private static Dictionary <string, EmailRecord> getNameValueByElementType(
HtmlAgilityPack.HtmlDocument source,
SalesForce salesForce
)
{
Dictionary <string, EmailRecord> output
= new Dictionary <string, EmailRecord>();
var document = source.DocumentNode;
foreach (KeyValuePair <string, bool> emailItem
in salesForce.emailHeaderIdentities)
{
if (emailItem.Value)
{
HtmlAgilityPack.HtmlNode editNode = source.GetElementbyId(emailItem.Key);
if (editNode.Attributes.ToList().Count(x => x.Name == "value") >= 1)
{
HtmlAgilityPack.HtmlAttribute attribute = editNode.Attributes["value"];
EmailRecord record = new EmailRecord();
record.emailAddress = attribute.Value;
output.Add(emailItem.Key, record);
}
}
}
return(output);
}
public HtmlAttributeWrapper(HtmlAttribute wrappedAttribute)
{
if(wrappedAttribute == null)
throw new ArgumentNullException("wrappedAttribute");
_wrappedAttribute = wrappedAttribute;
}
/// <summary>
/// Creates a duplicate of this attribute.
/// </summary>
/// <returns>The cloned attribute.</returns>
public HtmlAttribute Clone()
{
HtmlAttribute att = new HtmlAttribute(_ownerdocument);
att.Name = Name;
att.Value = Value;
return att;
}
private static void MakeAbsolute(HtmlAttribute attr, Uri baseUrl) {
var url = attr.Value;
if (!url.StartsWith("http:", StringComparison.OrdinalIgnoreCase)
&& !url.StartsWith("https:", StringComparison.OrdinalIgnoreCase)
&& Uri.IsWellFormedUriString(url, UriKind.Relative)) {
attr.Value = new Uri(baseUrl, url).ToString();
}
}
private string ExtractModel(string make, HtmlAttribute htmlAttribute)
{
if (htmlAttribute != null)
{
var stringToReplace = string.Format(@"/usedcar/{0}__", make);
return htmlAttribute.Value.Replace(stringToReplace, string.Empty).Replace(".html", string.Empty);
}
return string.Empty;
}
/// <summary>
///
/// </summary>
/// <param name="source"></param>
/// <param name="advanced"></param>
/// <returns></returns>
private static void getNameValueByElementTypeVoid(
HtmlAgilityPack.HtmlDocument source,
SalesForce salesForce
)
{
var document = source.DocumentNode;
HtmlAgilityPack.HtmlNode editNode = document.QuerySelector(salesForce.elementOne);
HtmlAgilityPack.HtmlAttribute attribute = editNode.Attributes["href"];
salesForce.endpointEditSearchFieldTemplate = attribute.Value;
}
/// <summary>
/// Inserts the specified attribute as the first node in the collection.
/// </summary>
/// <param name="newAttribute">The attribute to insert. May not be null.</param>
/// <returns>The prepended attribute.</returns>
public HtmlAttribute Prepend(HtmlAttribute newAttribute)
{
if (newAttribute == null)
{
throw new ArgumentNullException("newAttribute");
}
_hashitems[newAttribute.Name] = newAttribute;
newAttribute._ownernode = _ownernode;
_items.Insert(0, newAttribute);
_ownernode._innerchanged = true;
_ownernode._outerchanged = true;
return newAttribute;
}
/// <summary>
/// This removes the vulnerable keywords and make values safe by html encoding and html character escaping.
/// </summary>
/// <param name="attribute">Attribute that contain values that need to check and clean.</param>
private void CleanAttributeValues(HapHtmlAttribute attribute)
{
if (CleanAttributes)
{
attribute.Value = HttpUtility.HtmlEncode(attribute.Value);
attribute.Value = Regex.Replace(attribute.Value, @"\s*j\s*a\s*v\s*a\s*s\s*c\s*r\s*i\s*p\s*t\s*", "", RegexOptions.IgnoreCase);
attribute.Value = Regex.Replace(attribute.Value, @"\s*s\s*c\s*r\s*i\s*p\s*t\s*", "", RegexOptions.IgnoreCase);
if (attribute.Name.ToLower() == "style")
{
attribute.Value = Regex.Replace(attribute.Value, @"\s*e\s*x\s*p\s*r\s*e\s*s\s*s\s*i\s*o\s*n\s*", "", RegexOptions.IgnoreCase);
attribute.Value = Regex.Replace(attribute.Value, @"\s*b\s*e\s*h\s*a\s*v\s*i\s*o\s*r\s*", "", RegexOptions.IgnoreCase);
}
if (attribute.Name.ToLower() == "href" || attribute.Name.ToLower() == "src")
{
attribute.Value = Regex.Replace(attribute.Value, @"\s*m\s*o\s*c\s*h\s*a\s*", "", RegexOptions.IgnoreCase);
}
}
// HtmlEntity Escape
if (EncodeHtmlEntities)
{
// Ensure no double encoding goes on - reverse the ones done by the CreoleParser
string value = attribute.Value;
value = value.Replace("2", "\"");
value = value.Replace("<", "<");
value = value.Replace(">", ">");
value = value.Replace("&", "&");
value = value.Replace("'", "'");
attribute.Value = value;
StringBuilder sbAttributeValue = new StringBuilder();
