public List <LaywerModel> GetLaywerList(string html)
{
var laywerList = new List <LaywerModel>();
try
{
// html = File.ReadAllText(@"C:\IIS\test\log.txt", Encoding.UTF8);
html = html.Replace(@"\\\", "");
// File.AppendAllText(@"C:\IIS\test\normalLog.txt", html + "\r\n");
var doc = new HtmlDocument();
doc.LoadHtml(html);
var lawyerDivs = doc.DocumentNode.SelectNodes("//div[@class='row1 cloudlaw-profile user-menu-container square biz-card clearfix']");
if (lawyerDivs != null)
{
//File.AppendAllText(@"C:\IIS\test\normalLog.txt", "div load" + "\r\n");
foreach (HtmlNode lawyerDiv in lawyerDivs)
{
var laywer = new LaywerModel();
laywer.userId = lawyerDiv.Attributes["userid"]?.Value;
var laywerDetail = lawyerDiv.SelectSingleNode(".//div[@class='col-md-7 user-details']");
var h41 = laywerDetail.SelectSingleNode(".//h4");
laywer.Name = h41?.InnerText.Trim();
var ps = laywerDetail.SelectNodes(".//p");
foreach (HtmlNode p in ps)
{
var str = p.InnerText;
if (str.Contains("Licensed"))
{
laywer.LicensedIn = str.Replace("Licensed In:", "").Trim();
}
if (str.Contains("Reg #"))
{
laywer.Registration = str.Replace("Reg #:", "").Trim();
}
if (str.Contains("Date of Admission"))
{
laywer.DateOfAdmission = str.Replace("Date of Admission:", "").Trim();
}
if (str.Contains("County:"))
{
laywer.County = str.Replace("County:", "").Trim();
}
}
ReadFromVcard(laywer, laywer.userId);
laywerList.Add(laywer);
}
}
}
catch (Exception ex)
{
File.AppendAllText(@"C:\IIS\test\error.txt", "GetLaywerList exception:" + DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss") + "\r\n");
File.AppendAllText(@"C:\IIS\test\error.txt", ex.StackTrace + "\r\n");
}
return(laywerList);
}
public void ReadData(string html)
{
var laywerList = new List <LaywerModel>();
// html = File.ReadAllText(@"C:\IIS\test\log.txt", Encoding.UTF8);
// html = html.Replace(@"\\\", "");
var doc = new HtmlDocument();
doc.LoadHtml(html);
var lawyerDivs = doc.DocumentNode.SelectNodes("//div[@class='row1 cloudlaw-profile user-menu-container square biz-card clearfix']");
foreach (HtmlNode lawyerDiv in lawyerDivs)
{
var laywer = new LaywerModel();
var laywerDetail = lawyerDiv.SelectSingleNode("//div[@class='col-md-7 user-details']");
var h41 = laywerDetail.SelectSingleNode(".//h4");
laywer.Name = h41.InnerText.Trim();
HtmlNode profileLink = laywerDetail.SelectSingleNode("//a[@zb-role='profile-link']");
var url = profileLink.Attributes["href"].Value;
HtmlDocument detailDoc = web.Load(url);
ReadDetailData(detailDoc, laywer);
//var ps = laywerDetail.SelectNodes("//div[@class='user-pad biz-address']/p");
//foreach(HtmlNode p in ps)
//{
// if(p.InnerText.Contains("Admission"))
// {
// laywer.DateOfAdmission = p.InnerText;
// }
// if (p.InnerText.Contains("Licensed "))
// {
// laywer.LicensedIn = p.InnerText;
// }
// if (p.InnerText.Contains("Reg"))
// {
// laywer.Registration = p.InnerText;
// }
//}
laywerList.Add(laywer);
}
}
public void ReadFromVcard(LaywerModel laywer, string userId)
{
try
{
var html = "";
var url = "http://www.zeekbeek.com/vcard.ashx?userId=" + userId;
File.AppendAllText(@"C:\IIS\test\normalLog.txt", url + "\r\n");
HttpWebRequest request = (HttpWebRequest)WebRequest.Create(url);
request.Method = "GET";
using (var response = (HttpWebResponse)request.GetResponse())
{
using (var reader = new StreamReader(response.GetResponseStream()))
{
