public override ParsedResult Parse(string domain, string proxy = null)
{
try
{
ParsedResult result = new ParsedResult();
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
string html = getHTML(domain, proxy);
if (html == null)
{
result.Url = domain;
result.expiridate = "-";
result.creationdate = "-";
return(result);
}
doc.LoadHtml(html);
HtmlAgilityPack.HtmlNode node = doc.DocumentNode.SelectSingleNode("//span[@class=\'span8\']//h1[@class=\'h3\']");
if (node != null)
{
result.Url = node.InnerText.Split(' ')[0].Trim();
}
else
{
result.Url = domain;
result.expiridate = "-";
result.creationdate = "-";
return(result);
}
return(new ParsedResult());
}
catch (Exception) { return(new ParsedResult {
Url = domain, creationdate = "-", expiridate = "-"
}); }
}
public void insertParsed(ParsedResult parsed)
{
result.Add(parsed);
if (result.Count == 1 + Sources.Count)
{
WriteResult();
//isdone = true;
}
}
public void getDA_PA(string domain, ParsedResult pr)
{
WebClient webClient = new WebClient();
string json = webClient.DownloadString(apiurl + domain);
if (json != null)
{
ParsedResult timer = JsonConvert.DeserializeObject <ParsedResult>(json);
pr.data = timer.data;
}
}
public CsvReaderWriter(string filepath, string resultpath, bool dp, bool ct, bool write_headers, bool dateparsing)
{
Sources = GetCsvDomains(filepath);
nextposition = 0;
result = new List <ParsedResult>();
result.Add(ParsedResult.getHeaders());
fileresultpath = resultpath;
da_pa = dp;
cf_tf = ct;
headers = write_headers;
dateparse = dateparsing;
}
public override ParsedResult Parse(string domain, string proxy = null)
{
try
{
ParsedResult result = new ParsedResult();
result.Url = domain;
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
string html = getHTML(domain, proxy);
if (html == null)
{
result.Url = domain;
result.expiridate = "not found";
result.creationdate = "";
return(result);
}
doc.LoadHtml(html);
HtmlAgilityPack.HtmlNode node = doc.DocumentNode.SelectSingleNode("//div[@class=\'whois-response\']");
if (node != null)
{
var rg1 = new Regex(@"Registrar Registration Expiration Date: (.*?)T");
result.expiridate = rg1.Match(node.InnerHtml).Groups[1].Value;
if (result.expiridate == string.Empty)
{
result.expiridate = "not found";
result.creationdate = "";
return(result);
}
var rg2 = new Regex(@"Creation Date: (.*?)T");
result.creationdate = rg2.Match(node.InnerHtml).Groups[1].Value;
}
else
{
result.Url = domain;
result.expiridate = "not found";
result.creationdate = "";
return(result);
}
return(result);
}
catch (Exception) { return(new ParsedResult {
Url = domain, creationdate = "", expiridate = "not found"
}); }
}
public override ParsedResult Parse(string domain, string proxy = null)
{
try
{
ParsedResult result = new ParsedResult();
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
string html = getHTML(domain, proxy);
if (html == null)
{
result.Url = domain;
result.Rank = "-";
return(result);
}
doc.LoadHtml(html);
HtmlAgilityPack.HtmlNode node = doc.DocumentNode.SelectSingleNode("//h1[@class=\"domain text-center-xs text-left-not-xs\"]");
if (node != null)
{
result.Url = node.InnerText.Trim().ToLower();
}
else
{
result.Url = domain;
result.Rank = "-";
return(result);
}
node = doc.DocumentNode.SelectSingleNode("//div[@id=\"worldRanking-item\"]//div[@class=\"rankValue\"]");
if (node != null)
{
var rg = new Regex(@"#(.*?)<");
result.Rank = rg.Match(node.InnerHtml.TrimEnd()).Groups[1].Value;
}
HtmlAgilityPack.HtmlNodeCollection nc = doc.DocumentNode.SelectNodes("//td[@class=\"text-right\"]");
if (nc != null)
{
result.Overall_Visits = nc[0].InnerText;
result.Time_On_Site = nc[1].InnerText;
result.Pages_per_Visit = nc[2].InnerText;
result.Pages_per_Visit = result.Pages_per_Visit.Replace('.', ',');
result.Bounce_Rate = nc[3].InnerText;
}
node = doc.DocumentNode.SelectSingleNode("//div[@id=\"review\"]");
if (node != null)
{
var rg = new Regex("\"Organic Search\" [(](.*?)%");
result.Organic_Search = rg.Match(node.InnerText).Groups[1].Value.TrimEnd();
if (result.Organic_Search.Contains('.'))
