public static string GenerateDoubleMetaphone(this string self)
{
MetaphoneData metaphoneData = new MetaphoneData();
int current = 0;
if (String.IsNullOrEmpty(self)) {
return self;
}
int last = self.Length - 1; //zero based index
string workingString = self.ToUpperInvariant() + " ";
bool isSlavoGermanic = (self.IndexOf('W') > -1) || (self.IndexOf('K') > -1) || (self.IndexOf("CZ", StringComparison.OrdinalIgnoreCase) > -1)
|| (self.IndexOf("WITZ", StringComparison.OrdinalIgnoreCase) > -1);
//skip these when at start of word
if (workingString.StartsWith(StringComparison.OrdinalIgnoreCase, "GN", "KN", "PN", "WR", "PS")) {
current += 1;
}
//Initial 'X' is pronounced 'Z' e.g. 'Xavier'
if (workingString[0] == 'X') {
metaphoneData.Add("S"); //'Z' maps to 'S'
current += 1;
}
while ((metaphoneData.PrimaryLength < 4) || (metaphoneData.SecondaryLength < 4)) {
if (current >= self.Length) {
break;
}
switch (workingString[current]) {
case 'A':
case 'E':
case 'I':
case 'O':
case 'U':
case 'Y':
if (current == 0) {
//all init vowels now map to 'A'
metaphoneData.Add("A");
}
current += 1;
break;
case 'B':
//"-mb", e.g", "dumb", already skipped over...
metaphoneData.Add("P");
if (workingString[current + 1] == 'B') {
current += 2;
}
else {
current += 1;
}
break;
case 'Ã':
metaphoneData.Add("S");
current += 1;
break;
case 'C':
//various germanic
if ((current > 1) && !IsVowel(workingString[current - 2]) && StringAt(workingString, (current - 1), "ACH")
&& ((workingString[current + 2] != 'I') && ((workingString[current + 2] != 'E') || StringAt(workingString, (current - 2), "BACHER", "MACHER")))) {
metaphoneData.Add("K");
current += 2;
break;
}
//special case 'caesar'
if ((current == 0) && StringAt(workingString, current, "CAESAR")) {
metaphoneData.Add("S");
current += 2;
break;
}
//italian 'chianti'
if (StringAt(workingString, current, "CHIA")) {
metaphoneData.Add("K");
current += 2;
break;
}
if (StringAt(workingString, current, "CH")) {
//find 'michael'
if ((current > 0) && StringAt(workingString, current, "CHAE")) {
metaphoneData.Add("K", "X");
current += 2;
break;
}
//greek roots e.g. 'chemistry', 'chorus'
if ((current == 0) && (StringAt(workingString, (current + 1), "HARAC", "HARIS") || StringAt(workingString, (current + 1), "HOR", "HYM", "HIA", "HEM"))
&& !StringAt(workingString, 0, "CHORE")) {
metaphoneData.Add("K");
current += 2;
break;
}
//germanic, greek, or otherwise 'ch' for 'kh' sound
if ((StringAt(workingString, 0, "VAN ", "VON ") || StringAt(workingString, 0, "SCH")) // 'architect but not 'arch', 'orchestra', 'orchid'
|| StringAt(workingString, (current - 2), "ORCHES", "ARCHIT", "ORCHID") || StringAt(workingString, (current + 2), "T", "S")
|| ((StringAt(workingString, (current - 1), "A", "O", "U", "E") || (current == 0)) //e.g., 'wachtler', 'wechsler', but not 'tichner'
&& StringAt(workingString, (current + 2), "L", "R", "N", "M", "B", "H", "F", "V", "W", " "))) {
metaphoneData.Add("K");
}
else {
if (current > 0) {
if (StringAt(workingString, 0, "MC")) {
//e.g., "McHugh"
metaphoneData.Add("K");
}
else {
metaphoneData.Add("X", "K");
}
}
else {
metaphoneData.Add("X");
}
}
current += 2;
break;
}
//e.g, 'czerny'
if (StringAt(workingString, current, "CZ") && !StringAt(workingString, (current - 2), "WICZ")) {
metaphoneData.Add("S", "X");
current += 2;
break;
}
//e.g., 'focaccia'
if (StringAt(workingString, (current + 1), "CIA")) {
metaphoneData.Add("X");
current += 3;
break;
}
//double 'C', but not if e.g. 'McClellan'
if (StringAt(workingString, current, "CC") && !((current == 1) && (workingString[0] == 'M'))) {
//'bellocchio' but not 'bacchus'
if (StringAt(workingString, (current + 2), "I", "E", "H") && !StringAt(workingString, (current + 2), "HU")) {
//'accident', 'accede' 'succeed'
if (((current == 1) && (workingString[current - 1] == 'A')) || StringAt(workingString, (current - 1), "UCCEE", "UCCES")) {
metaphoneData.Add("KS");
}
//'bacci', 'bertucci', other italian
else {
metaphoneData.Add("X");
}
current += 3;
break;
}
else {
//Pierce's rule
metaphoneData.Add("K");
current += 2;
break;
}
}
if (StringAt(workingString, current, "CK", "CG", "CQ")) {
metaphoneData.Add("K");
current += 2;
break;
}
if (StringAt(workingString, current, "CI", "CE", "CY")) {
//italian vs. english
if (StringAt(workingString, current, "CIO", "CIE", "CIA")) {
metaphoneData.Add("S", "X");
}
else {
metaphoneData.Add("S");
}
current += 2;
break;
}
//else
metaphoneData.Add("K");
//name sent in 'mac caffrey', 'mac gregor
if (StringAt(workingString, (current + 1), " C", " Q", " G")) {
current += 3;
}
else if (StringAt(workingString, (current + 1), "C", "K", "Q") && !StringAt(workingString, (current + 1), "CE", "CI")) {
current += 2;
}
else {
current += 1;
}
break;
case 'D':
if (StringAt(workingString, current, "DG")) {
if (StringAt(workingString, (current + 2), "I", "E", "Y")) {
//e.g. 'edge'
metaphoneData.Add("J");
current += 3;
break;
}
else {
//e.g. 'edgar'
metaphoneData.Add("TK");
current += 2;
break;
}
}
if (StringAt(workingString, current, "DT", "DD")) {
metaphoneData.Add("T");
current += 2;
break;
}
//else
metaphoneData.Add("T");
current += 1;
break;
case 'F':
if (workingString[current + 1] == 'F') {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("F");
break;
case 'G':
if (workingString[current + 1] == 'H') {
if ((current > 0) && !IsVowel(workingString[current - 1])) {
metaphoneData.Add("K");
current += 2;
break;
}
if (current < 3) {
//'ghislane', ghiradelli
if (current == 0) {
if (workingString[current + 2] == 'I') {
metaphoneData.Add("J");
}
else {
metaphoneData.Add("K");
}
current += 2;
break;
}
}
//Parker's rule (with some further refinements) - e.g., 'hugh'
if (((current > 1) && StringAt(workingString, (current - 2), "B", "H", "D")) //e.g., 'bough'
|| ((current > 2) && StringAt(workingString, (current - 3), "B", "H", "D")) //e.g., 'broughton'
|| ((current > 3) && StringAt(workingString, (current - 4), "B", "H"))) {
current += 2;
break;
}
else {
//e.g., 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'
if ((current > 2) && (workingString[current - 1] == 'U') && StringAt(workingString, (current - 3), "C", "G", "L", "R", "T")) {
metaphoneData.Add("F");
}
else if ((current > 0) && workingString[current - 1] != 'I') {
metaphoneData.Add("K");
}
current += 2;
break;
}
}
if (workingString[current + 1] == 'N') {
if ((current == 1) && IsVowel(workingString[0]) && !isSlavoGermanic) {
metaphoneData.Add("KN", "N");
}
else
//not e.g. 'cagney'
if (!StringAt(workingString, (current + 2), "EY") && (workingString[current + 1] != 'Y') && !isSlavoGermanic) {
metaphoneData.Add("N", "KN");
}
else {
metaphoneData.Add("KN");
}
current += 2;
break;
}
//'tagliaro'
if (StringAt(workingString, (current + 1), "LI") && !isSlavoGermanic) {
metaphoneData.Add("KL", "L");
current += 2;
break;
}
//-ges-,-gep-,-gel-, -gie- at beginning
if ((current == 0)
&& ((workingString[current + 1] == 'Y') || StringAt(workingString, (current + 1), "ES", "EP", "EB", "EL", "EY", "IB", "IL", "IN", "IE", "EI", "ER"))) {
metaphoneData.Add("K", "J");
current += 2;
break;
}
// -ger-, -gy-
if ((StringAt(workingString, (current + 1), "ER") || (workingString[current + 1] == 'Y')) && !StringAt(workingString, 0, "DANGER", "RANGER", "MANGER")
&& !StringAt(workingString, (current - 1), "E", "I") && !StringAt(workingString, (current - 1), "RGY", "OGY")) {
metaphoneData.Add("K", "J");
current += 2;
break;
}
// italian e.g, 'biaggi'
if (StringAt(workingString, (current + 1), "E", "I", "Y") || StringAt(workingString, (current - 1), "AGGI", "OGGI")) {
//obvious germanic
if ((StringAt(workingString, 0, "VAN ", "VON ") || StringAt(workingString, 0, "SCH")) || StringAt(workingString, (current + 1), "ET")) {
metaphoneData.Add("K");
}
else
//always soft if french ending
if (StringAt(workingString, (current + 1), "IER ")) {
metaphoneData.Add("J");
}
else {
metaphoneData.Add("J", "K");
}
current += 2;
break;
}
if (workingString[current + 1] == 'G') {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("K");
break;
case 'H':
//only keep if first & before vowel or btw. 2 vowels
if (((current == 0) || IsVowel(workingString[current - 1])) && IsVowel(workingString[current + 1])) {
metaphoneData.Add("H");
current += 2;
}
else //also takes care of 'HH'
{
current += 1;
}
break;
case 'J':
//obvious spanish, 'jose', 'san jacinto'
if (StringAt(workingString, current, "JOSE") || StringAt(workingString, 0, "SAN ")) {
if (((current == 0) && (workingString[current + 4] == ' ')) || StringAt(workingString, 0, "SAN ")) {
metaphoneData.Add("H");
}
else {
metaphoneData.Add("J", "H");
}
current += 1;
break;
}
if ((current == 0) && !StringAt(workingString, current, "JOSE")) {
metaphoneData.Add("J", "A"); //Yankelovich/Jankelowicz
}
else
//spanish pron. of e.g. 'bajador'
if (IsVowel(workingString[current - 1]) && !isSlavoGermanic && ((workingString[current + 1] == 'A') || (workingString[current + 1] == 'O'))) {
metaphoneData.Add("J", "H");
}
else if (current == last) {
metaphoneData.Add("J", " ");
}
else if (!StringAt(workingString, (current + 1), "L", "T", "K", "S", "N", "M", "B", "Z") && !StringAt(workingString, (current - 1), "S", "K", "L")) {
metaphoneData.Add("J");
}
if (workingString[current + 1] == 'J') //it could happen!
