public HtmlNode GetElementByProperty(IEnumerable <HtmlNode> nodes, string propertyName, string propertyValue)
{
HtmlNode n = null;
foreach (HtmlNode node in nodes)
{
HtmlAttributeCollection attributes = node.Attributes;
if (attributes.Count > 0)
{
foreach (HtmlAttribute att in attributes)
{
if (att.Name != null && att.Name.Equals(propertyName) && att.Value.Equals(propertyValue))
{
n = node;
break;
}
}
if (n != null)
{
break;
}
}
}
return(n);
}
public LoginXPXB()
{
InitializeComponent();
ServicePointManager.ServerCertificateValidationCallback = new RemoteCertificateValidationCallback(ValidateServerCertificate);
loginUrl = BbQuery.BbLoginUrlNative;
loginUrl += "/?action=relogin";
loginDoc = getLoginDoc(loginUrl);
doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(loginDoc);
var forms = doc.DocumentNode.Descendants("FORM");//.Where(node => node.Name == "login");
foreach (var f in forms)
{
HtmlAttributeCollection atts = f.Attributes;
foreach (HtmlAttribute h in atts)
{
Debug.WriteLine(h.Name + " = " + h.Value);
}
Debug.WriteLine("form descs inputs " + f.Descendants("INPUT").Count().ToString());
}
var inputs = doc.DocumentNode.Descendants("INPUT");
foreach (var f in inputs)
{
HtmlAttributeCollection atts = f.Attributes;
foreach (HtmlAttribute h in atts)
{
Debug.WriteLine(h.Name + " = " + h.Value);
}
}
}
private void TruncateTag(HtmlNode node, string tagName)
{
HtmlAttributeCollection attributes = node.Attributes;
while (attributes.Count > 0)
{
AddError(Constants.ERROR_ATTRIBUTE_NOT_IN_POLICY,
HtmlEntityEncoder.HtmlEntityEncode(tagName),
HtmlEntityEncoder.HtmlEntityEncode(attributes[0].Name),
HtmlEntityEncoder.HtmlEntityEncode(attributes[0].Value));
node.Attributes.Remove(attributes[0].Name);
}
HtmlNodeCollection childNodes = node.ChildNodes;
int j = 0;
int length = childNodes.Count;
for (int i = 0; i < length; i++)
{
HtmlNode nodeToRemove = childNodes[j];
if (nodeToRemove.NodeType != HtmlNodeType.Text)
{
node.RemoveChild(nodeToRemove);
}
else
{
j++;
}
}
}
public void GetContent()
{
if (docPageParse != null)
{
HtmlNodeCollection pageNodes = docPageParse.DocumentNode.SelectNodes("//tr");
if (pageNodes != null)
{
foreach (var item in pageNodes)
{
HtmlAttributeCollection str = item.Attributes;
foreach (var item1 in str)
{
if (item1.Name == "style")
{
if (item1.Value == "cursor:pointer;")
{
VRXParsePage record = new VRXParsePage((item.Id).Replace("td", ""));
SaveRecord(record.saveRecord);
}
}
}
}
}
else
{
SaveError("Content is not null (NOT VIP) " + pageParse);
}
}
}
//________________________________________________________________________________________________________
//第二部分实验 (模块1：根据模板抽取信息)
private void divide_Click(object sender, EventArgs e) //得到模板节点
{
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
String txt = this.Domtext.Text;
doc.LoadHtml(txt);
HtmlNode node = doc.DocumentNode.FirstChild;
HtmlAttributeCollection attrcollection = node.Attributes;
String tag = node.Name;
HtmlAttributeCollection attrs = node.Attributes;
foreach (var item in attrs)
{
Console.WriteLine(item.Name + " : " + item.Value); //提取节点的属性集合
}
this.Domtext.Text = tag + " " + node.HasChildNodes;
Console.WriteLine("文本信息： " + node.InnerText);
/* foreach (HtmlNode link in doc.DocumentNode.ChildNodes)
* {
* // this.Domtext.Text=link.InnerText;
* // Console.Out.WriteLine(link.InnerText);
*
* } */
}
public void HtmlAttributeCollectionConstructorTest()
