public WebsiteStats GetAllData(string url)
{
try
{
var ImageList = new List <string>();
var WordFrequency = new Dictionary <string, int>();
//Utility class to get HTML document from http
var web = new HtmlWeb();
//Load() Method download the specified HTML document from Internet resource.
var doc = web.Load(url);
var rootNode = doc.DocumentNode;
var nodes = doc.DocumentNode.SelectNodes("//img");
foreach (var src in nodes)
{
if (src.Attributes["src"] != null)
{
ImageList.Add(src.Attributes["src"].Value);
}
}
ImageList = ImageList.Distinct().Where(s => !string.IsNullOrWhiteSpace(s)).ToList();
var text = string.Empty;
//fileter for words in the document which are not in the script or style tags and normalize the space
foreach (var node in doc.DocumentNode.SelectNodes("//*[not(self::script or self::style)]/text()[normalize-space()]"))
{
text += " " + node.InnerText;
}
//Remove carriage return, newline characters and empty strings
var wordsArray = Regex.Split(text.Replace("\r", "").Replace("\n", ""), @"\s+").Where(s => s != string.Empty);
var TotalWordsCount = 0;
if (wordsArray != null)
{
TotalWordsCount = wordsArray.Count();
foreach (var item in wordsArray)
{
if (WordFrequency.ContainsKey(item))
{
WordFrequency[item]++;
}
else
{
WordFrequency.Add(item, 1);
}
}
}
var WebPageStats = new WebsiteStats
{
Images = ImageList,
TotalWordsCount = TotalWordsCount,
WordFrequency = WordFrequency
};
return(WebPageStats);
}
catch (Exception ex)
{
// TODO; log error
return(null);
}
}
public IActionResult ShowStats(string host)
{
WebsiteStatsFactory factory = new WebsiteStatsFactory();
WebsiteStats stats = (WebsiteStats)factory[host];
if (stats == null)
{
ViewBag.Message = "Invalid Host Name!";
return(View("Index"));
}
else
{
return(View(stats));
}
}
public JsonResult GetImages(string WebUrl)
{
string JsonData = string.Empty;
var WebPageData = new WebsiteStats();
var UrlModel = new UrlModel {
urlWeb = WebUrl
};
var ApiUrl = string.Format("{0}://{1}{2}", Request.Url.Scheme, Request.Url.Authority, Url.Content("~"));
var Status = false;
WebPageData = ApiHelper.HttpPostResult <WebsiteStats>(ApiUrl + "api/websiteinfo/loadurl", UrlModel).Result.Result;
if (WebPageData != null)
{
Status = true;
WebPageData.WordFrequency = (from item in WebPageData.WordFrequency orderby item.Value descending select item)
.Take(10)
.ToDictionary(x => x.Key, x => x.Value);
JsonData = JsonConvert.SerializeObject(WebPageData);
}
return(Json(new { success = Status, ImageData = JsonData }));
}
