/// <summary>
/// _s the processing node.
/// </summary>
/// <param name="ProcessingNode">The processing node.</param>
/// <param name="item">The item.</param>
/// <returns></returns>
private bool _ProcessingNode(XmlNode ProcessingNode, XmlNamespaceManager nm, ref ItemMasterRow item)
{
try
{
item.ItemTypeID = ItemTypes.JournalArticle;
XmlNode node = ProcessingNode.SelectSingleNode("//rss:title", nm);
if(node!=null)
item.Title = node.InnerText;
node = ProcessingNode.SelectSingleNode("//rss:link", nm);
if (node != null)
item.Links += "mainLink|" + node.InnerText + "|";
node = ProcessingNode.SelectSingleNode("//rss:description", nm);
if (node != null)
{
string strContent = node.InnerText;
strContent = Regex.Replace(strContent, @"[\n|\t|\r]", "");
strContent = Regex.Replace(strContent, @"<i>.*?</i>", "");
strContent = Regex.Replace(strContent, @"<.*?>", "");
item.Abstract = strContent;
}
node = ProcessingNode.SelectSingleNode("//prism:publicationYear", nm);
if (node != null)
{
item.PubDate = node.InnerText;
item.PubYear = Regex.Match(item.PubDate, @"\d{4}").Value;
}
node = ProcessingNode.SelectSingleNode("//prism:publicationName", nm);
if (node != null)
item.Title2 = node.InnerText;
node = ProcessingNode.SelectSingleNode("//prism:volume", nm);
if (node != null)
item.Volume = node.InnerText;
node = ProcessingNode.SelectSingleNode("//prism:number", nm);
if (node != null)
item.Volume2 = node.InnerText;
node = ProcessingNode.SelectSingleNode("//dc:identifier", nm);
if (node != null)
{
string strContent = node.InnerText;
item.DOI = Regex.Replace(strContent, @"^doi:", "", RegexOptions.IgnoreCase);
}
int count = 0;
foreach(XmlNode sNode in ProcessingNode.SelectNodes("//dc:creator", nm))
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Author;
nmr.SequenceNo = count;
count++;
string strContent = sNode.InnerText;
int index = strContent.LastIndexOf(" ");
nmr.LastName = strContent.Substring(index);
nmr.ForeName = strContent.Substring(0, index);
item.Authors.Add(nmr);
}
node = ProcessingNode.SelectSingleNode("//prism:publisher", nm);
if(node!=null)
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Publisher;
nmr.LastName = node.InnerText;
nmr.ForeName = "";
item.Authors.Add(nmr);
}
return true;
}
catch
{ }
return false;
}
private bool _ProcessBibTexBlock(ref ItemMasterRow item, string strBibTexBlock)
{
_InitializeBibTexParser();
Dictionary<string, string> dataBibTex = new Dictionary<string, string>();
try
{
if (_ParseBibTexBlock(ref dataBibTex, strBibTexBlock))
{
item = new ItemMasterRow();
item.ItemTypeID = ItemTypes.JournalArticle;
if (dataBibTex.ContainsKey("ItemType"))
_ConvertBibTex2WizFolioItemType(ref item, dataBibTex["ItemType"].ToLower());
item.Title = dataBibTex.ContainsKey("Title") ? dataBibTex["Title"] : item.Title;
item.Title2 = dataBibTex.ContainsKey("Title2") ? dataBibTex["Title2"] : item.Title2;
item.Affiliation = dataBibTex.ContainsKey("Affiliation") ? dataBibTex["Affiliation"] : item.Affiliation;
item.Edition = dataBibTex.ContainsKey("Edition") ? dataBibTex["Edition"] : item.Edition;
item.Volume = dataBibTex.ContainsKey("Volume") ? dataBibTex["Volume"] : item.Volume;
item.Volume2 = dataBibTex.ContainsKey("Volume2") ? dataBibTex["Volume2"] : item.Volume2;
item.Abstract = dataBibTex.ContainsKey("Abstract") ? dataBibTex["Abstract"] : item.Abstract;
item.Notes = dataBibTex.ContainsKey("Notes") ? dataBibTex["Notes"] : item.Notes;
item.PubPlace = dataBibTex.ContainsKey("PubPlace") ? dataBibTex["PubPlace"] : item.PubPlace;
item.PubDate = dataBibTex.ContainsKey("PubDate") ? dataBibTex["PubDate"] : item.PubDate;
item.Pages = dataBibTex.ContainsKey("Pages") ? dataBibTex["Pages"] : item.Pages;