{
HtmlElement element = new HtmlElement("root");
HtmlAttributeCollection target = new HtmlAttributeCollection(element);
StringAssert.Equals(target.HtmlElement.Name, "root");
}
private static void TryRemoveTagAttribute(HtmlAttributeCollection atts)
{
for (int i = atts.Count - 1; i >= 0; i--)
{
var item = atts[i];
if (DefaultAllowedAttributes.Contains(item.Name) == false)
{
item.Remove(); continue;
}
if (item.Value.Contains("&{"))
{
item.Remove(); continue;
}
if (DefaultUriAttributes.Contains(item.Name))
{
if (TryRemoveSchemes(item))
{
continue;
}
}
if (item.Name.ToLower() == "style")
{
TryRemoveStyle(item);
}
}
}
public bool Start(NBoilerpipeContentHandler instance, string localName, HtmlAttributeCollection atts)
{
string sizeAttr = atts ["size"].Value;
if (sizeAttr != null)
{
Matcher m = CommonTagActions.PAT_FONT_SIZE.Matcher(sizeAttr);
if (m.Matches())
{
string rel = m.Group(1);
int val = System.Convert.ToInt32(m.Group(2));
int size;
if (rel.Length == 0)
{
// absolute
size = val;
}
else
{
// relative
int?prevSize;
if (instance.fontSizeStack.IsEmpty())
{
prevSize = 3;
}
else
{
prevSize = 3;
foreach (int?s in instance.fontSizeStack)
{
if (s != null)
{
prevSize = s;
break;
}
}
}
if (rel[0] == '+')
{
size = (int)prevSize + val;
}
else
{
size = (int)prevSize - val;
}
}
instance.fontSizeStack.Add(0, size);
}
else
{
instance.fontSizeStack.Add(0, null);
}
}
else
{
instance.fontSizeStack.Add(0, null);
}
return(false);
}
private void ReadEventDetails(HtmlDocument doc)
{
// Read event table for event details
//HtmlNode eventTable = doc.DocumentNode.SelectNodes(eventTableTag)[0];
try
{
foreach (HtmlNode eventTable in doc.DocumentNode.SelectNodes(eventTableTag))
{
HtmlAttributeCollection attributes = eventTable.Attributes;
foreach (HtmlAttribute dateAttribute in attributes.AttributesWithName(dateTag))
{
dateString = dateAttribute.Value;
}
foreach (HtmlAttribute eventAttribute in attributes.AttributesWithName(eventTag))
{
eventName = eventAttribute.Value;
}
foreach (HtmlAttribute betTypeAttribute in attributes.AttributesWithName(betTag))
{
betType = betTypeAttribute.Value;
}
foreach (HtmlAttribute sportAttribute in attributes.AttributesWithName(sportTag))
{
sport = sportAttribute.Value;
}
foreach (HtmlAttribute sportBracketAttribute in attributes.AttributesWithName(sportBracketTag))
{
sportBracket = sportBracketAttribute.Value;
}
}
}
catch
{
}
if (dateString != "")
{
try
{
eventTime = Convert.ToDateTime(dateString);
if (eventTime < DateTime.Now)
{
inPlay = true;
}
}
catch
{
}
timeToEvent = eventTime - DateTime.Now;
}
}
private void LoadInOdds(HtmlDocument doc)
{
List <string> bookieCodesList = new List <string>();
HtmlNode headerRow = doc.DocumentNode.SelectNodes(eventTableHeaderTag)[0];
if (headerRow != null)
{
foreach (HtmlNode cell in headerRow.SelectNodes("th|td"))
{
HtmlAttributeCollection attributes = cell.Attributes;
foreach (HtmlAttribute bookieCode in attributes.AttributesWithName(bookieTag))
{
// If bookie code is not already in the list
if (!bookieCodesList.Contains(bookieCode.Value))
{
bookieCodesList.Add(bookieCode.Value);
}
}
}
}
HtmlNodeCollection eventRows = doc.DocumentNode.SelectNodes(eventTableRowTag);
foreach (HtmlNode eventRow in eventRows)
{
string name = "";
List <string> oddsList = new List <string>();
HtmlAttributeCollection attributes = eventRow.Attributes;
foreach (HtmlAttribute resultName in attributes.AttributesWithName(resultTag))