html = reader.ReadToEnd();
}
}
File.AppendAllText(@"C:\IIS\test\normalLog.txt", html + "\r\n");
// var file = @"C:\IIS\test\P36344.vcf";
var dataArr = html.Split(new string[] { "\r\n" }, StringSplitOptions.None).ToList();
//var dataArr = File.ReadAllLines(file);
foreach (var str in dataArr)
{
if (str.StartsWith("N:"))
{
var name = str.Replace("N:", "").Split(';');
if (name.Length >= 3)
{
laywer.Surname = name[0];
laywer.GivenName = name[1];
laywer.MiddleName = name[2];
}
}
if (str.StartsWith("FN:"))
{
laywer.Name = str.Replace("FN:", "");
}
if (str.StartsWith("ORG:"))
{
laywer.Org = str.Replace("ORG:", "");;
}
if (str.StartsWith("TITLE:"))
{
laywer.Title = str.Replace("TITLE:", "");;
}
if (str.StartsWith("EMAIL;PREF;INTERNET:"))
{
laywer.Email = str.Replace("EMAIL;PREF;INTERNET:", "");
}
if (str.StartsWith("URL;WORK:"))
{
laywer.WebUrl = str.Replace("URL;WORK:", "");
}
if (str.StartsWith("TEL;WORK;FAX:"))
{
laywer.Fax = str.Replace("TEL;WORK;FAX:", "");
}
if (str.StartsWith("TEL;WORK;VOICE:"))
{
laywer.Telphone = str.Replace("TEL;WORK;VOICE:", "");
}
if (str.StartsWith("TEL;CELL;VOICE:"))
{
laywer.Cellphone = str.Replace("TEL;CELL;VOICE:", "");
}
if (str.StartsWith("ADR;PREF;WORK;PARCEL;ENCODING=QUOTED-PRINTABLE:"))
{
var address = str.Replace("ADR;PREF;WORK;PARCEL;ENCODING=QUOTED-PRINTABLE:", "").Split(';');
if (address.Length >= 7)
{
laywer.Street = address[2];
laywer.AddressLocality = address[3];
laywer.Region = address[4];
laywer.PostalCode = address[5];
laywer.Country = address[6];
}
}
}
}
catch (Exception ex)
{
File.AppendAllText(@"C:\IIS\test\error.txt", "ReadFromVcard exception:" + DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss") + "\r\n");
File.AppendAllText(@"C:\IIS\test\error.txt", ex.ToString() + "\r\n");
}
}
public void ReadDetailDataNew(string html, LaywerModel laywer)
{
html = File.ReadAllText(@"C:\IIS\test\detail.txt", Encoding.UTF8);
var doc = new HtmlDocument();
doc.LoadHtml(html);
var uls = doc.DocumentNode.SelectNodes("//ul[@class='list-group']");
//var div= doc.DocumentNode.SelectSingleNode("//div[@class='panel-body profile-pa']");
//if(div!=null)
//{
// var alist = div.SelectNodes("//a[@class='tag tagdlg disabled']");
// foreach(HtmlNode practise in alist)
// {
// laywer.ParctiseArea += practise.InnerText + " | ";
// }
// if(laywer.ParctiseArea.Length>0)
// {
// laywer.ParctiseArea = laywer.ParctiseArea.Trim().Substring(0, laywer.ParctiseArea.Length - 1);
// }
//}
foreach (HtmlNode ul in uls)
{
HtmlNode org = ul.SelectSingleNode(".//span[@itemtype='http://schema.org/Place']");
if (org != null)
{
laywer.Org = org.InnerText.Trim();
}
HtmlNode addressLink = ul.SelectSingleNode(".//div[@itemprop='address']");
if (addressLink != null)
{
HtmlNode street = addressLink.SelectSingleNode(".//div[@class='btn-block']");
if (street != null)
{
laywer.Street = street.InnerText.Trim();
}
HtmlNode addressLocality = addressLink.SelectSingleNode(".//span[@itemprop='addressLocality']");
if (addressLocality != null)
{
laywer.AddressLocality = addressLocality.InnerText.Trim();
}
HtmlNode region = addressLink.SelectSingleNode(".//span[@itemprop='addressRegion']");
if (region != null)
{
laywer.Region = region.InnerText.Trim();
}
HtmlNode postalCode = addressLink.SelectSingleNode(".//span[@itemprop='postalCode']");
if (postalCode != null)
{
laywer.PostalCode = postalCode.InnerText.Trim();