foreach (char c in attribute.Value.ToCharArray())
{
sbAttributeValue.Append(EncodeCharacterToHtmlEntityEscape(c));
}
attribute.Value = sbAttributeValue.ToString();
}
}
private string RetrieveImageUrl(HtmlAttribute htmlAttribute)
{
try
{
var url = htmlAttribute.Value;
var htmlDocument = _htmlWeb.Load(url);
if (htmlDocument.DocumentNode != null)
{
var node = htmlDocument.DocumentNode.SelectSingleNode("//*[@id = 'frontCover']");
if (node != null)
{
return string.Format("{0}{1}", BaseUrl, node.Attributes["src"].Value);
}
}
return null;
}
catch (Exception)
{
return null;
}
}
private string _value(HtmlAttribute x)
{
if (x == null)
return "";
return x.Value;
}
public static void WithValue(this HtmlAttribute a, string expectedValue)
{
Assert.AreEqual(expectedValue, a.Value);
}
private void Parse()
{
int lastquote = 0;
if (OptionComputeChecksum)
{
_crc32 = new Crc32();
}
Lastnodes = new Dictionary<string, HtmlNode>();
_c = 0;
_fullcomment = false;
_parseerrors = new List<HtmlParseError>();
_line = 1;
_lineposition = 1;
_maxlineposition = 1;
_state = ParseState.Text;
_oldstate = _state;
_documentnode._innerlength = Text.Length;
_documentnode._outerlength = Text.Length;
_remainderOffset = Text.Length;
_lastparentnode = _documentnode;
_currentnode = CreateNode(HtmlNodeType.Text, 0);
_currentattribute = null;
_index = 0;
PushNodeStart(HtmlNodeType.Text, 0);
while (_index < Text.Length)
{
_c = Text[_index];
IncrementPosition();
switch (_state)
{
case ParseState.Text:
if (NewCheck())
continue;
break;
case ParseState.WhichTag:
if (NewCheck())
continue;
if (_c == '/')
{
PushNodeNameStart(false, _index);
}
else
{
PushNodeNameStart(true, _index - 1);
DecrementPosition();
}
_state = ParseState.Tag;
break;
case ParseState.Tag:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
{
PushNodeNameEnd(_index - 1);
if (_state != ParseState.Tag)
continue;
_state = ParseState.BetweenAttributes;
continue;
}
if (_c == '/')
{
PushNodeNameEnd(_index - 1);
if (_state != ParseState.Tag)
continue;
_state = ParseState.EmptyTag;
continue;
}
if (_c == '>')
{
PushNodeNameEnd(_index - 1);
if (_state != ParseState.Tag)
continue;
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.Tag)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
}
break;
case ParseState.BetweenAttributes:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
continue;
if ((_c == '/') || (_c == '?'))
{
_state = ParseState.EmptyTag;
continue;
}
if (_c == '>')
{
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.BetweenAttributes)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
PushAttributeNameStart(_index - 1);
_state = ParseState.AttributeName;
break;
case ParseState.EmptyTag:
if (NewCheck())
continue;
if (_c == '>')
{
if (!PushNodeEnd(_index, true))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.EmptyTag)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
_state = ParseState.BetweenAttributes;
break;
case ParseState.AttributeName:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
{
PushAttributeNameEnd(_index - 1);
_state = ParseState.AttributeBeforeEquals;
continue;
}
if (_c == '=')
{
PushAttributeNameEnd(_index - 1);
_state = ParseState.AttributeAfterEquals;
continue;
}
if (_c == '>')
{
PushAttributeNameEnd(_index - 1);
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.AttributeName)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
break;
case ParseState.AttributeBeforeEquals:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
continue;
if (_c == '>')
{
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.AttributeBeforeEquals)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
if (_c == '=')
{
_state = ParseState.AttributeAfterEquals;
continue;
}
// no equals, no whitespace, it's a new attrribute starting
_state = ParseState.BetweenAttributes;
DecrementPosition();
break;
case ParseState.AttributeAfterEquals:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
continue;
if ((_c == '\'') || (_c == '"'))
{
_state = ParseState.QuotedAttributeValue;
PushAttributeValueStart(_index, _c);
lastquote = _c;
continue;
}
if (_c == '>')
{
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.AttributeAfterEquals)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
PushAttributeValueStart(_index - 1);
_state = ParseState.AttributeValue;
break;
case ParseState.AttributeValue:
if (NewCheck())
continue;
if (IsWhiteSpace(_c))
