/// <summary>
/// Builds a hash of the information about word counts found in the page passed to the parser.
/// </summary>
/// <param name="wordsOnPage"></param>
/// <returns>Dictionary<string, int></returns>
public Dictionary <string, int> GetTextStats(List <string> wordsOnPage)
{
//This is designed this way so you can add other metrics without changing the ui.
allWordsTally = AllWordsTally(wordsOnPage);
statsDictionary.Add("Total Word Count", wordsOnPage.Count);
statsDictionary.Add("Unique Word Count", allWordsTally.Count());
allWordsTally.Remove("");
int TopFreqCount = Int32.Parse(ConfigurationManager.AppSettings["Word-Frequency-Count"]);
int rank = 1;
foreach (var kvp in allWordsTally.OrderByDescending(m => m.Value).Take(TopFreqCount))
{
statsDictionary.Add(string.Format("Word Rank {0} - {1}", rank, kvp.Key), kvp.Value);
rank++;
}
updateProgressBarDelegate.Invoke(250);
return(statsDictionary);
}
/// <summary>
/// Downloads all of the imageUrls, and saves them to the destination folder
/// </summary>
/// <param name="destination"></param>
/// <param name="imageUrls"></param>
/// <param name="sitename"></param>
/// <returns></returns>
public List <string> DownloadWithFilePaths(string destination, List <string> imageUrls, string sitename)
{
List <string> localPaths = new List <string>();
int count = imageUrls.Count;
int updateincrement = 500 / count;
string localFileName = "";
using (WebClient client = new WebClient())
{
foreach (string url in imageUrls)
{
try
{
string tempUrl = url;
if (IsRelativePath(url))
{
tempUrl = GetSiteRoot(sitename) + url;
}
localFileName = Path.Combine(destination, GetImageName(url));
if (!File.Exists(localFileName))
{
client.DownloadFile(tempUrl, localFileName);
}
localPaths.Add(localFileName);
updateProgressBarDelegate.Invoke(updateincrement);
}
catch (Exception e)
{
log.Error(e.StackTrace);
}
}
}
return(localPaths);
}