{
current += 2;
}
else {
current += 1;
}
break;
case 'K':
if (workingString[current + 1] == 'K') {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("K");
break;
case 'L':
if (workingString[current + 1] == 'L') {
//spanish e.g. 'cabrillo', 'gallegos'
if (((current == (self.Length - 3)) && StringAt(workingString, (current - 1), "ILLO", "ILLA", "ALLE"))
|| ((StringAt(workingString, (last - 1), "AS", "OS") || StringAt(workingString, last, "A", "O")) && StringAt(workingString, (current - 1), "ALLE"))) {
metaphoneData.Add("L", " ");
current += 2;
break;
}
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("L");
break;
case 'M':
if ((StringAt(workingString, (current - 1), "UMB") && (((current + 1) == last) || StringAt(workingString, (current + 2), "ER"))) //'dumb','thumb'
|| (workingString[current + 1] == 'M')) {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("M");
break;
case 'N':
if (workingString[current + 1] == 'N') {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("N");
break;
case 'Ð':
current += 1;
metaphoneData.Add("N");
break;
case 'P':
if (workingString[current + 1] == 'H') {
metaphoneData.Add("F");
current += 2;
break;
}
//also account for "campbell", "raspberry"
if (StringAt(workingString, (current + 1), "P", "B")) {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("P");
break;
case 'Q':
if (workingString[current + 1] == 'Q') {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("K");
break;
case 'R':
//french e.g. 'rogier', but exclude 'hochmeier'
if ((current == last) && !isSlavoGermanic && StringAt(workingString, (current - 2), "IE") && !StringAt(workingString, (current - 4), "ME", "MA")) {
metaphoneData.Add("", "R");
}
else {
metaphoneData.Add("R");
}
if (workingString[current + 1] == 'R') {
current += 2;
}
else {
current += 1;
}
break;
case 'S':
//special cases 'island', 'isle', 'carlisle', 'carlysle'
if (StringAt(workingString, (current - 1), "ISL", "YSL")) {
current += 1;
break;
}
//special case 'sugar-'
if ((current == 0) && StringAt(workingString, current, "SUGAR")) {
metaphoneData.Add("X", "S");
current += 1;
break;
}
if (StringAt(workingString, current, "SH")) {
//germanic
if (StringAt(workingString, (current + 1), "HEIM", "HOEK", "HOLM", "HOLZ")) {
metaphoneData.Add("S");
}
else {
metaphoneData.Add("X");
}
current += 2;
break;
}
//italian & armenian
if (StringAt(workingString, current, "SIO", "SIA") || StringAt(workingString, current, "SIAN")) {
if (!isSlavoGermanic) {
metaphoneData.Add("S", "X");
}
else {
metaphoneData.Add("S");
}
current += 3;
break;
}
//german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'
//also, -sz- in slavic language altho in hungarian it is pronounced 's'
if (((current == 0) && StringAt(workingString, (current + 1), "M", "N", "L", "W")) || StringAt(workingString, (current + 1), "Z")) {
metaphoneData.Add("S", "X");
if (StringAt(workingString, (current + 1), "Z")) {
current += 2;
}
else {
current += 1;
}
break;
}
if (StringAt(workingString, current, "SC")) {
//Schlesinger's rule
if (workingString[current + 2] == 'H') {
//dutch origin, e.g. 'school', 'schooner'
if (StringAt(workingString, (current + 3), "OO", "ER", "EN", "UY", "ED", "EM")) {
//'schermerhorn', 'schenker'
if (StringAt(workingString, (current + 3), "ER", "EN")) {
metaphoneData.Add("X", "SK");
}
else {
metaphoneData.Add("SK");
}
current += 3;
break;
}
else {
if ((current == 0) && !IsVowel(workingString[3]) && (workingString[3] != 'W')) {
metaphoneData.Add("X", "S");
}
else {
metaphoneData.Add("X");
}
current += 3;
break;
}
}
if (StringAt(workingString, (current + 2), "I", "E", "Y")) {
metaphoneData.Add("S");
current += 3;
break;
}
//else
metaphoneData.Add("SK");
current += 3;
break;
}
//french e.g. 'resnais', 'artois'
if ((current == last) && StringAt(workingString, (current - 2), "AI", "OI")) {
metaphoneData.Add("", "S");
}
else {
metaphoneData.Add("S");
}
if (StringAt(workingString, (current + 1), "S", "Z")) {
current += 2;
}
else {
current += 1;
}
break;
case 'T':
if (StringAt(workingString, current, "TION")) {
metaphoneData.Add("X");
current += 3;
break;
}
if (StringAt(workingString, current, "TIA", "TCH")) {
metaphoneData.Add("X");
current += 3;
break;
}
if (StringAt(workingString, current, "TH") || StringAt(workingString, current, "TTH")) {
//special case 'thomas', 'thames' or germanic
if (StringAt(workingString, (current + 2), "OM", "AM") || StringAt(workingString, 0, "VAN ", "VON ") || StringAt(workingString, 0, "SCH")) {
metaphoneData.Add("T");
}
else {
metaphoneData.Add("O", "T");
}
current += 2;
break;
}
if (StringAt(workingString, (current + 1), "T", "D")) {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("T");
break;
case 'V':
if (workingString[current + 1] == 'V') {
current += 2;
}
else {
current += 1;
}
metaphoneData.Add("F");
break;
case 'W':
//can also be in middle of word
if (StringAt(workingString, current, "WR")) {
metaphoneData.Add("R");
current += 2;
break;
}
if ((current == 0) && (IsVowel(workingString[current + 1]) || StringAt(workingString, current, "WH"))) {
//Wasserman should match Vasserman
if (IsVowel(workingString[current + 1])) {
metaphoneData.Add("A", "F");
}
else {
//need Uomo to match Womo
metaphoneData.Add("A");
}
}
//Arnow should match Arnoff
if (((current == last) && IsVowel(workingString[current - 1])) || StringAt(workingString, (current - 1), "EWSKI", "EWSKY", "OWSKI", "OWSKY")
|| StringAt(workingString, 0, "SCH")) {
metaphoneData.Add("", "F");
current += 1;
break;
}
//polish e.g. 'filipowicz'
if (StringAt(workingString, current, "WICZ", "WITZ")) {
metaphoneData.Add("TS", "FX");
current += 4;
break;
}
//else skip it
current += 1;
break;
case 'X':
//french e.g. breaux
if (!((current == last) && (StringAt(workingString, (current - 3), "IAU", "EAU") || StringAt(workingString, (current - 2), "AU", "OU")))) {
metaphoneData.Add("KS");
}
if (StringAt(workingString, (current + 1), "C", "X")) {
current += 2;
}
else {
current += 1;
}
break;
case 'Z':
//chinese pinyin e.g. 'zhao'
if (workingString[current + 1] == 'H') {
metaphoneData.Add("J");
current += 2;
break;
}
else if (StringAt(workingString, (current + 1), "ZO", "ZI", "ZA") || (isSlavoGermanic && ((current > 0) && workingString[current - 1] != 'T'))) {
metaphoneData.Add("S", "TS");
}
else {
metaphoneData.Add("S");
}
if (workingString[current + 1] == 'Z') {
current += 2;
}
else {
current += 1;
}
break;
default:
current += 1;
break;
}
}
return metaphoneData.ToString();
}
public static string GenerateDoubleMetaphone(this string self)
{
MetaphoneData metaphoneData = new MetaphoneData();
int current = 0;
if (self.Length < 1)
{
return(self);
}
int last = self.Length - 1; //zero based index
string workingString = self.ToUpperInvariant() + " ";
bool isSlavoGermanic = (self.IndexOf('W') > -1) || (self.IndexOf('K') > -1) || (self.IndexOf("CZ", StringComparison.OrdinalIgnoreCase) > -1) ||
(self.IndexOf("WITZ", StringComparison.OrdinalIgnoreCase) > -1);
//skip these when at start of word
if (workingString.StartsWith(StringComparison.OrdinalIgnoreCase, "GN", "KN", "PN", "WR", "PS"))
{
current += 1;
}
//Initial 'X' is pronounced 'Z' e.g. 'Xavier'
if (workingString[0] == 'X')
{
metaphoneData.Add("S"); //'Z' maps to 'S'
current += 1;
}
while ((metaphoneData.PrimaryLength < 4) || (metaphoneData.SecondaryLength < 4))
{
if (current >= self.Length)
{
break;
}
switch (workingString[current])
{
case 'A':
case 'E':
case 'I':
case 'O':
case 'U':
case 'Y':
if (current == 0)
{
//all init vowels now map to 'A'
metaphoneData.Add("A");
}
current += 1;
break;
case 'B':
//"-mb", e.g", "dumb", already skipped over...