{
PushAttributeValueEnd(_index - 1);
_state = ParseState.BetweenAttributes;
continue;
}
if (_c == '>')
{
PushAttributeValueEnd(_index - 1);
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
if (_state != ParseState.AttributeValue)
continue;
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
break;
case ParseState.QuotedAttributeValue:
if (_c == lastquote)
{
PushAttributeValueEnd(_index - 1);
_state = ParseState.BetweenAttributes;
continue;
}
if (_c == '<')
{
if (_index < Text.Length)
{
if (Text[_index] == '%')
{
_oldstate = _state;
_state = ParseState.ServerSideCode;
continue;
}
}
}
break;
case ParseState.Comment:
if (_c == '>')
{
if (_fullcomment)
{
if ((Text[_index - 2] != '-') ||
(Text[_index - 3] != '-'))
{
continue;
}
}
if (!PushNodeEnd(_index, false))
{
// stop parsing
_index = Text.Length;
break;
}
_state = ParseState.Text;
PushNodeStart(HtmlNodeType.Text, _index);
continue;
}
break;
case ParseState.ServerSideCode:
if (_c == '%')
{
if (_index < Text.Length)
{
if (Text[_index] == '>')
{
switch (_oldstate)
{
case ParseState.AttributeAfterEquals:
_state = ParseState.AttributeValue;
break;
case ParseState.BetweenAttributes:
PushAttributeNameEnd(_index + 1);
_state = ParseState.BetweenAttributes;
break;
default:
_state = _oldstate;
break;
}
IncrementPosition();
}
}
}
break;
case ParseState.PcData:
// look for </tag + 1 char
// check buffer end
if ((_currentnode._namelength + 3) <= (Text.Length - (_index - 1)))
{
if (string.Compare(Text.Substring(_index - 1, _currentnode._namelength + 2),
"</" + _currentnode.Name, StringComparison.OrdinalIgnoreCase) == 0)
{
int c = Text[_index - 1 + 2 + _currentnode.Name.Length];
if ((c == '>') || (IsWhiteSpace(c)))
{
// add the script as a text node
HtmlNode script = CreateNode(HtmlNodeType.Text,
_currentnode._outerstartindex +
_currentnode._outerlength);
script._outerlength = _index - 1 - script._outerstartindex;
_currentnode.AppendChild(script);
PushNodeStart(HtmlNodeType.Element, _index - 1);
PushNodeNameStart(false, _index - 1 + 2);
_state = ParseState.Tag;
IncrementPosition();
}
}
}
break;
}
}
// finish the current work
if (_currentnode._namestartindex > 0)
{
PushNodeNameEnd(_index);
}
PushNodeEnd(_index, false);
// we don't need this anymore
Lastnodes.Clear();
}
private string htmlValue(HtmlAttribute node)
{
if (node == null)
return String.Empty;
else
return node.Value;
}
internal HtmlAttributeAdapter(AP.HtmlAttribute attribute)
{
_attribute = attribute;
}
private void PushAttributeNameStart(int index)
{
this._currentattribute = this.CreateAttribute();
this._currentattribute._namestartindex = index;
this._currentattribute.Line = this._line;
this._currentattribute._lineposition = this._lineposition;
this._currentattribute._streamposition = index;
}
internal void WriteAttribute(TextWriter outText, HtmlAttribute att)
{
string name;
if (_ownerdocument.OptionOutputAsXml)
{
if (_ownerdocument.OptionOutputUpperCase)
{
name = att.XmlName.ToUpper();
}
else
{
name = att.XmlName;
}
outText.Write(" " + name + "=\"" + HtmlDocument.HtmlEncode(att.XmlValue) + "\"");
}
else
{
if (_ownerdocument.OptionOutputUpperCase)
{
name = att.Name.ToUpper();
}
else
{
name = att.Name;
}
if (att.Name.Length >= 4)
{
if ((att.Name[0] == '<') && (att.Name[1] == '%') &&
(att.Name[att.Name.Length-1] == '>') && (att.Name[att.Name.Length-2] == '%'))
{
outText.Write(" " + name);
return;
}
}
if (_ownerdocument.OptionOutputOptimizeAttributeValues)
{
if (att.Value.IndexOfAny(new Char[]{(char)10, (char)13, (char)9, ' '}) < 0)
{
outText.Write(" " + name + "=" + att.Value);
}
else
{
outText.Write(" " + name + "=\"" + att.Value + "\"");
}
}
else
{
outText.Write(" " + name + "=\"" + att.Value + "\"");
}
}
}
/// <summary>
/// Removes a given attribute from the list.
/// </summary>
/// <param name="attribute">The attribute to remove. May not be null.</param>
public void Remove(HtmlAttribute attribute)
{
if (attribute == null)
{
throw new ArgumentNullException("attribute");
}
int index = GetAttributeIndex(attribute);
if (index == -1)
{
throw new IndexOutOfRangeException();
}
RemoveAt(index);
}
private void ProcessAttrubute(HtmlAttribute imgSrcAttribute)
{
var newSrcValue = ProcessUrl(imgSrcAttribute.Value);
imgSrcAttribute.Value = newSrcValue;
}
/// <summary>
/// This removes the vulnerable keywords and make values safe by html encoding and html character escaping.
