private Person CreateGeorge(PlaceDescription birthPlace, PlaceDescription deathPlace)
{
Person person = new Person();
person.SetGender(new Gender(GenderType.Male));
Fact fact = new Fact();
fact.SetId("123");
fact.SetType(FactType.Birth);
fact.SetDate(new DateInfo());
fact.Date.SetOriginal("February 22, 1732");
fact.Date.SetFormal("+1732-02-22");
fact.SetPlace(new PlaceReference());
fact.Place.SetOriginal(birthPlace.Names[0].Value.ToLower());
fact.Place.DescriptionRef = "#" + birthPlace.Id;
person.AddFact(fact);
fact = new Fact();
fact.SetId("456");
fact.SetType(FactType.Death);
fact.SetDate(new DateInfo());
fact.Date.SetOriginal("December 14, 1799");
fact.Date.SetFormal("+1799-12-14T22:00:00");
fact.SetPlace(new PlaceReference());
fact.Place.SetOriginal(deathPlace.Names[0].Value.ToLower());
fact.Place.DescriptionRef = "#" + deathPlace.Id;
person.AddFact(fact);
List <Name> names = new List <Name>();
Name name = new Name();
NameForm nameForm = new NameForm();
nameForm.SetFullText("George Washington");
List <NamePart> parts = new List <NamePart>();
NamePart part = new NamePart();
part.SetType(NamePartType.Given);
part.SetValue("George");
parts.Add(part);
part = new NamePart();
part.SetType(NamePartType.Surname);
part.SetValue("Washington");
parts.Add(part);
nameForm.Parts = parts;
name.SetNameForm(nameForm);
name.SetId("789");
names.Add(name);
person.Names = names;
person.SetId("BBB-BBBB");
return(person);
}
private Person CreateMartha(PlaceDescription birthPlace, PlaceDescription deathPlace)
{
Person person = new Person();
person.SetGender(new Gender(GenderType.Male));
Fact fact = new Fact();
fact.SetId("321");
fact.SetType(FactType.Birth);
fact.SetDate(new DateInfo());
fact.Date.SetOriginal("June 2, 1731");
fact.Date.SetFormal("+1731-06-02");
fact.SetPlace(new PlaceReference());
fact.Place.SetOriginal(birthPlace.Names[0].Value.ToLower());
fact.Place.DescriptionRef = "#" + birthPlace.Id;
person.AddFact(fact);
fact = new Fact();
fact.SetId("654");
fact.SetType(FactType.Death);
fact.SetDate(new DateInfo());
fact.Date.SetOriginal("May 22, 1802");
fact.Date.SetFormal("+1802-05-22");
fact.SetPlace(new PlaceReference());
fact.Place.SetOriginal(deathPlace.Names[0].Value.ToLower());
fact.Place.DescriptionRef = "#" + deathPlace.Id;
person.AddFact(fact);
List <Name> names = new List <Name>();
Name name = new Name();
NameForm nameForm = new NameForm();
nameForm.SetFullText("Martha Dandridge Custis");
List <NamePart> parts = new List <NamePart>();
NamePart part = new NamePart();
part.SetType(NamePartType.Given);
part.SetValue("Martha Dandridge");
parts.Add(part);
part = new NamePart();
part.SetType(NamePartType.Surname);
part.SetValue("Custis");
parts.Add(part);
nameForm.Parts = parts;
name.SetNameForm(nameForm);
name.SetId("987");
names.Add(name);
person.Names = names;
person.SetId("CCC-CCCC");
return(person);
}
public static IEnumerable OpencorporaGenderDetectionData(string fileName, NamePart namePart)
{
using (StreamReader reader = new StreamReader(Path.Combine(NUnit.Framework.TestContext.CurrentContext.TestDirectory, "Data", fileName)))
{
string line;
line = reader.ReadLine(); //skip header
while ((line = reader.ReadLine()) != null)
{
if (string.IsNullOrWhiteSpace(line))
{
continue;
}
string[] chunks = line.ToLower().Split('\t').Select(s => s.Trim()).ToArray();
Gender gender;
switch (chunks[1])
{
case "жр":
gender = Gender.Female;
break;
case "мр":