metaphoneData.Add("P");
if (workingString[current + 1] == 'B')
{
current += 2;
}
else
{
current += 1;
}
break;
case 'Ã':
metaphoneData.Add("S");
current += 1;
break;
case 'C':
//various germanic
if ((current > 1) && !IsVowel(workingString[current - 2]) && StringAt(workingString, (current - 1), "ACH") &&
((workingString[current + 2] != 'I') && ((workingString[current + 2] != 'E') || StringAt(workingString, (current - 2), "BACHER", "MACHER"))))
{
metaphoneData.Add("K");
current += 2;
break;
}
//special case 'caesar'
if ((current == 0) && StringAt(workingString, current, "CAESAR"))
{
metaphoneData.Add("S");
current += 2;
break;
}
//italian 'chianti'
if (StringAt(workingString, current, "CHIA"))
{
metaphoneData.Add("K");
current += 2;
break;
}
if (StringAt(workingString, current, "CH"))
{
//find 'michael'
if ((current > 0) && StringAt(workingString, current, "CHAE"))
{
metaphoneData.Add("K", "X");
current += 2;
break;
}
//greek roots e.g. 'chemistry', 'chorus'
if ((current == 0) && (StringAt(workingString, (current + 1), "HARAC", "HARIS") || StringAt(workingString, (current + 1), "HOR", "HYM", "HIA", "HEM")) &&
!StringAt(workingString, 0, "CHORE"))
{
metaphoneData.Add("K");
current += 2;
break;
}
//germanic, greek, or otherwise 'ch' for 'kh' sound
if ((StringAt(workingString, 0, "VAN ", "VON ") || StringAt(workingString, 0, "SCH")) || // 'architect but not 'arch', 'orchestra', 'orchid'
StringAt(workingString, (current - 2), "ORCHES", "ARCHIT", "ORCHID") || StringAt(workingString, (current + 2), "T", "S") ||
((StringAt(workingString, (current - 1), "A", "O", "U", "E") || (current == 0)) && //e.g., 'wachtler', 'wechsler', but not 'tichner'
StringAt(workingString, (current + 2), "L", "R", "N", "M", "B", "H", "F", "V", "W", " ")))
{
metaphoneData.Add("K");
}
else
{
if (current > 0)
{
if (StringAt(workingString, 0, "MC"))
{
//e.g., "McHugh"
metaphoneData.Add("K");
}
else
{
metaphoneData.Add("X", "K");
}
}
else
{
metaphoneData.Add("X");
}
}
current += 2;
break;
}
//e.g, 'czerny'
if (StringAt(workingString, current, "CZ") && !StringAt(workingString, (current - 2), "WICZ"))
{
metaphoneData.Add("S", "X");
current += 2;
break;
}
//e.g., 'focaccia'
if (StringAt(workingString, (current + 1), "CIA"))
{
metaphoneData.Add("X");
current += 3;
break;
}
//double 'C', but not if e.g. 'McClellan'
if (StringAt(workingString, current, "CC") && !((current == 1) && (workingString[0] == 'M')))
{
//'bellocchio' but not 'bacchus'
if (StringAt(workingString, (current + 2), "I", "E", "H") && !StringAt(workingString, (current + 2), "HU"))
{
//'accident', 'accede' 'succeed'
if (((current == 1) && (workingString[current - 1] == 'A')) || StringAt(workingString, (current - 1), "UCCEE", "UCCES"))
{
metaphoneData.Add("KS");
}
//'bacci', 'bertucci', other italian
else
{
metaphoneData.Add("X");
}
current += 3;
break;
}
else
{
//Pierce's rule
metaphoneData.Add("K");
current += 2;
break;
}
}
if (StringAt(workingString, current, "CK", "CG", "CQ"))
{
metaphoneData.Add("K");
current += 2;
break;
}
if (StringAt(workingString, current, "CI", "CE", "CY"))
{
//italian vs. english
if (StringAt(workingString, current, "CIO", "CIE", "CIA"))
{
metaphoneData.Add("S", "X");
}
else
{
metaphoneData.Add("S");
}
current += 2;
break;
}
//else
metaphoneData.Add("K");
//name sent in 'mac caffrey', 'mac gregor
if (StringAt(workingString, (current + 1), " C", " Q", " G"))
{
current += 3;
}
else if (StringAt(workingString, (current + 1), "C", "K", "Q") && !StringAt(workingString, (current + 1), "CE", "CI"))
{
current += 2;
}
else
{
current += 1;
}
break;
case 'D':
if (StringAt(workingString, current, "DG"))
{
if (StringAt(workingString, (current + 2), "I", "E", "Y"))
{
//e.g. 'edge'
metaphoneData.Add("J");
current += 3;
break;
}
else
{
//e.g. 'edgar'
metaphoneData.Add("TK");
current += 2;
break;
}
}
if (StringAt(workingString, current, "DT", "DD"))
{
metaphoneData.Add("T");
current += 2;
break;
}
//else
metaphoneData.Add("T");
current += 1;
break;
case 'F':
if (workingString[current + 1] == 'F')
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("F");
break;
case 'G':
if (workingString[current + 1] == 'H')
{
if ((current > 0) && !IsVowel(workingString[current - 1]))
{
metaphoneData.Add("K");
current += 2;
break;
}
if (current < 3)
{
//'ghislane', ghiradelli
if (current == 0)
{
if (workingString[current + 2] == 'I')
{
metaphoneData.Add("J");
}
else
{
metaphoneData.Add("K");
}
current += 2;
break;
}
}
//Parker's rule (with some further refinements) - e.g., 'hugh'
if (((current > 1) && StringAt(workingString, (current - 2), "B", "H", "D")) || //e.g., 'bough'
((current > 2) && StringAt(workingString, (current - 3), "B", "H", "D")) || //e.g., 'broughton'
((current > 3) && StringAt(workingString, (current - 4), "B", "H")))
{
current += 2;
break;
}
else
{
//e.g., 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'
if ((current > 2) && (workingString[current - 1] == 'U') && StringAt(workingString, (current - 3), "C", "G", "L", "R", "T"))
{
metaphoneData.Add("F");
}
else if ((current > 0) && workingString[current - 1] != 'I')
{
metaphoneData.Add("K");
}
current += 2;
break;
}
}
if (workingString[current + 1] == 'N')
{
if ((current == 1) && IsVowel(workingString[0]) && !isSlavoGermanic)
{
metaphoneData.Add("KN", "N");
}
else
//not e.g. 'cagney'
if (!StringAt(workingString, (current + 2), "EY") && (workingString[current + 1] != 'Y') && !isSlavoGermanic)
{
metaphoneData.Add("N", "KN");
}
else
{
metaphoneData.Add("KN");
}
current += 2;
break;
}
//'tagliaro'
if (StringAt(workingString, (current + 1), "LI") && !isSlavoGermanic)
{
metaphoneData.Add("KL", "L");
current += 2;
break;
}
//-ges-,-gep-,-gel-, -gie- at beginning
if ((current == 0) &&
((workingString[current + 1] == 'Y') || StringAt(workingString, (current + 1), "ES", "EP", "EB", "EL", "EY", "IB", "IL", "IN", "IE", "EI", "ER")))
{
metaphoneData.Add("K", "J");
current += 2;
break;
}
// -ger-, -gy-
if ((StringAt(workingString, (current + 1), "ER") || (workingString[current + 1] == 'Y')) && !StringAt(workingString, 0, "DANGER", "RANGER", "MANGER") &&
!StringAt(workingString, (current - 1), "E", "I") && !StringAt(workingString, (current - 1), "RGY", "OGY"))
{
metaphoneData.Add("K", "J");
current += 2;
break;
}
// italian e.g, 'biaggi'
if (StringAt(workingString, (current + 1), "E", "I", "Y") || StringAt(workingString, (current - 1), "AGGI", "OGGI"))
{
//obvious germanic
if ((StringAt(workingString, 0, "VAN ", "VON ") || StringAt(workingString, 0, "SCH")) || StringAt(workingString, (current + 1), "ET"))
{
metaphoneData.Add("K");
}
else
//always soft if french ending
if (StringAt(workingString, (current + 1), "IER "))
{
metaphoneData.Add("J");
}
else
{
metaphoneData.Add("J", "K");
}
current += 2;
break;
}
if (workingString[current + 1] == 'G')
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("K");
break;
case 'H':
//only keep if first & before vowel or btw. 2 vowels
if (((current == 0) || IsVowel(workingString[current - 1])) && IsVowel(workingString[current + 1]))
{
metaphoneData.Add("H");
current += 2;
}
else //also takes care of 'HH'
{
current += 1;
}
break;
case 'J':
//obvious spanish, 'jose', 'san jacinto'
if (StringAt(workingString, current, "JOSE") || StringAt(workingString, 0, "SAN "))
{
if (((current == 0) && (workingString[current + 4] == ' ')) || StringAt(workingString, 0, "SAN "))
{
metaphoneData.Add("H");
}
else
{
metaphoneData.Add("J", "H");
}
current += 1;
break;
}
if ((current == 0) && !StringAt(workingString, current, "JOSE"))
{
metaphoneData.Add("J", "A"); //Yankelovich/Jankelowicz
}
else
//spanish pron. of e.g. 'bajador'
if (IsVowel(workingString[current - 1]) && !isSlavoGermanic && ((workingString[current + 1] == 'A') || (workingString[current + 1] == 'O')))
{
metaphoneData.Add("J", "H");
}
else if (current == last)
{
metaphoneData.Add("J", " ");
}
else if (!StringAt(workingString, (current + 1), "L", "T", "K", "S", "N", "M", "B", "Z") && !StringAt(workingString, (current - 1), "S", "K", "L"))
{
metaphoneData.Add("J");
}
if (workingString[current + 1] == 'J') //it could happen!