/// </summary>
/// <param name="attribute">Attribute that contain values that need to check and clean.</param>
private void CleanAttributeValues(HapHtmlAttribute attribute)
{
if (CleanAttributes)
{
attribute.Value = HttpUtility.HtmlEncode(attribute.Value);
attribute.Value = Regex.Replace(attribute.Value, @"\s*j\s*a\s*v\s*a\s*s\s*c\s*r\s*i\s*p\s*t\s*", "", RegexOptions.IgnoreCase);
attribute.Value = Regex.Replace(attribute.Value, @"\s*s\s*c\s*r\s*i\s*p\s*t\s*", "", RegexOptions.IgnoreCase);
if (attribute.Name.ToLower() == "style")
{
attribute.Value = Regex.Replace(attribute.Value, @"\s*e\s*x\s*p\s*r\s*e\s*s\s*s\s*i\s*o\s*n\s*", "", RegexOptions.IgnoreCase);
attribute.Value = Regex.Replace(attribute.Value, @"\s*b\s*e\s*h\s*a\s*v\s*i\s*o\s*r\s*", "", RegexOptions.IgnoreCase);
}
if (attribute.Name.ToLower() == "href" || attribute.Name.ToLower() == "src")
{
attribute.Value = Regex.Replace(attribute.Value, @"\s*m\s*o\s*c\s*h\s*a\s*", "", RegexOptions.IgnoreCase);
}
}
// HtmlEntity Escape
if (EncodeHtmlEntities)
{
// Ensure no double encoding goes on - reverse the ones done by the CreoleParser
string value = attribute.Value;
value = value.Replace("2", "\"");
value = value.Replace("<", "<");
value = value.Replace(">", ">");
value = value.Replace("&", "&");
value = value.Replace("'", "'");
attribute.Value = value;
StringBuilder sbAttributeValue = new StringBuilder();
foreach (char c in attribute.Value.ToCharArray())
{
sbAttributeValue.Append(EncodeCharacterToHtmlEntityEscape(c));
}
attribute.Value = sbAttributeValue.ToString();
}
}
/// <summary>
/// Creates and inserts a new attribute as the last attribute in the collection.
/// </summary>
/// <param name="name">The name of the attribute to insert.</param>
/// <param name="value">The value of the attribute to insert.</param>
/// <returns>The appended attribute.</returns>
public HtmlAttribute Append(string name, string value)
{
HtmlAttribute att = _ownernode._ownerdocument.CreateAttribute(name, value);
return(Append(att));
}
public HtmlAttributeFacade(HtmlAttribute attrib)
{
_attrib = attrib;
}
/// <summary>
/// This removes the vulnerable keywords and make values safe by html encoding and html character escaping.
/// </summary>
/// <param name="attribute">Attribute that contain values that need to check and clean.</param>
private static void CleanAttributeValues(HtmlAttribute attribute)
{
attribute.Value = HttpUtility.HtmlEncode(attribute.Value);
attribute.Value = Regex.Replace(attribute.Value, @"\s*j\s*a\s*v\s*a\s*s\s*c\s*r\s*i\s*p\s*t\s*", "", RegexOptions.IgnoreCase);
attribute.Value = Regex.Replace(attribute.Value, @"\s*s\s*c\s*r\s*i\s*p\s*t\s*", "", RegexOptions.IgnoreCase);
if (attribute.Name.ToLower() == "style")
{
attribute.Value = Regex.Replace(attribute.Value, @"\s*e\s*x\s*p\s*r\s*e\s*s\s*s\s*i\s*o\s*n\s*", "", RegexOptions.IgnoreCase);
attribute.Value = Regex.Replace(attribute.Value, @"\s*b\s*e\s*h\s*a\s*v\s*i\s*o\s*r\s*", "", RegexOptions.IgnoreCase);
}
if (attribute.Name.ToLower() == "href" || attribute.Name.ToLower() == "src")
{
//if (!attribute.Value.StartsWith("http://") || attribute.Value.StartsWith("/"))
// attribute.Value = "";
attribute.Value = Regex.Replace(attribute.Value, @"\s*m\s*o\s*c\s*h\s*a\s*", "", RegexOptions.IgnoreCase);
}
// HtmlEntity Escape
StringBuilder sbAttriuteValue = new StringBuilder();
foreach (char c in attribute.Value.ToCharArray())
{
sbAttriuteValue.Append(EncodeCharacterToHtmlEntityEscape(c));
}
attribute.Value = sbAttriuteValue.ToString();
}
private void ParsePrefixAttribute(RdfAParserContext context, RdfAEvaluationContext evalContext, HtmlAttribute attr, String baseUri, Dictionary<string,Uri> hiddenPrefixes, List<String> inScopePrefixes)
{
//Do nothing if the @prefix attribute is empty
if (attr.Value.Equals(String.Empty)) return;
StringReader reader = new StringReader(attr.Value);
char next;
bool canExit = false;
do
{
StringBuilder prefixData = new StringBuilder();
StringBuilder uriData = new StringBuilder();
//Grab a Prefix - characters up to the next colon
next = (char)reader.Peek();
while (next != ':')
{
//Add the Character and discard it
prefixData.Append(next);
reader.Read();
if (reader.Peek() == -1)
{
this.OnWarning("Aborted parsing a prefix attribute since failed to find a prefix of the form prefix: from the following content: " + prefixData.ToString());
return;
}
else
{
next = (char)reader.Peek();
}
}
//Discard the colon
reader.Read();
//Discard the whitespace
next = (char)reader.Peek();
while (Char.IsWhiteSpace(next))
{
reader.Read();
if (reader.Peek() == -1)
{