item.ID1 = dataBibTex.ContainsKey("ID1") ? dataBibTex["ID1"] : item.ID1;
item.Links = dataBibTex.ContainsKey("Links") ? ("Source|" + dataBibTex["Links"] + "|") : item.Links;
item.Keywords = dataBibTex.ContainsKey("Keywords") ? dataBibTex["Keywords"] : item.Keywords;
if (dataBibTex.ContainsKey("Authors"))
{
string[] Contributors = dataBibTex["Authors"].Split('|');
for (int i = 0; i < Contributors.Length - 1; i = i + 2)
{
NameMasterRow Contributor = new NameMasterRow();
Contributor.LastName = Contributors[i];
Contributor.ForeName = Contributors[i + 1];
Contributor.SequenceNo = i / 2;
item.Authors.Add(Contributor);
}
}
if (item.ItemTypeID == ItemTypes.BookWhole || item.ItemTypeID == ItemTypes.BookChapter)
{
if (dataBibTex.ContainsKey("Editors"))
{
string[] editors = dataBibTex["Editors"].Split('|');
for (int i = 0; i < editors.Length - 1; i = i + 2)
{
NameMasterRow editor = new NameMasterRow();
editor.NameTypeID = NameTypes.Editor;
editor.LastName = editors[i];
editor.ForeName = editors[i + 1];
editor.SequenceNo = i / 2;
item.Authors.Add(editor);
}
}
}
if (item.ItemTypeID == ItemTypes.BookWhole || item.ItemTypeID == ItemTypes.BookChapter || item.ItemTypeID == ItemTypes.Proceeding)
{
if (dataBibTex.ContainsKey("Publisher"))
{
NameMasterRow publisher = new NameMasterRow();
publisher.NameTypeID = NameTypes.Publisher;
publisher.LastName = dataBibTex["Publisher"];
publisher.ForeName = "";
item.Authors.Add(publisher);
}
}
item.Trim();
}
}
catch
{
return false;
}
return true;
}
/// <summary>
/// Return Author name in desired format
/// </summary>
/// <param name="name">NameMasterRow authorname</param>
/// <param name="format">AuthorFormat format</param>
/// <returns></returns>
public static string FormatName(NameMasterRow name, AuthorFormat format)
{
string formatname = string.Empty;
string[] temp = Regex.Split(name.ForeName, @"[\s|-]+");
name.Initials = string.Empty;
for (int i = 0; i < temp.Length; i++)
{
if(Regex.Replace(temp[i], @"\s+", "").Length > 0)
name.Initials += temp[i].Substring(0, 1) + "|";
}
name.Initials = name.Initials.ToUpper();
switch(format)
{
//William Jefferson Clinton
case AuthorFormat.FullNameFirstNameFirst:
formatname = name.ForeName;
if (formatname.Length > 0)
formatname += " " + name.LastName;
else
formatname = name.LastName;
break;
//W. J. Clinton
case AuthorFormat.InitialWithDot:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", ". ");
if (formatname.Length > 0)
formatname += name.LastName;
else
formatname = name.LastName;
break;
//W.J. Clinton
case AuthorFormat.InitialWithDotWithoutSpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", ".");
if (formatname.Length > 0)
formatname += " " + name.LastName;
else
formatname = name.LastName;
break;
//W J Clinton
case AuthorFormat.InitialWithoutDot:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", " ");
if (formatname.Length > 0)
formatname += name.LastName;
else
formatname = name.LastName;
break;
//WJ Clinton
case AuthorFormat.InitialWithoutDotWithoutSpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", "") + " " + name.LastName;
else
formatname = name.LastName;
break;
//Clinton, William Jefferson
case AuthorFormat.FullNameLastNameFirstWithComma:
formatname = name.LastName;
if (name.ForeName.Length > 0)
formatname += ", " + name.ForeName;
break;
//Clinton, W. J.
case AuthorFormat.InitialLastNameFirstWithCommaWithDot:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", ". ");
if (formatname.Length > 0)
formatname = name.LastName + ", " + Regex.Replace(formatname, @"^\s+|\s+$", "");
else
formatname = name.LastName;
break;
//Clinton, W.J.