{
name = resultName.Value;
}
foreach (HtmlNode cell in eventRow.SelectNodes("th|td"))
{
attributes = cell.Attributes;
bool isNoBet = (cell.Attributes.Contains("class") && cell.Attributes["class"].Value.Contains(noBetTag));
if (!isNoBet)
{
foreach (HtmlAttribute odds in attributes.AttributesWithName(oddsTag))
{
oddsList.Add(odds.Value);
}
}
else
{
oddsList.Add("0");
}
}
results.Add(new Result(name, bookieCodesList, oddsList, bookmakers));
}
}
protected override void RenderBeginTag(string tagName, HtmlAttributeCollection attributes)
{
var ctx = CurrentContext;
ctx.Document.AddStyleAttribute("white-space", "nowrap");
ctx.Document.RenderBeginTag("span");
base.RenderBeginTag(tagName, attributes);
}
/// <summary>
/// 删除节点上所有的属性
/// </summary>
/// <param name="node"></param>
public static void RemoveAttributes(this HtmlNode node)
{
HtmlAttributeCollection attrs = node.Attributes;
for (var i = 0; i < attrs.Count; i++)
{
node.Attributes.Remove(attrs[i]);
}
}
private void TagContentSearch_Click(object sender, RoutedEventArgs e)
{
string TagName = GUITagBox.Text;
string[] Attributes = GUITagAttributeBox.Text.Split(new string[] { " " }, StringSplitOptions.RemoveEmptyEntries);;
List <HtmlNode> acceptedNodes = new List <HtmlNode>(); // those among matchingNodes that have all required attributes
FolderArea.Text = "";
foreach (string filePath in LocalFolderFiles)
{
string fileContents = File.ReadAllText(filePath);
HtmlAgilityPack.HtmlDocument theHTML = new HtmlDocument();
theHTML.LoadHtml(fileContents);
IEnumerable <HtmlNode> matchingNodes = theHTML.DocumentNode.Descendants(TagName);
foreach (var matchingNode in matchingNodes)
{
HtmlAttributeCollection matchingAttrs = matchingNode.Attributes;
bool AcceptThisNode = true;
foreach (string attribute in Attributes)
{
HtmlAttribute attr = matchingAttrs[0];
int Number = matchingAttrs.Where(x => x.Name == attribute).Count();
if (Number == 0)
{
//this matchingNode did not make it - go to next outer loop
AcceptThisNode = false;
break;
}
}
if (AcceptThisNode)
{
if (matchingNode.InnerHtml != "" && matchingNode.InnerHtml != null)
{
acceptedNodes.Add(matchingNode);
}
}
}
}
if (acceptedNodes.Count() == 0)
{
FolderArea.Text = "No matching tags were found.";
}
else
{
foreach (HtmlNode node in acceptedNodes)
{
FolderArea.Text += node.InnerHtml;
}
}
}//end fn
internal static JObject NodeAttributesToJson(HtmlAttributeCollection attributeCollection)
{
var json = new JObject();
foreach (var attribute in attributeCollection)
{
json.Add(attribute.Name, attribute.Value);
}
return(json);
}
public static Hashtable fromSetting(string html)
{
Hashtable ht = new Hashtable();
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(html);
string data = null;
string cUrl = null;
foreach (HtmlNode form in doc.DocumentNode.SelectNodes("//form"))
{
cUrl = form.Attributes["action"].Value;
ht["url"] = cUrl;
//Console.WriteLine("Found: " + form.Attributes["action"].Value);
//HtmlAttributeCollection attrs = form.Attributes;
//// Console.Write(attrs.Count.ToString());
//Console.WriteLine("Found: " + form.ChildAttributes("action"));
foreach (HtmlNode row in form.SelectNodes("//input"))//这里只是遍历所有的INPUT元素
{
HtmlAttributeCollection input = row.Attributes;
Console.WriteLine("row");
string cInputName = null;
string cInputValue = null;
foreach (var item in input)//这里只是遍历一个元素所有属性
{
if (item.Name.ToString().ToLower().Equals("name"))