{
current += 2;
}
else
{
current += 1;
}
break;
case 'K':
if (workingString[current + 1] == 'K')
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("K");
break;
case 'L':
if (workingString[current + 1] == 'L')
{
//spanish e.g. 'cabrillo', 'gallegos'
if (((current == (self.Length - 3)) && StringAt(workingString, (current - 1), "ILLO", "ILLA", "ALLE")) ||
((StringAt(workingString, (last - 1), "AS", "OS") || StringAt(workingString, last, "A", "O")) && StringAt(workingString, (current - 1), "ALLE")))
{
metaphoneData.Add("L", " ");
current += 2;
break;
}
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("L");
break;
case 'M':
if ((StringAt(workingString, (current - 1), "UMB") && (((current + 1) == last) || StringAt(workingString, (current + 2), "ER"))) || //'dumb','thumb'
(workingString[current + 1] == 'M'))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("M");
break;
case 'N':
if (workingString[current + 1] == 'N')
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("N");
break;
case 'Ð':
current += 1;
metaphoneData.Add("N");
break;
case 'P':
if (workingString[current + 1] == 'H')
{
metaphoneData.Add("F");
current += 2;
break;
}
//also account for "campbell", "raspberry"
if (StringAt(workingString, (current + 1), "P", "B"))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("P");
break;
case 'Q':
if (workingString[current + 1] == 'Q')
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("K");
break;
case 'R':
//french e.g. 'rogier', but exclude 'hochmeier'
if ((current == last) && !isSlavoGermanic && StringAt(workingString, (current - 2), "IE") && !StringAt(workingString, (current - 4), "ME", "MA"))
{
metaphoneData.Add("", "R");
}
else
{
metaphoneData.Add("R");
}
if (workingString[current + 1] == 'R')
{
current += 2;
}
else
{
current += 1;
}
break;
case 'S':
//special cases 'island', 'isle', 'carlisle', 'carlysle'
if (StringAt(workingString, (current - 1), "ISL", "YSL"))
{
current += 1;
break;
}
//special case 'sugar-'
if ((current == 0) && StringAt(workingString, current, "SUGAR"))
{
metaphoneData.Add("X", "S");
current += 1;
break;
}
if (StringAt(workingString, current, "SH"))
{
//germanic
if (StringAt(workingString, (current + 1), "HEIM", "HOEK", "HOLM", "HOLZ"))
{
metaphoneData.Add("S");
}
else
{
metaphoneData.Add("X");
}
current += 2;
break;
}
//italian & armenian
if (StringAt(workingString, current, "SIO", "SIA") || StringAt(workingString, current, "SIAN"))
{
if (!isSlavoGermanic)
{
metaphoneData.Add("S", "X");
}
else
{
metaphoneData.Add("S");
}
current += 3;
break;
}
//german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'
//also, -sz- in slavic language altho in hungarian it is pronounced 's'
if (((current == 0) && StringAt(workingString, (current + 1), "M", "N", "L", "W")) || StringAt(workingString, (current + 1), "Z"))
{
metaphoneData.Add("S", "X");
if (StringAt(workingString, (current + 1), "Z"))
{
current += 2;
}
else
{
current += 1;
}
break;
}
if (StringAt(workingString, current, "SC"))
{
//Schlesinger's rule
if (workingString[current + 2] == 'H')
{
//dutch origin, e.g. 'school', 'schooner'
if (StringAt(workingString, (current + 3), "OO", "ER", "EN", "UY", "ED", "EM"))
{
//'schermerhorn', 'schenker'
if (StringAt(workingString, (current + 3), "ER", "EN"))
{
metaphoneData.Add("X", "SK");
}
else
{
metaphoneData.Add("SK");
}
current += 3;
break;
}
else
{
if ((current == 0) && !IsVowel(workingString[3]) && (workingString[3] != 'W'))
{
metaphoneData.Add("X", "S");
}
else
{
metaphoneData.Add("X");
}
current += 3;
break;
}
}
if (StringAt(workingString, (current + 2), "I", "E", "Y"))
{
metaphoneData.Add("S");
current += 3;
break;
}
//else
metaphoneData.Add("SK");
current += 3;
break;
}
//french e.g. 'resnais', 'artois'
if ((current == last) && StringAt(workingString, (current - 2), "AI", "OI"))
{
metaphoneData.Add("", "S");
}
else
{
metaphoneData.Add("S");
}
if (StringAt(workingString, (current + 1), "S", "Z"))
{
current += 2;
}
else
{
current += 1;
}
break;
case 'T':
if (StringAt(workingString, current, "TION"))
{
metaphoneData.Add("X");
current += 3;
break;
}
if (StringAt(workingString, current, "TIA", "TCH"))
{
metaphoneData.Add("X");
current += 3;
break;
}
if (StringAt(workingString, current, "TH") || StringAt(workingString, current, "TTH"))
{
//special case 'thomas', 'thames' or germanic
if (StringAt(workingString, (current + 2), "OM", "AM") || StringAt(workingString, 0, "VAN ", "VON ") || StringAt(workingString, 0, "SCH"))
{
metaphoneData.Add("T");
}
else
{
metaphoneData.Add("O", "T");
}
current += 2;
break;
}
if (StringAt(workingString, (current + 1), "T", "D"))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("T");
break;
case 'V':
if (workingString[current + 1] == 'V')
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("F");
break;
case 'W':
//can also be in middle of word
if (StringAt(workingString, current, "WR"))
{
metaphoneData.Add("R");
current += 2;
break;
}
if ((current == 0) && (IsVowel(workingString[current + 1]) || StringAt(workingString, current, "WH")))
{
//Wasserman should match Vasserman
if (IsVowel(workingString[current + 1]))
{
metaphoneData.Add("A", "F");
}
else
{
//need Uomo to match Womo
metaphoneData.Add("A");
}
}
//Arnow should match Arnoff
if (((current == last) && IsVowel(workingString[current - 1])) || StringAt(workingString, (current - 1), "EWSKI", "EWSKY", "OWSKI", "OWSKY") ||
StringAt(workingString, 0, "SCH"))
{
metaphoneData.Add("", "F");
current += 1;
break;
}
//polish e.g. 'filipowicz'
if (StringAt(workingString, current, "WICZ", "WITZ"))
{
metaphoneData.Add("TS", "FX");
current += 4;
break;
}
//else skip it
current += 1;
break;
case 'X':
//french e.g. breaux
if (!((current == last) && (StringAt(workingString, (current - 3), "IAU", "EAU") || StringAt(workingString, (current - 2), "AU", "OU"))))
{
metaphoneData.Add("KS");
}
if (StringAt(workingString, (current + 1), "C", "X"))
{
current += 2;
}
else
{
current += 1;
}
break;
case 'Z':
//chinese pinyin e.g. 'zhao'
if (workingString[current + 1] == 'H')
{
metaphoneData.Add("J");
current += 2;
break;
}
else if (StringAt(workingString, (current + 1), "ZO", "ZI", "ZA") || (isSlavoGermanic && ((current > 0) && workingString[current - 1] != 'T')))
{
metaphoneData.Add("S", "TS");
}
else
{
metaphoneData.Add("S");
}
if (workingString[current + 1] == 'Z')
{
current += 2;
}
else
{
current += 1;
}
break;
default:
current += 1;
break;
}
}
return(metaphoneData.ToString());
}
public static string ToDoubleMetaphone(this string input)
{
MetaphoneData metaphoneData = new MetaphoneData();
int current = 0;
if (input.Length < 1)
{
return(input);
}
int last = input.Length - 1; //zero based index
string workingString = input.ToUpperInvariant() + " ";
bool isSlavoGermanic = (input.IndexOf(charW) > -1) ||
(input.IndexOf(charK) > -1) ||
(input.IndexOf(strCZ, StringComparison.OrdinalIgnoreCase) > -1) ||
(input.IndexOf(strWITZ, StringComparison.OrdinalIgnoreCase) > -1);
//skip these when at start of word
if (workingString.StartsWith(StringComparison.OrdinalIgnoreCase, strGN, strKN, strPN, strWR, strPS))
{
current += 1;
}
//Initial 'X' is pronounced 'Z' e.g. 'Xavier'
if (workingString[0] == charX)
{
metaphoneData.Add(strS); //'Z' maps to 'S'
current += 1;
}
while ((metaphoneData.PrimaryLength < 4) || (metaphoneData.SecondaryLength < 4))
{
if (current >= input.Length)
{
break;
}
switch (workingString[current])
{
case charA:
case charE:
case charI:
case charO:
case charU:
case charY:
if (current == 0)
{
//all init vowels now map to 'A'
metaphoneData.Add("A");
}
current += 1;
break;
case charB:
//"-mb", e.g", "dumb", already skipped over...