}
}
//HtmlNode tel = ul.SelectSingleNode(".//a[@zb-role='phone-vpl-link']");
//if (tel != null)
//{
// laywer.Tel = tel.InnerText.Trim();
//}
HtmlNode web = ul.SelectSingleNode(".//a[@zb-role='web-vpl-link']");
if (web != null)
{
laywer.WebUrl = web.InnerText.Trim();
}
var lis = ul.SelectNodes(".//li[@class='list-group-item']");
foreach (HtmlNode li in lis)
{
//HtmlNode job = li.SelectSingleNode(".//div[@itemprop='jobTitle']");
//if (job != null)
//{
// laywer.Org = li.InnerText.Replace(job.InnerText, " | " + job.InnerText).Trim().Replace(" ", "");
//}
if (li.InnerText.Contains("County"))
{
laywer.County = li.InnerText.Replace("County:", "").Trim();
}
if (li.InnerText.Contains("Country"))
{
laywer.Country = li.InnerText.Replace("Country:", "").Trim();
}
if (li.InnerText.Contains("(T)"))
{
laywer.Telphone = li.InnerText.Replace("(T)", "").Trim();
}
if (li.InnerText.Contains("(F)"))
{
laywer.Fax = li.InnerText.Replace("(F)", "").Trim();
}
if (li.InnerHtml.Contains("Javascript"))
{
}
if (li.InnerText.ToLower().Contains("registration") || li.InnerText.ToLower().Contains("admission"))
{
var text = li.InnerHtml;
foreach (var str in Regex.Split(text, "<br>"))
{
if (str.Contains("State of Admission") || str.Contains("Licensed"))
{
laywer.LicensedIn = str.Replace("State of Admission:", "").Trim();
}
if (str.Contains("Registration"))
{
laywer.Registration = str.Replace("Registration #:", "").Trim();
}
if (str.Contains("Date of Admission"))
{
laywer.DateOfAdmission = str.Replace("Date of Admission:", "").Trim();
}
}
}
}
}
}
public void ReadDetailData(HtmlDocument doc, LaywerModel laywer)
{
//html = File.ReadAllText(@"C:\IIS\test\detail.txt", Encoding.UTF8);
// var doc = new HtmlDocument();
// doc.LoadHtml(html);
var uls = doc.DocumentNode.SelectNodes("//ul[@class='list-group']");
//var div= doc.DocumentNode.SelectSingleNode("//div[@class='panel-body profile-pa']");
//if(div!=null)
//{
// var alist = div.SelectNodes("//a[@class='tag tagdlg disabled']");
// foreach(HtmlNode practise in alist)
// {
// laywer.ParctiseArea += practise.InnerText + " | ";
// }
// if(laywer.ParctiseArea.Length>0)
// {
// laywer.ParctiseArea = laywer.ParctiseArea.Trim().Substring(0, laywer.ParctiseArea.Length - 1);
// }
//}
var metaDiv = doc.DocumentNode.SelectSingleNode("//div[@class='DnnModule DnnModule-ZBUserProfileViewer DnnModule-3386']");
if (metaDiv != null)
{
var emailMeta = metaDiv.SelectSingleNode(".//meta[@itemprop='email']");
if (emailMeta != null)
{
laywer.Email = emailMeta.Attributes["content"].Value;
}
var faxMeta = metaDiv.SelectSingleNode(".//meta[@itemprop='faxNumber']");
if (faxMeta != null)
{
laywer.Fax = faxMeta.Attributes["content"].Value;
}
var telephoneMeta = metaDiv.SelectSingleNode(".//meta[@itemprop='telephone']");
if (telephoneMeta != null)
{
laywer.Telphone = telephoneMeta.Attributes["content"].Value;
}
}
foreach (HtmlNode ul in uls)
{
HtmlNode org = ul.SelectSingleNode(".//span[@itemtype='http://schema.org/Place']");
if (org != null)
{
laywer.Org = org.InnerText.Trim();
}
HtmlNode addressLink = ul.SelectSingleNode(".//div[@itemprop='address']");
if (addressLink != null)
{
HtmlNode street = addressLink.SelectSingleNode(".//div[@class='btn-block']");
if (street != null)
{
laywer.Street = street.InnerText.Trim();
}
HtmlNode addressLocality = addressLink.SelectSingleNode(".//span[@itemprop='addressLocality']");
