public static List <PDFDocument> FindDuplicates(PDFDocument pdf_document, TitleCombinedCache cache)
{
List <PDFDocument> duplicate_pdf_documents = new List <PDFDocument>();
// If a document is already marked as duplicate, then it doesn nt have any duplicates
if (Choices.ReadingStages_DUPLICATE == pdf_document.ReadingStage)
{
return(duplicate_pdf_documents);
}
string title_this = pdf_document.TitleCombined.ToLower();
if (title_this.Length < 128)
{
if (PDFDocument.TITLE_UNKNOWN != title_this)
{
foreach (var entry in cache.Entries)
{
// Don't match the document to itself
if (entry.pdf_document == pdf_document)
{
continue;
}
// Don't match the document to duplicates
if (Choices.ReadingStages_DUPLICATE == entry.pdf_document.ReadingStage)
{
continue;
}
// Check if we are almost similar to the other document
string title_other = entry.title_combined;
if (PDFDocument.TITLE_UNKNOWN != title_other)
{
if ((LetterPairSimilarity.CompareStrings(title_this, title_other) > 0.95) && (StringTools.LewensteinSimilarity(title_this, title_other) > 0.9))
{
duplicate_pdf_documents.Add(entry.pdf_document);
}
}
}
}
}
return(duplicate_pdf_documents);
}
private void FindDuplicates(PDFDocument pdf_document_this)
{
// Invoke the GUI
this.Dispatcher.Invoke(new Action(() =>
{
ClearDuplicates();
}
));
Stopwatch stopwatch = new Stopwatch();
stopwatch.Start();
if (null == pdf_document_this)
{
Logging.Info("Not doing duplicate detection if we don't have a PDFDocument to work with.");
return;
}
string title_this = pdf_document_this.TitleCombined.ToLower();
if (PDFDocument.TITLE_UNKNOWN == title_this)
{
Logging.Info("Not doing duplicate detection for an unknown title.");
return;
}
// Find all the document in the library that have the same title as this document's
TitleCombinedCache cache = new TitleCombinedCache(pdf_document_this.Library.PDFDocuments);
List <PDFDocument> duplicate_pdf_documents = FindDuplicates(pdf_document_this, cache);
Logging.Info("It took {0}ms to run the duplicate detection.", stopwatch.ElapsedMilliseconds);
// Invoke the GUI
this.Dispatcher.Invoke(new Action(() =>
{
RenderDuplicates(duplicate_pdf_documents);
}
));
}
private void FindDuplicates(PDFDocument pdf_document_this)
{
WPFDoEvents.AssertThisCodeIs_NOT_RunningInTheUIThread();
// Invoke the GUI
WPFDoEvents.InvokeInUIThread(() =>
{
ClearDuplicates();
});
Stopwatch stopwatch = Stopwatch.StartNew();
if (null == pdf_document_this)
{
Logging.Info("Not doing duplicate detection if we don't have a PDFDocument to work with.");
return;
}
string title_this = pdf_document_this.TitleCombined.ToLower();
if (Constants.TITLE_UNKNOWN == title_this)
{
Logging.Info("Not doing duplicate detection for an unknown title.");
return;
}
// Find all the document in the library that have the same title as this document's
TitleCombinedCache cache = new TitleCombinedCache(pdf_document_this.LibraryRef.Xlibrary.PDFDocuments);
List <PDFDocument> duplicate_pdf_documents = FindDuplicates(pdf_document_this, cache);
Logging.Info("It took {0}ms to run the duplicate detection.", stopwatch.ElapsedMilliseconds);
// Invoke the GUI
WPFDoEvents.InvokeInUIThread(() =>
{
RenderDuplicates(duplicate_pdf_documents);
}
);
}