this.OnWarning("Aborted parsing a prefix attribute since reached the end of the attribute without finding a URI to go with the prefix '" + prefixData.ToString() + ":'");
return;
}
else
{
next = (char)reader.Peek();
}
}
//Grab the URI - characters up to the next whitespace or end of string
next = (char)reader.Peek();
while (!Char.IsWhiteSpace(next))
{
uriData.Append(next);
reader.Read();
if (reader.Peek() == -1)
{
//End of string so will exit after this
canExit = true;
break;
}
else
{
next = (char)reader.Peek();
}
}
//Now resolve the URI and apply it
String uri = Tools.ResolveUri(uriData.ToString(), baseUri);
if (!(uri.EndsWith("/") || uri.EndsWith("#"))) uri += "#";
String prefix = prefixData.ToString();
if (evalContext.NamespaceMap.HasNamespace(prefix))
{
if (hiddenPrefixes == null) hiddenPrefixes = new Dictionary<string, Uri>();
hiddenPrefixes.Add(prefix, new Uri(uri));
}
evalContext.NamespaceMap.AddNamespace(prefix, new Uri(uri));
inScopePrefixes.Add(prefix);
} while (!canExit);
}
private void Parse()
{
int num = 0;
if (this.OptionComputeChecksum)
this._crc32 = new Crc32();
this.Lastnodes = new Dictionary<string, HtmlNode>();
this._c = 0;
this._fullcomment = false;
this._parseerrors = new List<HtmlParseError>();
this._line = 1;
this._lineposition = 1;
this._maxlineposition = 1;
this._state = HtmlDocument.ParseState.Text;
this._oldstate = this._state;
this._documentnode._innerlength = this.Text.Length;
this._documentnode._outerlength = this.Text.Length;
this._remainderOffset = this.Text.Length;
this._lastparentnode = this._documentnode;
this._currentnode = this.CreateNode(HtmlNodeType.Text, 0);
this._currentattribute = (HtmlAttribute) null;
this._index = 0;
this.PushNodeStart(HtmlNodeType.Text, 0);
while (this._index < this.Text.Length)
{
this._c = (int) this.Text[this._index];
this.IncrementPosition();
switch (this._state)
{
case HtmlDocument.ParseState.Text:
if (!this.NewCheck())
continue;
continue;
case HtmlDocument.ParseState.WhichTag:
if (!this.NewCheck())
{
if (this._c == 47)
{
this.PushNodeNameStart(false, this._index);
}
else
{
this.PushNodeNameStart(true, this._index - 1);
this.DecrementPosition();
}
this._state = HtmlDocument.ParseState.Tag;
continue;
}
continue;
case HtmlDocument.ParseState.Tag:
if (!this.NewCheck())
{
if (HtmlDocument.IsWhiteSpace(this._c))
{
this.PushNodeNameEnd(this._index - 1);
if (this._state == HtmlDocument.ParseState.Tag)
{
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
continue;
}
if (this._c == 47)
{
this.PushNodeNameEnd(this._index - 1);
if (this._state == HtmlDocument.ParseState.Tag)
{
this._state = HtmlDocument.ParseState.EmptyTag;
continue;
}
continue;
}
if (this._c == 62)
{
this.PushNodeNameEnd(this._index - 1);
if (this._state == HtmlDocument.ParseState.Tag)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.Tag)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
continue;
}
continue;
}
continue;
case HtmlDocument.ParseState.BetweenAttributes:
if (!this.NewCheck() && !HtmlDocument.IsWhiteSpace(this._c))
{
if (this._c == 47 || this._c == 63)
{
this._state = HtmlDocument.ParseState.EmptyTag;
continue;
}
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.BetweenAttributes)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
this.PushAttributeNameStart(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeName;
continue;
}
continue;
case HtmlDocument.ParseState.EmptyTag:
if (!this.NewCheck())
{
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, true))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.EmptyTag)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
continue;
case HtmlDocument.ParseState.AttributeName:
if (!this.NewCheck())
{
if (HtmlDocument.IsWhiteSpace(this._c))
{
this.PushAttributeNameEnd(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeBeforeEquals;
continue;
}
if (this._c == 61)
{
this.PushAttributeNameEnd(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeAfterEquals;
continue;
}
if (this._c == 62)
{
this.PushAttributeNameEnd(this._index - 1);
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeName)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
continue;
}
continue;
case HtmlDocument.ParseState.AttributeBeforeEquals:
if (!this.NewCheck() && !HtmlDocument.IsWhiteSpace(this._c))
{
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeBeforeEquals)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
if (this._c == 61)
{
this._state = HtmlDocument.ParseState.AttributeAfterEquals;
continue;
}
this._state = HtmlDocument.ParseState.BetweenAttributes;
this.DecrementPosition();
continue;
}
continue;
case HtmlDocument.ParseState.AttributeAfterEquals:
