private XmlDocument GetXmlFromStream(MIMETypedStream stream)
{
if (stream.MIMEType == "text/xml")
{
XmlDocument doc = new XmlDocument();
doc.Load(new MemoryStream(stream.stream));
return(doc);
}
return(null);
}
// retorna o histórico de um pid por ordem decrescente (do mais recente para o mais antigo) da data de registo
public List <Historico> GetHistoric(string pid)
{
List <Historico> result = new List <Historico>();
foreach (Datastream ds in manager.getDatastreamHistory(pid, "METS"))
{
XmlDocument metsDoc;
XmlNamespaceManager nameSpaceManager;
MIMETypedStream stream = service.getDatastreamDissemination(pid, "METS", ds.createDate);
metsDoc = GetXmlFromStream(stream);
nameSpaceManager = new XmlNamespaceManager(metsDoc.NameTable);
nameSpaceManager.AddNamespace("mets", "http://www.loc.gov/METS/");
nameSpaceManager.AddNamespace("xlink", "http://www.w3.org/TR/xlink/");
// Get author and log message from METS file
string user = "";
List <string> logMsg = new List <string>();
XmlNode headerNode = metsDoc.SelectSingleNode("//mets:agent", nameSpaceManager);
if (headerNode != null)
{
if (headerNode.SelectSingleNode("mets:name", nameSpaceManager) != null)
{
user = headerNode.SelectSingleNode("mets:name", nameSpaceManager).InnerText;
}
if (headerNode.SelectNodes("mets:note", nameSpaceManager) != null)
{
foreach (XmlNode node in headerNode.SelectNodes("mets:note", nameSpaceManager))
{
logMsg.Add(node.InnerText);
}
}
}
result.Add(new Historico(ds.createDate, user, logMsg.ToArray()));
}
return(result);
}
private XmlDocument GetXmlFromStream(MIMETypedStream stream) {
if (stream.MIMEType == "text/xml") {
XmlDocument doc = new XmlDocument();
doc.Load(new MemoryStream(stream.stream));
return doc;
}
return null;
}
public ObjDigital GetStructureForPid(string pid, string originalTimestamp, bool loadVersions)
{
XmlDocument metsDoc;
XmlDocument dcDoc;
XmlNamespaceManager nameSpaceManager;
XmlNamespaceManager dcNameSpaceManager;
string versionTimestamp = null;
try {
versionTimestamp = manager.getDatastream(pid, "METS", originalTimestamp).createDate;
MIMETypedStream stream = service.getDatastreamDissemination(pid, "METS", originalTimestamp);
metsDoc = GetXmlFromStream(stream);
nameSpaceManager = new XmlNamespaceManager(metsDoc.NameTable);
nameSpaceManager.AddNamespace("mets", "http://www.loc.gov/METS/");
nameSpaceManager.AddNamespace("xlink", "http://www.w3.org/TR/xlink/");
// Temos de pedir o DC mais próximo da versão mais actualizada do METS (logo passamos o versionTimestamp)
MIMETypedStream dcStream = service.getDatastreamDissemination(pid, "DC", versionTimestamp);
dcDoc = GetXmlFromStream(dcStream);
dcNameSpaceManager = new XmlNamespaceManager(dcDoc.NameTable);
dcNameSpaceManager.AddNamespace("dc", "http://purl.org/dc/elements/1.1/");
dcNameSpaceManager.AddNamespace("oai_dc", "http://www.openarchives.org/OAI/2.0/oai_dc/");
dcNameSpaceManager.AddNamespace("xsi", "http://www.w3.org/2001/XMLSchema-instance");
} catch (Exception ex) { throw new Exception("Datastreams METS e/ou DC inválidos para o objecto " + pid + ".", ex); }
try {
XmlNode structMap = metsDoc.SelectSingleNode("mets:mets/mets:structMap", nameSpaceManager);
XmlNode mainDiv = structMap.SelectSingleNode("mets:div", nameSpaceManager);
XmlNode recordID = metsDoc.SelectSingleNode("mets:mets/mets:metsHdr/mets:altRecordID", nameSpaceManager);
string type;
try { type = structMap.Attributes["TYPE"].Value.ToString().ToUpper(); } catch { type = "PHYSICAL"; }
ObjDigital newDocumentType = null;
if (type == "LOGICAL")
{
newDocumentType = new ObjDigComposto();
newDocumentType.state = State.unchanged;
newDocumentType.serverState = GetState(pid);
ObjDigSimples[] parts = new ObjDigSimples[mainDiv.SelectNodes("mets:div", nameSpaceManager).Count];
foreach (XmlNode subDiv in mainDiv.SelectNodes("mets:div", nameSpaceManager))
{
// Ler objecto directamente
XmlNode fptr = subDiv.SelectSingleNode("mets:fptr", nameSpaceManager);
ObjDigSimples objSimples = GetStructureForPid(fptr.Attributes["FILEID"].Value, originalTimestamp, true) as ObjDigSimples;
parts[int.Parse(subDiv.Attributes["ORDER"].Value) - 1] = objSimples;
}
((ObjDigComposto)newDocumentType).objSimples.AddRange(parts);
}
else
{
newDocumentType = new ObjDigSimples();
newDocumentType.state = State.unchanged;
newDocumentType.serverState = GetState(pid);
Anexo[] parts = new Anexo[mainDiv.SelectNodes("mets:div", nameSpaceManager).Count];
foreach (XmlNode subDiv in mainDiv.SelectNodes("mets:div", nameSpaceManager))
{
XmlNode fptr = subDiv.SelectSingleNode("mets:fptr", nameSpaceManager);
Anexo anexo = new Anexo();
anexo.pid = pid;
anexo.dataStreamID = fptr.Attributes["FILEID"].Value;
anexo.mimeType = subDiv.Attributes["TYPE"].Value;
parts[int.Parse(subDiv.Attributes["ORDER"].Value) - 1] = anexo;
try
{
Datastream data = manager.getDatastream(pid, anexo.dataStreamID, null);
anexo.url = data.location;
anexo.checksum = data.checksum;
}
catch (Exception ex) { Trace.WriteLine(ex.ToString()); throw new Exception("URL de datastream " + anexo.dataStreamID + " no objecto " + pid + " não foi encontrado.", ex); }
}
((ObjDigSimples)newDocumentType).nextDatastreamId = service.listDatastreams(pid, null).Count(ds => ds.ID.Contains("IMG")) + 1;
((ObjDigSimples)newDocumentType).fich_associados.AddRange(parts);
if (loadVersions)
{
((ObjDigSimples)newDocumentType).historico.AddRange(GetHistoric(pid));
}
}
newDocumentType.pid = pid;
newDocumentType.version = versionTimestamp;
newDocumentType.titulo = mainDiv.Attributes["LABEL"].Value;
if (mainDiv.Attributes["TYPE"] != null)
{
newDocumentType.tipologia = mainDiv.Attributes["TYPE"].Value;
}
if (recordID != null) // há objetos que não têm gisa_id
{
newDocumentType.gisa_id = recordID.InnerText;
}
else
{
newDocumentType.gisa_id = "";
}
XmlNodeList assuntos = dcDoc.SelectNodes("oai_dc:dc/dc:subject", dcNameSpaceManager);
if (assuntos.Count > 0)
{
List <string> listAssuntos = new List <string>();
foreach (XmlNode node in assuntos)
{
listAssuntos.Add(node.InnerText);
}
newDocumentType.assuntos = listAssuntos;
}
return(newDocumentType);
}
catch (Exception ex) { Trace.WriteLine(ex.ToString()); throw new Exception("Erro ao analisar objecto " + pid + ".", ex); }
}