{
cInputName = item.Value;
}
if (item.Name.ToString().ToLower().Equals("value"))
{
cInputValue = item.Value;
}
}
if (cInputName == null)
{
continue;
}
if (cInputName.Equals("key"))
{
ht["key"] = cInputValue;
}
;
data += cInputName + "=" + cInputValue;
data += "&";
}
}
if (data != null)
{
ht["data"] = data;
}
return(ht);
}
public static void AddAttribute(HtmlAttributeCollection atributos, string key, string value, FieldTypes fieldType)
{
if (atributos.Contains(key))
{
atributos[key].Value += ", " + value;
}
else
{
atributos.Add(key, value);
}
}
/// <summary>
/// Gets the value of an HTML attribute safely from an HTMLAttributeCollection
/// </summary>
/// <param name="p_hacCollection"></param>
/// <param name="p_strItem"></param>
/// <returns></returns>
private static string GetHtmlAttributeValue(HtmlAttributeCollection p_hacCollection, string p_strItem)
{
if (p_hacCollection.Contains(p_strItem))
{
return(((HtmlAttribute)p_hacCollection[p_strItem]).Value);
}
else
{
return("");
}
}
public string ExtractQueryHashScriptLink()
{
HtmlNode link = _documentNode.QuerySelector("[href*=\"" + QueryHashSource + "\"]");
HtmlAttributeCollection attributes = link.Attributes;
HtmlAttribute href = attributes.FirstOrDefault(attr => attr.Name == "href");
string scriptSourceRelative = href.Value;
string scriptSourceAbsolute = Utility.BaseUrl + scriptSourceRelative;
return(scriptSourceAbsolute);
}
private bool MatchesAttributes(HtmlAttributeCollection attributes, TextSearch search)
{
foreach (var attribute in attributes)
{
if (search.IsMatch(attribute.Name) || search.IsMatch(attribute.Value))
{
return(true);
}
}
return(false);
}
public static bool AttributeExists(this HtmlAttributeCollection collection, string name, string value)
{
if (collection == null)
{
return(false);
}
var attr = collection.AttributesWithName(name)?.FirstOrDefault() ?? null;
return(attr != null && attr.Value == value);
}
public static void AppendOrCreate(this HtmlAttributeCollection helper, string name, string value)
{
if (!helper.Contains(name))
{
helper.Append(name, value);
}
else
{
helper[name].Value += value;
}
}
public static T GetValue <T>(this HtmlAttributeCollection attributes, string key, IFormatProvider formatProvider)
{
var value = attributes[key]?.Value;
if (string.IsNullOrEmpty(value))
{
return(default(T));
}
T convertedValue = (T)Convert.ChangeType(value, typeof(T), formatProvider);
return(convertedValue);
}
public static void RemoveAttribute(this HtmlNode node, string attrName)
{
HtmlAttributeCollection attrs = node.Attributes;
for (var i = 0; i < attrs.Count; i++)
{
var attr = attrs[i];
if (string.Equals(attr.Name, attrName, StringComparison.OrdinalIgnoreCase))
{
node.Attributes.Remove(attr);
break;
}
}
}
//根据文本框中信息抽取指定节点
private void extract_info_Click(object sender, EventArgs e)
{
Template tm = new Template(); //获取模板
tm.createTem(this.Domtext.Text); //将模板放在Domtext中；
attrs = tm.getattr(); //设置抽取属性；
tag = tm.gettag(); //设置抽取标签；
// tm.Extract("http://www.walmart.com/ip/VTech-CS6919-15-DECT-6.0-Expandable-Cordless-Phone-with-Caller-ID-and-Handset-Speakerphone-Red/45074431");
// this.webBrowser1.Navigate("http://www.walmart.com/ip/VTech-CS6919-15-DECT-6.0-Expandable-Cordless-Phone-with-Caller-ID-and-Handset-Speakerphone-Red/45074431");
nextextraturl(extractid);
// partition_timer.Start();
}
public static Hashtable HtmlTable(string s)
{
