void DivideNameAndOccupation()
{
var nameCell = GetDeclarationField(DeclarationField.NameAndOccupationOrRelativeType);
NameDocPosition = adapter.GetDocumentPosition(GetRowIndex(), nameCell.Col);
string v = nameCell.GetText(true);
if (DataHelper.IsEmptyValue(v))
{
return;
}
if (DataHelper.IsRelativeInfo(v))
{
SetRelative(v);
}
else
{
string pattern = @"\s+\p{Pd}\s+"; // UnicodeCategory.DashPunctuation
v = Regex.Replace(v, @"\d+\.\s+", "");
string[] two_parts = Regex.Split(v, pattern);
string clean_v = Regex.Replace(v, pattern, " ");
string[] words = Regex.Split(clean_v, @"[\,\s\n]+");
if (words.Length >= 3 && TextHelpers.CanBePatronymic(words[2]) &&
!TextHelpers.MayContainsRole(words[0]) &&
!TextHelpers.MayContainsRole(words[1]))
{
// ex: "Рутенберг Дмитрий Анатольевич начальник управления"
PersonName = String.Join(" ", words.Take(3)).Trim();
Occupation = String.Join(" ", words.Skip(3)).Trim();
}
else if (TextHelpers.CanBePatronymic(words.Last()))
{
// ex: "начальник управления Рутенберг Дмитрий Анатольевич"
PersonName = String.Join(" ", words.Skip(words.Length - 3)).Trim();
Occupation = String.Join(" ", words.Take(words.Length - 3)).Trim();
}
else if (words.Length >= 2 && TextHelpers.CanBeInitials(words[1]) && TextHelpers.MayContainsRole(String.Join(" ", words.Skip(2)).Trim()))
{
// ex: "Головачева Н.В., заместитель"
PersonName = String.Join(" ", words.Take(2)).Trim();
Occupation = String.Join(" ", words.Skip(2)).Trim();
}
else if (two_parts.Length == 2)
{
PersonName = two_parts[0].Trim();
Occupation = String.Join(" - ", two_parts.Skip(1)).Trim();
}
else
{
throw new SmartParserException(
string.Format("Cannot parse name+occupation value {0} at row {1}", v, GetRowIndex()));
}
}
}
bool DivideNameAndOccupation(Cell nameCell)
{
NameDocPosition = adapter.GetDocumentPosition(GetRowIndex(), nameCell.Col);
string v = nameCell.GetText(true);
if (DataHelper.IsEmptyValue(v))
{
return(true);
}
if (DataHelper.IsRelativeInfo(v))
{
SetRelative(v);
}
else
{
string pattern = @"\s+\p{Pd}\s+"; // UnicodeCategory.DashPunctuation
v = Regex.Replace(v, @"\d+\.\s+", "");
string[] two_parts = Regex.Split(v, pattern);
string clean_v = Regex.Replace(v, pattern, " ");
string[] words = Regex.Split(clean_v, @"[\,\s\n]+");
if (words.Length >= 3 && TextHelpers.CanBePatronymic(words[2]) &&
!TextHelpers.MayContainsRole(words[0]) &&
!TextHelpers.MayContainsRole(words[1]))
{
// ex: "Рутенберг Дмитрий Анатольевич начальник управления"
PersonName = String.Join(" ", words.Take(3)).Trim();
Occupation = String.Join(" ", words.Skip(3)).Trim();
}
else if (TextHelpers.CanBePatronymic(words.Last()))
{
// ex: "начальник управления Рутенберг Дмитрий Анатольевич"
PersonName = String.Join(" ", words.Skip(words.Length - 3)).Trim();
Occupation = String.Join(" ", words.Take(words.Length - 3)).Trim();
}
else if (Regex.Match(v, @"\w\.\w\.,").Success)
{
// ex: "Головачева Н.В., заместитель"
var match = Regex.Match(v, @"\w\.\w\.,");
PersonName = v.Substring(0, match.Index + match.Length - 1).Trim();
Occupation = v.Substring(match.Index + match.Length).Trim();
}
else if (words.Length >= 2 && TextHelpers.CanBeInitials(words[1]) &&
TextHelpers.MayContainsRole(String.Join(" ", words.Skip(2)).Trim()))
{
// ex: "Головачева Н.В., заместитель"
PersonName = String.Join(" ", words.Take(2)).Trim();
Occupation = String.Join(" ", words.Skip(2)).Trim();
}
else if (two_parts.Length == 2)
{
PersonName = two_parts[0].Trim();
Occupation = String.Join(" - ", two_parts.Skip(1)).Trim();
}
else
{
// maybe PDF has split cells (table on different pages)
// example file: "5966/14 Upravlenie delami.pdf" converted to docx
Logger.Error(string.Format("Cannot parse name+occupation value {0} at row {1}", v, GetRowIndex()));
return(false);
}
}
return(true);
}