public MicrosoftScholarParser()
{
client = new APIService();
sett = new SettingsRecord();
set = sett.ReadSettings();
masMaxResults = set.MASMaxResults;
}
public bool getCitationsNext(string url,ref List<Paper> p)
{
HtmlWeb web; web = new HtmlWeb();
HtmlDocument doc;
HtmlNode n;
if (url.Contains("viewdoc"))//e.g. http://citeseer.ist.psu.edu/viewdoc/summary?doi=10.1.1.31.3487
{
doc = web.Load(url);
HtmlNode citUrl = doc.DocumentNode.SelectSingleNode("//*[@id=\"docCites\"]/td[2]/a");
if (citUrl == null)
return false;
url = "http://citeseer.ist.psu.edu" + citUrl.GetAttributeValue("href", "");
}
SettingsRecord sr = new SettingsRecord();
Settings s = sr.ReadSettings();
if (p.Count % 10 != 0 || p.Count >= s.CiteSeerMaxResults)
return false;
int pagen = p.Count;
url = url + "&sort=cite&start=" + p.Count;
doc = web.Load(url);
if (doc != null)
Console.WriteLine("Document Loaded!");
else
{
Console.WriteLine("Load Error!");
return false;
}
HtmlNodeCollection rows;
Paper tempPaperObj;
rows = doc.DocumentNode.SelectNodes("//*[@id=\"result_list\"]/div");
if(rows==null)
{
Console.WriteLine("Load Error! Next page doesn't exist! Returned.");
return false;
}
for (int i = 0; i < rows.Count; i++)
{
int numCit=0, year=0;
string title="", authNames="", abs, pUrl="";
if (rows[i].SelectSingleNode("div[3]/a[@title=\"number of citations\"]") != null)
{
try
{
int comI = rows[i].SelectSingleNode("div[3]/a[@title=\"number of citations\"]").InnerText.Substring(9).IndexOf(' ');
if (rows[i].SelectSingleNode("div[3]/a[@title=\"number of citations\"]").InnerText.Substring(9).Remove(comI) != null)
numCit = Convert.ToInt32((rows[i].SelectSingleNode("div[3]/a[@title=\"number of citations\"]").InnerText.Substring(9).Remove(comI)));
}
catch (Exception e) { }
}
else
numCit = 0;
try
{
title = rows[i].SelectSingleNode("h3/a").InnerText.Trim();
authNames = rows[i].SelectSingleNode("div[1]/span[1]").InnerText.Substring(3).Trim();
}
catch (Exception e) { }
String tempYear;
if (rows[i].SelectSingleNode("div[1]/span[@class=\"pubyear\"]") != null)
{
try
{
tempYear = rows[i].SelectSingleNode("div[1]/span[@class=\"pubyear\"]").InnerText;
if (tempYear != null)
year = Convert.ToInt32(tempYear.Substring(2));
}
catch (Exception e) { }
}
else year = 0;
if (rows[i].SelectSingleNode("div[2]") != null)
abs = rows[i].SelectSingleNode("div[2]").InnerText;
else
abs = "";
try
{
pUrl = "http://citeseer.ist.psu.edu" + rows[i].SelectSingleNode("h3/a").GetAttributeValue("href", "");
}
catch (Exception e) { }
tempPaperObj = new Paper(title, pUrl, authNames, abs, year, "", "", numCit, pUrl, 0);
//if (tempPaperObj.NumberOfCitations > 0)
p.Add(tempPaperObj);
}
return true;
}
public CiteSeerJournal_FinalAuthorSearch(string searchEle, int searchTy, string affISSN, string key) //searchType 0 for authorSearch, 1 for journalSearch
{
SettingsRecord sett = new SettingsRecord();
Settings set = sett.ReadSettings();
maxResults = set.CiteSeerMaxResults;
searchElement = searchEle;
searchType = searchTy;
keyword = key;
affiliation = "";
ISSN = "";
auth1 = new Author(searchEle);
journ1 = new Journal(searchEle);
keyword = keyword.Trim();
keyword = Regex.Replace(keyword, @"\s+", " ");
keyword = keyword.Replace(" ", "+");
searchElement = searchElement.Trim();
searchElement = Regex.Replace(searchElement, @"\s+", " ");
searchElement = searchElement.Replace(" ", "+");
if (searchType == 0)
{
//initialURL = "http://citeseerx.ist.psu.edu/search?q=author%3A%28" + searchElement + "%29&submit=Search&ic=1&sort=cite&t=doc";
affiliation = affISSN;
affiliation = affiliation.Trim();
affiliation = Regex.Replace(affiliation, @"\s+", " ");
affiliation = affiliation.Replace(" ", "+");
initialURL = "http://citeseerx.ist.psu.edu/search?q=author%3A%28" + searchElement + "%29&submit=Search&ic=1&sort=cite&t=doc";
}
else if (searchType == 1)
{
ISSN = ISSN.Trim();
ISSN = ISSN.Replace(" ", "");
//Console.WriteLine(searchElement + ", " + keyword + ", " + ISSN);
initialURL = "http://citeseerx.ist.psu.edu/search?q=venue%3A%28" + searchElement + "%29&submit=Search&ic=1&sort=cite&t=doc";
//Console.WriteLine(initialURL);
}
else
{
initialURL = "";
}
CitePage = new HtmlWeb();
noResult = 10;
try
{
CiteDoc = CitePage.Load(initialURL);
}
catch (Exception e)
{
noResult = 0;
}
PageNo = 1;
//Console.WriteLine("Document opened");
getNoResult();
}
