private void ExtractExperience(string content, ref LinkedInProfile profile)
{
string[] titleSplitWord = Regex.Split(content, "<span class=\"title\">");
string title = Regex.Split(titleSplitWord[1], "</span>")[0].Trim();
profile.ExperiencePosition = WebUtility.HtmlDecode(title);
logger.Info(title);
string org = string.Empty;
if (content.Contains("<span class=\"org summary\">"))
{
string[] orgSplitWord = Regex.Split(content, "<span class=\"org summary\">");
org = Regex.Split(orgSplitWord[1], "</span>")[0].Trim();
profile.ExperienceCompany = WebUtility.HtmlDecode(org);
logger.Info(org);
}
string[] orgDetailSplitWord = Regex.Split(content, "<p class=\"orgstats organization-details");
string orgDetail = string.Empty;
if (orgDetailSplitWord.Count() > 1)
{
string[] orgDetailTemp = Regex.Split(orgDetailSplitWord[1], "position\">");
orgDetail = Regex.Split(orgDetailTemp[1], "</p>")[0].Trim();
//profile.ExperienceCompanyDetail = WebUtility.HtmlDecode(orgDetail);
string[] orgDetailSplit = Regex.Split(orgDetail, ";");
switch (orgDetailSplit.Count())
{
case 1:
profile.ExperienceCompanyBusinessSector = WebUtility.HtmlDecode(orgDetailSplit[0]).Trim();
break;
case 4:
profile.ExperienceCompanyType = WebUtility.HtmlDecode(orgDetailSplit[0]).Trim();
profile.ExperienceCompanySize = WebUtility.HtmlDecode(orgDetailSplit[1]).Trim();
profile.ExperienceCompanyBusinessSector = WebUtility.HtmlDecode(orgDetailSplit[3]).Trim();
break;
case 3:
profile.ExperienceCompanyType = WebUtility.HtmlDecode(orgDetailSplit[0]).Trim();
profile.ExperienceCompanySize = WebUtility.HtmlDecode(orgDetailSplit[1]).Trim();
profile.ExperienceCompanyBusinessSector = WebUtility.HtmlDecode(orgDetailSplit[2]).Trim();
break;
}
logger.Info(orgDetail);
}
string[] periodSplitWord = new string[] { };
if (content.Contains("<p class=\"period\">"))
{
periodSplitWord = Regex.Split(content, "<p class=\"period\">");
}
if (content.Contains("<div class=\"period\">"))
{
periodSplitWord = Regex.Split(content, "<div class=\"period\">");
}
string[] timeSplitWord = Regex.Split(periodSplitWord[1], "</abbr>");
if (timeSplitWord.Count() == 1)
{
return;
}
string firstTime = Regex.Split(timeSplitWord[0], ">")[1].Trim();
string secondTime = Regex.Split(timeSplitWord[1], ">")[1].Trim();
string duration = Regex.Split(periodSplitWord[1], "</span>")[1].Trim();
profile.ExperiencePeriod = WebUtility.HtmlDecode(string.Format(firstTime + " - " + secondTime + " " + duration));
logger.Info(profile.ExperiencePeriod);
if (content.Contains("<span class=\"location\">"))
{
string[] locationSplitWord = Regex.Split(content, "<span class=\"location\">");
profile.ExperienceCompanyLocation =
WebUtility.HtmlDecode(Regex.Split(locationSplitWord[1], "</span>")[0]);
logger.Info(profile.ExperienceCompanyLocation);
}
}
private void WriteToDatabase(LinkedInProfile profile, string link)
{
string username = profile.Username.Replace("'", "\"");
string position = profile.Position.Replace("'", "\"");
string company = profile.Company.Replace("'", "\"");
string summary = profile.Summary.Replace("'", "\"");
string expCompany = profile.ExperienceCompany.Replace("'", "\"");
string expPosition = profile.ExperiencePosition.Replace("'", "\"");
string expCompanyType = profile.ExperienceCompanyType.Replace("'", "\"");
string expCompanySize = profile.ExperienceCompanySize.Replace("'", "\"");
string expCompanyBusinessSector = profile.ExperienceCompanyBusinessSector.Replace("'", "\"");
string expCompanyLocation = profile.ExperienceCompanyLocation.Replace("'", "\"");
string expPeriod = profile.ExperiencePeriod.Replace("'", "\"");
string language = profile.Language.Count > 0 ? profile.Language[0] : string.Empty;
for (int i = 1; i < profile.Language.Count; i++)
{
language += ";" + profile.Language[i].Replace("'", "\"");
}
string skill = profile.SkillAndExpertise.Count > 0 ? profile.SkillAndExpertise[0] : string.Empty;
for (int i = 1; i < profile.SkillAndExpertise.Count; i++)
{
skill += ";" + profile.SkillAndExpertise[i].Replace("'", "\"");
}
link = link.Replace("'", "\"");
mExe.ExecQuery("INSERT INTO UserDB (Username,Position,Company,Description,ExperienceCompany,ExperiencePosition," +
