public static List <string> GettingAllUrls2(string PageSource, string MustMatchString)
{
List <string> suburllist1 = new List <string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
try
{
string[] Dataconnection = System.Text.RegularExpressions.Regex.Split(PageSource, "2nd degree contact\",");
string DataImage = string.Empty;
foreach (string item in Dataconnection)
{
if (!item.Contains("!DOCTYPE "))
{
if (item.Contains("&pid="))
{
int startindex = item.IndexOf("&pid=");
string start = item.Substring(startindex).Replace("&pid=", "");
int endIndex = start.IndexOf(",");
string finalurl = "http://www.linkedin.com/profile/view?id=" + start.Substring(0, endIndex).Replace("\"", string.Empty);
suburllist1.Add(finalurl);
}
}
}
}
catch { }
return(suburllist1.Distinct().ToList());
}
public static List<string> GettingAllUrls2(string PageSource, string MustMatchString)
{
List<string> suburllist1 = new List<string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
try
{
string[] Dataconnection = System.Text.RegularExpressions.Regex.Split(PageSource, "2nd degree contact\",");
string DataImage = string.Empty;
foreach (string item in Dataconnection)
{
if (!item.Contains("!DOCTYPE "))
{
if (item.Contains("&pid="))
{
int startindex = item.IndexOf("&pid=");
string start = item.Substring(startindex).Replace("&pid=","");
int endIndex = start.IndexOf(",");
string finalurl = "http://www.linkedin.com/profile/view?id=" + start.Substring(0, endIndex).Replace("\"", string.Empty);
suburllist1.Add(finalurl);
}
}
}
}
catch { }
return suburllist1.Distinct().ToList();
}
public static List<string> GettingAllprname(string PageSource, string MustMatchString)
{
List<string> suburllist1 = new List<string>();
Dictionary<string, string> categoryDictonsry = new Dictionary<string, string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
try
{
string[] Dataconnection = System.Text.RegularExpressions.Regex.Split(PageSource, "<select id=\"cid");
string DataImage = string.Empty;
string[] datacon = System.Text.RegularExpressions.Regex.Split(PageSource, "<select id=\"cid");
string[] Arr = Regex.Split(datacon[1], "</option>");
foreach (string item in Arr)
{
if (!item.Contains("Show All Companies"))
{
if (item.Contains("option value"))//(item.Contains("search/profile/person?"))
{
string[] category = Regex.Split(item, ">");
string catId = category[0].Replace("<option value=", "").Replace("\n", "").Replace("\"", "").Replace("/", "").Trim();
string catname = category[1];
//string value = item.Substring(item.IndexOf("<option value="), item.IndexOf("\"") - item.IndexOf("<option value=")).Trim().Replace("<option value=", "").Replace("\"", "");
string finalurl = catId + "," + catname;
//categoryDict
// string finalurl = item.Substring(item.IndexOf("\""), item.IndexOf(">") - item.IndexOf("\"")).Trim().Replace("class=", "").Replace("\"","");
//string finalurl1 = "http://subscriber.zoominfo.com/zoominfo/" + finalurl;
suburllist1.Add(catname);
if (item.Contains("<select id=\"status"))
{
break;
}
}
}
}
}
catch { }
return suburllist1.Distinct().ToList();
}
public static List <string> GettingAllprname(string PageSource, string MustMatchString)
{
List <string> suburllist1 = new List <string>();
Dictionary <string, string> categoryDictonsry = new Dictionary <string, string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
try
{
string[] Dataconnection = System.Text.RegularExpressions.Regex.Split(PageSource, "<select id=\"cid");
string DataImage = string.Empty;
string[] datacon = System.Text.RegularExpressions.Regex.Split(PageSource, "<select id=\"cid");
string[] Arr = Regex.Split(datacon[1], "</option>");
foreach (string item in Arr)
{
if (!item.Contains("Show All Companies"))
{
if (item.Contains("option value"))//(item.Contains("search/profile/person?"))
