public void SafelyChangeParent(PageElement parent)
{
if (this.Parent != null)
{
this.Parent.Children.Remove(this);
}
this.Parent = parent;
}
public void SafelyAddChild(PageObject child)
{
PageElement tempPE = child as PageElement;
if (tempPE != null)
{
tempPE.SafelyChangeParent(this);
}
Children.Add(child);
}
public override PageObject Create(char?lastChar, char currentChar, char?nextChar)
{
if (this.IsPossibleStart(lastChar, currentChar, nextChar))
{
PageElement result = new PageElement(this.Page);
result.html.Append(currentChar);
//if (nextChar.HasValue && nextChar.Value != '/')
// result.OpeningTag = true;
return(result);
}
return(null);
}
string recursiveGetTextualObjects(PageElement currentObject, StringBuilder stringBuilder)
{
foreach (var child in currentObject.Children)
{
PageElement tempPE = child as PageElement;
if (tempPE != null)
{
if (tempPE.ElementName == "script")
{
continue;
}
recursiveGetTextualObjects(tempPE, stringBuilder);
}
else
{
stringBuilder.Append(child.html);
}
}
return(stringBuilder.ToString());
}
public override void AddCharacter(char?lastChar, char currentChar, char?nextChar)
{
html.Append(currentChar);
if (currentChar == '=')
{
nextIsValue = true;
}
if (IsEndingCharacter(lastChar, currentChar, nextChar))
{
IsCompleted = true;
string tempString = html.ToString();
// First check if it is a closing element, if not check if it is a self closed, if not it is a opening element
if (tempString[1] == '/')
{
TagType = PageElementTagType.Closing;
ElementName = tempString.Substring(2, tempString.Length - 3).ToLower();
}
else if (tempString[tempString.Length - 2] == '/')
{
TagType = PageElementTagType.SelfClosed;
ElementName = tempString.Substring(1, tempString.IndexOfAny(new char[] { ' ', '/' })).ToLower();
}
else
{
TagType = PageElementTagType.Opening;
ElementName = tempString.Substring(1, tempString.IndexOfAny(new char[] { ' ', '/', '>' }) - 1).ToLower();
}
if (TagType == PageElementTagType.Opening)
{
this.Page.ElementsStack.Push(this);
}
else if (TagType == PageElementTagType.Closing)
{
PageElement current = this.Page.ElementsStack.Pop();
while (current.ElementName != this.ElementName)
{
PageElement tempElement = current;
tempElement.TagType = PageElementTagType.Broken;
current = this.Page.ElementsStack.Pop();
current.AddChildren(tempElement.Children);
tempElement.Children.Clear();
}
current.Parent.SafelyAddChild(this);
}
if (hasLink && ElementName == "a")
{
bool nextLink = false;
foreach (var attribute in Attributes)
{
if (nextLink && attribute as ElementAttributeValue != null)
{
string tempLink = fixLink(attribute.html.ToString());
if (tempLink != null)
{
Page.PageLinks.Add(tempLink);
}
nextLink = false;
}
if (attribute.html.ToString().ToLower() == "href")
{
nextLink = true;
}
}
}
}
for (int j = possibles.Count - 1; j >= 0; j--)
{
possibles[j].AddCharacter(lastChar, currentChar, nextChar);
if (possibles[j].IsCompleted)
{
if (possibles[j].html.ToString().ToLower() == "href")
{
hasLink = true;
}
if (possibles[j] as ElementAttributeValue != null)
{
inValue = false;
nextIsValue = false;
}
Attributes.Add(possibles[j]);
possibles.Remove(possibles[j]);
}
}
if (!inValue)
{
possibles.AddRange(ObjectFactory(lastChar, currentChar, nextChar));
}
}
internal void Parse(string pageString)
{
// This goes one character at a time through the entire html string
// Its initial use is to determine text areas as either elements or element values
// It will also handle parenting these elements
Page.Children.Clear();
PageElement tempContainer = new PageElement(Page);
Page.ElementsStack.Push(tempContainer);
List <PageObject> currentObjects = new List <FoogleEngine.PageObject>();
//List<PageObject> completedObjects = new List<PageObject>();
char?lastChar = null;
char currentChar;
char?nextChar = null;
for (int i = 0; i < pageString.Length; i++)
{
currentChar = pageString[i];
if ((i + 1) == pageString.Length)
{
nextChar = null;
}
else
{
nextChar = pageString[i + 1];
}
for (int j = currentObjects.Count - 1; j >= 0; j--)
{
currentObjects[j].AddCharacter(lastChar, currentChar, nextChar);
if (currentObjects[j].IsCompleted)
{
if (currentObjects[j] as PageElement == null || ((PageElement)currentObjects[j]).TagType != PageElement.PageElementTagType.Closing)
{
tempContainer.SafelyAddChild(currentObjects[j]);
if (currentObjects[j] as PageElementValue != null)
{
if (tempContainer.ElementName != "script")
{
Page.SearchableText += currentObjects[j].html.ToString().ToLower();
}
if (tempContainer.ElementName == "title")
{
Page.Title += currentObjects[j].html.ToString();
}
}
}
tempContainer = Page.ElementsStack.Peek();
//Page.ElementsStack.Peek().Children.Add(currentObjects[j]);
// if currentParent is null set it to first page element
}
if (currentObjects[j].IsCompleted || !currentObjects[j].IsOK)
{
currentObjects.Remove(currentObjects[j]);
}
}
currentObjects.AddRange(ObjectFactory(lastChar, currentChar, nextChar));
lastChar = currentChar;
}
tempContainer.Parent = null;
tempContainer.TagType = PageElement.PageElementTagType.Broken;
for (int i = tempContainer.Children.Count - 1; i >= 0; i--)
{
Page.Children.Add(tempContainer.Children[i]);
PageElement tempPE = tempContainer.Children[i] as PageElement;
if (tempPE != null)
{
tempPE.Parent = null;
}
tempContainer.Children.Remove(tempContainer.Children[i]);
}
Page.Children.Add(tempContainer);
Page.Children.Reverse();
}