{
result.Organic_Search = result.Organic_Search.Remove(result.Organic_Search.IndexOf('.'));
}
}
HtmlAgilityPack.HtmlNodeCollection htmlNodes =
doc.DocumentNode.SelectNodes("//table[@id=\"countriesBreakdownTable\"]//tr");
if (htmlNodes != null)
{
foreach (HtmlAgilityPack.HtmlNode row in htmlNodes)
{
string country = row.SelectSingleNode("td").InnerText;
string procent = row.SelectSingleNode("td/div[@class=\"shareValue\"]").InnerText;
result.CountresAdd(country, procent);
}
}
result.CountraseRelease();
return(result);
}
catch (Exception ex) { MessageBox.Show(ex.ToString()); return(new ParsedResult {
Url = domain, Rank = "-"
}); }
}
public override ParsedResult Parse(string domain, string proxy = null)
{
try
{
ParsedResult result = new ParsedResult();
HtmlAgilityPack.HtmlDocument doc = new HtmlAgilityPack.HtmlDocument();
string html = getHTML(domain, proxy);
if (html == null)
{
result.Url = domain;
result.Rank = "-";
return(result);
}
doc.LoadHtml(html);
HtmlAgilityPack.HtmlNode node = doc.DocumentNode.SelectSingleNode("//span[@class=\'span8\']//h1[@class=\'h3\']");
if (node != null)
{
result.Url = node.InnerText.Split(' ')[0].Trim();
}
else
{
result.Url = domain;
result.Rank = "-";
return(result);
}
//span[@class='col-pad']//strong[@class='metrics-data align-vmiddle']
node = doc.DocumentNode.SelectSingleNode("//section[@id=\'rank-panel-content\']//span[@class=\'globleRank\']//strong[@class=\'metrics-data align-vmiddle\']");
if (node != null)
{
result.Rank = node.InnerText.Trim();
}
if (result.Rank == "-")
{
result.Url = domain;
result.Rank = "-";
return(result);
}
HtmlAgilityPack.HtmlNodeCollection nc = doc.DocumentNode.SelectNodes("//section[@id=\'engagement-content\']//strong[@class=\'metrics-data align-vmiddle\']");
if (nc != null)
{
result.Bounce_Rate = nc[0].InnerText.Trim();
result.Pages_per_Visit = nc[1].InnerText.Trim();
result.Pages_per_Visit = result.Pages_per_Visit.Replace('.', ',');
result.Time_On_Site = nc[2].InnerText.Trim();
}
decimal firstvisitors = 0, addvisitors = 0;
node =
doc.DocumentNode.SelectSingleNode("//section[@id=\'keyword-content\']//strong[@class=\'metrics-data align-vmiddle\']");
if (node != null)
{
result.Organic_Search = node.InnerText.Trim();
if (result.Organic_Search != "-")
{
string s = result.Organic_Search.Remove(node.InnerText.Trim().Length - 2);
s = s.Replace('.', ',');
try
{
firstvisitors = Convert.ToDecimal(s);
}
catch (Exception)
{
firstvisitors = 0;
}
node =
doc.DocumentNode.SelectSingleNode("//section[@id=\'keyword-content\']//span[@class=\'align-vmiddle change-wrapper change-down color-gen2 \']");
if (node == null)
{
node =
doc.DocumentNode.SelectSingleNode("//section[@id=\'keyword-content\']//span[@class=\'align-vmiddle change-wrapper change-up \']");
}
if (node != null)
{
s = node.InnerText.Trim().Remove(node.InnerText.Trim().Length - 2);
s = s.Replace('.', ',');
try { addvisitors = Convert.ToDecimal(s); }
catch (Exception)
{
addvisitors = 0;
}
}
result.Overall_Visits = (firstvisitors + addvisitors).ToString();
result.Overall_Visits.Remove(result.Overall_Visits.Length - 1);
result.Overall_Visits = result.Overall_Visits.Replace(',', '.');
result.Overall_Visits = result.Overall_Visits.TrimEnd('0').TrimEnd('.');
result.Overall_Visits += "K";
}
else
{
result.Overall_Visits = "-";
}
if (result.Organic_Search.Contains('.'))
{
result.Organic_Search = result.Organic_Search.Remove(result.Organic_Search.IndexOf('.'));
}
}
//table[@id='demographics_div_country_table']/tbody//tr/td[2]
HtmlAgilityPack.HtmlNodeCollection htmlNodes =
doc.DocumentNode.SelectNodes("//table[@id=\'demographics_div_country_table\']/tbody//tr");
if (htmlNodes != null & htmlNodes.Count > 1)
{
foreach (HtmlAgilityPack.HtmlNode row in htmlNodes)
{
string country = row.SelectSingleNode("td[1]").InnerText.Split(';')[1].Trim();
string procent = row.SelectSingleNode("td[2]").InnerText.Trim();
result.CountresAdd(country, procent);
}
}
result.CountraseRelease();
return(result);
}
catch (Exception) { return(new ParsedResult {
Url = domain, Rank = "-"
}); }
}