case AuthorFormat.InitialLastNameFirstWithCommaWithDotWithoutSpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", ".");
if (formatname.Length > 0)
formatname = name.LastName + ", " + formatname;
else
formatname = name.LastName;
break;
//Clinton, W J
case AuthorFormat.InitialLastNameFirstWithCommaWithoutDot:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", " ");
if (formatname.Length > 0)
formatname = name.LastName + ", " + Regex.Replace(formatname, @"^\s+|\s+$", "");
else
formatname = name.LastName;
break;
//Clinton, WJ
case AuthorFormat.InitialLastNameFirstWithCommaWithoutDotWithoutSpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = name.LastName + ", " + Regex.Replace(formatname, @"\|", "");
else
formatname = name.LastName;
break;
//Clinton
case AuthorFormat.LastNameOnly:
formatname = name.LastName;
break;
//Clinton W. J.
case AuthorFormat.InitialLastNameFirstWithoutCommaWithDot:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", ". ");
if (formatname.Length > 0)
formatname = name.LastName + " " + Regex.Replace(formatname, @"^\s+|\s+$", "");
else
formatname = name.LastName;
break;
//Clinton W.J.
case AuthorFormat.InitialLastNameFirstWithoutCommaWithDotWithoutSpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", ".");
if (formatname.Length > 0)
formatname = name.LastName + " " + formatname;
else
formatname = name.LastName;
break;
//Clinton W J
case AuthorFormat.InitialLastNameFirstWithoutCommaWithoutDot:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = Regex.Replace(formatname, @"\|", " ");
if (formatname.Length > 0)
formatname = name.LastName + " " + Regex.Replace(formatname, @"^\s+|\s+$", "");
else
formatname = name.LastName;
break;
//Clinton WJ
case AuthorFormat.InitialLastNameFirstWithoutCommaWithoutDotWithoutSpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname = name.LastName + " " + Regex.Replace(formatname, @"\|", "");
else
formatname = name.LastName;
break;
//William J. Clinton
case AuthorFormat.FirstNameFirstMidNameInitialWithDot:
formatname = name.ForeName + " ";
if(name.Initials.Length > 0)
{
formatname += name.Initials.Substring(0, 1).ToUpper();
}
if (formatname.Length > 0)
formatname += ". ";
formatname += name.LastName;
break;
//Clinton,W. J.
case AuthorFormat.InitialLastNameFirstWithCommaWithDotWithoutSpaceAfterComma:
formatname = name.Initials;
if (formatname.Length > 0)
formatname += Regex.Replace(formatname, @"\|", ". ");
if (formatname.Length > 0)
formatname = name.LastName + "," + Regex.Replace(formatname, @"^\s+|\s+$", "");
else
formatname = name.LastName;
break;
//Clinton,W.J.
case AuthorFormat.InitialLastNameFirstWithCommaWithDotWithoutAnySpace:
formatname = name.Initials;
if (formatname.Length > 0)
formatname += Regex.Replace(formatname, @"\|", ".");
if (formatname.Length > 0)
formatname = name.LastName + "," + formatname;
else
formatname = name.LastName;
break;
}
return formatname;
}
private bool _ProcessingNode(XmlNode ProcessingNode, XmlNamespaceManager nm, ref ItemMasterRow item)
{
try
{
item.ItemTypeID = ItemTypes.BookWhole;
XmlNode node = ProcessingNode.SelectSingleNode("//atom:title", nm);
if (node != null)
item.Title = node.InnerText;
node = ProcessingNode.SelectSingleNode("//atom:link", nm);
if (node != null)
item.Links += "mainLink|" + node.Attributes["href"].Value + "|";
node = ProcessingNode.SelectSingleNode("//atom:summary", nm);
if (node != null)
{
string strContent = node.InnerText;
item.Abstract = strContent;
}
node = ProcessingNode.SelectSingleNode("//oclcterms:recordIdentifier", nm);
if (node != null)
{
string id = node.InnerText;
string strItemDetailsURL = WORLDCAT_ITEMURL.Replace("{identifier}", id);
strItemDetailsURL = strItemDetailsURL.Replace("{key}", WORLDCAT_APIKEY);
XmlDocument xmlItemDoc = new XmlDocument();
xmlItemDoc.Load(strItemDetailsURL);
XmlNamespaceManager nmItem = new XmlNamespaceManager(xmlItemDoc.NameTable);
nmItem.AddNamespace("dc", "http://purl.org/dc/elements/1.1/");
int count = 0;
foreach (XmlNode sNode in xmlItemDoc.SelectNodes("//dc:creator", nmItem))
{
string text = sNode.InnerText;