HtmlDocument doc = new HtmlDocument();
doc.LoadHtml(s);
Hashtable htc = new Hashtable();
int index = 1;
foreach (HtmlNode table in doc.DocumentNode.SelectNodes("//table"))
{
// table.Attributes
// Console.WriteLine("Found: " + table.Id);
HtmlAttributeCollection attrs = table.Attributes;
// Console.Write(attrs.Count.ToString());
if (attrs.Count != 6)
{
continue;
}
foreach (HtmlNode row in table.SelectNodes("tr"))
{
// Console.WriteLine("row");
// Console.WriteLine(row.HasAttributes);
int i = 0;
Hashtable ht = new Hashtable();
foreach (HtmlNode cell in row.SelectNodes("th|td"))
{
string val = null;
//Console.WriteLine(i);
val = cell.InnerText;
// Console.WriteLine("cell: " + cell.InnerText);
//string sp = SelectSP(cell.InnerText.ToString());
if (i == 9)
{
string ss = Selectvalue(cell.InnerHtml, "a");
string u = "http://vip.win007.com/changeDetail/overunder.aspx?id=" + ss + "&companyid=3";
val = u;
}
ht.Add(i, val);
i++;
}
htc.Add(index, ht);
index++;
}
}
return(htc);
}
private static void addProps(StringBuilder dynamicObject, HtmlAttributeCollection attributes)
{
foreach (var attribute in attributes)
{
if (attribute.Value.StartsWith("@"))
{
dynamicObject.Append($".Add(\"{attribute.Name}\", {attribute.Value.Substring(1)})");
}
else
{
dynamicObject.Append($".Add(\"{attribute.Name}\", \"{attribute.Value}\")");
}
}
}
private static List <string> GetCourses()
{
HtmlDocument boardHtml = new HtmlDocument();
boardHtml.LoadHtml(Driver.GetInstance.WebDrive.PageSource);
List <string> Courses = new List <string>();
foreach (var Nodo in boardHtml.DocumentNode.CssSelect(".ic-DashboardCard__link"))
{
HtmlAttributeCollection atts = Nodo.Attributes;
Courses.Add(atts.Where(a => a.Name.ToLower() == "href").FirstOrDefault().Value);
}
return(Courses);
}
protected IEnumerable <ViewLayerControlAttribute> ConvertAttributes(HtmlAttributeCollection attributeCollection)
{
var convertedAttributes = attributeCollection
.Where(attr => AttributeMap.ContainsKey(attr.Name))
.Select(attr =>
{
if (attr.QuoteType == AttributeValueQuote.DoubleQuote)
{
return(new ViewLayerControlAttribute($"{AttributeMap[attr.Name]}", $"\"{attr.Value}\""));
}
return(new ViewLayerControlAttribute($"{AttributeMap[attr.Name]}", $"'{attr.Value}'"));
});
return(convertedAttributes);
}
protected void InitXmlNode(XmlNode xmlNode, HtmlDocument htmlDoc)
{
Debug.Assert(xmlNode != null);
Debug.Assert(
xmlNode.OwnerDocument == htmlDoc.m_XmlNode.OwnerDocument ||
xmlNode.OwnerDocument == htmlDoc.m_XmlNode
);
m_XmlNode = xmlNode;
m_UpperName = null;
m_Doc = htmlDoc;
m_Attributes = new HtmlAttributeCollection(this);
m_ChildNodeList = new HtmlNodeList(m_XmlNode.ChildNodes);
}
public void createTem(String s) //根据纯文本得到节点属性
{
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
doc.LoadHtml(s);
HtmlNode node = doc.DocumentNode.FirstChild;
tag = node.Name;
attrs = node.Attributes;
foreach (var item in attrs)
{
attrString = " and " + "@" + item.Name + "=" + "'" + item.Value + "'"; //提取节点的属性集合
}
attrString = attrString.Substring(5);
attrString = "//" + tag + "[" + attrString + "]";
}
protected bool isAttributeValueEquals(HtmlAttributeCollection attributes, string attr, string str)
{
try
{
if (attributes.Contains(attr))
{
if (attributes[attr].Value == str)
{
return(true);
}
}
} catch (Exception) { }
return(false);
}
/// <summary>
/// Initializes a new instance of the <see cref="MimeKit.Text.HtmlTagToken"/> class.
