode main1_img = main1.SelectSingleNode("img");
HtmlAgilityPack.HtmlNode main2 = item.SelectSingleNode("div[2]");
var main2_table = main2.SelectNodes("div/table/tbody/tr");
SubtitleWebList zimu1 = new SubtitleWebList
{
title = main2.SelectSingleNode("p[1]/a").InnerText,
intro = main2.SelectSingleNode("p[2]/a").InnerText,
img = "http:" + main1_img.GetAttributeValue("data-original", null),
url = urlbase + main1.GetAttributeValue("href", null),
data = new List <SubtitleWebItem>(),
};
foreach (HtmlAgilityPack.HtmlNode table in main2_table)
{
try
{
HtmlAgilityPack.HtmlNode td1 = table.SelectSingleNode("td[1]");
HtmlAgilityPack.HtmlNode td2 = table.SelectSingleNode("td[2]/div/i");
HtmlAgilityPack.HtmlNode td11 = td1.SelectSingleNode("img");
if (td2 != null && td11 != null)
{
HtmlAgilityPack.HtmlNode td1a = td1.SelectSingleNode("a");
string type = td2.GetAttributeValue("title", null).TrimEnd('分');
SubtitleWebItem zimu2 = new SubtitleWebItem
{
imgalt = td11.GetAttributeValue("alt", null).Replace(" ", " ").Trim(),
imgsrc = urlbase + td11.GetAttributeValue("src", null),
title = td1a.InnerText,
url = urlbase + td1a.GetAttributeValue("href", null),
down = table.SelectSingleNode("td[3]").InnerText,
};
if (type.StartsWith("字幕质量:"))
{
type = type.Remove(0, 5).Trim();
int _type;
if (int.TryParse(type, out _type))
{
zimu2.type = _type;
}
}
zimu1.data.Add(zimu2);
}
}
catch { }
}
if (zimu1.data.Count > 0)
{
zimu1s.Add(zimu1);
}
}
catch { }
}
}
}
return(zimu1s);
}
public static bool HasClass(HtmlAgilityPack.HtmlNode node, params string[] arrClassName)
{
return(arrClassName.All(x => node.GetAttributeValue("class", "").Split(' ').Contains(x)));
}
/// <summary>
/// 抓取用户信息
/// </summary>
/// <remarks>
/// 步骤：
/// 1、加载用户名片地址，获取用户名片中的加userid密文(b89034eecee41f9b87b1eac2a0faac43)（如果获取不到，也就是它有可能跳转到用户店铺首页，此时获取另一种密文UvGx0OmvyvGIu）
/// 2、根据第1步中的密文，转到用户个人信誉页面提取数据
/// </remarks>
private void GrabeBuyerInfo(XCLShouCang.Model.TB_SearchKey keyModel, Action <string> action_ShowProcess, XCLNetTools.Message.MessageModel msgModel)
{
action_ShowProcess("正准备查询账号信息...");
string searchUrl = string.Empty;
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
HtmlAgilityPack.HtmlNode tempHTMLNode = null;
HtmlAgilityPack.HtmlNodeCollection tempHTMLNodeCon = null;
string pageHtml = string.Empty;
XCLShouCang.Model.TB_BuyerInfo buyerModel = new XCLShouCang.Model.TB_BuyerInfo();
Regex regNumber = new Regex(@"[-\+]?\d+(\.\d+)?");
#region 基本信息
action_ShowProcess("正在查询基本信息...");
Regex regUserID = new Regex("(user_id=)(.*)");
Regex regArea = new Regex("(：)(.*)");
Regex regReg = new Regex(@"(\d{4}.\d{2}.\d{2})");
searchUrl = Web.Common.CommonHelper.WebInfo.TB_UserProfileURL.Replace("{UID}", System.Web.HttpUtility.UrlEncode(keyModel.KeyName, System.Text.Encoding.GetEncoding("GB2312")));
pageHtml = Web.Common.CommonHelper.DownLoadHTMLString(searchUrl);
doc.LoadHtml(pageHtml);