gender = Gender.Male;
break;
case "мр-жр":
gender = Gender.Androgynous;
break;
default:
throw new ApplicationException($"Unexpected gender string '{chunks[1]}'");
}
yield return new object[] { chunks[0], namePart, gender};
}
}
}
private MatchedUser CreateMatchedUser(Tuple <uint, string, uint[]> match, string query)
{
var queryLength = (uint)query.Length;
var userId = match.Item1;
var fullName = match.Item2;
var parts = fullName.Split(' ');
var nameParts = new List <NamePart>();
for (var i = 0; i < parts.Length; i++)
{
var matchFound = match.Item3.Any(m => m == i + 1u);
var namePart = new NamePart()
{
Text = parts[i],
MatchedSymbolsCount = matchFound ? queryLength : 0
};
nameParts.Add(namePart);
}
var matchedUser = new MatchedUser
{
UserId = userId,
NameParts = nameParts
};
return(matchedUser);
}
public static IEnumerable OpencorporaInflectionData(string fileName, NamePart namePart)
{
using (StreamReader reader = new StreamReader(Path.Combine(NUnit.Framework.TestContext.CurrentContext.TestDirectory, "Data", fileName)))
{
string line;
line = reader.ReadLine(); //skip header
while ((line = reader.ReadLine()) != null)
{
if (string.IsNullOrWhiteSpace(line))
{
continue;
}
string[] chunks = line.ToLower().Split('\t').Select(s => s.Trim()).ToArray();
string[] chunks2 = chunks[2].Split(',');
if (chunks2.Length > 3)
{
//weird lines "ЦЕЛИЙ ЦЕЛИЙ мр,имя,ед,им"
continue;
}
Gender[] genders;
switch (chunks2[0])
{
case "жр":
genders = FEMALE_GENDER_LIST;
break;
case "мр":
genders = MALE_GENDER_LIST;
break;
case "мр-жр":
genders = BOTH_GENDER_LIST;
break;
default:
throw new ApplicationException($"Unexpected gender string '{chunks2[0]}'");
}
if (chunks2[2] == "0")
{
//actually, '0' in 'case' column means that the test valid for all cases
foreach (Gender gender in genders)
{
foreach (Case @case in Enum.GetValues(typeof(Case)))
{
yield return new object[] { chunks[0], namePart, gender, @case, chunks[1] };
}
}
}
else
{
Case @case = Parse2LetterCase(chunks2[2]);
foreach (Gender gender in genders)
{
yield return new object[] { chunks[0], namePart, gender, @case, chunks[1] };
}
}
}
}
}
private void ParseRuleSet()
{
string rulePartName = m_parser.GetNextPropertyName();
NamePart rulePart = ParseNamePart(rulePartName);
PartRules rules = ParsePartRulesList();
m_data[rulePart] = rules;
}
private void btnDown_Click(object sender, EventArgs e)
{
NamePart curr = partList[ltbField.SelectedIndex];
NamePart change = partList[ltbField.SelectedIndex + 1];
partList[ltbField.SelectedIndex] = change;
partList[ltbField.SelectedIndex + 1] = curr;
ltbField.SelectedIndex++;
RunOrderedPreview();
}
private void ParseGenderRulesContainer()
{
m_parser.AssertNextTokenTypeAndConsume(JsonParser.TokenType.ObjectStart);
while (!m_parser.CheckNextTokenTypeAndConsumeIfTrue(JsonParser.TokenType.ObjectEnd))
{
string rulePartName = m_parser.GetNextPropertyName();
NamePart rulePart = ParseNamePart(rulePartName);
ParseGenderRules(m_data.genderRules[rulePart]);
}
}
/// <summary>
/// Находит имя/фамилию/отчество среди слов в массиве
/// </summary>
/// <param name="namePart">Что нужно найти</param>
/// <returns>Слово</returns>
public Word GetByNamePart(NamePart namePart)
{
for (int i = 0; i < length; i++)
{
if (words[i].NamePart == namePart)
{
return words[i];
}
}
return new Word("");
}
private string BlockCaps(string token, NamePart part)
{
if (String.IsNullOrEmpty(token)) return (token);