string[] split = Regex.Split(text, @"\s*\,\s*");
if(split.Length > 1)
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Author;
nmr.SequenceNo = count;
count++;
nmr.LastName = Regex.Replace(split[0], @"^\s+|\s+$|\.\s*$", "");
nmr.ForeName = Regex.Replace(split[1], @"^\s+|\s+$|\.\s*$", "");
item.Authors.Add(nmr);
}
}
foreach (XmlNode sNode in xmlItemDoc.SelectNodes("//dc:contributor", nmItem))
{
string text = sNode.InnerText;
string[] split = Regex.Split(text, @"\s*\,\s*");
if (split.Length > 1)
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Author;
nmr.SequenceNo = count;
count++;
nmr.LastName = Regex.Replace(split[0], @"^\s+|\s+$|\.\s*$", "");
nmr.ForeName = Regex.Replace(split[1], @"^\s+|\s+$|\.\s*$", "");
item.Authors.Add(nmr);
}
}
foreach (XmlNode sNode in xmlItemDoc.SelectNodes("//dc:publisher", nmItem))
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Publisher;
nmr.SequenceNo = count;
count++;
nmr.LastName = Regex.Replace(sNode.InnerText, @"^\s+|\s+$|\.\s*$", "");
item.Authors.Add(nmr);
}
XmlNode tempNode = xmlItemDoc.SelectSingleNode("//dc:date", nmItem);
if(tempNode != null)
{
if (Regex.Match(tempNode.InnerText, @"\d{4}").Success)
{
item.PubYear = item.PubDate = Regex.Match(tempNode.InnerText, @"\d{4}").Value;
}
}
tempNode = xmlItemDoc.SelectSingleNode("//dc:format", nmItem);
if(tempNode != null)
{
if (Regex.Match(tempNode.InnerText, @"\d+\sp\.\s").Success)
{
string temp = Regex.Match(tempNode.InnerText, @"\d+\sp\.\s").Value;
item.Pages = Regex.Match(temp, @"\d+").Value;
}
}
//XmlNodeList listKeyNodes = xmlItemDoc.SelectNodes("//dc:subject", nmItem);
//foreach (XmlNode sNode in listKeyNodes)
//{
// if (sNode.Attributes["xsi:type"].Value.Contains("LCSH"))
// item.Keywords += sNode.InnerText + "|";
//}
}
return true;
}
catch
{ }
return false;
}
/// <summary>
/// _s the fetch item.
/// </summary>
/// <param name="item">The item.</param>
/// <param name="strPMID">The STR PMID.</param>
/// <returns></returns>
private bool _ProcessingNode(XmlDocument xmlDoc, ref ItemMasterRow item)
{
Dictionary<string, string> Data = new Dictionary<string, string>();
item.ItemTypeID = ItemTypes.JournalArticle;
XmlNodeList authorsList = xmlDoc.SelectNodes("//Article/AuthorList/Author");
int count = 0;
foreach (XmlNode author in authorsList)
{
XmlNode lastname = author.SelectSingleNode("LastName");
XmlNode firstname = author.SelectSingleNode("ForeName");
if(lastname != null && firstname != null){
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Author;
nmr.SequenceNo = count;
count++;
nmr.LastName = lastname.InnerText;
nmr.ForeName = firstname.InnerText;
item.Authors.Add(nmr);
}
}
XmlNode node = xmlDoc.SelectSingleNode("//Article/Abstract/AbstractText");
if(node != null)
item.Abstract = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/Affiliation");
if (node != null)
item.Affiliation = node.InnerText;
node = xmlDoc.SelectSingleNode("//PMID");
if (node != null)
item.ID2 = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/ArticleTitle");
if (node != null)
item.Title = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/Journal/ISSN");
if (node != null)
item.ID1 = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/Journal//Volume");
if (node != null)
item.Volume = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/Journal//Issue");
if (node != null)
item.Volume2 = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/Journal//PubDate/Year");
if (node != null)
{
item.PubDate = node.InnerText;
item.PubYear = node.InnerText;
}
node = xmlDoc.SelectSingleNode("//Article/Journal/Title");
if (node != null)
item.Title2 = node.InnerText;
node = xmlDoc.SelectSingleNode("//Article/Pagination/MedlinePgn");
if (node != null)
item.Pages = node.InnerText;
node = xmlDoc.SelectSingleNode("//ArticleIdList/ArticleId[@IdType=doi]");
if (node != null)
item.DOI = node.InnerText;
item.Links += "mainLink|" + PUBMED_LINK.Replace("{pmid}", item.ID2) + "|";
item.Links += "relatedLinks|" + PUBMED_RELATEDLINK.Replace("{pmid}", item.ID2);
return true;
}
/// <summary>
/// _s the processing node.
