public DocTypeMatchResult GetMatchingDocType(ScanPages scanPages, List<DocTypeMatchResult> listOfPossibleMatches = null)
{
// Get list of types
DocTypeMatchResult bestMatchResult = new DocTypeMatchResult();
var collection_doctypes = GetDocTypesCollection();
MongoCursor<DocType> foundSdf = collection_doctypes.Find(Query.EQ("isEnabled", true));
#if TEST_PERF_GETMATCHINGDOCTYPE
Stopwatch stopWatch1 = new Stopwatch();
Stopwatch stopWatch2 = new Stopwatch();
#endif
foreach (DocType doctype in foundSdf)
{
#if TEST_PERF_GETMATCHINGDOCTYPE
stopWatch1.Start();
#endif
// Check if document matches
DocTypeMatchResult matchResult = CheckIfDocMatches(scanPages, doctype, false, null);
#if TEST_PERF_GETMATCHINGDOCTYPE
stopWatch1.Stop();
stopWatch2.Start();
#endif
// Find the best match
bool bThisIsBestMatch = false;
if (bestMatchResult.matchCertaintyPercent < matchResult.matchCertaintyPercent)
bThisIsBestMatch = true;
else if (bestMatchResult.matchCertaintyPercent == matchResult.matchCertaintyPercent)
if (bestMatchResult.matchFactor < matchResult.matchFactor)
bThisIsBestMatch = true;
// Redo match to get date and time info
if (bThisIsBestMatch)
{
matchResult = CheckIfDocMatches(scanPages, doctype, true, null);
bestMatchResult = matchResult;
}
// Check if this should be returned in the list of best matches
if (listOfPossibleMatches != null)
if ((matchResult.matchCertaintyPercent > 0) || (matchResult.matchFactor > 0))
listOfPossibleMatches.Add(matchResult);
#if TEST_PERF_GETMATCHINGDOCTYPE
stopWatch2.Stop();
#endif
}
#if TEST_PERF_GETMATCHINGDOCTYPE
logger.Info("T1 : {0}ms, T2 : {1}ms", stopWatch1.ElapsedMilliseconds, stopWatch2.ElapsedMilliseconds);
#endif
// If no exact match get date info from entire doc
if (bestMatchResult.matchCertaintyPercent != 100)
{
int bestDateIdx = 0;
List<ExtractedDate> extractedDates = DocTextAndDateExtractor.ExtractDatesFromDoc(scanPages, "", out bestDateIdx);
bestMatchResult.datesFoundInDoc = extractedDates;
if (extractedDates.Count > 0)
bestMatchResult.docDate = extractedDates[bestDateIdx].dateTime;
}
// If list of best matches to be returned then sort that list now
if (listOfPossibleMatches != null)
{
listOfPossibleMatches = listOfPossibleMatches.OrderByDescending(o => o.matchCertaintyPercent).ThenBy(o => o.matchFactor).ToList();
}
return bestMatchResult;
}
public DocTypeMatchResult CheckIfDocMatches(ScanPages scanPages, DocType docType, bool extractDates, List<DocMatchingTextLoc> matchingTextLocs)
{
// Setup check info
DocTypeMatchResult matchResult = new DocTypeMatchResult();
matchResult.matchCertaintyPercent = 0;
matchResult.matchResultCode = DocTypeMatchResult.MatchResultCodes.NOT_FOUND;
if (!docType.isEnabled)
{
matchResult.matchResultCode = DocTypeMatchResult.MatchResultCodes.DISABLED;
return matchResult;
}
if (docType.matchExpression == null)
{
matchResult.matchResultCode = DocTypeMatchResult.MatchResultCodes.NO_EXPR;
return matchResult;
}
// Check the expression
double matchFactorTotal = 0;
if (MatchAgainstDocText(docType.matchExpression, scanPages, ref matchFactorTotal, matchingTextLocs))
{
matchResult.matchCertaintyPercent = 100;
matchResult.matchResultCode = DocTypeMatchResult.MatchResultCodes.FOUND_MATCH;
}
matchResult.docTypeName = docType.docTypeName;
matchResult.matchFactor = matchFactorTotal;
// Extract date
if (extractDates)
{
int bestDateIdx = 0;
List<ExtractedDate> extractedDates = DocTextAndDateExtractor.ExtractDatesFromDoc(scanPages, docType.dateExpression, out bestDateIdx);
matchResult.datesFoundInDoc = extractedDates;
if (extractedDates.Count > 0)
matchResult.docDate = extractedDates[bestDateIdx].dateTime;
}
return matchResult;
}
private void CheckDisplayedDocForMatchAndShowResult()
{
if (_curDocDisplay_scanPages == null)
{
_curDocDisplay_lastMatchResult = null;
return;
}
DocType chkDocType = GetDocTypeFromForm(new DocType());
chkDocType.isEnabled = true;
List<DocMatchingTextLoc> matchingTextLocs = new List<DocMatchingTextLoc>();
DocTypeMatchResult matchRslt = _docTypesMatcher.CheckIfDocMatches(_curDocDisplay_scanPages, chkDocType, true, matchingTextLocs);
_curDocDisplay_lastMatchResult = matchRslt;
DisplayMatchResultForDoc(matchRslt, matchingTextLocs);
}
private void DisplayMatchResultForDoc(DocTypeMatchResult matchRslt, List<DocMatchingTextLoc> matchingTextLocs)