if (!this.NewCheck() && !HtmlDocument.IsWhiteSpace(this._c))
{
if (this._c == 39 || this._c == 34)
{
this._state = HtmlDocument.ParseState.QuotedAttributeValue;
this.PushAttributeValueStart(this._index, this._c);
num = this._c;
continue;
}
if (this._c == 62)
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeAfterEquals)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
this.PushAttributeValueStart(this._index - 1);
this._state = HtmlDocument.ParseState.AttributeValue;
continue;
}
continue;
case HtmlDocument.ParseState.AttributeValue:
if (!this.NewCheck())
{
if (HtmlDocument.IsWhiteSpace(this._c))
{
this.PushAttributeValueEnd(this._index - 1);
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
if (this._c == 62)
{
this.PushAttributeValueEnd(this._index - 1);
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
if (this._state == HtmlDocument.ParseState.AttributeValue)
{
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
}
continue;
}
continue;
case HtmlDocument.ParseState.Comment:
if (this._c == 62 && (!this._fullcomment || (int) this.Text[this._index - 2] == 45 && (int) this.Text[this._index - 3] == 45))
{
if (!this.PushNodeEnd(this._index, false))
{
this._index = this.Text.Length;
continue;
}
this._state = HtmlDocument.ParseState.Text;
this.PushNodeStart(HtmlNodeType.Text, this._index);
continue;
}
continue;
case HtmlDocument.ParseState.QuotedAttributeValue:
if (this._c == num)
{
this.PushAttributeValueEnd(this._index - 1);
this._state = HtmlDocument.ParseState.BetweenAttributes;
continue;
}
if (this._c == 60 && this._index < this.Text.Length && (int) this.Text[this._index] == 37)
{
this._oldstate = this._state;
this._state = HtmlDocument.ParseState.ServerSideCode;
continue;
}
continue;
case HtmlDocument.ParseState.ServerSideCode:
if (this._c == 37 && this._index < this.Text.Length && (int) this.Text[this._index] == 62)
{
switch (this._oldstate)
{
case HtmlDocument.ParseState.BetweenAttributes:
this.PushAttributeNameEnd(this._index + 1);
this._state = HtmlDocument.ParseState.BetweenAttributes;
break;
case HtmlDocument.ParseState.AttributeAfterEquals:
this._state = HtmlDocument.ParseState.AttributeValue;
break;
default:
this._state = this._oldstate;
break;
}
this.IncrementPosition();
continue;
}
continue;
case HtmlDocument.ParseState.PcData:
if (this._currentnode._namelength + 3 <= this.Text.Length - (this._index - 1) && string.Compare(this.Text.Substring(this._index - 1, this._currentnode._namelength + 2), "</" + this._currentnode.Name, StringComparison.OrdinalIgnoreCase) == 0)
{
int c = (int) this.Text[this._index - 1 + 2 + this._currentnode.Name.Length];
if (c == 62 || HtmlDocument.IsWhiteSpace(c))
{
HtmlNode node = this.CreateNode(HtmlNodeType.Text, this._currentnode._outerstartindex + this._currentnode._outerlength);
node._outerlength = this._index - 1 - node._outerstartindex;
this._currentnode.AppendChild(node);
this.PushNodeStart(HtmlNodeType.Element, this._index - 1);
this.PushNodeNameStart(false, this._index - 1 + 2);
this._state = HtmlDocument.ParseState.Tag;
this.IncrementPosition();
continue;
}
continue;
}
continue;
default:
continue;
}
}
if (this._currentnode._namestartindex > 0)
this.PushNodeNameEnd(this._index);
this.PushNodeEnd(this._index, false);
this.Lastnodes.Clear();
}
private void ParseVocabAttribute(RdfAParserContext context, RdfAEvaluationContext evalContext, HtmlAttribute attr)
{
if (attr.Value.Equals(String.Empty))
{
//Reset Local Vocabulary
evalContext.LocalVocabulary = new TermMappings(context.DefaultVocabulary);
}
else
{
evalContext.LocalVocabulary.VocabularyUri = attr.Value;
}
}
private bool ExtendLastAttributeValue(int index, bool close)
{
System.Diagnostics.Debug.Assert(_currentnode != null && _currentnode.Attributes.Count > 0);
var lastattribute = _currentattribute;
// find an attribute with a value to be extended
var attrs = _currentnode.Attributes;
int candidate = -1;
for (int i = attrs.Count - 1; i >= 0; i--)
if (attrs[i]._valuelength > 0 || attrs[i]._valuestartindex > (attrs[i]._namestartindex + attrs[i]._namelength)) // an attribute with a value specified
{
candidate = i;
break;
}
if (candidate < 0)
return false; // no candidate to be extended
// set new last attribute
_currentattribute = attrs[candidate];
while (attrs.Count > candidate + 1)
attrs.RemoveAt(attrs.Count - 1);
//
if (close)
{
PushAttributeValueEnd(index);
_state = ParseState.BetweenAttributes;
}
else
{
_currentattribute._valuelength = 0;
_state = ParseState.QuotedAttributeValue;
}
return true;
}
/// <summary>
/// Construct a HTML element attribute wrapper object.