buyerModel.NickName = (keyModel.KeyName ?? "").Trim();
//名片
var nameCardNode = doc.DocumentNode.SelectSingleNode("//div[@id='NameCard']");
if (null != nameCardNode)
{
//店铺
var shopNode = nameCardNode.SelectSingleNode("//li[contains(text(),'淘宝店铺')]/a");
if (null != shopNode)
{
buyerModel.ShopName = (shopNode.InnerText ?? "").Trim();
buyerModel.ShopURL = shopNode.GetAttributeValue("href", string.Empty);
}
//UID
buyerModel.UserID = regUserID.Match(buyerModel.ShopURL).Groups[2].Value;
//注册时间
tempHTMLNode = nameCardNode.SelectSingleNode("//li[contains(text(),'注册时间')]");
if (null != tempHTMLNode)
{
buyerModel.RegTime = XCLNetTools.StringHander.Common.GetDateTimeNullable(regReg.Match(tempHTMLNode.InnerText.Trim()).Groups[1].Value);
}
//认证
tempHTMLNode = nameCardNode.SelectSingleNode("//li[contains(text(),'认证情况')]/a/img");
if (null != tempHTMLNode)
{
buyerModel.CertificateLinkImg = tempHTMLNode.GetAttributeValue("src", string.Empty);
buyerModel.CertificateType = (tempHTMLNode.GetAttributeValue("title", string.Empty) ?? "").Trim();
}
//所在地
tempHTMLNode = nameCardNode.SelectSingleNode("//li[contains(text(),'所 在 地')]");
if (null != tempHTMLNode)
{
buyerModel.Area = (regArea.Match(tempHTMLNode.InnerText.Trim()).Groups[2].Value ?? "").Trim();
}
}
else
{
//此时可能被跳转到店铺首页去了
Match userIDMT = new Regex(@"user-rate-([A-Za-z0-9]+)\.htm").Match(pageHtml);//UvGx0OmvyvGIu
if (userIDMT.Success)
{
buyerModel.UserID = (userIDMT.Groups[1].Value ?? "").Trim();
buyerModel.ShopName = new Regex("首页|-|淘宝网").Replace(doc.DocumentNode.SelectSingleNode("//title").InnerText, "").Trim();
}
}
if (string.IsNullOrEmpty(buyerModel.UserID))
{
msgModel.Message = "未找到该用户信息！";
msgModel.IsSuccess = false;
return;
}
#endregion
#region 信用首页获取其它信息
action_ShowProcess("正在查询信用信息...");
pageHtml = Web.Common.CommonHelper.DownLoadHTMLString(Web.Common.CommonHelper.WebInfo.TB_UserRateHomeURL.Replace("{UID}", buyerModel.UserID.ToString()));
doc.LoadHtml(pageHtml);
var docNode = doc.DocumentNode;
//买家 http://rate.taobao.com/user-rate-49095164e83c97b0a713aa536ea1cbc5.htm
//卖家 http://rate.taobao.com/user-rate-b89034eecee41f9b87b1eac2a0faac43.htm
var personNode = docNode.SelectSingleNode("//div[contains(@class,'personal-info-box')]");
buyerModel.ShopCreateTime = XCLNetTools.StringHander.Common.GetDateTimeNullable(docNode.SelectSingleNode("//input[@id='J_showShopStartDate']").GetAttributeValue("value", string.Empty));
buyerModel.IsSeller = null != buyerModel.ShopCreateTime;
if (buyerModel.IsSeller)
{
tempHTMLNode = docNode.SelectSingleNode("//li[contains(text(),'当前主营')]");
if (null != tempHTMLNode)
{
buyerModel.ShopMainType = (tempHTMLNode.SelectSingleNode("a").InnerText ?? "").Trim();
}
tempHTMLNode = docNode.SelectSingleNode("//input[@id='J_ShopIdHidden']");
if (null != tempHTMLNode)
{
buyerModel.ShopID = XCLNetTools.StringHander.Common.GetLong(tempHTMLNode.GetAttributeValue("value", string.Empty));
}
