/// <summary>
/// Extrae los metadatos del documento
/// </summary>
public override FileMetadata AnalyzeFile()
{
try
{
this.foundMetadata = new FileMetadata();
using (PdfDocument doc = PdfReader.Open(this.fileStream, PdfDocumentOpenMode.InformationOnly))
{
int imageNumber = 0;
//Read embedded images
foreach (PdfDictionary item in doc.Internals.GetAllObjects().Where(p => p is PdfDictionary d && d.Stream != null && "/Image".Equals(d.Elements["/Subtype"]?.ToString())))
{
try
{
using (MemoryStream msJPG = new MemoryStream(item.Stream.Value))
{
using (EXIFDocument eDoc = new EXIFDocument(msJPG))
{
FileMetadata exifMetadata = eDoc.AnalyzeFile();
//Ignore images which only contain 'Adobe JPEG' makernotes
if (exifMetadata != null && exifMetadata.HasMetadata() && !exifMetadata.Makernotes.All(p => p.Key == "Adobe JPEG"))
{
foundMetadata.EmbeddedImages.Add(imageNumber.ToString(), exifMetadata);
imageNumber++;
this.foundMetadata.AddRange(exifMetadata.Users.ToArray());
this.foundMetadata.AddRange(exifMetadata.Applications.ToArray());
}
}
}
}
catch (Exception)
{
}
}
ReadXMPMetadata(doc);
if (doc.Info.Title != string.Empty)
{
this.foundMetadata.Title = Functions.ToPlainText(doc.Info.Title);
if (Uri.IsWellFormedUriString(doc.Info.Title, UriKind.Absolute))
{
this.foundMetadata.Add(new Diagrams.Path(PathAnalysis.CleanPath(doc.Info.Title), true));
}
}
if (doc.Info.Subject != string.Empty)
{
this.foundMetadata.Subject = Functions.ToPlainText(doc.Info.Subject);
}
if (doc.Info.Author != string.Empty)
{
this.foundMetadata.Add(new User(Functions.ToPlainText(doc.Info.Author), true));
}
if (doc.Info.Keywords != string.Empty)
{
this.foundMetadata.Keywords = Functions.ToPlainText(doc.Info.Keywords);
}
if (doc.Info.Creator != string.Empty)
{
string strSoftware = ApplicationAnalysis.GetApplicationsFromString(Functions.ToPlainText(doc.Info.Creator));
if (strSoftware.Trim() != string.Empty)
{
this.foundMetadata.Add(new Application(strSoftware));
}
//No se ha localizado ninguna aplicación conocida, aun así mostrar la aplicación encontrada
else if (!String.IsNullOrWhiteSpace(Functions.ToPlainText(doc.Info.Creator)))
{
this.foundMetadata.Add(new Application(Functions.ToPlainText(doc.Info.Creator).Trim()));
}
}
if (!String.IsNullOrWhiteSpace(doc.Info.Producer))
{
string strSoftware = ApplicationAnalysis.GetApplicationsFromString(Functions.ToPlainText(doc.Info.Producer));
if (!String.IsNullOrWhiteSpace(strSoftware))
{
this.foundMetadata.Add(new Application(strSoftware));
}
//No se ha localizado ninguna aplicación conocida, aun así mostrar la aplicación encontrada
else if (!String.IsNullOrWhiteSpace(Functions.ToPlainText(doc.Info.Producer)))
{
this.foundMetadata.Add(new Application(Functions.ToPlainText(doc.Info.Producer).Trim()));
}
}
try
{
if (doc.Info.CreationDate != DateTime.MinValue)
{
this.foundMetadata.Dates.CreationDate = doc.Info.CreationDate;
}
}
catch (InvalidCastException)
{
}
try
{
if (doc.Info.ModificationDate != DateTime.MinValue)
{
this.foundMetadata.Dates.ModificationDate = doc.Info.ModificationDate;
}
}
catch (InvalidCastException)
{
}
}
SearchPathsLinksAndEmails(this.fileStream);
//Find users in paths
foreach (Diagrams.Path path in this.foundMetadata.Paths)
{
string strUser = PathAnalysis.ExtractUserFromPath(path.Value);
this.foundMetadata.Add(new User(strUser, path.IsComputerFolder));
}
//Also search software in the title (only pdf). It is added only if the software is known.
if (!String.IsNullOrEmpty(foundMetadata.Title))
{
string strSoftware = ApplicationAnalysis.GetApplicationsFromString(foundMetadata.Title);
if (!String.IsNullOrWhiteSpace(strSoftware))
{
this.foundMetadata.Add(new Application(strSoftware));
}
}
}
catch (PdfReaderException)
{ }
catch (Exception ex)
{
System.Diagnostics.Debug.WriteLine(ex.ToString());
}
finally
{
if (foundMetadata == null)
{
this.foundMetadata = new FileMetadata();
}
if (fileStream != null)
{
this.fileStream.Dispose();
}
}
return(this.foundMetadata);
}
