public void doAllCheck(MatchingResult.VerificationStatus vs, int vnm, int vam, MatchingResult result)
{
Console.Out.WriteLine(result.verificationStatus);
Console.Out.WriteLine(result.namePart);
Console.Out.WriteLine(result.addressPart);
Assert.AreEqual(vs, result.verificationStatus);
Assert.AreEqual(vnm, (int) result.nameMatchingResult);
Assert.AreEqual(vam, (int)(result.addressMatchingResult));
}
//other methods
public MatchingResult Match()
{
nameSet = new List<String>();
registrationAddressSet = new AddressSet();
mailAddressSet = new AddressSet();
MatchingResult result = new MatchingResult();
maskedAddressLine = transactionAddressLine.Mask();
transactionAddressLine = transactionAddressLine.PrepareString(WORDSTOREMOVE);
ChangeNullStringToEmpty(leaderEntry);
//Przygotowanie zbioru imion
GetWordsFromField(nameSet,leaderEntry.firstName.PrepareString(WORDSTOREMOVE));
GetWordsFromField(nameSet, leaderEntry.lastName.PrepareString(WORDSTOREMOVE));
ArrangeAddressSet(registrationAddressSet, leaderEntry.street, leaderEntry.postCode, leaderEntry.houseNumber, leaderEntry.apartmentNumber, leaderEntry.city);
ArrangeAddressSet(mailAddressSet, leaderEntry.streetMail, leaderEntry.postCodeMail, leaderEntry.houseNumberMail, leaderEntry.apartmentNumberMail, leaderEntry.cityMail);
MatchName(result);
double registrationAddressMatchingResult;
String registrationAddressMasked = maskedAddressLine;
String registrationAddressLeftovers = MatchAddress(out registrationAddressMatchingResult, registrationAddressSet, ref registrationAddressMasked);
double mailAddressMatchingResult;
String mailAddressMasked = maskedAddressLine;
String mailAddressLeftOvers = MatchAddress(out mailAddressMatchingResult, mailAddressSet, ref mailAddressMasked);
if (registrationAddressMatchingResult > mailAddressMatchingResult)
{
result.addressMatchingResult = registrationAddressMatchingResult;
addressMatchingKind = AddressMatchingKind.REGISTRATION;
transactionAddressLine = registrationAddressLeftovers;
maskedAddressLine = registrationAddressMasked;
}
else
{
result.addressMatchingResult = mailAddressMatchingResult;
addressMatchingKind = AddressMatchingKind.MAIL;
transactionAddressLine = mailAddressLeftOvers;
maskedAddressLine = mailAddressMasked;
}
moreThanTwoWordsLeft = transactionAddressLine.Split(' ').Where(x => x.Length > 1).Count() >= 2;
int maximumNameIndex = maskedAddressLine.LastIndexOf('#');
int minimumAddressIndex = maskedAddressLine.IndexOf('%');
String between = string.Empty;
if (maximumNameIndex > minimumAddressIndex)
{
//Tu bym obniżył współczynnik matchingu bo coś jest nie tak i wysłałbym cały adres
result.namePart = String.Empty;
result.addressPart = String.Empty;
}
else
{
between = originalAddressLine.Mid(maximumNameIndex + 1, minimumAddressIndex - maximumNameIndex - 1); //originalAddressLine.Substring(maximumNameIndex+1, minimumAddressIndex - maximumNameIndex -1);
}
if (!Regex.IsMatch(between, @"[a-zA-Z/d]+")) //w środku nie ma żadnych znaków alfanumerycznych
{
result.namePart = originalAddressLine.Left(maximumNameIndex + 1);
result.addressPart = originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
else // w środku są znaki alfanumeryczne
{
if (!Regex.IsMatch(between, @"[a-zA-Z]+")) //w środku są same cyfry - na pewno nie jest to część imienia i nazwiska
{
result.namePart = originalAddressLine.Left(maximumNameIndex + 1);
result.addressPart = between + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
else if (!Regex.IsMatch(between, @"[/d]+")) //w środku są same litery
{
//sprawdzam czy zaczyna się to od myślnika - zwykle jest to drugi człon imienia i nazwiska
if(Regex.IsMatch(between,@"([A-Za-z0-9]+\.)+[A-Za-z0-9]+"))
{
Regex rx = new Regex(@"([A-Za-z0-9]+\.)+[A-Za-z0-9]+");
MatchCollection mx = rx.Matches(between);
result.namePart = originalAddressLine.Left(maximumNameIndex + 1) + between.Left(mx[mx.Count - 1].Index + mx[mx.Count - 1].Length);
result.addressPart = between.Right(between.Length - mx[mx.Count - 1].Index - mx[mx.Count - 1].Length) + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
//sprawdzam czy zaczyna się to od myślnika - zwykle jest to drugi człon imienia i nazwiska
else if (between.Replace(" ", "").StartsWith("-"))
{
String[] split = between.Split(new char[] { ' ' },StringSplitOptions.RemoveEmptyEntries);
if(split.Count() > 0)
{
result.namePart = originalAddressLine.Left(maximumNameIndex + 1) + split[0];