"ExperienceCompanyType,ExperienceCompanySize,ExperienceCompanyBusinessSector,ExperienceCompanyLocation," +
"ExperiencePeriod,Language,Skill,Link) " + "VALUES" + " (" +
"N'" + username + "'," +
"N'" + position + "'," +
"N'" + company + "'," +
"N'" + summary + "'," +
"N'" + expCompany + "'," +
"N'" + expPosition + "'," +
"N'" + expCompanyType + "'," +
"N'" + expCompanySize + "'," +
"N'" + expCompanyBusinessSector + "'," +
"N'" + expCompanyLocation + "'," +
"N'" + expPeriod + "'," +
"N'" + language + "'," +
"N'" + skill + "'," +
"N'" + link + "'" +
")");
}
// <span class="full-name">: tag for user full name
// <p class="headline-title title" style="display:block">: tag for position
// <p class=" description summary">: tag for description
// <div class="postitle">: tag for begin experience
// <li class="competency language">: tag for language
// <li class="competency show-bean ">: tag for skill
// <li class="with-photo">: tag for link profile
public LinkedInProfile Process()
{
var profile = new LinkedInProfile();
logger.Info("Start parse HTML");
// Get username
logger.Info("Get username");
string[] usernameSplitWord = Regex.Split(Content, "<span class=\"full-name\">");
profile.Username = ExtractUsername(usernameSplitWord[1]);
logger.Info(profile.Username);
// Get position
logger.Info("Get position");
string[] postionSplitWord = Regex.Split(Content, "<p class=\"headline-title title\" style=\"display:block\">");
if (postionSplitWord.Count() > 1)
{
string position = ExtractPosition(postionSplitWord[1]);
string[] positionSplit = Regex.Split(position, " at ");
profile.Position = positionSplit[0];
profile.Company = positionSplit.Count() > 1 ? positionSplit[1] : string.Empty;
}
else
{
profile.Position = string.Empty;
profile.Company = string.Empty;
}
logger.Info(profile.Position);
// Get summary
logger.Info("Get summary");
string[] summarySplitWord = Regex.Split(Content, "<p class=\" description summary\">");
profile.Summary = summarySplitWord.Count() > 1 ? ExtractSummary(summarySplitWord[1]) : string.Empty;
// Get experience
logger.Info("Get experience");
string[] expSplitWord = Regex.Split(Content, "<div class=\"postitle\">");
profile.ExperienceCompany = string.Empty;
profile.ExperienceCompanyLocation = string.Empty;
profile.ExperienceCompanyType = string.Empty;
profile.ExperienceCompanySize = string.Empty;
profile.ExperienceCompanyBusinessSector = string.Empty;
profile.ExperiencePeriod = string.Empty;
profile.ExperiencePosition = string.Empty;
if (expSplitWord.Count() > 1) ExtractExperience(expSplitWord[1], ref profile);
//logger.Info(profile.Experience);
// Get language
logger.Info("Get language");
string[] langSplitWord = Regex.Split(Content, "<li class=\"competency language\">");
if (langSplitWord.Count() > 1)
{
for (int i = 1; i < langSplitWord.Count(); i++)
{
string lang = ExtractLanguage(langSplitWord[i]);
profile.Language.Add(lang);
logger.Info(lang);
}
}
// Get skill and expertise
logger.Info("Get skill and expertise");
if (Content.Contains("<li class=\"competency show-bean \">"))
{
string[] skillSplitWord = Regex.Split(Content, "<li class=\"competency show-bean \">");
for (int i = 1; i < skillSplitWord.Count(); i++)
{
var content = Regex.Split(skillSplitWord[i], "</span>")[0];
string skill = ExtractSkillAndExpertise(content);
profile.SkillAndExpertise.Add(skill);
logger.Info(skill);
}
}
if (Content.Contains("<li class=\"competency show-bean extra-skill\">"))
{
string[] extraSkillSplitWord = Regex.Split(Content, "<li class=\"competency show-bean extra-skill\">");
for (int i = 1; i < extraSkillSplitWord.Count(); i++)
{
var content = Regex.Split(extraSkillSplitWord[i], "</span>")[0];
string skill = ExtractSkillAndExpertise(content);
profile.SkillAndExpertise.Add(skill);
logger.Info(skill);
}
}
// Get next profile link
logger.Info("Get next profile");
string[] nextSplitWord = Regex.Split(Content, "<li class=\"with-photo\">");
if (nextSplitWord.Count() > 1)
{
for (int i = 1; i < nextSplitWord.Count(); i++)
{
string profileLink = ExtractNextProfile(nextSplitWord[i]);
profile.NextProfile.Add(profileLink);
logger.Info(profileLink);
}
}
logger.Info("End parse HTML");
return profile;
}