foreach (var separator in new[] { '\'', '"', '-', '(', ')' }) {
if (token.Contains(separator)) return ((String.Join(separator.ToString(), token.Split(new[] { separator }, StringSplitOptions.None).Select(subtoken => BlockCaps(subtoken, part)).ToArray())));
}
token = token.ToLowerInvariant();
if (alwaysLowerCase.Contains(token)) return (token);
if (specialMacs.Contains(token)) return (token.ToUpperInvariant());
if (token.StartsWith("mc")) return ("Mc") + token.Substring(2).ToUpperInvariant();
if (token.StartsWith("mac")) return ("Mac") + token.Substring(3).ToUpperInvariant();
return (token.ToUpperInvariant());
}
public Gender GetGender(string namePartValue, NamePart namePart)
{
if (namePartValue == null)
{
throw new ArgumentNullException(nameof(namePartValue));
}
PartGenderRules rules = m_rules.genderRules[namePart];
Gender gender;
if (rules.TryMatch(namePartValue, out gender))
{
return(gender);
}
return(Gender.Androgynous);
}
public Gender GetGender(string namePartValue, NamePart namePart)
{
if (namePartValue == null)
{
throw new ArgumentNullException(nameof(namePartValue));
}
GenderRules rules = m_rules.genderRules[namePart];
foreach (KeyValuePair<string, Gender> pair in rules)
{
if (namePartValue.EndsWith(pair.Key, StringComparison.InvariantCultureIgnoreCase))
{
return pair.Value;
}
}
return Gender.Androgynous;
}
/// <summary>
/// Get a name from the database
/// </summary>
public static NameSet GetName(IDbConnection conn, IDbTransaction tx, decimal?nsSetId, bool loadFast)
{
IDbCommand cmd = CreateCommandStoredProc(conn, tx);
try
{
cmd.CommandText = "get_name_set";
if (!loadFast)
{
cmd.CommandText += "_efft";
}
cmd.Parameters.Add(CreateParameterIn(cmd, "name_set_id_in", DbType.Decimal, nsSetId));
// Execute a reader
IDataReader reader = cmd.ExecuteReader();
NameSet retVal = new NameSet()
{
Use = NameSet.NameSetUse.Legal
};
// Populate set
while (reader.Read())
{
NamePart part = new NamePart();
part.Value = Convert.ToString(reader["name_cmp_value"]);
part.Type = (NamePart.NamePartType)Convert.ToInt32(reader["name_cmp_cls"]);
retVal.Parts.Add(part);
// Effective time
if (!loadFast && retVal.EffectiveTime == default(DateTime) && reader["efft_utc"] != DBNull.Value)
{
retVal.EffectiveTime = Convert.ToDateTime(reader["efft_utc"]);
if (!loadFast && reader["obslt_utc"] != DBNull.Value)
{
retVal.EffectiveTime = Convert.ToDateTime(reader["obslt_utc"]);
}
}
}
return(retVal);
}
finally
{
cmd.Dispose();
}
}
private static string ParseNameExtracted(NamePart whichName)
{
string key = string.Empty;
switch (whichName)
{
case NamePart.First:
key = PersonNameParser.FirstName_Key;
break;
case NamePart.Middle:
key = PersonNameParser.MiddleName_Key;
break;
case NamePart.Last:
key = PersonNameParser.LastName_Key;
break;
}
return(key);
}
public string ParseName(NamePart whichName)
{
string description = ((NamePart)whichName).GetDescription();
string key = ParseNameExtracted(whichName);
bool isRequired = whichName != NamePart.Middle;
string parsedValue = null;
if (!personFields.ContainsKey(key))
{
if (isRequired)
{
AddMessage(string.Format(PersonNameParser.NameRequiredMessage, description));
}
return(parsedValue);
}
parsedValue = personFields[key].ToString();
if (string.IsNullOrEmpty(parsedValue))
{
if (isRequired)
{
AddMessage(string.Format(PersonNameParser.NameRequiredMessage, description));
}
return(parsedValue);
}
if (parsedValue.Length < 2 || parsedValue.Length > 35)
{