metaphoneData.Add("P");
if (workingString[current + 1] == charB)
{
current += 2;
}
else
{
current += 1;
}
break;
case charAdash:
metaphoneData.Add(strS);
current += 1;
break;
case charC:
//various germanic
if ((current > 1) &&
!IsVowel(workingString[current - 2]) &&
StringAt(workingString, (current - 1), strACH) &&
((workingString[current + 2] != charI) &&
((workingString[current + 2] != charE) ||
StringAt(workingString, (current - 2), strBACHER, strMACHER))))
{
metaphoneData.Add(strK);
current += 2;
break;
}
//special case 'caesar'
if ((current == 0) && StringAt(workingString, current, strCAESAR))
{
metaphoneData.Add(strS);
current += 2;
break;
}
//italian 'chianti'
if (StringAt(workingString, current, strCHIA))
{
metaphoneData.Add(strK);
current += 2;
break;
}
if (StringAt(workingString, current, strCH))
{
//find 'michael'
if ((current > 0) && StringAt(workingString, current, strCHAE))
{
metaphoneData.Add(strK, strX);
current += 2;
break;
}
//greek roots e.g. 'chemistry', 'chorus'
if ((current == 0) &&
(StringAt(workingString, (current + 1), strHARAC, strHARIS) ||
StringAt(workingString, (current + 1), strHOR, strHYM, strHIA, strHEM)) &&
!StringAt(workingString, 0, strCHORE))
{
metaphoneData.Add(strK);
current += 2;
break;
}
//germanic, greek, or otherwise 'ch' for 'kh' sound
if ((StringAt(workingString, 0, strVANsp, strVONsp) ||
StringAt(workingString, 0, strSCH)) || // 'architect but not 'arch', 'orchestra', 'orchid'
StringAt(workingString, (current - 2), strORCHES, strARCHIT, strORCHID) ||
StringAt(workingString, (current + 2), strT, strS) ||
((StringAt(workingString, (current - 1), strA, strO, strU, strE) ||
(current == 0)) && //e.g., 'wachtler', 'wechsler', but not 'tichner'
StringAt(workingString, (current + 2), strL, strR, strN, strM, strB, strH, strF, strV, strW, sp)))
{
metaphoneData.Add(strK);
}
else
{
if (current > 0)
{
if (StringAt(workingString, 0, strMC))
{
//e.g., "McHugh"
metaphoneData.Add(strK);
}
else
{
metaphoneData.Add(strX, strK);
}
}
else
{
metaphoneData.Add(strX);
}
}
current += 2;
break;
}
//e.g, 'czerny'
if (StringAt(workingString, current, strCZ) && !StringAt(workingString, (current - 2), strWICZ))
{
metaphoneData.Add(strS, strX);
current += 2;
break;
}
//e.g., 'focaccia'
if (StringAt(workingString, (current + 1), strCIA))
{
metaphoneData.Add(strX);
current += 3;
break;
}
//double 'C', but not if e.g. 'McClellan'
if (StringAt(workingString, current, strCC) && !((current == 1) && (workingString[0] == charM)))
{
//'bellocchio' but not 'bacchus'
if (StringAt(workingString, (current + 2), strI, strE, strH) &&
!StringAt(workingString, (current + 2), strHU))
{
//'accident', 'accede' 'succeed'
if (((current == 1) && (workingString[current - 1] == charA)) ||
StringAt(workingString, (current - 1), strUCCEE, strUCCES))
{
metaphoneData.Add(strKS);
}
//'bacci', 'bertucci', other italian
else
{
metaphoneData.Add(strX);
}
current += 3;
break;
}
else
{
//Pierce's rule
metaphoneData.Add(strK);
current += 2;
break;
}
}
if (StringAt(workingString, current, strCK, strCG, strCQ))
{
metaphoneData.Add(strK);
current += 2;
break;
}
if (StringAt(workingString, current, strCI, strCE, strCY))
{
//italian vs. english
if (StringAt(workingString, current, strCIO, strCIE, strCIA))
{
metaphoneData.Add(strS, strX);
}
else
{
metaphoneData.Add(strS);
}
current += 2;
break;
}
//else
metaphoneData.Add(strK);
//name sent in 'mac caffrey', 'mac gregor
if (StringAt(workingString, (current + 1), strspC, strspQ, strspG))
{
current += 3;
}
else if (StringAt(workingString, (current + 1), strC, strK, strQ) &&
!StringAt(workingString, (current + 1), strCE, strCI))
{
current += 2;
}
else
{
current += 1;
}
break;
case charD:
if (StringAt(workingString, current, strDG))
{
if (StringAt(workingString, (current + 2), strI, strE, strY))
{
//e.g. 'edge'
metaphoneData.Add(strJ);
current += 3;
break;
}
else
{
//e.g. 'edgar'
metaphoneData.Add(strTK);
current += 2;
break;
}
}
if (StringAt(workingString, current, strDT, strDD))
{
metaphoneData.Add(strT);
current += 2;
break;
}
//else
metaphoneData.Add(strT);
current += 1;
break;
case charF:
if (workingString[current + 1] == charF)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strF);
break;
case charG:
if (workingString[current + 1] == charH)
{
if ((current > 0) && !IsVowel(workingString[current - 1]))
{
metaphoneData.Add(strK);
current += 2;
break;
}
if (current < 3)
{
//'ghislane', ghiradelli
if (current == 0)
{
if (workingString[current + 2] == charI)
{
metaphoneData.Add(strJ);
}
else
{
metaphoneData.Add(strK);
}
current += 2;
break;
}
}
//Parker's rule (with some further refinements) - e.g., 'hugh'
if (((current > 1) && StringAt(workingString, (current - 2), strB, strH, strD)) || //e.g., 'bough'
((current > 2) && StringAt(workingString, (current - 3), strB, strH, strD)) || //e.g., 'broughton'
((current > 3) && StringAt(workingString, (current - 4), strB, strH)))
{
current += 2;
break;
}
else
{
//e.g., 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'
if ((current > 2) && (workingString[current - 1] == charU) &&
StringAt(workingString, (current - 3), strC, strG, strL, strR, strT))
{
metaphoneData.Add(strF);
}
else if ((current > 0) && workingString[current - 1] != charI)
{
metaphoneData.Add(strK);
}
current += 2;
break;
}
}
if (workingString[current + 1] == charN)
{
if ((current == 1) && IsVowel(workingString[0]) && !isSlavoGermanic)
{
metaphoneData.Add(strKN, strN);
}
else
//not e.g. 'cagney'
if (!StringAt(workingString, (current + 2), strEY) &&
(workingString[current + 1] != charY) && !isSlavoGermanic)
{
metaphoneData.Add(strN, strKN);
}
else
{
metaphoneData.Add(strKN);
}
current += 2;
break;
}
//'tagliaro'
if (StringAt(workingString, (current + 1), strLI) && !isSlavoGermanic)
{
metaphoneData.Add(strKL, strL);
current += 2;
break;
}
//-ges-,-gep-,-gel-, -gie- at beginning
if ((current == 0) &&
((workingString[current + 1] == charY) ||
StringAt(workingString, (current + 1), strES, strEP, strEB, strEL, strEY, strIB, strIL, strIN, strIE, strEI, strER)))
{
metaphoneData.Add(strK, strJ);
current += 2;
break;
}
// -ger-, -gy-
if ((StringAt(workingString, (current + 1), strER) ||
(workingString[current + 1] == charY)) &&
!StringAt(workingString, 0, strDANGER, strRANGER, strMANGER) &&
!StringAt(workingString, (current - 1), strE, strI) &&
!StringAt(workingString, (current - 1), strRGY, strOGY))
{
metaphoneData.Add(strK, strJ);
current += 2;
break;
}
// italian e.g, 'biaggi'
if (StringAt(workingString, (current + 1), strE, strI, strY) ||
StringAt(workingString, (current - 1), strAGGI, strOGGI))
{
//obvious germanic
if ((StringAt(workingString, 0, strVANsp, strVONsp) ||
StringAt(workingString, 0, strSCH)) ||
StringAt(workingString, (current + 1), strET))
{
metaphoneData.Add(strK);
}
else
//always soft if french ending
if (StringAt(workingString, (current + 1), strIERsp))
{
metaphoneData.Add(strJ);
}
else
{
metaphoneData.Add(strJ, strK);
}
current += 2;
break;
}
if (workingString[current + 1] == charG)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strK);
break;
case 'H':
//only keep if first & before vowel or btw. 2 vowels
if (((current == 0) || IsVowel(workingString[current - 1])) && IsVowel(workingString[current + 1]))
{
metaphoneData.Add(strH);
current += 2;
}
else //also takes care of 'HH'
{
current += 1;
}
break;
case 'J':
//obvious spanish, 'jose', 'san jacinto'
if (StringAt(workingString, current, strJOSE) || StringAt(workingString, 0, strSANsp))
{
if (((current == 0) && (workingString[current + 4] == ' ')) || StringAt(workingString, 0, strSANsp))
{
metaphoneData.Add(strH);
}
else
{
metaphoneData.Add(strJ, strH);
}
current += 1;
break;
}
if ((current == 0) && !StringAt(workingString, current, strJOSE))
{
metaphoneData.Add(strJ, strA); //Yankelovich/Jankelowicz
}
else
//spanish pron. of e.g. 'bajador'
if (current > 0 &&
IsVowel(workingString[current - 1]) &&
!isSlavoGermanic && ((workingString[current + 1] == charA) ||
(workingString[current + 1] == charO)))
{
metaphoneData.Add(strJ, strH);
}
else if (current == last)
{
metaphoneData.Add(strJ, sp);
}
else if (!StringAt(workingString, (current + 1), strL, strT, strK, strS, strN, strM, strB, strZ) &&
!StringAt(workingString, (current - 1), strS, strK, strL))
{
metaphoneData.Add(strJ);
}
if (workingString[current + 1] == charJ) //it could happen!
