// RESULTS FROM AUTHOR PROFILE PAGE
public SG.Author getAuthStatistics(string authUrl)
{
bool isOK = true;
if (authUrl == null) return null;
authUrl += "&pagesize=100";
GSAuthScraper authScraper = new GSAuthScraper(authUrl, 0,ref isOK);
if (!isOK) return null;
authScraper.getCitationStats();
SG.Author author = new SG.Author(authScraper.getName(), authScraper.getAffiliation(), authScraper.getHomePage(), authScraper.getHIndex(), authScraper.getIIndex());
//Console.WriteLine(author.Name + "," + author.getHIndex() + "," + author.getI10Index());
List<SG.Paper> papers = authScraper.getPapersOfCurrentPage();
if (papers == null) return null;
if (papers.Count == 0) return author;
foreach (SG.Paper paper in papers) author.addPaper(paper);
return author;
}
// SEARCH PAGE RESULTS
public SG.Author getAuthors(string authName, string affiliation, string keywords, ref string next_url)
{
// CONNECTIONS
SG.Author result = new SG.Author(authName);
if (authName == null) authName = "";
else
{
authName.Trim();
authName = Regex.Replace(authName, @"\s+", "+");
//authName = authName.Insert(0, "author:");
//Console.WriteLine(authName);
}
if (affiliation == null) affiliation = "";
else
{
affiliation = affiliation.Trim();
affiliation = Regex.Replace(affiliation, @"\s+", "+");
}
if (keywords == null) keywords = "";
else
{
keywords = keywords.Trim();
keywords = Regex.Replace(keywords, @"\s+", "+");
}
//string url = "http://scholar.google.com/scholar?q=" + authName + "&btnG=&hl=en&as_sdt=1.";//keywords + "&as_epq=&as_oq=" //affiliation + "&as_eq=&as_occt=any&as_sauthors=" + authName;// + "&as_publication=&as_ylo=&as_yhi=&btnG=&hl=en&as_sdt=0%2C5";
string url = "http://scholar.google.com/scholar?as_q=&as_epq=&as_oq=&as_eq=&as_occt=any&as_sauthors=" + authName + "&as_publication=&as_ylo=&as_yhi=&btnG=&hl=en&as_sdt=0,5&num=20";
Console.WriteLine(url);
HtmlWeb web = new HtmlWeb();
try
{
doc = web.Load(url);
}
catch (Exception e) {
return null;
}
//Console.WriteLine(doc.DocumentNode.InnerHtml);
string xpath = "//div[@class=\"gs_ri\"]";
string title, titleLink, authors, publication, publisher, cited_by_url, summary;
int year, rank = 1, no_of_citations;
HtmlNodeCollection searchResults = doc.DocumentNode.SelectNodes(xpath);
if (searchResults == null)
{
if (checkForCaptcha())
{
Console.WriteLine("Captcha problem ...");
return null;
}
Console.WriteLine("No results ...");
return result;
}
else
{
Console.WriteLine(url);
foreach (HtmlNode n in searchResults)
{
// TITLE AND TITLE LINK
HtmlNode child = n.SelectSingleNode(".//*[@class=\"gs_rt\"]");
title = child.InnerText;
titleLink = "";
HtmlNode url_node = child.SelectSingleNode(".//a");
if (url_node != null)
{
titleLink = url_node.GetAttributeValue("href", "");
if (!titleLink.Equals(""))
{
//titleLink = "http://scholar.google.com" + titleLink;
titleLink = titleLink.Replace("amp;", "");
}
}
// AUTHORS AND PUBLICATION
child = n.SelectSingleNode(".//*[@class=\"gs_a\"]");
authors = "";
publication = "";
publisher = "";
year = 1970;
if (child != null)
{
string[] names = child.InnerText.Split('-');
if (names.Length == 1) authors = names[0];
else if (names.Length == 2)
{
authors = names[0];
bool flag = false;
names[1].Trim();
try { year = Convert.ToInt32(names[1]); }
catch (FormatException fe) { flag = true; }
if (flag)
{
string[] p = names[1].Split(',');
try { year = Convert.ToInt32(p[1]); }
catch (Exception e) { }
publication = p[0];
}
}
else
{
authors = names[0];
publisher = names[2];
bool flag = false;
names[1].Trim();
try { year = Convert.ToInt32(names[1]); }
catch (FormatException fe) { flag = true; }
if (flag)
{
string[] p = names[1].Split(',');
try { year = Convert.ToInt32(p[1]); }
catch (Exception e) { }
publication = p[0];
}
}
}
// SUMMARY
child = n.SelectSingleNode(".//*[@class=\"gs_rs\"]");
summary = "";
if (child != null)
{
summary = child.InnerText;
}
// CITATION STUFF
no_of_citations = 0;
cited_by_url = "";
child = n.SelectSingleNode(".//*[@class=\"gs_fl\"]");
if (child != null) child = child.FirstChild;
if (child != null)
{
string text = child.InnerText;
try
{
text = text.Replace("Cited by", "");
text = text.Trim();
no_of_citations = Convert.ToInt32(text);
}
catch (Exception e) { }
cited_by_url = no_of_citations != 0 ? child.GetAttributeValue("href", "") : "";
if (!cited_by_url.Equals(""))
{
cited_by_url = "http://scholar.google.com" + cited_by_url;
cited_by_url = cited_by_url.Replace("amp;", "");
}
}
publisher.Trim();
publication.Trim();
SG.Paper paper = new SG.Paper(title, titleLink, authors, summary, year, publication, publisher, no_of_citations, cited_by_url, rank);
result.addPaper(paper);
rank++;
}
//NEXT PAGE URL
HtmlNode bottom = doc.DocumentNode.SelectSingleNode(".//*[@id=\"gs_n\"]//table//td[@align=\"left\"]//a");
if (bottom != null)
{
url = bottom.GetAttributeValue("href", "");
if (!url.Equals(""))
{
url = "http://scholar.google.com" + url;
url = url.Replace("amp;", "");
next_url = url;
}
else next_url = null;
}
else next_url = null;
}
return result;
}