{
List<Brush> exprColrBrushes = new List<Brush>
{
Brushes.DarkMagenta, Brushes.Green, Brushes.Red, Brushes.Orange, Brushes.Purple, Brushes.Peru, Brushes.Purple
};
if (matchRslt.docDate != DateTime.MinValue)
txtDateResult.Text = matchRslt.docDate.ToLongDateString();
else
txtDateResult.Text = "";
// Display match status
string matchFactorStr = String.Format("{0}", (int)matchRslt.matchFactor);
if (matchRslt.matchCertaintyPercent == 100)
{
txtCheckResult.Text = "MATCHES (" + matchFactorStr + "%)";
txtCheckResult.Foreground = Brushes.White;
txtCheckResult.Background = Brushes.Green;
}
else
{
txtCheckResult.Text = "FAILED (" + matchFactorStr + "%)";
txtCheckResult.Foreground = Brushes.White;
txtCheckResult.Background = Brushes.Red;
}
// Display matching text locations
locRectHandler.ClearTextMatchRect("txtMatch");
foreach (DocMatchingTextLoc txtLoc in matchingTextLocs)
{
if (txtLoc.pageIdx+1 == _curDocDisplay_pageNum)
{
DocRectangle inRect = _curDocDisplay_scanPages.scanPagesText[txtLoc.pageIdx][txtLoc.elemIdx].bounds;
Brush colrBrush = exprColrBrushes[txtLoc.exprIdx % exprColrBrushes.Count];
DocRectangle computedLocation = new DocRectangle(inRect.X, inRect.Y, inRect.Width, inRect.Height);
double wid = computedLocation.Width;
double inx = computedLocation.X;
computedLocation.X = inx + txtLoc.posInText * wid / txtLoc.foundInTxtLen;
computedLocation.Width = txtLoc.matchLen * wid / txtLoc.foundInTxtLen;
locRectHandler.DrawTextMatchRect(_curDocDisplay_scanPages.scanPagesText[txtLoc.pageIdx][txtLoc.elemIdx].bounds, colrBrush, "txtMatch");
}
}
}
private void HandleDocMatchingAndDisplay()
{
#if TEST_PERF_SHOWDOCFIRSTTIME
Stopwatch stopWatch1 = new Stopwatch();
Stopwatch stopWatch2 = new Stopwatch();
Stopwatch stopWatch3 = new Stopwatch();
Stopwatch stopWatch4 = new Stopwatch();
Stopwatch stopWatch5 = new Stopwatch();
Stopwatch stopWatch6 = new Stopwatch();
stopWatch1.Start();
#endif
#if TEST_PERF_SHOWDOCFIRSTTIME
stopWatch1.Stop();
stopWatch2.Start();
#endif
// Re-check the document
DocTypeMatchResult latestMatchResult;
List<DocTypeMatchResult> possMatches = new List<DocTypeMatchResult>();
if (_curDocScanPages != null)
latestMatchResult = _docTypesMatcher.GetMatchingDocType(_curDocScanPages, possMatches);
else
latestMatchResult = new DocTypeMatchResult();
// Check for a new doc - so cancel processing this one
if (!_newCurDocProcessingCancel)
{
#if TEST_PERF_SHOWDOCFIRSTTIME
stopWatch2.Stop();
stopWatch3.Start();
#endif
// Update the doc type list view for popup
List<DocTypeMatchResult> possDocMatches = new List<DocTypeMatchResult>();
foreach (DocTypeMatchResult res in possMatches)
possDocMatches.Add(res);
// Check for a new doc - so cancel processing this one
if (!_newCurDocProcessingCancel)
{
#if TEST_PERF_SHOWDOCFIRSTTIME
stopWatch3.Stop();
stopWatch4.Start();
#endif
// Add list of previously used doctypes
List<string> lastUsedDocTypes = _scanDocHandler.GetLastNDocTypesUsed(10);
foreach (string s in lastUsedDocTypes)
{
DocTypeMatchResult mr = new DocTypeMatchResult();
mr.docTypeName = s;
possDocMatches.Add(mr);
}
// Check for a new doc - so cancel processing this one
if (!_newCurDocProcessingCancel)
{
#if TEST_PERF_SHOWDOCFIRSTTIME
stopWatch4.Stop();
stopWatch5.Start();
#endif
this.Dispatcher.BeginInvoke((Action)delegate()
{
_listOfPossibleDocMatches.Clear();
foreach (DocTypeMatchResult dtmr in possDocMatches)
_listOfPossibleDocMatches.Add(dtmr);
});
// Check for a new doc - so cancel processing this one
if (!_newCurDocProcessingCancel)
{
#if TEST_PERF_SHOWDOCFIRSTTIME
stopWatch5.Stop();
stopWatch6.Start();
#endif
// Show type and date
ShowDocumentTypeAndDate(latestMatchResult.docTypeName);
}
}
}
}
#if TEST_PERF_SHOWDOCFIRSTTIME
stopWatch6.Stop();
logger.Info("ShowDocFirstTime: {6} A {0:0.00}, B {1:0.00}, C {2:0.00}, D {3:0.00}, E {4:0.00}, F {5:0.00}", stopWatch1.ElapsedTicks * 1000.0 / Stopwatch.Frequency,
stopWatch2.ElapsedTicks * 1000.0 / Stopwatch.Frequency, stopWatch3.ElapsedTicks * 1000.0 / Stopwatch.Frequency,
stopWatch4.ElapsedTicks * 1000.0 / Stopwatch.Frequency, stopWatch5.ElapsedTicks * 1000.0 / Stopwatch.Frequency,
stopWatch6.ElapsedTicks * 1000.0 / Stopwatch.Frequency, _newCurDocProcessingCancel ? "CANCELLED" : "");
#endif
}