{
current += 2;
}
else
{
current += 1;
}
break;
case charK:
if (workingString[current + 1] == charK)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strK);
break;
case charL:
if (workingString[current + 1] == charL)
{
//spanish e.g. 'cabrillo', 'gallegos'
if (((current == (input.Length - 3)) &&
StringAt(workingString, (current - 1), strILLO, strILLA, strALLE)) ||
((StringAt(workingString, (last - 1), strAS, strOS) ||
StringAt(workingString, last, strA, strO)) &&
StringAt(workingString, (current - 1), strALLE)))
{
metaphoneData.Add(strL, sp);
current += 2;
break;
}
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("L");
break;
case charM:
if ((StringAt(workingString, (current - 1), strUMB) &&
(((current + 1) == last) ||
StringAt(workingString, (current + 2), strER))) || //'dumb','thumb'
(workingString[current + 1] == charM))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("M");
break;
case charN:
if (workingString[current + 1] == charN)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strN);
break;
case charOdash:
current += 1;
metaphoneData.Add(strN);
break;
case charP:
if (workingString[current + 1] == charH)
{
metaphoneData.Add(strF);
current += 2;
break;
}
//also account for "campbell", "raspberry"
if (StringAt(workingString, (current + 1), strP, strB))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strP);
break;
case charQ:
if (workingString[current + 1] == charQ)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strK);
break;
case charR:
//french e.g. 'rogier', but exclude 'hochmeier'
if ((current == last) && !isSlavoGermanic &&
StringAt(workingString, (current - 2), strIE) &&
!StringAt(workingString, (current - 4), strME, strMA))
{
metaphoneData.Add(string.Empty, strR);
}
else
{
metaphoneData.Add(strR);
}
if (workingString[current + 1] == charR)
{
current += 2;
}
else
{
current += 1;
}
break;
case charS:
//special cases 'island', 'isle', 'carlisle', 'carlysle'
if (StringAt(workingString, (current - 1), strISL, strYSL))
{
current += 1;
break;
}
//special case 'sugar-'
if ((current == 0) && StringAt(workingString, current, strSUGAR))
{
metaphoneData.Add(strX, strS);
current += 1;
break;
}
if (StringAt(workingString, current, strSH))
{
//germanic
if (StringAt(workingString, (current + 1), strHEIM, strHOEK, strHOLM, strHOLZ))
{
metaphoneData.Add(strS);
}
else
{
metaphoneData.Add(strX);
}
current += 2;
break;
}
//italian & armenian
if (StringAt(workingString, current, strSIO, strSIA) || StringAt(workingString, current, strSIAN))
{
if (!isSlavoGermanic)
{
metaphoneData.Add(strS, strX);
}
else
{
metaphoneData.Add(strS);
}
current += 3;
break;
}
//german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'
//also, -sz- in slavic language altho in hungarian it is pronounced 's'
if (((current == 0) &&
StringAt(workingString, (current + 1), strM, strN, strL, strW)) ||
StringAt(workingString, (current + 1), strZ))
{
metaphoneData.Add(strS, strX);
if (StringAt(workingString, (current + 1), strZ))
{
current += 2;
}
else
{
current += 1;
}
break;
}
if (StringAt(workingString, current, strSC))
{
//Schlesinger's rule
if (workingString[current + 2] == charH)
{
//dutch origin, e.g. 'school', 'schooner'
if (StringAt(workingString, (current + 3), strOO, strER, strEN, strUY, strED, strEM))
{
//'schermerhorn', 'schenker'
if (StringAt(workingString, (current + 3), strER, strEN))
{
metaphoneData.Add(strX, strSK);
}
else
{
metaphoneData.Add(strSK);
}
current += 3;
break;
}
else
{
if ((current == 0) && !IsVowel(workingString[3]) && (workingString[3] != charW))
{
metaphoneData.Add(strX, strS);
}
else
{
metaphoneData.Add(strX);
}
current += 3;
break;
}
}
if (StringAt(workingString, (current + 2), strI, strE, strY))
{
metaphoneData.Add(strS);
current += 3;
break;
}
//else
metaphoneData.Add(strSK);
current += 3;
break;
}
//french e.g. 'resnais', 'artois'
if ((current == last) && StringAt(workingString, (current - 2), strAI, strOI))
{
metaphoneData.Add(string.Empty, strS);
}
else
{
metaphoneData.Add(strS);
}
if (StringAt(workingString, (current + 1), strS, strZ))
{
current += 2;
}
else
{
current += 1;
}
break;
case charT:
if (StringAt(workingString, current, strTION))
{
metaphoneData.Add(strX);
current += 3;
break;
}
if (StringAt(workingString, current, strTIA, strTCH))
{
metaphoneData.Add(strX);
current += 3;
break;
}
if (StringAt(workingString, current, strTH) || StringAt(workingString, current, strTTH))
{
//special case 'thomas', 'thames' or germanic
if (StringAt(workingString, (current + 2), strOM, strAM) ||
StringAt(workingString, 0, strVANsp, strVONsp) || StringAt(workingString, 0, strSCH))
{
metaphoneData.Add(strT);
}
else
{
metaphoneData.Add(strO, strT);
}
current += 2;
break;
}
if (StringAt(workingString, (current + 1), strT, strD))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strT);
break;
case charV:
if (workingString[current + 1] == charV)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strF);
break;
case charW:
//can also be in middle of word
if (StringAt(workingString, current, strWR))
{
metaphoneData.Add(strR);
current += 2;
break;
}
if ((current == 0) && (IsVowel(workingString[current + 1]) ||
StringAt(workingString, current, strWH)))
{
//Wasserman should match Vasserman
if (IsVowel(workingString[current + 1]))
{
metaphoneData.Add(strA, strF);
}
else
{
//need Uomo to match Womo
metaphoneData.Add(strA);
}
}
//Arnow should match Arnoff
if ((current == last && current > 0 && IsVowel(workingString[current - 1])) ||
StringAt(workingString, (current - 1), strEWSKI, strEWSKY, strOWSKI, strOWSKY) ||
StringAt(workingString, 0, strSCH))
{
metaphoneData.Add(string.Empty, strF);
current += 1;
break;
}
//polish e.g. 'filipowicz'
if (StringAt(workingString, current, strWICZ, strWITZ))
{
metaphoneData.Add(strTS, strFX);
current += 4;
break;
}
//else skip it
current += 1;
break;
case charX:
//french e.g. breaux
if (!((current == last) &&
(StringAt(workingString, (current - 3), strIAU, strEAU) ||
StringAt(workingString, (current - 2), strAU, strOU))))
{
metaphoneData.Add(strKS);
}
if (StringAt(workingString, (current + 1), strC, strX))
{
current += 2;
}
else
{
current += 1;
}
break;
case charZ:
//chinese pinyin e.g. 'zhao'
if (workingString[current + 1] == charH)
{
metaphoneData.Add(strJ);
current += 2;
break;
}
else if (StringAt(workingString, (current + 1), strZO, strZI, strZA) ||
(isSlavoGermanic && ((current > 0) && workingString[current - 1] != charT)))
{
metaphoneData.Add(strS, strTS);
}
else
{
metaphoneData.Add(strS);
}
if (workingString[current + 1] == charZ)
{
current += 2;
}
else
{
current += 1;
}
break;
default:
current += 1;
break;
}
}
return(metaphoneData.ToString());
}
public static string ToDoubleMetaphone(this string input)
{
MetaphoneData metaphoneData = new MetaphoneData();
int current = 0;
if (input.Length < 1)
{
return input;
}
int last = input.Length - 1; //zero based index
string workingString = input.ToUpperInvariant() + " ";
bool isSlavoGermanic = (input.IndexOf(charW) > -1)
|| (input.IndexOf(charK) > -1)
|| (input.IndexOf(strCZ, StringComparison.OrdinalIgnoreCase) > -1)
|| (input.IndexOf(strWITZ, StringComparison.OrdinalIgnoreCase) > -1);
//skip these when at start of word
if (workingString.StartsWith(StringComparison.OrdinalIgnoreCase, strGN, strKN, strPN, strWR, strPS))
{
current += 1;
}
//Initial 'X' is pronounced 'Z' e.g. 'Xavier'
if (workingString[0] == charX)
{
metaphoneData.Add(strS); //'Z' maps to 'S'
current += 1;
}
while ((metaphoneData.PrimaryLength < 4) || (metaphoneData.SecondaryLength < 4))
{
if (current >= input.Length)
{
break;
}
switch (workingString[current])
{
case charA:
case charE:
case charI:
case charO:
case charU:
case charY:
if (current == 0)
{
//all init vowels now map to 'A'
metaphoneData.Add("A");
}
current += 1;
break;
case charB:
//"-mb", e.g", "dumb", already skipped over...
metaphoneData.Add("P");
if (workingString[current + 1] == charB)
{
current += 2;
}
else
{
current += 1;
}
break;
case charAdash:
metaphoneData.Add(strS);
current += 1;
break;
case charC:
//various germanic
if ((current > 1)
&& !IsVowel(workingString[current - 2])
&& StringAt(workingString, (current - 1), strACH)
&& ((workingString[current + 2] != charI)
&& ((workingString[current + 2] != charE)
|| StringAt(workingString, (current - 2), strBACHER, strMACHER))))
{
metaphoneData.Add(strK);
current += 2;
break;
}
//special case 'caesar'
if ((current == 0) && StringAt(workingString, current, strCAESAR))
{
metaphoneData.Add(strS);
current += 2;
break;
}
//italian 'chianti'
if (StringAt(workingString, current, strCHIA))
{
metaphoneData.Add(strK);
current += 2;
break;
}
if (StringAt(workingString, current, strCH))
{
//find 'michael'
if ((current > 0) && StringAt(workingString, current, strCHAE))
{
metaphoneData.Add(strK, strX);
current += 2;
break;
}
//greek roots e.g. 'chemistry', 'chorus'
if ((current == 0)
&& (StringAt(workingString, (current + 1), strHARAC, strHARIS)
|| StringAt(workingString, (current + 1), strHOR, strHYM, strHIA, strHEM))
&& !StringAt(workingString, 0, strCHORE))
{
metaphoneData.Add(strK);
current += 2;
break;
}
//germanic, greek, or otherwise 'ch' for 'kh' sound
if ((StringAt(workingString, 0, strVANsp, strVONsp)
|| StringAt(workingString, 0, strSCH)) // 'architect but not 'arch', 'orchestra', 'orchid'
|| StringAt(workingString, (current - 2), strORCHES, strARCHIT, strORCHID)
|| StringAt(workingString, (current + 2), strT, strS)
|| ((StringAt(workingString, (current - 1), strA, strO, strU, strE)
|| (current == 0)) //e.g., 'wachtler', 'wechsler', but not 'tichner'