/// </summary>
/// <param name="ProcessingNode">The processing node.</param>
/// <param name="nm">The nm.</param>
/// <param name="item">The item.</param>
/// <returns></returns>
private bool _ProcessingNode(XmlNode ProcessingNode, XmlNamespaceManager nm, ref ItemMasterRow item)
{
try
{
item.ItemTypeID = ItemTypes.JournalArticle;
XmlNode node = ProcessingNode.SelectSingleNode("//atom:title", nm);
if (node != null)
item.Title = node.InnerText;
node = ProcessingNode.SelectSingleNode("//atom:link", nm);
if (node != null)
item.Links += "mainLink|" + node.Attributes["href"].Value + "|";
nm.AddNamespace("atom", "http://scholarsportal.info/metadata");
node = ProcessingNode.SelectSingleNode("//atom:article/atom:publication-date", nm);
if (node != null)
{
item.PubDate = node.InnerText;
item.PubYear = Regex.Match(item.PubDate, @"\d{4}").Value;
}
node = ProcessingNode.SelectSingleNode("//atom:journal/atom:title", nm);
if (node != null)
item.Title2 = node.InnerText;
node = ProcessingNode.SelectSingleNode("//atom:journal/atom:volume", nm);
if (node != null)
item.Volume = node.InnerText;
node = ProcessingNode.SelectSingleNode("//atom:journal/atom:issue", nm);
if (node != null)
item.Volume2 = node.InnerText;
node = ProcessingNode.SelectSingleNode("//atom:pages", nm);
if (node != null)
{
XmlNode fpageNode = node.SelectSingleNode("atom:fpage", nm);
XmlNode lpageNode = node.SelectSingleNode("atom:lpage", nm);
if (fpageNode != null && lpageNode != null)
{
item.Pages = fpageNode.InnerText + "-" + lpageNode.InnerText;
}
}
node = ProcessingNode.SelectSingleNode("//atom:journal/atom:issn-ppub", nm);
if (node != null)
{
item.ID1 = node.InnerText;
}
int count = 0;
foreach (XmlNode sNode in ProcessingNode.SelectNodes("//atom:article/atom:authors/atom:author", nm))
{
XmlNode givenNameNode = sNode.SelectSingleNode("atom:givenname", nm);
XmlNode surNameNode = sNode.SelectSingleNode("atom:surname", nm);
if (givenNameNode != null && surNameNode != null)
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Author;
nmr.SequenceNo = count;
count++;
nmr.LastName = surNameNode.InnerText;
nmr.ForeName = givenNameNode.InnerText;
item.Authors.Add(nmr);
}
}
node = ProcessingNode.SelectSingleNode("//atom:journal/atom:publisher", nm);
if (node != null)
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameTypeID = NameTypes.Publisher;
nmr.LastName = node.InnerText;
nmr.ForeName = "";
item.Authors.Add(nmr);
}
node = ProcessingNode.SelectSingleNode("//atom:article/atom:abstract", nm);
if (node != null)
{
string strContent = node.InnerText;
item.Abstract = strContent;
}
XmlNodeList nodesKeyword = ProcessingNode.SelectNodes("//atom:article/atom:keywords/atom:keyword", nm);
foreach (XmlNode keyword in nodesKeyword)
{
item.Keywords += keyword.InnerText + "|";
}
return true;
}
catch
{ }
return false;
}
public static List<NameMasterRow> ConvertWithoutNameID(DataView dv)
{
List<NameMasterRow> NameList = new List<NameMasterRow>();
string[] strColumns = TableColumns;
if (dv.Count > 0)
{
for (int i = 0; i < strColumns.Length; i++)
{
if (strColumns[i] == NameMasterColumns.NameID.ToString())
continue;
if (!dv.Table.Columns.Contains(strColumns[i]))
return NameList;
}
}
for (int i = 0; i < dv.Count; i++)
{
NameMasterRow nmr = new NameMasterRow();
nmr.NameID = int.MinValue;
nmr.DisplayName = dv[i][NameMasterColumns.DisplayName.ToString()] + "";
nmr.ForeName = dv[i][NameMasterColumns.ForeName.ToString()] + "";
nmr.Initials = dv[i][NameMasterColumns.Initials.ToString()] + "";
nmr.LastName = dv[i][NameMasterColumns.LastName.ToString()] + "";
nmr.NameTypeID = (NameTypes)int.Parse(dv[i][NameMasterColumns.NameTypeID.ToString()] + "");
NameList.Add(nmr);
}
return NameList;
}