{
string[] category = Regex.Split(item, ">");
string catId = category[0].Replace("<option value=", "").Replace("\n", "").Replace("\"", "").Replace("/", "").Trim();
string catname = category[1];
//string value = item.Substring(item.IndexOf("<option value="), item.IndexOf("\"") - item.IndexOf("<option value=")).Trim().Replace("<option value=", "").Replace("\"", "");
string finalurl = catId + "," + catname;
//categoryDict
// string finalurl = item.Substring(item.IndexOf("\""), item.IndexOf(">") - item.IndexOf("\"")).Trim().Replace("class=", "").Replace("\"","");
//string finalurl1 = "http://subscriber.zoominfo.com/zoominfo/" + finalurl;
suburllist1.Add(catname);
if (item.Contains("<select id=\"status"))
{
break;
}
}
}
}
}
catch { }
return(suburllist1.Distinct().ToList());
}
public static List<string> GettingAllUrls1_writtenBysharan(string PageSource, string MustMatchString)
{
List<string> suburllist1 = new List<string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
try
{
string[] Dataconnection = System.Text.RegularExpressions.Regex.Split(PageSource, "/profile/view?");
string DataImage = string.Empty;
foreach (string item in Dataconnection)
{
if (!item.Contains("!DOCTYPE "))
{
if (item.Contains("vsrp_people_res_name"))
{
//string finalurl = item.Substring(item.IndexOf("/profile/view?id="), item.IndexOf("url_unfollow_infl") - item.IndexOf(",\"link_nprofile_view")).Trim();
//string finalurls1 = finalurl.Substring(finalurl.IndexOf("/profile/view?id="), finalurl.IndexOf("%3Aprimary")+10 - finalurl.IndexOf("")).Trim();
//int startindex = item.IndexOf("?");
//string start = item.Substring(startindex);
//int endIndex = start.IndexOf(",");
string id = Utils.getBetween(item, "pid=", "&");
// string finalurl = "http://www.linkedin.com/profile/view" + start.Substring(0, endIndex).Replace("\"", string.Empty);
string finalurl = "https://www.linkedin.com/profile/view?id=" + id;
suburllist1.Add(finalurl);
//http://www.linkedin.com?id=230385129&authType=OUT_OF_NETWORK&authToken=OWRO&locale=en_US&srchid=2732756961375079939471&srchindex=1&srchtotal=10&trk=vsrp_people_res_name&trkInfo=VSRPsearchId%3A2732756961375079939471%2CVSRPtargetId%3A230385129%2CVSRPcmpt%3Aprimary"
//http://www.linkedin.com/profile/view?id=6741047&authType=OUT_OF_NETWORK&authToken=MWyZ&locale=en_US&srchid=2732758471375077875757&srchindex=1&srchtotal=55&trk=vsrp_people_res_name&trkInfo=VSRPsearchId%3A2732758471375077875757%2CVSRPtargetId%3A6741047%2CVSRPcmpt%3Aprimary
}
}
}
}
catch { }
return suburllist1.Distinct().ToList();
}
public static List<string> GettingAllUrls(string PageSource, string MustMatchString)
{
List<string> suburllist = new List<string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
StringArray datagoogle = htmlUtil.GetHyperlinkedUrls(PageSource);
for (int i = 0; i < datagoogle.Length; i++)
{
string hreflink = datagoogle.GetString(i);
if (hreflink.Contains(MustMatchString) && hreflink.Contains("goback"))
{
suburllist.Add(hreflink);
}
}
return suburllist.Distinct().ToList();
}
public static List <string> GettingAllUrls(string PageSource, string MustMatchString)
{
List <string> suburllist = new List <string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
StringArray datagoogle = htmlUtil.GetHyperlinkedUrls(PageSource);
for (int i = 0; i < datagoogle.Length; i++)
{
string hreflink = datagoogle.GetString(i);
if (hreflink.Contains(MustMatchString)) //&& hreflink.Contains("goback"))
{
suburllist.Add(hreflink);
}
}
return(suburllist.Distinct().ToList());
}
public static List <string> GettingAllUrls1_writtenBysharan(string PageSource, string MustMatchString)
{
List <string> suburllist1 = new List <string>();
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
try
{
string[] Dataconnection = System.Text.RegularExpressions.Regex.Split(PageSource, "/profile/view?");
string DataImage = string.Empty;
foreach (string item in Dataconnection)
{
if (!item.Contains("!DOCTYPE "))
{
if (item.Contains("vsrp_people_res_name"))
{
//string finalurl = item.Substring(item.IndexOf("/profile/view?id="), item.IndexOf("url_unfollow_infl") - item.IndexOf(",\"link_nprofile_view")).Trim();
//string finalurls1 = finalurl.Substring(finalurl.IndexOf("/profile/view?id="), finalurl.IndexOf("%3Aprimary")+10 - finalurl.IndexOf("")).Trim();
//int startindex = item.IndexOf("?");
//string start = item.Substring(startindex);
//int endIndex = start.IndexOf(",");
string id = Utils.getBetween(item, "pid=", "&");
// string finalurl = "http://www.linkedin.com/profile/view" + start.Substring(0, endIndex).Replace("\"", string.Empty);
string finalurl = "https://www.linkedin.com/profile/view?id=" + id;
suburllist1.Add(finalurl);
//http://www.linkedin.com?id=230385129&authType=OUT_OF_NETWORK&authToken=OWRO&locale=en_US&srchid=2732756961375079939471&srchindex=1&srchtotal=10&trk=vsrp_people_res_name&trkInfo=VSRPsearchId%3A2732756961375079939471%2CVSRPtargetId%3A230385129%2CVSRPcmpt%3Aprimary"
//http://www.linkedin.com/profile/view?id=6741047&authType=OUT_OF_NETWORK&authToken=MWyZ&locale=en_US&srchid=2732758471375077875757&srchindex=1&srchtotal=55&trk=vsrp_people_res_name&trkInfo=VSRPsearchId%3A2732758471375077875757%2CVSRPtargetId%3A6741047%2CVSRPcmpt%3Aprimary
}
}
}
}
catch { }
return(suburllist1.Distinct().ToList());
}
public List<string> GettingAllUrls(string PageSource, string MustMatchString)
{
List<string> suburllist = new List<string>();
try
{
HtmlUtil htmlUtil = new HtmlUtil();
PageSource = htmlUtil.EntityDecode(PageSource);
StringArray datagoogle = htmlUtil.GetHyperlinkedUrls(PageSource);
for (int i = 0; i < datagoogle.Length; i++)
{
try
{
string hreflink = datagoogle.GetString(i);
if (hreflink.Contains(MustMatchString) && hreflink.Contains("&authType="))
{
if (hreflink.Contains("http://www.linkedin.com"))
{
suburllist.Add(hreflink);
Log("[ " + DateTime.Now + " ] => [ URL >>> " + hreflink + " ]");
}
else
{
suburllist.Add("http://www.linkedin.com" + hreflink);
Log("[ " + DateTime.Now + " ] => [ URL >>> http://www.linkedin.com" + hreflink + " ]");
}
}
}
catch
{
}
}
}
catch
{
}
suburllist = suburllist.Distinct().ToList();
return suburllist.Distinct().ToList();
}