public List<PubMedItem> Parse(string xml)
{
List<PubMedItem> pmItemList = new List<PubMedItem>();
if (string.IsNullOrEmpty(xml))
return pmItemList;
if (xml.IndexOf("XML not found for id") > -1)
return pmItemList;
xml = xml.Replace("http://www.ncbi.nlm.nih.gov/corehtml/query/DTD/pubmed_140101.dtd", "");
XmlDocument xdoc = new XmlDocument();
xdoc.LoadXml(xml);
XmlNodeList articleNodeList = xdoc.SelectNodes("/PubmedArticleSet/PubmedArticle");
if (articleNodeList.Count < 1)
return pmItemList;
foreach (XmlNode articleNode in articleNodeList)
{
PubMedItem it = new PubMedItem();
it.Pmid = articleNode.SelectSingleNode("MedlineCitation/PMID").InnerText;
it.Title = articleNode.SelectSingleNode("MedlineCitation/Article/ArticleTitle").InnerText;
XmlNode node = articleNode.SelectSingleNode("MedlineCitation/Article/Abstract/AbstractText");
if (node != null)
{
it.Abstract = node.InnerText;
}
else
it.Abstract = "";
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/ISOAbbreviation");
if (node != null)
{
it.Journal = node.InnerText;
}
else
{
it.Journal = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/Title").InnerText;
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/ISSN");
if (node != null)
{
it.JIssn = node.InnerText;
}
else
{
it.JIssn = "";
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/Volume");
if (node != null)
{
it.Volume = node.InnerText;
}
else
it.Volume = "";
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/Issue");
if (node != null)
{
it.Issue = node.InnerText;
}
else
{
it.Issue = "";
}
//if (it.Volssue.Length > 98)
// it.Volssue = it.Volssue.Substring(0, 90);
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/PubDate/Year");
if (node != null)
it.Pubdate = node.InnerText;
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/PubDate/Month");
if (node != null)
it.Pubdate = it.Pubdate + "," + node.InnerText;
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/PubDate/Day");
if (node != null)
it.Pubdate += " " + node.InnerText;
XmlNodeList authorNodelist = articleNode.SelectNodes("MedlineCitation/Article/AuthorList/Author");
if (authorNodelist == null)
{
it.Authors = "";
}
else if (authorNodelist.Count == 0)
{
it.Authors = "";
}
else
{
string authors = "";
foreach (XmlNode authorNode in authorNodelist)
{
node=authorNode.SelectSingleNode("./LastName");
string lastname = "",forename="";
if (node != null)
lastname = node.InnerText;
node = authorNode.SelectSingleNode("./ForeName");
if (node != null)
forename= node.InnerText;
authors += forename+" "+lastname+",";
}
authors = authors.Substring(0,authors.Length-1);
it.Authors = authors;
/*string author = "";
node = nodelist[0].SelectSingleNode("./LastName");
if (node != null)
author = node.InnerText;
node = nodelist[0].SelectSingleNode("./Initials");
if (node != null)
author = author + ", " + node.InnerText;
it.Authors = author;
if (nodelist.Count > 1)
{
author = " and ";
node = nodelist[1].SelectSingleNode("./LastName");
if (node != null)
author += node.InnerText;
node = nodelist[1].SelectSingleNode("./Initials");
if (node != null)
author = author + ", " + node.InnerText;
it.Authors += author;
}*/
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Pagination/MedlinePgn");
if (node != null)
{
it.Pagination = node.InnerText;
}
else
it.Pagination = "";
node=articleNode.SelectSingleNode("PubmedData/ArticleIdList/ArticleId[@IdType='doi']");
if (node != null)
{
it.Doi = node.InnerText;
}
XmlNodeList pubTypeNodeList = articleNode.SelectNodes("MedlineCitation/Article/PublicationTypeList/PublicationType");
if(pubTypeNodeList==null)
{
it.PublicationType = "";
}
else if (pubTypeNodeList.Count < 1)
{
it.PublicationType = "";
}
else
{
string pubtype = "";
foreach (XmlNode pubTypeNode in pubTypeNodeList)
{
pubtype+=pubTypeNode.InnerText+"|";
}
pubtype = pubtype.Substring(0, pubtype.Length - 1);
it.PublicationType = pubtype;
}
pmItemList.Add(it);
}
return pmItemList;
}
public List <PubMedItem> Parse(string xml)
{