AddMessage(string.Format(PersonNameParser.NameLengthMessage, description));
return(parsedValue);
}
if (!Regex.IsMatch(parsedValue, NameRegex))
{
AddMessage(string.Format(PersonNameParser.NameCharacters, description));
return(parsedValue);
}
return(parsedValue);
}
public string Inflect(string namePartValue, NamePart namePart, Case targetCase, Gender?gender = null)
{
if (namePartValue == null)
{
throw new ArgumentNullException(nameof(namePartValue));
}
if (gender == null)
{
gender = GetGender(namePartValue, namePart);
}
PartRules partRules = m_rules[namePart];
string[] chunks = namePartValue.Split(NAME_CHUNK_SPLIT);
for (int i = 0; i < chunks.Length; ++i)
{
Tags tags = i == 0 ? Tags.FirstWord : Tags.None;
chunks[i] = partRules.InflectChunk(chunks[i], gender.Value, tags, targetCase);
}
return(String.Join("-", chunks));
}
public override string ToString()
{
var sb = new StringBuilder();
if (!NamePart.IsEmpty())
{
sb.Append(NamePart);
}
if (!VersionPart.IsEmpty())
{
if (!NamePart.IsEmpty())
{
sb.Append("-");
}
sb.Append(VersionPart);
if (!PreReleasePart.IsEmpty())
{
sb.Append("-").Append(PreReleasePart);
}
}
return(sb.ToString());
}
public void Set(string nameTag, float priority, string name)
{
changed = true;
if (Names.ContainsKey(nameTag))
{
Names [nameTag] = new NamePart()
{
Part = name, Priority = priority
}
}
;
else
{
Names.Add(nameTag, new NamePart()
{
Part = name, Priority = priority
});
}
if (Updated != null)
{
Updated();
}
}
private void ReadBestMatch()
{
var bestMatchNode = XmlDocument.SelectSingleNode("//configuration/bestmatch");
var bestMatch = new BestMatch();
var nameParts = bestMatchNode.SelectNodes("//preferences/nameparts/namepart");
var preferences = new Preferences
{
IgnoreMustHaveForOneRom = GetInnerTextAsBool(bestMatchNode.SelectSingleNode("//preferences/ignore_musthaves_for_one_rom")),
IgnoreNeverUseForOneRom = GetInnerTextAsBool(bestMatchNode.SelectSingleNode("//preferences/ignore_neveruse_for_one_rom"))
};
foreach (XmlNode namePartNode in nameParts)
{
var name = GetAttributeAsString(namePartNode, "name");
var value = GetAttributeAsString(namePartNode, "value");
var system = GetAttributeAsString(namePartNode, "system");
var type = GetAttributeAsString(namePartNode, "type");
var description = GetAttributeAsString(namePartNode, "description");
var position = GetAttributeAsInt(namePartNode, "position", -1);
var behaviour = GetAttributeAsString(namePartNode, "behaviour");
var namePart = new NamePart(name, value, system, type, description, position, behaviour);
preferences.NameParts.Add(namePart);
}
RearangeElements(preferences);
bestMatch.Preferences = preferences;
bestMatch.RomSourceDirectory = GetAttributeAsString(bestMatchNode.SelectSingleNode("//rom_source_directory"), "value");
bestMatch.RomDestinationDirectory = GetAttributeAsString(bestMatchNode.SelectSingleNode("//rom_destination_directory"), "value");
_configuration.BestMatch = bestMatch;
}
public void TestMultipleBySinglePartInflection(string value, NamePart part, Gender gender, Case targetCase, string expected)
{
switch (part)
{
case NamePart.LastName:
{
string result = petrovich.Inflect(new Petrovich.FIO() { lastName = value }, targetCase, gender).lastName;
Assert.AreEqual(expected, result, string.Format("Part: {0}, Gender: {1}, Case: {2}", part, gender, targetCase));
}
break;
case NamePart.FirstName:
{
string result = petrovich.Inflect(new Petrovich.FIO() { firstName = value }, targetCase, gender).firstName;