tempHTMLNode = docNode.SelectSingleNode(string.Format("//a[contains(text(),'{0}')]", buyerModel.NickName));
if (null != tempHTMLNode)
{
buyerModel.ShopURL = (tempHTMLNode.GetAttributeValue("href", string.Empty) ?? "").Trim();
}
tempHTMLNode = docNode.SelectSingleNode("//li[contains(text(),'卖家信用')]");
if (null != tempHTMLNode)
{
buyerModel.SellerCreditValue = XCLNetTools.StringHander.Common.GetInt(regNumber.Match(tempHTMLNode.InnerText).Value);
if (buyerModel.SellerCreditValue > 0)
{
var sellerCreditImgNode = tempHTMLNode.SelectSingleNode("child::a/img");
if (null != sellerCreditImgNode)
{
buyerModel.SellerCreditImg = (sellerCreditImgNode.GetAttributeValue("src", string.Empty) ?? "").Trim();
}
}
}
var buyerCreditNode = docNode.SelectSingleNode("//li[contains(text(),'买家信用')]");
buyerModel.BuyerCreditValue = XCLNetTools.StringHander.Common.GetInt(regNumber.Match(buyerCreditNode.InnerText).Value);
if (buyerModel.BuyerCreditValue > 0)
{
tempHTMLNode = buyerCreditNode.SelectSingleNode("child::a/img");
if (null != tempHTMLNode)
{
buyerModel.BuyerCreditImg = (tempHTMLNode.GetAttributeValue("src", string.Empty) ?? "").Trim();
}
}
}
else
{
var buyerCreditNode = docNode.SelectSingleNode("//li[contains(text(),'买家信用')]");
buyerModel.BuyerCreditValue = XCLNetTools.StringHander.Common.GetInt(buyerCreditNode.SelectSingleNode("child::a[@id='J_BuyerRate']").InnerText.Trim());
if (buyerModel.BuyerCreditValue > 0)
{
buyerModel.BuyerCreditImg = (buyerCreditNode.SelectSingleNode(".//img").GetAttributeValue("src", string.Empty) ?? "").Trim();
}
}
//认证信息
if (string.IsNullOrEmpty(buyerModel.CertificateType))
{
tempHTMLNodeCon = docNode.SelectNodes("//li[contains(text(),'认证信息')]/a/img | //dt[contains(text(),'认证信息')]");
if (null != tempHTMLNodeCon && tempHTMLNodeCon.Count > 0)
{
foreach (var m in tempHTMLNodeCon)
{
buyerModel.CertificateLinkImg = (m.GetAttributeValue("src", string.Empty) ?? "").Trim();
buyerModel.CertificateType = (m.GetAttributeValue("title", string.Empty) ?? "").Trim();
if (string.IsNullOrEmpty(buyerModel.CertificateType) && string.Equals(m.Name, "dt", StringComparison.OrdinalIgnoreCase))
{
var cerImgNode = m.SelectSingleNode("following-sibling::dd/a/img");
if (null != cerImgNode)
{
buyerModel.CertificateLinkImg = (cerImgNode.GetAttributeValue("src", string.Empty) ?? "").Trim();
buyerModel.CertificateType = (cerImgNode.GetAttributeValue("title", string.Empty) ?? "").Trim();
}
}
if (!string.IsNullOrEmpty(buyerModel.CertificateType))
{
break;
}
}
}
}
//所在地
if (string.IsNullOrEmpty(buyerModel.Area))
{
tempHTMLNodeCon = docNode.SelectNodes("//li[contains(text(),'所在地区')] | //dt[contains(text(),'所在地区')]");
if (null != tempHTMLNodeCon && tempHTMLNodeCon.Count > 0)
{
foreach (var m in tempHTMLNodeCon)
{
buyerModel.Area = (new Regex(@"所在地区：|\s+").Replace(m.InnerText, "") ?? "").Trim();
if (string.IsNullOrEmpty(buyerModel.Area) && string.Equals(m.Name, "dt", StringComparison.OrdinalIgnoreCase))
