/// <summary>
/// Method which returns a list of words for every links on the website
/// </summary>
/// <param name="Creator">TXTCreating object which will take the HTML and returns it to the HTMLTxtAnalyser</param>
/// <param name="URL">URL of the HTML to analyse</param>
/// <returns>List of words</returns>
public List <string> WordSearching(TXTCreating Creator, string URL)
{
//Variables
char chrOne = ' ';
char chrTwo = ' ';
char chrThree = ' ';
char chrFour = ' ';
char chrFive = ' ';
char chrSix = ' ';
char chrSeven = ' ';
bool takeChr = false;
bool scripted = false;
string word = "";
//Objects
List <string> lstWords = new List <string>();
//Taking HTML and cleaning it
string HTML = Creator.TakeHTML(URL);
if (HTML != null)
{
HTML = Regex.Replace(HTML, @"\n", " ");
HTML = Regex.Replace(HTML, @"\r", " ");
HTML = Regex.Replace(HTML, @"\t", " ");
//Loop to search for words
for (int x = 0; x < HTML.Length; ++x)
{
char currentChr = Convert.ToChar(HTML.Substring(x, 1));
chrOne = chrTwo;
chrTwo = chrThree;
chrThree = chrFour;
chrFour = chrFive;
chrFive = chrSix;
chrSix = chrSeven;
chrSeven = currentChr;
if (chrTwo == 's' && chrThree == 'c' && chrFour == 'r' && chrFive == 'i' && chrSix == 'p' && chrSeven == 't')
{
scripted = IsScript(chrOne);
}
if (currentChr == '>')
{
takeChr = true;
}
else if (currentChr == '<' && takeChr)
{
takeChr = false;
}
else if (takeChr && !scripted)
{
if (currentChr == ' ')
{
if (word != "")
{
word = word.ToLower();
lstWords.Add(word);
word = "";
}
}
else
{
word += currentChr;
}
}
}
}
else
{
lstWords = null;
}
return(lstWords);
}
/// <summary>
/// Method to search links
/// </summary>
/// <param name="Creator">TXTCreating object which will take the HTML and returns it to the HTMLLinkFinder</param>
/// <param name="URL">URL of the HTML to analyse</param>
/// <returns>Return a list of links</returns>
public List <string> Finder(TXTCreating Creator, string URL)
{
//Variables
char chrOne = ' ';
char chrTwo = ' ';
char chrThree = ' ';
char chrFour = ' ';
string Link = "";
bool takeCar = false;
bool isLink = false;
int chrRemaining = 0;
//Objects
List <string> lstLinks = new List <string>();
//Taking HTML
string HTML = Creator.TakeHTML(URL);
if (HTML != null)
{
//Loop to search for links
for (int x = 0; x < HTML.Length; ++x)
{
chrOne = chrTwo;
chrTwo = chrThree;
chrThree = chrFour;
chrFour = Convert.ToChar(HTML.Substring(x, 1));
if (chrFour == 'a' && chrThree == '<')
{
isLink = true;
}
else if (chrOne == '<' && chrTwo == '/' && chrThree == 'a' && chrFour == '>')
{
isLink = false;
}
if (isLink)
{
if (chrOne == 'h' && chrTwo == 'r' && chrThree == 'e' && chrFour == 'f')
{
takeCar = true;
chrRemaining = 2;
}
else if (chrFour == '"' && chrRemaining == 0 && takeCar)
{
takeCar = false;
lstLinks.Add(Link);
Link = "";
}
else if (takeCar)
{
if (chrRemaining != 0)
{
chrRemaining--;
}
else
{
Link += chrFour;
}
}
}
}
}
else
{
lstLinks = null;
}
return(lstLinks);
}