if (addressLocality != null)
{
laywer.AddressLocality = addressLocality.InnerText.Trim();
}
HtmlNode region = addressLink.SelectSingleNode(".//span[@itemprop='addressRegion']");
if (region != null)
{
laywer.Region = region.InnerText.Trim();
}
HtmlNode postalCode = addressLink.SelectSingleNode(".//span[@itemprop='postalCode']");
if (postalCode != null)
{
laywer.PostalCode = postalCode.InnerText.Trim();
}
}
HtmlNode web = ul.SelectSingleNode(".//a[@zb-role='web-vpl-link']");
if (web != null)
{
laywer.WebUrl = web.InnerText.Trim();
}
var lis = ul.SelectNodes(".//li[@class='list-group-item']");
foreach (HtmlNode li in lis)
{
if (li.InnerText.Contains("County"))
{
laywer.County = li.InnerText.Replace("County:", "").Trim();
}
if (li.InnerText.Contains("Country"))
{
laywer.Country = li.InnerText.Replace("Country:", "").Trim();
}
if (string.IsNullOrEmpty(laywer.Telphone))
{
if (li.InnerText.Contains("(C)") || li.InnerText.Contains("(T)"))
{
laywer.Telphone = li.InnerText.Replace("(C)", "").Replace("(T)", "").Trim();
}
}
if (li.InnerText.ToLower().Contains("registration") || li.InnerText.ToLower().Contains("admission"))
{
var text = li.InnerHtml;
foreach (var str in Regex.Split(text, "<br>"))
{
if (str.Contains("State of Admission") || str.Contains("Licensed"))
{
laywer.LicensedIn = str.Replace("State of Admission:", "").Trim();
}
if (str.Contains("Registration"))
{
laywer.Registration = str.Replace("Registration #:", "").Trim();
}
if (str.Contains("Date of Admission"))
{
laywer.DateOfAdmission = str.Replace("Date of Admission:", "").Trim();
}
}
}
}
}
}
public void ReadFromVcard(LaywerModel laywer, string file)
{
try
{
var dataArr = File.ReadAllLines(file);
foreach (var str in dataArr)
{
if (str.StartsWith("N:"))
{
var name = str.Replace("N:", "").Split(';');
if (name.Length >= 3)
{
laywer.Surname = name[0];
laywer.GivenName = name[1];
laywer.MiddleName = name[2];
}
}
// if (str.StartsWith("FN:"))
// {
// laywer.Name = str.Replace("FN:", "");
// }
if (str.StartsWith("ORG:"))
{
laywer.Org = str.Replace("ORG:", "").Replace(";", "");
}
if (str.StartsWith("TITLE:"))
{
laywer.Title = str.Replace("TITLE:", "");;
}
if (str.StartsWith("EMAIL;PREF;INTERNET:"))
{
laywer.Email = str.Replace("EMAIL;PREF;INTERNET:", "");
}
if (str.StartsWith("URL;WORK:"))
{
laywer.WebUrl = str.Replace("URL;WORK:", "");
}
if (str.StartsWith("URL;PREF;WORK:"))
{
laywer.ZeekBeekUrl = str.Replace("URL;PREF;WORK:", "");
}
if (str.StartsWith("TEL;WORK;FAX:"))
{
laywer.Fax = str.Replace("TEL;WORK;FAX:", "");
}
if (str.StartsWith("TEL;WORK;VOICE:"))
{
laywer.Telphone = str.Replace("TEL;WORK;VOICE:", "");
}
if (str.StartsWith("TEL;CELL;VOICE:"))
{
laywer.Cellphone = str.Replace("TEL;CELL;VOICE:", "");
}
if (str.StartsWith("ADR;PREF;WORK;PARCEL;ENCODING=QUOTED-PRINTABLE:"))
{
var address = str.Replace("ADR;PREF;WORK;PARCEL;ENCODING=QUOTED-PRINTABLE:", "").Split(';');
if (address.Length >= 7)
{
laywer.Street = address[2];
laywer.AddressLocality = address[3];
laywer.Region = address[4];
laywer.PostalCode = address[5];
laywer.Country = address[6];
}
}
}
}
catch (Exception ex)
{
File.AppendAllText(@"C:\IIS\zeekbeek\error.txt", "ReadFromVcard exception:" + DateTime.Now.ToString("yyyy-MM-dd HH:mm:ss") + "\r\n");
File.AppendAllText(@"C:\IIS\zeekbeek\error.txt", ex.ToString() + "\r\n");
}
}