public void JFK()
{
var jfk = new HumanName("president john 'jack' fitzgerald kennedy");
Assert.AreEqual("president", jfk.Title);
Assert.AreEqual("john", jfk.First);
Assert.AreEqual("fitzgerald", jfk.Middle);
Assert.AreEqual("kennedy", jfk.Last);
Assert.AreEqual(string.Empty, jfk.Suffix);
Assert.AreEqual("jack", jfk.Nickname);
Assert.AreEqual("president john fitzgerald kennedy", jfk.FullName);
Assert.AreEqual("kennedy", jfk.LastBase);
Assert.AreEqual(string.Empty, jfk.LastPrefixes);
jfk.Normalize();
Assert.AreEqual("President", jfk.Title);
Assert.AreEqual("John", jfk.First);
Assert.AreEqual("Fitzgerald", jfk.Middle);
Assert.AreEqual("Kennedy", jfk.Last);
Assert.AreEqual(string.Empty, jfk.Suffix);
Assert.AreEqual("Jack", jfk.Nickname);
Assert.AreEqual("President John Fitzgerald Kennedy", jfk.FullName);
Assert.AreEqual("Kennedy", jfk.LastBase);
Assert.AreEqual(string.Empty, jfk.LastPrefixes);
}
public void DifferentInputsSameValues()
{
var fml = new HumanName("john x smith");
var lfm = new HumanName("smith, john x");
Assert.IsTrue(fml == lfm);
}
public void BlankInput()
{
var parsed = new HumanName(string.Empty);
Assert.AreEqual(string.Empty, parsed.First);
Assert.AreEqual(string.Empty, parsed.Middle);
Assert.AreEqual(string.Empty, parsed.Last);
Assert.AreEqual(string.Empty, parsed.Title);
Assert.AreEqual(string.Empty, parsed.Nickname);
Assert.AreEqual(string.Empty, parsed.Suffix);
}
public void LastBaseAndPrefixes()
{
var parsed = new HumanName("John Smith");
Assert.AreEqual("Smith", parsed.Last);
Assert.AreEqual(string.Empty, parsed.LastPrefixes);
Assert.AreEqual("Smith", parsed.LastBase);
parsed = new HumanName("johannes van der waals");
Assert.AreEqual("johannes", parsed.First);
Assert.AreEqual("van der", parsed.LastPrefixes); // specifically, the prefixes to the last name
Assert.AreEqual("waals", parsed.LastBase); // only the base component of the last name
Assert.AreEqual("van der waals", parsed.Last); // the full last name, combined
parsed.Normalize();
Assert.AreEqual("Johannes", parsed.First);
Assert.AreEqual("van der", parsed.LastPrefixes);
Assert.AreEqual("Waals", parsed.LastBase);
Assert.AreEqual("van der Waals", parsed.Last);
}
/// <summary>
/// The main parse method for the parser. This method is run upon assignment to the
/// fullName attribute or instantiation.
///
/// Basic flow is to hand off to `pre_process` to handle nicknames. It
/// then splits on commas and chooses a code path depending on the number of commas.
/// `parsePieces` then splits those parts on spaces and
/// `joinOnConjunctions` joins any pieces next to conjunctions.
/// </summary>
private void ParseFullName()
{
if (ParseMultipleNames)
{
if (_FullName.Contains('&'))
{
var split = _FullName.IndexOf('&');
var primary = _FullName.Substring(0, split);
var secondary = _FullName.Substring(split + 1);
AdditionalName = new HumanName(secondary);
_FullName = primary;
}
else if (_FullName.ToLowerInvariant().Contains(" and "))
{
var split = _FullName.IndexOf(" and ", StringComparison.InvariantCultureIgnoreCase);
var primary = _FullName.Substring(0, split);
var secondary = _FullName.Substring(split + 5 /* length of " and " */);
AdditionalName = new HumanName(secondary);
_FullName = primary;
}
}
ParseNicknames(ref _FullName, out _NicknameList);
// break up fullName by commas
var parts = _FullName
.Split(new[] { ',' }, StringSplitOptions.RemoveEmptyEntries)
.Select(part => part.Trim())
.ToList();
if (parts.Count == 0)
{
// Edge case where the input was all in parens and has become the nickname
// See https://github.com/aeshirey/NameParserSharp/issues/8
}
else if (parts.Count == 1)
{
// no commas, title first middle middle middle last suffix
// part[0]
var pieces = ParsePieces(parts);
for (var i = 0; i < pieces.Length; i++)
{
var piece = pieces[i];
var nxt = i == pieces.Length - 1 ? string.Empty : pieces[i + 1];
// title must have a next piece, unless it's just a title
if (IsTitle(piece) && (!string.IsNullOrEmpty(nxt) || pieces.Length == 1))
{
// some last names appear as titles (https://github.com/aeshirey/NameParserSharp/issues/9)
// if we've already parsed out titles, first, or middle names, something appearing as a title may in fact be a last name
if (_FirstList.Count > 0 || _MiddleList.Count > 0)
{
_LastList.Add(piece);
}
else
{
_TitleList.Add(piece);
}
}
else if (string.IsNullOrEmpty(First))
{
_FirstList.Add(piece);
}
else if (AreSuffixes(pieces.Skip(i + 1)))
{
_LastList.Add(piece);
_SuffixList = _SuffixList.Concat(pieces.Skip(i + 1)).ToList();
break;
}
else if (!string.IsNullOrEmpty(nxt))
{
// another component exists, so this is likely a middle name
_MiddleList.Add(piece);
}
else if (!ParseMultipleNames || AdditionalName == null)
{
// no additional name. some last names can appear to be suffixes. try to figure this out
if (_LastList.Count > 0 && IsSuffix(piece))
{
_SuffixList.Add(piece);
}
else
{
_LastList.Add(piece);
}
}
else if (AdditionalName._LastList.Any() && IsAnInitial(piece))
{
// the additional name has a last, and this one looks like a middle. we'll save as a middle and later will copy the last name.