/// </summary>
/// <param name="attribute"></param>
protected HtmlAttributeBase(HtmlAgilityPack.HtmlAttribute attribute)
{
this.attribute = attribute;
}
internal int GetAttributeIndex(HtmlAttribute attribute)
{
if (attribute == null)
{
throw new ArgumentNullException("attribute");
}
for (int i = 0; i < _items.Count; i++)
{
if (((HtmlAttribute)_items[i]) == attribute)
return i;
}
return -1;
}
private bool IsStaticResource(HtmlNode node, HtmlAttribute attr)
{
var nodeName = node.Name.ToLower();
if (nodeName == "link" || nodeName == "script" || nodeName == "img")
{
return true;
}
var attValue = (attr.Value ?? "").ToLower();
if (attr.Value.StartsWith("#") || attr.Value.StartsWith("javascript:"))
{
return true;
}
if (attValue.EndsWith(".js") || attValue.EndsWith(".css") || attValue.EndsWith(".png") || attValue.EndsWith(".jpg")
|| attValue.EndsWith(".jpeg") || attValue.EndsWith(".ico") || attValue.EndsWith(".gif"))
{
return true;
}
return false;
}
public static IHtmlAttribute AsAttribute(this AP.HtmlAttribute attribute)
{
return(new HtmlAttributeAdapter(attribute));
}
private bool IsStaticResource(HtmlNode node, HtmlAttribute attr)
{
var nodeName = node.Name.ToLower();
if (nodeName == "link" || nodeName == "script" || nodeName == "img")
{
return true;
}
var extension = Path.GetExtension(attr.Value).ToLower();
if (extension == ".js" || extension == ".css" || extension == ".png" || extension == ".jpg" || extension == ".jpeg" || extension == ".ico" || extension == ".gif")
{
return true;
}
return false;
}
private bool ParseProfileAttribute(RdfAParserContext context, RdfAEvaluationContext evalContext, HtmlAttribute attr)
{
String[] profiles;
if (attr.Value.Contains(" "))
{
profiles = attr.Value.Split(' ');
}
else
{
profiles = new String[] { attr.Value };
}
String prefixQuery = "PREFIX rdfa: <" + RdfANamespace + "> SELECT SAMPLE(?prefix) AS ?NamespacePrefix SAMPLE(?uri) AS ?NamespaceURI WHERE { ?s rdfa:prefix ?prefix ; rdfa:uri ?uri } GROUP BY ?s HAVING (COUNT(?prefix) = 1 && COUNT(?uri) = 1)";
String termQuery = "PREFIX rdfa: <" + RdfANamespace + "> SELECT SAMPLE(?term) AS ?Term SAMPLE(?uri) AS ?URI WHERE {?s rdfa:term ?term ; rdfa:uri ?uri } GROUP BY ?s HAVING (COUNT(?term) = 1 && COUNT(?uri) = 1)";
foreach (String profile in profiles)
{
try
{
Graph g = new Graph();
if (profile.Equals(XHtmlVocabNamespace) || profile.Equals(XHtmlVocabNamespace.Substring(0, XHtmlVocabNamespace.Length-1)))
{
//XHTML Vocabulary is a fixed vocabulary
evalContext.LocalVocabulary.Merge(new XHtmlRdfAVocabulary());
}
else
{
try
{
UriLoader.Load(g, new Uri(profile));
}
catch
{
//If we fail then we return false which indicates that the DOM subtree is ignored
this.OnWarning("Unable to retrieve a Profile document which the library could parse from the URI '" + profile + "'");
return false;
}
//Namespace Mappings
Object results = g.ExecuteQuery(prefixQuery);
if (results is SparqlResultSet)
{
SparqlResultSet rset = (SparqlResultSet)results;
foreach (SparqlResult r in rset.Results)
{
INode prefixNode = r["NamespacePrefix"];
INode nsNode = r["NamespaceURI"];
if (prefixNode.NodeType == NodeType.Literal && nsNode.NodeType == NodeType.Literal)
{
String prefix = ((ILiteralNode)prefixNode).Value.ToLower();
String ns = ((ILiteralNode)nsNode).Value;
evalContext.LocalVocabulary.AddNamespace(prefix, ns);
}
}
}
//Term Mappings
results = g.ExecuteQuery(termQuery);
if (results is SparqlResultSet)
{
SparqlResultSet rset = (SparqlResultSet)results;
foreach (SparqlResult r in rset.Results)
{
INode termNode = r["Term"];
INode uriNode = r["URI"];
if (termNode.NodeType == NodeType.Literal && uriNode.NodeType == NodeType.Literal)
{
String term = ((ILiteralNode)termNode).Value;