&& StringAt(workingString, (current + 2), strL, strR, strN, strM, strB, strH, strF, strV, strW, sp)))
{
metaphoneData.Add(strK);
}
else
{
if (current > 0)
{
if (StringAt(workingString, 0, strMC))
{
//e.g., "McHugh"
metaphoneData.Add(strK);
}
else
{
metaphoneData.Add(strX, strK);
}
}
else
{
metaphoneData.Add(strX);
}
}
current += 2;
break;
}
//e.g, 'czerny'
if (StringAt(workingString, current, strCZ) && !StringAt(workingString, (current - 2), strWICZ))
{
metaphoneData.Add(strS, strX);
current += 2;
break;
}
//e.g., 'focaccia'
if (StringAt(workingString, (current + 1), strCIA))
{
metaphoneData.Add(strX);
current += 3;
break;
}
//double 'C', but not if e.g. 'McClellan'
if (StringAt(workingString, current, strCC) && !((current == 1) && (workingString[0] == charM)))
{
//'bellocchio' but not 'bacchus'
if (StringAt(workingString, (current + 2), strI, strE, strH)
&& !StringAt(workingString, (current + 2), strHU))
{
//'accident', 'accede' 'succeed'
if (((current == 1) && (workingString[current - 1] == charA))
|| StringAt(workingString, (current - 1), strUCCEE, strUCCES))
{
metaphoneData.Add(strKS);
}
//'bacci', 'bertucci', other italian
else
{
metaphoneData.Add(strX);
}
current += 3;
break;
}
else
{
//Pierce's rule
metaphoneData.Add(strK);
current += 2;
break;
}
}
if (StringAt(workingString, current, strCK, strCG, strCQ))
{
metaphoneData.Add(strK);
current += 2;
break;
}
if (StringAt(workingString, current, strCI, strCE, strCY))
{
//italian vs. english
if (StringAt(workingString, current, strCIO, strCIE, strCIA))
{
metaphoneData.Add(strS, strX);
}
else
{
metaphoneData.Add(strS);
}
current += 2;
break;
}
//else
metaphoneData.Add(strK);
//name sent in 'mac caffrey', 'mac gregor
if (StringAt(workingString, (current + 1), strspC, strspQ, strspG))
{
current += 3;
}
else if (StringAt(workingString, (current + 1), strC, strK, strQ)
&& !StringAt(workingString, (current + 1), strCE, strCI))
{
current += 2;
}
else
{
current += 1;
}
break;
case charD:
if (StringAt(workingString, current, strDG))
{
if (StringAt(workingString, (current + 2), strI, strE, strY))
{
//e.g. 'edge'
metaphoneData.Add(strJ);
current += 3;
break;
}
else
{
//e.g. 'edgar'
metaphoneData.Add(strTK);
current += 2;
break;
}
}
if (StringAt(workingString, current, strDT, strDD))
{
metaphoneData.Add(strT);
current += 2;
break;
}
//else
metaphoneData.Add(strT);
current += 1;
break;
case charF:
if (workingString[current + 1] == charF)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strF);
break;
case charG:
if (workingString[current + 1] == charH)
{
if ((current > 0) && !IsVowel(workingString[current - 1]))
{
metaphoneData.Add(strK);
current += 2;
break;
}
if (current < 3)
{
//'ghislane', ghiradelli
if (current == 0)
{
if (workingString[current + 2] == charI)
{
metaphoneData.Add(strJ);
}
else
{
metaphoneData.Add(strK);
}
current += 2;
break;
}
}
//Parker's rule (with some further refinements) - e.g., 'hugh'
if (((current > 1) && StringAt(workingString, (current - 2), strB, strH, strD)) //e.g., 'bough'
|| ((current > 2) && StringAt(workingString, (current - 3), strB, strH, strD)) //e.g., 'broughton'
|| ((current > 3) && StringAt(workingString, (current - 4), strB, strH)))
{
current += 2;
break;
}
else
{
//e.g., 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'
if ((current > 2) && (workingString[current - 1] == charU)
&& StringAt(workingString, (current - 3), strC, strG, strL, strR, strT))
{
metaphoneData.Add(strF);
}
else if ((current > 0) && workingString[current - 1] != charI)
{
metaphoneData.Add(strK);
}
current += 2;
break;
}
}
if (workingString[current + 1] == charN)
{
if ((current == 1) && IsVowel(workingString[0]) && !isSlavoGermanic)
{
metaphoneData.Add(strKN, strN);
}
else
//not e.g. 'cagney'
if (!StringAt(workingString, (current + 2), strEY)
&& (workingString[current + 1] != charY) && !isSlavoGermanic)
{
metaphoneData.Add(strN, strKN);
}
else
{
metaphoneData.Add(strKN);
}
current += 2;
break;
}
//'tagliaro'
if (StringAt(workingString, (current + 1), strLI) && !isSlavoGermanic)
{
metaphoneData.Add(strKL, strL);
current += 2;
break;
}
//-ges-,-gep-,-gel-, -gie- at beginning
if ((current == 0)
&& ((workingString[current + 1] == charY)
|| StringAt(workingString, (current + 1), strES, strEP, strEB, strEL, strEY, strIB, strIL, strIN, strIE, strEI, strER)))
{
metaphoneData.Add(strK, strJ);
current += 2;
break;
}
// -ger-, -gy-
if ((StringAt(workingString, (current + 1), strER)
|| (workingString[current + 1] == charY))
&& !StringAt(workingString, 0, strDANGER, strRANGER, strMANGER)
&& !StringAt(workingString, (current - 1), strE, strI)
&& !StringAt(workingString, (current - 1), strRGY, strOGY))
{
metaphoneData.Add(strK, strJ);
current += 2;
break;
}
// italian e.g, 'biaggi'
if (StringAt(workingString, (current + 1), strE, strI, strY)
|| StringAt(workingString, (current - 1), strAGGI, strOGGI))
{
//obvious germanic
if ((StringAt(workingString, 0, strVANsp, strVONsp)
|| StringAt(workingString, 0, strSCH))
|| StringAt(workingString, (current + 1), strET))
{
metaphoneData.Add(strK);
}
else
//always soft if french ending
if (StringAt(workingString, (current + 1), strIERsp))
{
metaphoneData.Add(strJ);
}
else
{
metaphoneData.Add(strJ, strK);
}
current += 2;
break;
}
if (workingString[current + 1] == charG)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strK);
break;
case 'H':
//only keep if first & before vowel or btw. 2 vowels
if (((current == 0) || IsVowel(workingString[current - 1])) && IsVowel(workingString[current + 1]))
{
metaphoneData.Add(strH);
current += 2;
}
else //also takes care of 'HH'
{
current += 1;
}
break;
case 'J':
//obvious spanish, 'jose', 'san jacinto'
if (StringAt(workingString, current, strJOSE) || StringAt(workingString, 0, strSANsp))
{
if (((current == 0) && (workingString[current + 4] == ' ')) || StringAt(workingString, 0, strSANsp))
{
metaphoneData.Add(strH);
}
else
{
metaphoneData.Add(strJ, strH);
}
current += 1;
break;
}
if ((current == 0) && !StringAt(workingString, current, strJOSE))
{
metaphoneData.Add(strJ, strA); //Yankelovich/Jankelowicz
}
else
//spanish pron. of e.g. 'bajador'
if (IsVowel(workingString[current - 1])
&& !isSlavoGermanic && ((workingString[current + 1] == charA)
|| (workingString[current + 1] == charO)))
{
metaphoneData.Add(strJ, strH);
}
else if (current == last)
{
metaphoneData.Add(strJ, sp);
}
else if (!StringAt(workingString, (current + 1), strL, strT, strK, strS, strN, strM, strB, strZ)
&& !StringAt(workingString, (current - 1), strS, strK, strL))
{
metaphoneData.Add(strJ);
}
if (workingString[current + 1] == charJ) //it could happen!
{
current += 2;
}
else
{
current += 1;
}
break;
case charK:
if (workingString[current + 1] == charK)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strK);
break;
case charL:
if (workingString[current + 1] == charL)
{
//spanish e.g. 'cabrillo', 'gallegos'
if (((current == (input.Length - 3))
&& StringAt(workingString, (current - 1), strILLO, strILLA, strALLE))
|| ((StringAt(workingString, (last - 1), strAS, strOS)
|| StringAt(workingString, last, strA, strO))
&& StringAt(workingString, (current - 1), strALLE)))
{
metaphoneData.Add(strL, sp);
current += 2;
break;
}
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("L");
break;
case charM:
if ((StringAt(workingString, (current - 1), strUMB)
&& (((current + 1) == last)
|| StringAt(workingString, (current + 2), strER))) //'dumb','thumb'
|| (workingString[current + 1] == charM))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("M");
break;
case charN:
if (workingString[current + 1] == charN)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strN);
break;
case charOdash:
current += 1;
metaphoneData.Add(strN);
break;
case charP:
if (workingString[current + 1] == charH)
{
metaphoneData.Add(strF);
current += 2;
break;
}
//also account for "campbell", "raspberry"
if (StringAt(workingString, (current + 1), strP, strB))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strP);
break;
case charQ:
if (workingString[current + 1] == charQ)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strK);
break;
case charR:
//french e.g. 'rogier', but exclude 'hochmeier'
if ((current == last) && !isSlavoGermanic
&& StringAt(workingString, (current - 2), strIE)
&& !StringAt(workingString, (current - 4), strME, strMA))
{
metaphoneData.Add(string.Empty, strR);
}
else
{
metaphoneData.Add(strR);
}
if (workingString[current + 1] == charR)
{
current += 2;
}
else
{
current += 1;
}
break;
case charS:
//special cases 'island', 'isle', 'carlisle', 'carlysle'
if (StringAt(workingString, (current - 1), strISL, strYSL))
{
current += 1;
break;
}
//special case 'sugar-'
if ((current == 0) && StringAt(workingString, current, strSUGAR))
{
metaphoneData.Add(strX, strS);
current += 1;
break;
}
if (StringAt(workingString, current, strSH))
{
//germanic
if (StringAt(workingString, (current + 1), strHEIM, strHOEK, strHOLM, strHOLZ))
{
metaphoneData.Add(strS);
}
else
{
metaphoneData.Add(strX);
}
current += 2;
break;
}
//italian & armenian
if (StringAt(workingString, current, strSIO, strSIA) || StringAt(workingString, current, strSIAN))
{
if (!isSlavoGermanic)
{
metaphoneData.Add(strS, strX);
}
else
{
metaphoneData.Add(strS);
}
current += 3;
break;
}
//german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'
//also, -sz- in slavic language altho in hungarian it is pronounced 's'
if (((current == 0)
&& StringAt(workingString, (current + 1), strM, strN, strL, strW))
|| StringAt(workingString, (current + 1), strZ))
{
metaphoneData.Add(strS, strX);
if (StringAt(workingString, (current + 1), strZ))
