/// <summary>
/// Init processor by passed options
/// </summary>
/// <param name="options"></param>
/// <returns></returns>
public bool Init(ReportOptions options)
{
stopWordsList = new List <string>()
{
"and", "or", "a", "the", "on", "in", "from", "to", "of", "is", "are", "there", "it", "by", "that", "but", "this"
};
currentOptions = options;
string sourceString = currentOptions.SourceString;
//1. Validate some passed options
if (String.IsNullOrWhiteSpace(sourceString))
{
isInited = false;
return(false);
}
sourceString = sourceString.Trim();
//2. Determine text
if (isUrl(sourceString))
{
this.currentUrl = sourceString;
string urlContent = getUrlContent(sourceString);
//Make a single line
urlContent = urlContent.Replace(Environment.NewLine, "").Replace("\n", " ").Replace("\t", " ");
//Checking for html markup
if (String.IsNullOrWhiteSpace(urlContent) || !isHtml(urlContent))
{
throw new Exception("Downloaded url content is empty or is not html markup or has wrong encoding!");
}
;
currentText = urlContent.ToLower().Trim();
}
else
{
//Source string is text
currentText = sourceString.ToLower().Trim().Replace(Environment.NewLine, " ").Replace("\n", "").Replace("\t", " ");
}
isInited = true;
return(isInited);
}
public SeoTextProcessor()
{
this.currentOptions = new ReportOptions();
this.currentText = string.Empty;
this.currentUrl = string.Empty;
}