String uri = ((ILiteralNode)uriNode).Value;
evalContext.LocalVocabulary.AddTerm(term, uri);
}
}
}
}
}
catch
{
//Ignore errors and continue processing
this.OnWarning("Ignoring the value '" + profile + "' since this is not a valid URI or a profile document was not successfully retrieved and parsed from this URI");
return false;
}
}
return true;
}
private static string ExtractMake(HtmlAttribute htmlAttribute)
{
if (htmlAttribute != null)
return htmlAttribute.Value.Replace(@"/usedcar/", string.Empty).Replace(@"/index.html", string.Empty).ToUpper();
return string.Empty;
}
internal void WriteAttribute(TextWriter outText, HtmlAttribute att)
{
string name;
string quote = att.QuoteType == AttributeValueQuote.DoubleQuote ? "\"" : "'";
if (_ownerdocument.OptionOutputAsXml)
{
if (_ownerdocument.OptionOutputOriginalCase)
name = att.OriginalName;
else
name = _ownerdocument.OptionOutputUpperCase ? att.XmlName.ToUpper() : att.XmlName;
if (!string.IsNullOrEmpty(name))
{
char firstChar = name[0];
if (firstChar == '-' || (firstChar >= '0' && firstChar <= '9')) // an invalid XML attribute name
name = '_' + name; // // fix such attribute name by prepending '_'
}
outText.Write(" " + name + "=" + quote + HtmlDocument.HtmlEncode(att.XmlValue) + quote);
}
else
{
if (_ownerdocument.OptionOutputUpperCase)
{
name = att.Name.ToUpper();
}
else
{
name = att.Name;
}
if (att.Name.Length >= 4)
{
if ((att.Name[0] == '<') && (att.Name[1] == '%') &&
(att.Name[att.Name.Length - 1] == '>') && (att.Name[att.Name.Length - 2] == '%'))
{
outText.Write(" " + name);
return;
}
}
if (_ownerdocument.OptionOutputOptimizeAttributeValues)
{
if (att.Value.IndexOfAny(new Char[] {(char) 10, (char) 13, (char) 9, ' '}) < 0)
{
outText.Write(" " + name + "=" + att.Value);
}
else
{
outText.Write(" " + name + "=" + quote + att.Value + quote);
}
}
else
{
outText.Write(" " + name + "=" + quote + att.Value + quote);
}
}
}
/// <summary>
/// Creates a duplicate of this attribute.
/// </summary>
/// <returns>The cloned attribute.</returns>
public HtmlAttribute Clone()
{
var att = new HtmlAttribute(_ownerdocument) {
Name = Name,
Value = Value
};
return att;
}
internal void WriteAttribute(TextWriter outText, HtmlAttribute att)
{
string name;
string quote = att.QuoteType == AttributeValueQuote.DoubleQuote ? "\"" : "'";
if (_ownerdocument.OptionOutputAsXml)
{
name = _ownerdocument.OptionOutputUpperCase ? att.XmlName.ToUpper() : att.XmlName;
if (_ownerdocument.OptionOutputOriginalCase)
name = att.OriginalName;
outText.Write(" " + name + "=" + quote + HtmlDocument.HtmlEncode(att.XmlValue) + quote);
}
else
{
name = _ownerdocument.OptionOutputUpperCase ? att.Name.ToUpper() : att.Name;
if (att.Name.Length >= 4)
{
if ((att.Name[0] == '<') && (att.Name[1] == '%') &&
(att.Name[att.Name.Length - 1] == '>') && (att.Name[att.Name.Length - 2] == '%'))
{
outText.Write(" " + name);
return;
}
}
if (_ownerdocument.OptionOutputOptimizeAttributeValues)
if (att.Value.IndexOfAny(new[] {(char) 10, (char) 13, (char) 9, ' '}) < 0)
outText.Write(" " + name + "=" + att.Value);
else
outText.Write(" " + name + "=" + quote + att.Value + quote);
else
outText.Write(" " + name + "=" + quote + att.Value + quote);
}
}
internal HtmlAttributeAdapter( AP.HtmlAttribute attribute )
{
_attribute = attribute;
}
private void PushAttributeNameStart(int index)
{
_currentattribute = CreateAttribute();
_currentattribute._namestartindex = index;
_currentattribute.Line = _line;
_currentattribute._lineposition = _lineposition;
_currentattribute._streamposition = index;
}
public static HtmlAttribute ValueShouldContain(this HtmlAttribute a, string expectedSubstring)
{
a.Value.ShouldContain(expectedSubstring);
return(a);
}