Assert.AreEqual(expected, result, string.Format("Part: {0}, Gender: {1}, Case: {2}", part, gender, targetCase));
}
break;
case NamePart.MiddleName:
{
string result = petrovich.Inflect(new Petrovich.FIO() { midName = value }, targetCase, gender).midName;
Assert.AreEqual(expected, result, string.Format("Part: {0}, Gender: {1}, Case: {2}", part, gender, targetCase));
}
break;
}
}
/// <summary>Helper: create a new instance to deserializing into</summary>
public static NamePart DeserializeLengthDelimited(Stream stream)
{
NamePart instance = new NamePart();
DeserializeLengthDelimited(stream, instance);
return instance;
}
/// <summary>Helper: create a new instance to deserializing into</summary>
public static NamePart DeserializeLength(Stream stream, int length)
{
NamePart instance = new NamePart();
DeserializeLength(stream, length, instance);
return instance;
}
/// <summary>Helper: put the buffer into a MemoryStream and create a new instance to deserializing into</summary>
public static NamePart Deserialize(byte[] buffer)
{
NamePart instance = new NamePart();
using (var ms = new MemoryStream(buffer))
Deserialize(ms, instance);
return instance;
}
/// <summary>Helper: create a new instance to deserializing into</summary>
public static NamePart Deserialize(Stream stream)
{
NamePart instance = new NamePart();
Deserialize(stream, instance);
return instance;
}
/// <summary>Serialize the instance into the stream</summary>
public static void Serialize(Stream stream, NamePart instance)
{
var msField = global::SilentOrbit.ProtocolBuffers.ProtocolParser.Stack.Pop();
if (instance.NamePartField == null)
throw new global::SilentOrbit.ProtocolBuffers.ProtocolBufferException("NamePartField is required by the proto specification.");
// Key for field: 1, LengthDelimited
stream.WriteByte(10);
global::SilentOrbit.ProtocolBuffers.ProtocolParser.WriteBytes(stream, Encoding.UTF8.GetBytes(instance.NamePartField));
// Key for field: 2, Varint
stream.WriteByte(16);
global::SilentOrbit.ProtocolBuffers.ProtocolParser.WriteBool(stream, instance.IsExtension);
global::SilentOrbit.ProtocolBuffers.ProtocolParser.Stack.Push(msField);
}
private static string ParseNameExtracted(NamePart whichName)
{
string key = string.Empty;
switch (whichName)
{
case NamePart.First:
key = PersonNameParser.FirstName_Key;
break;
case NamePart.Middle:
key = PersonNameParser.MiddleName_Key;
break;
case NamePart.Last:
key = PersonNameParser.LastName_Key;
break;
}
return key;
}
public void TestGenderDetection(string value, NamePart part, Gender expected)
{
Gender gender = petrovich.GetGender(value, part);
Assert.AreEqual(expected, gender, $"Value: {value} Part: {part}");
}
/// <summary>Helper: Serialize into a MemoryStream and return its byte array</summary>
public static byte[] SerializeToBytes(NamePart instance)
{
using (var ms = new MemoryStream())
{
Serialize(ms, instance);
return ms.ToArray();
}
}
public string ToTitleCase(string name, NamePart part)
{
return String.IsNullOrEmpty(name) ? (name) : (String.Join(" ", name.Split(' ').Select(token => TitleCase(token, part)).ToArray()));
}
public void TestSinglePartInflection(string value, NamePart part, Gender gender, Case targetCase, string expected)
{
string result = petrovich.Inflect(value, part, targetCase, gender);
Assert.AreEqual(expected, result, string.Format("Part: {0}, Gender: {1}, Case: {2}", part, gender, targetCase));
}