{
buyerModel.Area = m.SelectSingleNode("following-sibling::dd").InnerText.Trim();
}
if (!string.IsNullOrEmpty(buyerModel.Area))
{
break;
}
}
}
}
//评价信息节点
var appraiseNode = docNode.SelectSingleNode("//div[contains(@class,'personal-rating')]//div[contains(@class,'rate-box')]");
//好评率
var goodRateNode = appraiseNode.SelectSingleNode("//em[contains(text(),'好评率')]");
if (null != goodRateNode)
{
buyerModel.GoodAppraiseRate = XCLNetTools.StringHander.Common.GetDecimal(regNumber.Match(goodRateNode.InnerText).Value);
}
//评价统计节点
var appraiseInfoNode = appraiseNode.SelectNodes("//div[@id='J_menu_list']//ul//li");
if (null != appraiseInfoNode && appraiseInfoNode.Count > 0)
{
var appraiseCountList = new List <XCLShouCang.Model.TB_AppraiseCount>();
var regGetCount = new Regex(@"\[(\d+),");
foreach (var m in appraiseInfoNode)
{
//[[5,{'rater':'1','direction':'0','result':'1','timeLine':'-7'}],[0,{'rater':'1','direction':'0','result':'0','timeLine':'-7'}],[0,{'rater':'1','direction':'0','result':'-1','timeLine':'-7'}]]
string relValue = m.GetAttributeValue("rel", string.Empty);
string mText = m.InnerText.Trim();
var matchs = regGetCount.Matches(relValue);
if (mText.Contains("最近一周"))
{
appraiseCountList.Add(new XCLShouCang.Model.TB_AppraiseCount()
{
AppraiseTimeType = Web.Common.CommonHelper.AppraiseTimeTypeEnum.最近一周.ToString(),
GoodCount = XCLNetTools.StringHander.Common.GetInt(matchs[0].Groups[1].Value),
MiddleCount = XCLNetTools.StringHander.Common.GetInt(matchs[1].Groups[1].Value),
BadCount = XCLNetTools.StringHander.Common.GetInt(matchs[2].Groups[1].Value)
});
}
else if (mText.Contains("最近一月"))
{
appraiseCountList.Add(new XCLShouCang.Model.TB_AppraiseCount()
{
AppraiseTimeType = Web.Common.CommonHelper.AppraiseTimeTypeEnum.最近一月.ToString(),
GoodCount = XCLNetTools.StringHander.Common.GetInt(matchs[0].Groups[1].Value),
MiddleCount = XCLNetTools.StringHander.Common.GetInt(matchs[1].Groups[1].Value),
BadCount = XCLNetTools.StringHander.Common.GetInt(matchs[2].Groups[1].Value)
});
}
else if (mText.Contains("最近半年"))
{
appraiseCountList.Add(new XCLShouCang.Model.TB_AppraiseCount()
{
AppraiseTimeType = Web.Common.CommonHelper.AppraiseTimeTypeEnum.最近半年.ToString(),
GoodCount = XCLNetTools.StringHander.Common.GetInt(matchs[0].Groups[1].Value),
MiddleCount = XCLNetTools.StringHander.Common.GetInt(matchs[1].Groups[1].Value),
BadCount = XCLNetTools.StringHander.Common.GetInt(matchs[2].Groups[1].Value)
});
}
else if (mText.Contains("半年以前"))
{
if (buyerModel.IsSeller)
{
var tmpRate = m.SelectNodes("//a[@data-point-val][contains(@href,'timeLine|-211')]");
if (null != tmpRate && tmpRate.Count == 3)
{
appraiseCountList.Add(new XCLShouCang.Model.TB_AppraiseCount()
{
AppraiseTimeType = Web.Common.CommonHelper.AppraiseTimeTypeEnum.半年以前.ToString(),
GoodCount = XCLNetTools.StringHander.Common.GetInt(tmpRate[0].InnerText.Trim()),