/// </summary>
/// <remarks>
/// Creates a new <see cref="HtmlTagToken"/>.
/// </remarks>
/// <param name="name">The name of the tag.</param>
/// <param name="attributes">The attributes.</param>
/// <param name="isEmptyElement"><c>true</c> if the tag is an empty element; otherwise, <c>false</c>.</param>
/// <exception cref="System.ArgumentNullException">
/// <para><paramref name="name"/> is <c>null</c>.</para>
/// <para>-or-</para>
/// <para><paramref name="attributes"/> is <c>null</c>.</para>
/// </exception>
public HtmlTagToken (string name, IEnumerable<HtmlAttribute> attributes, bool isEmptyElement) : base (HtmlTokenKind.Tag)
{
if (name == null)
throw new ArgumentNullException (nameof (name));
if (attributes == null)
throw new ArgumentNullException (nameof (attributes));
Attributes = new HtmlAttributeCollection (attributes);
IsEmptyElement = isEmptyElement;
Id = name.ToHtmlTagId ();
Name = name;
}
internal HtmlAttributeCollection CreateTestCollection()
{
HtmlElement root = new HtmlElement("root");
HtmlAttributeCollection target = new HtmlAttributeCollection(root);
HtmlAttribute attribute = new HtmlAttribute("first");
target.Add(attribute);
attribute = new HtmlAttribute("second", "value");
target.Add(attribute);
attribute = new HtmlAttribute("third", "\"another value\"");
target.Add(attribute);
return target;
}
private static void Entitize(HtmlAttributeCollection collection)
{
foreach (HtmlAttribute at in collection)
{
at.Value = Entitize(at.Value);
}
}
public FlowedToHtmlTagContext (HtmlTagId tag) : base (tag)
{
attrs = HtmlAttributeCollection.Empty;
}
public FlowedToHtmlTagContext (HtmlTagId tag, HtmlAttribute attr) : base (tag)
{
attrs = new HtmlAttributeCollection (new [] { attr });
}
/// <summary>
/// Initializes a new instance of the <see cref="MimeKit.Text.HtmlTagToken"/> class.
/// </summary>
/// <remarks>
/// Creates a new <see cref="HtmlTagToken"/>.
/// </remarks>
/// <param name="name">The name of the tag.</param>
/// <param name="isEndTag"><c>true</c> if the tag is an end tag; otherwise, <c>false</c>.</param>
/// <exception cref="System.ArgumentNullException">
/// <paramref name="name"/> is <c>null</c>.
/// </exception>
public HtmlTagToken (string name, bool isEndTag) : base (HtmlTokenKind.Tag)
{
if (name == null)
throw new ArgumentNullException (nameof (name));
Attributes = new HtmlAttributeCollection ();
Id = name.ToHtmlTagId ();
IsEndTag = isEndTag;
Name = name;
}
/// <summary>
/// This constructs a new Html element with the specified tag attributeName.
/// </summary>
public HtmlElement(string name)
{
this.nodes = new HtmlNodeCollection(this);
this.attributes = new HtmlAttributeCollection(this);
this.name = name;
}
