private int MarkupProcess(int startOffset, int cutOffIndex)
{
int count = 0;
int offset = startOffset;
int character = 0;
characterList.Clear();
PageData pageData = new PageData();
pageData.StartOffset = offset;
// If the markup was previously open, the markup tag should be attached to the front.
string tag = "";
foreach (TagData data in tagList)
{
tag += "<" + data.TagName + data.AttributeName + ">";
}
pageData.PreviousTag = tag;
bool isTag = false;
while ((character = stream.Read()) != -1)
{
offset++;
characterList.Add((char)character);
TagData tagData = new TagData();
isTag = false;
if (LESS_THAN == character) // '<'
{
isTag = IsTag(tagData, ref offset);
}
if (isTag)
{
if (tagData.IsEndTag)
{
int lastIndex = tagList.Count;
tagList.RemoveAt(lastIndex - 1);
}
else
{
tagList.Add(tagData);
}
}
else
{
count++;
}
if (count >= cutOffIndex)
{
break;
}
}
// If the markup was previously open, you should attach the label tag.
tag = "";
foreach (TagData data in tagList)
{
tag = "</" + data.TagName + ">" + tag;
}
pageData.EndTag = tag;
pageData.EndOffset = offset;
pageList.Add(pageData);
if (character == -1)
{
offset = -1;
}
return(offset);
}
private bool IsTag(TagData tag, ref int offset)
{
List <char> tagChaList = new List <char>();
bool isTag = false;
bool isQuotationOpen = false;
bool attributesFound = false;
tag.IsEndTag = false;
bool isPreviousLessThan = true;
bool isPreviousSlash = false;
int character;
tag.TagName = "";
tag.AttributeName = "";
// SkipWhiteSpace(ref offset);
while ((!isTag) && ((character = stream.Read()) != -1))
{
offset++;
characterList.Add((char)character);
if (!isQuotationOpen && (SLASH == character)) // '/'
{
if (isPreviousLessThan)
{
tag.IsEndTag = true;
}
else
{
// if the tag has a '/' it may be an end tag.
isPreviousSlash = true;
}
isPreviousLessThan = false;
// SkipWhiteSpace(ref offset);
}
else if (GREATER_THAN == character) // '>'
{
isTag = true;
if (isPreviousSlash)
{
tag.IsEndTag = true;
}
if (!attributesFound)
{
tag.TagName = new String(tagChaList.ToArray());
}
else
{
tag.AttributeName = new String(tagChaList.ToArray());
}
isPreviousSlash = false;
isPreviousLessThan = false;
}
else if (QUOTATION_MARK == character)
{
tagChaList.Add((char)character);
isQuotationOpen = !isQuotationOpen;
isPreviousSlash = false;
isPreviousLessThan = false;
}
else if (WHITE_SPACE >= character || EQUAL == character) // ' ', '='
{
// Let's save tag name.
if (!attributesFound)
{
tag.TagName = new String(tagChaList.ToArray());
tagChaList.Clear();
}
tagChaList.Add((char)character);
// If the tag contains white spaces then it may have attributes.
if (!isQuotationOpen)
{
attributesFound = true;
}
}
else
{
tagChaList.Add((char)character);
isPreviousSlash = false;
isPreviousLessThan = false;
}
}
return(isTag);
}