MiddleCount = XCLNetTools.StringHander.Common.GetInt(tmpRate[1].InnerText.Trim()),
BadCount = XCLNetTools.StringHander.Common.GetInt(tmpRate[2].InnerText.Trim())
});
}
}
else
{
//[[260,'http://ratehis.taobao.com/user-rate-UvFkbMmQyMGHL--isarchive|true--detailed|1--goodNeutralOrBad|1--timeLine|-211--receivedOrPosted|0--buyerOrSeller|1.htm#RateType'],[0,'http://ratehis.taobao.com/user-rate-UvFkbMmQyMGHL--isarchive|true--detailed|1--goodNeutralOrBad|0--timeLine|-211--receivedOrPosted|0--buyerOrSeller|1.htm#RateType'],[0,'http://ratehis.taobao.com/user-rate-UvFkbMmQyMGHL--isarchive|true--detailed|1--goodNeutralOrBad|-1--timeLine|-211--receivedOrPosted|0--buyerOrSeller|1.htm#RateType']]
appraiseCountList.Add(new XCLShouCang.Model.TB_AppraiseCount()
{
AppraiseTimeType = Web.Common.CommonHelper.AppraiseTimeTypeEnum.半年以前.ToString(),
GoodCount = XCLNetTools.StringHander.Common.GetInt(matchs[0].Groups[1].Value),
MiddleCount = XCLNetTools.StringHander.Common.GetInt(matchs[1].Groups[1].Value),
BadCount = XCLNetTools.StringHander.Common.GetInt(matchs[2].Groups[1].Value)
});
}
}
}
if (buyerModel.IsSeller)
{
buyerModel.SellerAppraiseCountList = appraiseCountList;
}
else
{
buyerModel.BuyerAppraiseCountList = appraiseCountList;
}
}
#endregion
#region 向数据库中保存数据
XCLShouCang.BLL.TB_SearchKey keyBLL = new XCLShouCang.BLL.TB_SearchKey();
keyBLL.Add(keyModel);
#endregion
msgModel.CustomObject = buyerModel;
msgModel.IsSuccess = true;
return;
}
public static XRefDetails From(HtmlAgilityPack.HtmlNode node)
{
if (node.Name != "xref")
{
throw new NotSupportedException("Only xref node is supported!");
}
var xref = new XRefDetails();
var rawUid = GetRawUid(node);
NameValueCollection queryString = null;
if (!string.IsNullOrEmpty(rawUid))
{
string others;
var anchorIndex = rawUid.IndexOf("#");
if (anchorIndex == -1)
{
xref.Anchor = string.Empty;
others = rawUid;
}
else
{
xref.Anchor = rawUid.Substring(anchorIndex);
others = rawUid.Remove(anchorIndex);
}
var queryIndex = others.IndexOf("?");
if (queryIndex == -1)
{
xref.Uid = HttpUtility.UrlDecode(others);
}
else
{
xref.Uid = HttpUtility.UrlDecode(others.Remove(queryIndex));
queryString = HttpUtility.ParseQueryString(others.Substring(queryIndex));
}
}
xref.InnerHtml = node.InnerHtml;
xref.DisplayProperty = node.GetAttributeValue("displayProperty", queryString?.Get("displayProperty") ?? XRefSpec.NameKey);
xref.AltProperty = node.GetAttributeValue("altProperty", queryString?.Get("altProperty") ?? "fullName");
xref.Text = node.GetAttributeValue("text", node.GetAttributeValue("name", StringHelper.HtmlEncode(queryString?.Get("text"))));
xref.Alt = node.GetAttributeValue("alt", node.GetAttributeValue("fullname", StringHelper.HtmlEncode(queryString?.Get("alt"))));
xref.Title = node.GetAttributeValue("title", queryString?.Get("title"));
// Both `data-raw-html` and `data-raw` are html encoded. Use `data-raw-html` with higher priority.