_MiddleList.Add(piece);
}
else
{
_LastList.Add(piece);
}
}
}
else if (AreSuffixes(parts[1].Split(' ')))
{
// suffix comma: title first middle last [suffix], suffix [suffix] [, suffix]
// parts[0], parts[1:...]
_SuffixList = _SuffixList.Concat(parts.Skip(1)).ToList();
var pieces = ParsePieces(parts[0].Split(new[] { ' ' }));
for (var i = 0; i < pieces.Length; i++)
{
var piece = pieces[i];
var nxt = i == pieces.Length - 1 ? string.Empty : pieces[i + 1];
if (IsTitle(piece) && (!string.IsNullOrEmpty(nxt) || pieces.Length == 1))
{
_TitleList.Add(piece);
continue;
}
if (string.IsNullOrEmpty(First))
{
_FirstList.Add(piece);
continue;
}
if (AreSuffixes(pieces.Skip(i + 1)))
{
_LastList.Add(piece);
_SuffixList = pieces.Skip(i + 1).Concat(_SuffixList).ToList();
break;
}
// correct for when we have "John D" with an AdditionalName={Catherine, T, MacArthur}.
// We should not see this as being First=John, Last=D; rather, First=John, Middle=D, Last=<AdditionalName.Last>
if (!string.IsNullOrEmpty(nxt))
{
// another component exists, so this is likely a middle name
_MiddleList.Add(piece);
}
else if (!ParseMultipleNames || AdditionalName == null)
{
// no additional name, so treat this as a last
_LastList.Add(piece);
}
else if (AdditionalName._LastList.Any() && IsAnInitial(piece))
{
// the additional name has a last, and this one looks like a middle. we'll save as a middle and later will copy the last name.
_MiddleList.Add(piece);
}
else
{
_LastList.Add(piece);
}
}
}
else
{
// lastname comma: last [suffix], title first middles[,] suffix [,suffix]
// parts[0], parts[1], parts[2:...]
var pieces = ParsePieces(parts[1].Split(new[] { ' ' }), 1);
// lastname part may have suffixes in it
var lastnamePieces = ParsePieces(parts[0].Split(new[] { ' ' }), 1);
foreach (var piece in lastnamePieces)
{
// the first one is always a last name, even if it look like a suffix
if (IsSuffix(piece) && _LastList.Any())
{
_SuffixList.Add(piece);
}
else
{
_LastList.Add(piece);
}
}
for (var i = 0; i < pieces.Length; i++)
{
var piece = pieces[i];
var nxt = i == pieces.Length - 1 ? string.Empty : pieces[i + 1];
if (IsTitle(piece) && (!string.IsNullOrEmpty(nxt) || pieces.Length == 1))
{
_TitleList.Add(piece);
}
else if (string.IsNullOrEmpty(First))
{
_FirstList.Add(piece);
}
else if (IsSuffix(piece))
{
_SuffixList.Add(piece);
}
else
{
_MiddleList.Add(piece);
}
}
if (parts.Count() >= 3 && !string.IsNullOrEmpty(parts[2]))
{
_SuffixList = _SuffixList.Concat(parts.Skip(2)).ToList();
}
}
IsUnparsable = !_TitleList.Any() &&
!_FirstList.Any() &&
!_MiddleList.Any() &&
!_LastList.Any() &&
!_SuffixList.Any() &&
!_NicknameList.Any();
PostProcessFirstnames();
PostProcessLastname();
PostProcessAdditionalName();
}
public void NicknameAtBeginning_SingleQuote()
{
var parsed = new HumanName("'TREY' ROBERT HENRY BUSH III");
Assert.AreEqual(parsed.First, "ROBERT");
Assert.AreEqual(parsed.Middle, "HENRY");
Assert.AreEqual(parsed.Last, "BUSH");
Assert.AreEqual(parsed.Nickname, "TREY");
Assert.AreEqual(parsed.Suffix, "III");
}
public void TwoNames_TitleFirstInitialMiddleInitialLast()
{
HumanName.ParseMultipleNames = true;
var parsed = new HumanName("Mr S R Bloggs and Miss L B Jones");
Assert.AreEqual("Mr", parsed.Title);
Assert.AreEqual("S", parsed.First);
Assert.AreEqual("R", parsed.Middle);
Assert.AreEqual("Bloggs", parsed.Last);
Assert.IsNotNull(parsed.AdditionalName);
Assert.AreEqual("Miss", parsed.AdditionalName.Title);
Assert.AreEqual("L", parsed.AdditionalName.First);
Assert.AreEqual("B", parsed.AdditionalName.Middle);
