public async Task Perform()
{
int len = _links.Count;
await Notify("Finished:", true);
await Task.Run(() => {
Parallel.For(0, 5, i => {
while (_links.Exists(x => !x.Checked))
{
Link link;
lock (_locker)
{
link = _links.FirstOrDefault(x => !x.Checked);
if (link != null)
{
link.Checked = true;
}
}
string content = HttpClientX.Find(link.Url).Result;
string result = string.Empty;
try
{
HtmlAgilityPack.HtmlDocument html = new HtmlAgilityPack.HtmlDocument();
html.LoadHtml(content);
var hs = html.DocumentNode.SelectNodes(".//*");
foreach (var h in hs)
{
string tag = h.Name;
if (tag.Equals("h1") || tag.Equals("h2") || tag.Equals("h3") || tag.Equals("h4") || tag.Equals("h5") || tag.Equals("h6"))
{
result += "\r\n" + h.OuterHtml + "\r\n";
}
}
}
catch
{
}
Notify(link.Number.ToString() + " ; ", false).Wait();
Application.Current.Dispatcher.Invoke(() => { link.Content = link.Url + "\r\n" + result; });
}
});
});
}
private async Task <string> GetPages(string key)
{
await Notify("getting google pages", true);
string searchurl = "http://google.com/search?q=" + key;
string resp = await HttpClientX.Find(searchurl);
var html = new HtmlAgilityPack.HtmlDocument();
html.LoadHtml(resp);
var table = html.GetElementbyId("foot");
foreach (var td in table.QuerySelectorAll("a"))
{
string href = td.GetAttributeValue("href", "");
_pages.Add("http://google.com" + href);
}
return(resp);
}
public async Task <List <string> > GetTop(string key)
{
string resp = await GetPages(key);
List <string> result = new List <string>();
string searchurl = "http://google.com/search?q=" + key;
int page = 0;
while (true)
{
await Notify("getting website links", true);
HtmlAgilityPack.HtmlDocument html;
try
{
if (page != 0)
{
resp = await HttpClientX.Find(_pages[page]);
}
html = new HtmlAgilityPack.HtmlDocument();
html.LoadHtml(resp);
var hs = html.DocumentNode.QuerySelectorAll("#search div[class=\"g\"] div[class=\"r\"] a");
foreach (var h in hs)
{
if (h.Name == "a" && h.QuerySelectorAll("h3").Count != 0)
{
string href = h.GetAttributeValue("href", "");
if (href != "")
{
result.Add(href);
}
}
}
if (_pages.Count > page && result.Count < Depth)
{
page++;
continue;
}
else
{
break;
}
}
catch
{
}
finally
{
}
}
await Notify("list created", true);
return(result.Where(x => x.Contains("http:") || x.Contains("https:")).ToList());
}