// `data-raw-html` will be decoded then displayed, while `data-raw` will be displayed directly.
var raw = node.GetAttributeValue("data-raw-html", null);
if (!string.IsNullOrEmpty(raw))
{
xref.Raw = StringHelper.HtmlDecode(raw);
}
else
{
xref.Raw = node.GetAttributeValue("data-raw", null);
}
xref.ThrowIfNotResolved = node.GetAttributeValue("data-throw-if-not-resolved", false);
return(xref);
}
/// <summary>
/// Prepare Html
/// </summary>
/// <param name="input"></param>
/// <param name="output"></param>
private void PrepareHtml(HtmlAgilityPack.HtmlNode input, HtmlAgilityPack.HtmlNode output)
{
HtmlAgilityPack.HtmlNode parent = output;
switch (input.NodeType)
{
case HtmlAgilityPack.HtmlNodeType.Document:
break;
case HtmlAgilityPack.HtmlNodeType.Element:
if (!availableTags.Contains(input.OriginalName))
{
return;
}
if (input.OriginalName.Equals("img"))
{
output.AppendChild(input.CloneNode(true));
return;
}
parent = output.AppendChild(input.CloneNode(false));
string newStyle = string.Empty;
string style = input.GetAttributeValue("style", string.Empty);
string href = input.GetAttributeValue("href", string.Empty);
parent.Attributes.RemoveAll();
if (style != string.Empty)
{
foreach (string item in style.Split(';'))
{
if ((item.Contains("font-weight") && item.Contains("bold")) ||
(item.Contains("font-style") && item.Contains("italic")) ||
(item.Contains("text-decoration") && item.Contains("underline")))
{
newStyle += string.Format("{0};", item);
}
}
}
if (newStyle != string.Empty)
{
parent.SetAttributeValue("style", newStyle);
}
if (href != string.Empty)
{
parent.SetAttributeValue("href", href);
}
break;
case HtmlAgilityPack.HtmlNodeType.Text:
output.AppendChild(input.CloneNode(true));
return;
default:
return;
}
foreach (HtmlAgilityPack.HtmlNode child in input.ChildNodes)
{
PrepareHtml(child, parent);
}
}
public string Convert(HtmlAgilityPack.HtmlNode node)
{
var x = node.GetAttributeValue("checked", false) ? "x" : " ";
return($"- [{x}] ");
}
public static string GetHref(this HtmlAgilityPack.HtmlNode htmlNode, string defaultValue = null)
{
return(htmlNode.GetAttributeValue("href", defaultValue));
}
public bool RecognitionPattern(HtmlAgilityPack.HtmlNode node)
{
return(node.GetAttributeValue("class", string.Empty).Equals("last_packages"));
}
public static bool ParseGoogleLoginPage(string strEmail, string strPassword, string strHTMLLoginPage, string strReferer, out string strData, out string strPostUri)
{
// Expected strData BEGIN
// ======================
// GALX=K2kN1B3RhzI