Assert.AreEqual("Jones", parsed.AdditionalName.Last);
Assert.IsNull(parsed.AdditionalName.AdditionalName);
}
public void TwoNames_MacAthur()
{
HumanName.ParseMultipleNames = true;
var parsed = new HumanName("John D. and Catherine T. MacArthur");
Assert.AreEqual("John", parsed.First);
Assert.AreEqual("D.", parsed.Middle);
Assert.AreEqual("MacArthur", parsed.Last);
Assert.IsNotNull(parsed.AdditionalName);
Assert.AreEqual("Catherine", parsed.AdditionalName.First);
Assert.AreEqual("T.", parsed.AdditionalName.Middle);
Assert.AreEqual("MacArthur", parsed.AdditionalName.Last);
Assert.IsNull(parsed.AdditionalName.AdditionalName);
}
public void TwoCommaWithMiddleName()
{
var parsed = new HumanName("Surname, John Middle, III");
Assert.AreEqual(parsed.First, "John");
Assert.AreEqual(parsed.Middle, "Middle");
Assert.AreEqual(parsed.Last, "Surname");
Assert.AreEqual(parsed.Suffix, "III");
}
public void TitleFirstOrLastName()
{
var mrJones = new HumanName("Mr. Jones");
Assert.AreEqual("Mr.", mrJones.Title);
Assert.AreEqual(string.Empty, mrJones.First);
Assert.AreEqual(string.Empty, mrJones.Middle);
Assert.AreEqual("Jones", mrJones.Last);
Assert.AreEqual(string.Empty, mrJones.Suffix);
Assert.AreEqual(string.Empty, mrJones.Nickname);
Assert.AreEqual("Jones", mrJones.LastBase);
Assert.AreEqual(string.Empty, mrJones.LastPrefixes);
var uncleAdam = new HumanName("Uncle Adam");
Assert.AreEqual("Uncle", uncleAdam.Title);
Assert.AreEqual("Adam", uncleAdam.First);
Assert.AreEqual(string.Empty, uncleAdam.Middle);
Assert.AreEqual(string.Empty, uncleAdam.Last);
Assert.AreEqual(string.Empty, uncleAdam.Suffix);
Assert.AreEqual(string.Empty, uncleAdam.Nickname);
Assert.AreEqual(string.Empty, uncleAdam.LastBase);
Assert.AreEqual(string.Empty, uncleAdam.LastPrefixes);
}
public void ThreeNames()
{
HumanName.ParseMultipleNames = true;
var johnSmith = new HumanName("Mr John Smith and Mrs Jane Doe and President Abraham Lincoln");
Assert.IsNotNull(johnSmith.AdditionalName);
var janeDoe = johnSmith.AdditionalName;
Assert.IsNotNull(janeDoe.AdditionalName);
var abrahamLincoln = janeDoe.AdditionalName;
Assert.AreEqual("Mr", johnSmith.Title);
Assert.AreEqual("John", johnSmith.First);
Assert.AreEqual("Smith", johnSmith.Last);
Assert.AreEqual("Mrs", janeDoe.Title);
Assert.AreEqual("Jane", janeDoe.First);
Assert.AreEqual("Doe", janeDoe.Last);
Assert.AreEqual("President", abrahamLincoln.Title);
Assert.AreEqual("Abraham", abrahamLincoln.First);
Assert.AreEqual("Lincoln", abrahamLincoln.Last);
}
public void NullInput()
{
var parsed = new HumanName(null);
}
public void Nixon()
{
var nixon = new HumanName("mr president richard (dick) nixon");
Assert.AreEqual("mr president", nixon.Title);
Assert.AreEqual("richard", nixon.First);
Assert.AreEqual(string.Empty, nixon.Middle);
Assert.AreEqual("nixon", nixon.Last);
Assert.AreEqual(string.Empty, nixon.Suffix);
Assert.AreEqual("dick", nixon.Nickname);
Assert.AreEqual("mr president richard nixon", nixon.FullName);
Assert.AreEqual("nixon", nixon.LastBase);
Assert.AreEqual(string.Empty, nixon.LastPrefixes);
nixon.Normalize();
Assert.AreEqual("Mr President", nixon.Title);
Assert.AreEqual("Richard", nixon.First);
Assert.AreEqual(string.Empty, nixon.Middle);
Assert.AreEqual("Nixon", nixon.Last);
Assert.AreEqual(string.Empty, nixon.Suffix);
Assert.AreEqual("Dick", nixon.Nickname);
Assert.AreEqual("Mr President Richard Nixon", nixon.FullName);
Assert.AreEqual("Nixon", nixon.LastBase);
Assert.AreEqual(string.Empty, nixon.LastPrefixes);
}