result.addressPart = String.Join(" ", split.Skip(1).Take(split.Count() - 1).ToArray())+ " " + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
else
{
result.namePart = originalAddressLine.Left(maximumNameIndex + 1) + between;
result.addressPart = originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
}
//sprawdzam czy zaczyna się to od przecinka - zwykle kończy on człon imienia i nazwiska i rozpoczyna adres
else if (between.Replace(" ", "").StartsWith(","))
{
result.namePart = originalAddressLine.Left(maximumNameIndex + 1);
result.addressPart = between + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
else
{
//jak zawiera skrót, to wszystko po lewej do name, od skrótu począwszy do addres
bool abbrFound = false;
String[] skroty = new String[] { "OS", "AL", "PL", "UL","ULICA","OSIEDLE","ALEJA","PLAC" };
foreach (String skrot in skroty)
{
String exact = String.Empty;
if (between.ContainsWord(skrot, ref exact))
{
String temp = between.ReplaceWords("|", skrot);
result.namePart = originalAddressLine.Left(maximumNameIndex + 1) + " " + temp.Left(temp.IndexOf("|"));
result.addressPart = exact + temp.Right(temp.Length - temp.IndexOf("|") - 1) + " " + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
abbrFound = true;
break;
}
}
if (!abbrFound)//w środku same litery, nie zaczyna się od myślnika ani od przecinka - do adresu
{
//STARA OBSŁUGA : DO ADRESU
//result.namePart = originalAddressLine.Left(maximumNameIndex + 1);
//result.addressPart = between + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
//NOWA OBSŁUGA : DO NAZWISKA
result.namePart = originalAddressLine.Left(maximumNameIndex + 1) + between;
result.addressPart = originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
}
}
else
{
// mamy litery i cyfry
// od cyfry począwszy w prawo do adresu
// przed cyframi do imienia
int firstOccurrenceOfDigit = between.IndexOfAny(new char [] {'0','1','2','3','4','5','6','7','8','9'});
result.namePart = originalAddressLine.Left(maximumNameIndex + 1) + between.Left(firstOccurrenceOfDigit);
result.addressPart = between.Right(between.Length - firstOccurrenceOfDigit) + originalAddressLine.Right(originalAddressLine.Length - minimumAddressIndex);
}
}
//TU PRZEROBIC, ABY ZGADZALO SIE Z TYM CO PL WYSYLAL DZIS DO BANKU
if (result.nameMatchingResult == 100 && result.addressMatchingResult == 100)
{
result.verificationStatus = MatchingResult.VerificationStatus.POSITIVE;
//entry.status = 2;
}
else if (result.nameMatchingResult == 100 || (result.nameMatchingResult > 49 && result.addressMatchingResult > 49))
{
result.verificationStatus = MatchingResult.VerificationStatus.TOFURTHERVERIFICATION;
//entry.status = 1;
}
else
{
result.verificationStatus = MatchingResult.VerificationStatus.NEGATIVE;
//entry.status = 3;
}
if (result.verificationStatus == MatchingResult.VerificationStatus.POSITIVE && moreThanTwoWordsLeft)
{
result.verificationStatus = MatchingResult.VerificationStatus.TOFURTHERVERIFICATION;
}
//ostatnia weryfikacja - do tej pory zapewnione jest, że wszystkie człony imienia i nazwiska znajdujące się w polach
//fistName i lastName znajdują się w linii adresowej
//teraz sprawdzam czy wszystkie czlony namePart znajdują się w polach firstName i lastName
if (result.verificationStatus == MatchingResult.VerificationStatus.POSITIVE)
{
Regex rx = new Regex("[A-Za-ząęćóźżśńłĄĘÓŹŻŚŃŁ]+", RegexOptions.CultureInvariant);
foreach (Match mt in rx.Matches(result.namePart))
{
if (!nameSet.Contains(mt.Value.PrepareString()))
{
result.verificationStatus = MatchingResult.VerificationStatus.TOFURTHERVERIFICATION;
break;
}
}
}
//Ostateczne wyczyszczenie - usunięcie wielokrotnych spacji i przycięcie
result.namePart = result.namePart.RemoveMultipleWhiteSpaces().Trim();
result.addressPart = result.addressPart.RemoveMultipleWhiteSpaces().Trim();
return result;
}
private void MatchName(MatchingResult result)
{
double nameElementWeight = nameSet.Count > 0 ? 100.0 / nameSet.Count : 0;
if (nameElementWeight > 0)
{
Regex rx;
foreach (var word in nameSet)
{
rx = new Regex(@"(^|[\s])" + word + @"([\s]|$)", RegexOptions.IgnoreCase);
if (rx.Matches(transactionAddressLine).Count > 0)
{
transactionAddressLine = transactionAddressLine.RemoveWords(word).RemoveMultipleWhiteSpaces();
maskedAddressLine = MaskString(word, '#', maskedAddressLine);
result.nameMatchingResult += nameElementWeight;
}
}
}
else
{
result.nameMatchingResult = 0;
}
}