List <PubMedItem> pmItemList = new List <PubMedItem>();
if (string.IsNullOrEmpty(xml))
{
return(pmItemList);
}
if (xml.IndexOf("XML not found for id") > -1)
{
return(pmItemList);
}
xml = xml.Replace("http://www.ncbi.nlm.nih.gov/corehtml/query/DTD/pubmed_140101.dtd", "");
XmlDocument xdoc = new XmlDocument();
xdoc.LoadXml(xml);
XmlNodeList articleNodeList = xdoc.SelectNodes("/PubmedArticleSet/PubmedArticle");
if (articleNodeList.Count < 1)
{
return(pmItemList);
}
foreach (XmlNode articleNode in articleNodeList)
{
PubMedItem it = new PubMedItem();
it.Pmid = articleNode.SelectSingleNode("MedlineCitation/PMID").InnerText;
it.Title = articleNode.SelectSingleNode("MedlineCitation/Article/ArticleTitle").InnerText;
XmlNode node = articleNode.SelectSingleNode("MedlineCitation/Article/Abstract/AbstractText");
if (node != null)
{
it.Abstract = node.InnerText;
}
else
{
it.Abstract = "";
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/ISOAbbreviation");
if (node != null)
{
it.Journal = node.InnerText;
}
else
{
it.Journal = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/Title").InnerText;
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/ISSN");
if (node != null)
{
it.JIssn = node.InnerText;
}
else
{
it.JIssn = "";
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/Volume");
if (node != null)
{
it.Volume = node.InnerText;
}
else
{
it.Volume = "";
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/Issue");
if (node != null)
{
it.Issue = node.InnerText;
}
else
{
it.Issue = "";
}
//if (it.Volssue.Length > 98)
// it.Volssue = it.Volssue.Substring(0, 90);
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/PubDate/Year");
if (node != null)
{
it.Pubdate = node.InnerText;
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/PubDate/Month");
if (node != null)
{
it.Pubdate = it.Pubdate + "," + node.InnerText;
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Journal/JournalIssue/PubDate/Day");
if (node != null)
{
it.Pubdate += " " + node.InnerText;
}
XmlNodeList authorNodelist = articleNode.SelectNodes("MedlineCitation/Article/AuthorList/Author");
if (authorNodelist == null)
{
it.Authors = "";
}
else if (authorNodelist.Count == 0)
{
it.Authors = "";
}
else
{
string authors = "";
foreach (XmlNode authorNode in authorNodelist)
{
node = authorNode.SelectSingleNode("./LastName");
string lastname = "", forename = "";
if (node != null)
{
lastname = node.InnerText;
}
node = authorNode.SelectSingleNode("./ForeName");
if (node != null)
{
forename = node.InnerText;
}
authors += forename + " " + lastname + ",";
}
authors = authors.Substring(0, authors.Length - 1);
it.Authors = authors;
/*string author = "";
* node = nodelist[0].SelectSingleNode("./LastName");
* if (node != null)
* author = node.InnerText;
* node = nodelist[0].SelectSingleNode("./Initials");
* if (node != null)
* author = author + ", " + node.InnerText;
*
* it.Authors = author;
*
* if (nodelist.Count > 1)
* {
* author = " and ";
* node = nodelist[1].SelectSingleNode("./LastName");
* if (node != null)
* author += node.InnerText;
* node = nodelist[1].SelectSingleNode("./Initials");
*
* if (node != null)
* author = author + ", " + node.InnerText;
*
* it.Authors += author;
* }*/
}
node = articleNode.SelectSingleNode("MedlineCitation/Article/Pagination/MedlinePgn");
if (node != null)
{
it.Pagination = node.InnerText;
}
else
{
it.Pagination = "";
}
node = articleNode.SelectSingleNode("PubmedData/ArticleIdList/ArticleId[@IdType='doi']");
if (node != null)
{
it.Doi = node.InnerText;
}
XmlNodeList pubTypeNodeList = articleNode.SelectNodes("MedlineCitation/Article/PublicationTypeList/PublicationType");
if (pubTypeNodeList == null)
{
it.PublicationType = "";
}
else if (pubTypeNodeList.Count < 1)
{
it.PublicationType = "";
}
else
{
string pubtype = "";
foreach (XmlNode pubTypeNode in pubTypeNodeList)
{
pubtype += pubTypeNode.InnerText + "|";
}
pubtype = pubtype.Substring(0, pubtype.Length - 1);
it.PublicationType = pubtype;
}
pmItemList.Add(it);
}
return(pmItemList);
}