/// <summary>Serialize the instance into the stream</summary>
public static void Serialize(Stream stream, NamePart instance)
{
if (instance.NamePartField == null)
throw new ArgumentNullException("NamePartField", "Required by proto specification.");
// Key for field: 1, LengthDelimited
stream.WriteByte(10);
global::SilentOrbit.ProtocolBuffers.ProtocolParser.WriteBytes(stream, Encoding.UTF8.GetBytes(instance.NamePartField));
// Key for field: 2, Varint
stream.WriteByte(16);
global::SilentOrbit.ProtocolBuffers.ProtocolParser.WriteBool(stream, instance.IsExtension);
}
public string ParseName(NamePart whichName)
{
string description = ((NamePart) whichName).GetDescription();
string key = ParseNameExtracted(whichName);
bool isRequired = whichName != NamePart.Middle;
string parsedValue = null;
if (!personFields.ContainsKey(key))
{
if (isRequired)
{
AddMessage(string.Format(PersonNameParser.NameRequiredMessage, description));
}
return parsedValue;
}
parsedValue = personFields[key].ToString();
if (string.IsNullOrEmpty(parsedValue))
{
if (isRequired)
{
AddMessage(string.Format(PersonNameParser.NameRequiredMessage, description));
}
return parsedValue;
}
if (parsedValue.Length < 2 || parsedValue.Length > 35)
{
AddMessage(string.Format(PersonNameParser.NameLengthMessage, description));
return parsedValue;
}
if(!Regex.IsMatch(parsedValue, NameRegex))
{
AddMessage(string.Format(PersonNameParser.NameCharacters, description));
return parsedValue;
}
return parsedValue;
}
/// <summary>Helper: Serialize with a varint length prefix</summary>
public static void SerializeLengthDelimited(Stream stream, NamePart instance)
{
var data = SerializeToBytes(instance);
global::SilentOrbit.ProtocolBuffers.ProtocolParser.WriteUInt32(stream, (uint)data.Length);
stream.Write(data, 0, data.Length);
}
private static IEnumerable ReadSinglePartData(string fileName, NamePart part)
{
using (StreamReader reader = new StreamReader(Path.Combine(NUnit.Framework.TestContext.CurrentContext.TestDirectory, "Data", fileName)))
{
string line;
line = reader.ReadLine(); //skip header
while ((line = reader.ReadLine()) != null)
{
if (string.IsNullOrWhiteSpace(line))
{
continue;
}
string[] chunks = line.Split(',').Select(s => s.Trim()).ToArray();
Gender gender = (Gender)Enum.Parse(typeof(Gender), chunks[1]);
Case @case = (Case)Enum.Parse(typeof(Case), chunks[2]);
yield return new object[] { chunks[0], part, gender, @case, chunks[3] };
}
}
}
public void TestSinglePartGender(string value, NamePart part, Gender expected)
{
Gender? gender = petrovich.GetGender(value, part);
Assert.AreEqual(expected, gender, string.Format("Part: {0}, Value: {1}", part, value));
}
public override void VisitNamePart(NamePart part)
{
BindIfNeeded(part);
base.VisitNamePart(part);
}
public string Inflect(string namePartValue, NamePart namePart, Case targetCase, Gender? gender = null)
{
if (namePartValue == null)
{
throw new ArgumentNullException(nameof(namePartValue));
}
if (gender == null)
{
gender = GetGender(namePartValue, namePart);
}
PartRules partRules = m_rules[namePart];
string[] chunks = namePartValue.Split(NAME_CHUNK_SPLIT);
for (int i = 0; i < chunks.Length; ++i)
{
Tags tags = i == 0 ? Tags.FirstWord : Tags.None;
chunks[i] = partRules.InflectChunk(chunks[i], gender.Value, tags, targetCase);
}
return String.Join("-", chunks);
}
internal PartGenderRules this[NamePart part]
{
get { return(m_rulesByNamePart[(int)part]); }
}
