public static void ParseUser(HackerItem item, List<HtmlNode> links)
{
var userLink = links.Where(x => x.Attributes["href"].Value.Contains("user")).ToList();
if (userLink.Count >= 1){
item.User = userLink[0].InnerText;
}
}
public static void ParsePoints(HackerItem item, List<HtmlNode> spans)
{
var pointSpan = spans.Where(x => x.Attributes["id"] != null && x.Attributes["id"].Value.Contains("score")).ToList();
if (pointSpan.Count == 1)
{
item.Points = Regex.Replace(pointSpan[0].InnerText, " p(.*)", "");
}
}
public static void ParseHost(HackerItem item, List<HtmlNode> spans)
{
var hostSpan = spans.Where(x => x.Attributes["class"] != null && x.Attributes["class"].Value.Contains("comhead")).ToList();
if (hostSpan.Count == 1){
var host = hostSpan[0].InnerText.Trim();
item.Host = host.Replace("(", "").Replace(")", "");
}
}
public static void ParseIDAndCommentCount(HackerItem item, List<HtmlNode> links)
{
var itemLinks = links.Where(x => x.Attributes["href"].Value.Contains("item")).ToList();
if (itemLinks.Count >= 1)
{
var itemLink = itemLinks[0];
item.ID = itemLink.Attributes["href"].Value.Replace("item?id=", "");
item.CommentCount = itemLink.InnerText == "discuss" ? "0" : itemLink.InnerText.Replace(" c(.*)", "");
}
}
public static void ParseUrlTitle(HackerItem item, List<HtmlNode> links)
{
HtmlNode link;
if (links.Count == 2)
{
link = links[1];
}
else
{
link = links[0];//if there is no upvote link
}
item.URL = link.Attributes["href"].Value;
item.Title = link.InnerText;
}
protected List<HackerItem> ParseList(List<HtmlNode> rows)
{
List<HackerItem> items = new List<HackerItem>();
HackerItem newsItem = null;
for (int i = 0; i < rows.Count; i++)
{
try
{
var row = rows[i];
var tableData = row.Descendants("td");
Debug.WriteLine(i);
if (tableData != null && tableData.Count() > 0)
{
if (tableData.Count(x => x.Attributes["class"] != null && x.Attributes["class"].Value.Contains("title")) > 0)
{
if (i == rows.Count - 1)
{
nextUrl = row.Descendants("a").ToList()[0].Attributes["href"].Value;
break;
}
newsItem = new HackerItem();
ParsingHelpers.ParseUrlTitle(newsItem, row.Descendants("a").ToList());
ParsingHelpers.ParseHost(newsItem, row.Descendants("span").ToList());
}
else if (tableData.Count(x => x.Attributes["class"] != null && x.Attributes["class"].Value.Contains("subtext")) > 0)
{
//points & user
var links = row.Descendants("a").ToList();
ParsingHelpers.ParsePoints(newsItem, row.Descendants("span").ToList());
ParsingHelpers.ParseUser(newsItem, links);
ParsingHelpers.ParseIDAndCommentCount(newsItem, links);
items.Add(newsItem);
}
}
}
catch (Exception e)
{
throw new HackerException("Could not parse the news feed. See inner exception for details", e);
}
}
return items;
}