// &continue=https%3A%2F%2Faccounts.google.com%2Fo%2Foauth2%2Fauth%3Fscope%3Demail%26response_type%3Dcode%26redirect_uri%3Dhttps%3A%2F%2Fteechip.com%2Fgoogle%26client_id%3D759311199253-liule4h0ij4trokni0clghutesmfrpaq.apps.googleusercontent.com%26hl%3Dvi%26from_login%3D1%26as%3D6c13d32c4c484e0a
// &service=lso
// <mpl=popup
// &shdf=Cm4LEhF0aGlyZFBhcnR5TG9nb1VybBoADAsSFXRoaXJkUGFydHlEaXNwbGF5TmFtZRoHVGVlQ2hpcAwLEgZkb21haW4aB1RlZUNoaXAMCxIVdGhpcmRQYXJ0eURpc3BsYXlUeXBlGgdERUZBVUxUDBIDbHNvIhR9f65sYBLeiTyA3ksWmSKfvjjy-CgBMhR69Mfqff5tofmPaXiU63kbTQdAzw
// &scc=1
// &sarp=1
// &_utf8=%E2%98%83
// &bgresponse=%21qapCVouLleiO7EZE_x4-8E88lh4CAAAAMFIAAAAFKgEfCMGz3oizuMFi9s1elOqqg8JcOP5CRdkNPXH--weicJaEAyf-pqsUcGZzFbdUIPWD6hVYEdNFEz35CvkIrsyHVpQRGe5ksfhAee9ehS3pR6bzwXyNFX7Pap-J7SQaN1swC53RMscX-GgiAB7Bd4pINn3QSjFPup0K0JDhJsMqBdkFQikNK4TCndA3NFztVcnkdr91yQuMEwxpfG6CjcHt3ckSyopOpBsFuHRcRqeEorlpvItyd9ITIvZdWd6LNZJgkDdAVbI2TdP3bPWBv9Lu5ux73a-RC4Oc7ow5uAWWbcMQ2D_4hKd4xT66Ie6pV8U4JxgvxhegMPaylafpK6NrqFNbEtEHjTksqILJKLRInylFLU37UpdQuwqPtUWz240
// &pstMsg=1
// &dnConn=
// &checkConnection=youtube%3A3988%3A1
// &checkedDomains=youtube
// &Email=<strEmail>
// &Passwd=<strPassword>
// &signIn=Sign+in
// &PersistentCookie=yes
// ======================
// Expected strData END
strData = "";
strPostUri = "";
HtmlAgilityPack.HtmlDocument htmlDoc = new HtmlAgilityPack.HtmlDocument();
htmlDoc.LoadHtml(strHTMLLoginPage);
if (htmlDoc.DocumentNode != null)
{
HtmlAgilityPack.HtmlNode frmNode = htmlDoc.DocumentNode.SelectSingleNode("//form[@id=\"gaia_loginform\"]");
if (frmNode != null)
{
strPostUri = frmNode.GetAttributeValue("action", "");
BuildDataString(frmNode, "Page", ref strData, true);
BuildDataString(frmNode, "GALX", ref strData, string.IsNullOrEmpty(strData));
BuildDataString(frmNode, "gxf", ref strData, string.IsNullOrEmpty(strData));
BuildDataString(GetValueByName(frmNode, "input", "continue"), "continue", ref strData);
BuildDataString(GetValueByName(frmNode, "input", "followup"), "followup", ref strData);
BuildDataString(frmNode, "service", ref strData);
BuildDataString(frmNode, "ProfileInformation", ref strData);
BuildDataString(GetValueByName(frmNode, "input", "_utf8"), "_utf8", ref strData);
BuildDataString(frmNode, "bgresponse", ref strData);
BuildDataString(frmNode, "pstMsg", ref strData);
BuildDataString(frmNode, "dnConn", ref strData);
BuildDataString(frmNode, "checkConnection", ref strData);
BuildDataString(frmNode, "checkedDomains", ref strData);
BuildDataString(frmNode, "identifiertoken", ref strData);
BuildDataString(frmNode, "identifiertoken_audio", ref strData);
BuildDataString(strEmail, "Email", ref strData);
BuildDataString(strPassword, "Passwd", ref strData);
BuildDataString(frmNode, "rmShown", ref strData);
}
}
return(false);;
}