{
current += 2;
}
else
{
current += 1;
}
break;
}
if (StringAt(workingString, current, strSC))
{
//Schlesinger's rule
if (workingString[current + 2] == charH)
{
//dutch origin, e.g. 'school', 'schooner'
if (StringAt(workingString, (current + 3), strOO, strER, strEN, strUY, strED, strEM))
{
//'schermerhorn', 'schenker'
if (StringAt(workingString, (current + 3), strER, strEN))
{
metaphoneData.Add(strX, strSK);
}
else
{
metaphoneData.Add(strSK);
}
current += 3;
break;
}
else
{
if ((current == 0) && !IsVowel(workingString[3]) && (workingString[3] != charW))
{
metaphoneData.Add(strX, strS);
}
else
{
metaphoneData.Add(strX);
}
current += 3;
break;
}
}
if (StringAt(workingString, (current + 2), strI, strE, strY))
{
metaphoneData.Add(strS);
current += 3;
break;
}
//else
metaphoneData.Add(strSK);
current += 3;
break;
}
//french e.g. 'resnais', 'artois'
if ((current == last) && StringAt(workingString, (current - 2), strAI, strOI))
{
metaphoneData.Add(string.Empty, strS);
}
else
{
metaphoneData.Add(strS);
}
if (StringAt(workingString, (current + 1), strS, strZ))
{
current += 2;
}
else
{
current += 1;
}
break;
case charT:
if (StringAt(workingString, current, strTION))
{
metaphoneData.Add(strX);
current += 3;
break;
}
if (StringAt(workingString, current, strTIA, strTCH))
{
metaphoneData.Add(strX);
current += 3;
break;
}
if (StringAt(workingString, current, strTH) || StringAt(workingString, current, strTTH))
{
//special case 'thomas', 'thames' or germanic
if (StringAt(workingString, (current + 2), strOM, strAM)
|| StringAt(workingString, 0, strVANsp, strVONsp) || StringAt(workingString, 0, strSCH))
{
metaphoneData.Add(strT);
}
else
{
metaphoneData.Add(strO, strT);
}
current += 2;
break;
}
if (StringAt(workingString, (current + 1), strT, strD))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strT);
break;
case charV:
if (workingString[current + 1] == charV)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(strF);
break;
case charW:
//can also be in middle of word
if (StringAt(workingString, current, strWR))
{
metaphoneData.Add(strR);
current += 2;
break;
}
if ((current == 0) && (IsVowel(workingString[current + 1])
|| StringAt(workingString, current, strWH)))
{
//Wasserman should match Vasserman
if (IsVowel(workingString[current + 1]))
{
metaphoneData.Add(strA, strF);
}
else
{
//need Uomo to match Womo
metaphoneData.Add(strA);
}
}
//Arnow should match Arnoff
if (((current == last) && IsVowel(workingString[current - 1]))
|| StringAt(workingString, (current - 1), strEWSKI, strEWSKY, strOWSKI, strOWSKY)
|| StringAt(workingString, 0, strSCH))
{
metaphoneData.Add(string.Empty, strF);
current += 1;
break;
}
//polish e.g. 'filipowicz'
if (StringAt(workingString, current, strWICZ, strWITZ))
{
metaphoneData.Add(strTS, strFX);
current += 4;
break;
}
//else skip it
current += 1;
break;
case charX:
//french e.g. breaux
if (!((current == last)
&& (StringAt(workingString, (current - 3), strIAU, strEAU)
|| StringAt(workingString, (current - 2), strAU, strOU))))
{
metaphoneData.Add(strKS);
}
if (StringAt(workingString, (current + 1), strC, strX))
{
current += 2;
}
else
{
current += 1;
}
break;
case charZ:
//chinese pinyin e.g. 'zhao'
if (workingString[current + 1] == charH)
{
metaphoneData.Add(strJ);
current += 2;
break;
}
else if (StringAt(workingString, (current + 1), strZO, strZI, strZA)
|| (isSlavoGermanic && ((current > 0) && workingString[current - 1] != charT)))
{
metaphoneData.Add(strS, strTS);
}
else
{
metaphoneData.Add(strS);
}
if (workingString[current + 1] == charZ)
{
current += 2;
}
else
{
current += 1;
}
break;
default:
current += 1;
break;
}
}
return metaphoneData.ToString();
}
public static string ToDoubleMetaphone(this string input)
{
MetaphoneData metaphoneData = new MetaphoneData();
int current = 0;
if (input.Length < 1)
{
return input;
}
int last = input.Length - 1; //zero based index
string workingString = input.ToUpperInvariant() + " ";
bool isSlavoGermanic = (input.IndexOf(CharW) > -1)
|| (input.IndexOf(CharK) > -1)
|| (input.IndexOf(StrCz, StringComparison.OrdinalIgnoreCase) > -1)
|| (input.IndexOf(StrWitz, StringComparison.OrdinalIgnoreCase) > -1);
//skip these when at start of word
if (workingString.StartsWith(StringComparison.OrdinalIgnoreCase, StrGn, StrKn, StrPn, StrWr, StrPs))
{
current += 1;
}
//Initial 'X' is pronounced 'Z' e.g. 'Xavier'
if (workingString[0] == CharX)
{
metaphoneData.Add(StrS); //'Z' maps to 'S'
current += 1;
}
while ((metaphoneData.PrimaryLength < 4) || (metaphoneData.SecondaryLength < 4))
{
if (current >= input.Length)
{
break;
}
switch (workingString[current])
{
case CharA:
case CharE:
case CharI:
case CharO:
case CharU:
case CharY:
if (current == 0)
{
//all init vowels now map to 'A'
metaphoneData.Add("A");
}
current += 1;
break;
case CharB:
//"-mb", e.g", "dumb", already skipped over...
metaphoneData.Add("P");
if (workingString[current + 1] == CharB)
{
current += 2;
}
else
{
current += 1;
}
break;
case CharAdash:
metaphoneData.Add(StrS);
current += 1;
break;
case CharC:
//various germanic
if ((current > 1)
&& !IsVowel(workingString[current - 2])
&& StringAt(workingString, (current - 1), StrAch)
&& ((workingString[current + 2] != CharI)
&& ((workingString[current + 2] != CharE)
|| StringAt(workingString, (current - 2), StrBacher, StrMacher))))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
//special case 'caesar'
if ((current == 0) && StringAt(workingString, current, StrCaesar))
{
metaphoneData.Add(StrS);
current += 2;
break;
}
//italian 'chianti'
if (StringAt(workingString, current, StrChia))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
if (StringAt(workingString, current, StrCh))
{
//find 'michael'
if ((current > 0) && StringAt(workingString, current, StrChae))
{
metaphoneData.Add(StrK, StrX);
current += 2;
break;
}
//greek roots e.g. 'chemistry', 'chorus'
if ((current == 0)
&& (StringAt(workingString, (current + 1), StrHarac, StrHaris)
|| StringAt(workingString, (current + 1), StrHor, StrHym, StrHia, StrHem))
&& !StringAt(workingString, 0, StrChore))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
//germanic, greek, or otherwise 'ch' for 'kh' sound
if ((StringAt(workingString, 0, StrVaNsp, StrVoNsp)
|| StringAt(workingString, 0, StrSch)) // 'architect but not 'arch', 'orchestra', 'orchid'
|| StringAt(workingString, (current - 2), StrOrches, StrArchit, StrOrchid)
|| StringAt(workingString, (current + 2), StrT, StrS)
|| ((StringAt(workingString, (current - 1), StrA, StrO, StrU, StrE)
|| (current == 0)) //e.g., 'wachtler', 'wechsler', but not 'tichner'
&& StringAt(workingString, (current + 2), StrL, StrR, StrN, StrM, StrB, StrH, StrF, StrV, StrW, Sp)))
{
metaphoneData.Add(StrK);
}
else
{
if (current > 0)
{
if (StringAt(workingString, 0, StrMc))
{
//e.g., "McHugh"
metaphoneData.Add(StrK);
}
else
{
metaphoneData.Add(StrX, StrK);
}
}
else
{
metaphoneData.Add(StrX);
}
}
current += 2;
break;
}
//e.g, 'czerny'
if (StringAt(workingString, current, StrCz) && !StringAt(workingString, (current - 2), StrWicz))
{
metaphoneData.Add(StrS, StrX);
current += 2;
break;
}
//e.g., 'focaccia'
if (StringAt(workingString, (current + 1), StrCia))
{
metaphoneData.Add(StrX);
current += 3;
break;
}
//double 'C', but not if e.g. 'McClellan'
if (StringAt(workingString, current, StrCc) && !((current == 1) && (workingString[0] == CharM)))
{
//'bellocchio' but not 'bacchus'
if (StringAt(workingString, (current + 2), StrI, StrE, StrH)
&& !StringAt(workingString, (current + 2), StrHu))
{
//'accident', 'accede' 'succeed'
if (((current == 1) && (workingString[current - 1] == CharA))
|| StringAt(workingString, (current - 1), StrUccee, StrUcces))
{
metaphoneData.Add(StrKs);
}
//'bacci', 'bertucci', other italian
else
{
metaphoneData.Add(StrX);
}
current += 3;
break;
}
else
{
//Pierce's rule
metaphoneData.Add(StrK);
current += 2;
break;
}
}
if (StringAt(workingString, current, StrCk, StrCg, StrCq))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
if (StringAt(workingString, current, StrCi, StrCe, StrCy))
{
//italian vs. english
if (StringAt(workingString, current, StrCio, StrCie, StrCia))
{
metaphoneData.Add(StrS, StrX);
}
else
{
metaphoneData.Add(StrS);
}
current += 2;
break;
}
//else
metaphoneData.Add(StrK);
//name sent in 'mac caffrey', 'mac gregor
if (StringAt(workingString, (current + 1), StrspC, StrspQ, StrspG))
{
current += 3;
}
else if (StringAt(workingString, (current + 1), StrC, StrK, StrQ)
&& !StringAt(workingString, (current + 1), StrCe, StrCi))
{
current += 2;
}
else
{
current += 1;
}
break;
case CharD:
if (StringAt(workingString, current, StrDg))
{
if (StringAt(workingString, (current + 2), StrI, StrE, StrY))
{
//e.g. 'edge'
metaphoneData.Add(StrJ);
current += 3;
break;
}
else
{
//e.g. 'edgar'
metaphoneData.Add(StrTk);
current += 2;
break;
}
}
if (StringAt(workingString, current, StrDt, StrDd))
{
metaphoneData.Add(StrT);
current += 2;
break;
}
//else
metaphoneData.Add(StrT);
current += 1;
break;
case CharF:
if (workingString[current + 1] == CharF)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrF);
break;
case CharG:
if (workingString[current + 1] == CharH)
{
if ((current > 0) && !IsVowel(workingString[current - 1]))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
if (current < 3)
{
//'ghislane', ghiradelli
if (current == 0)
{
if (workingString[current + 2] == CharI)
{
metaphoneData.Add(StrJ);
}
else
{
metaphoneData.Add(StrK);
}
current += 2;
break;
}
}
//Parker's rule (with some further refinements) - e.g., 'hugh'
if (((current > 1) && StringAt(workingString, (current - 2), StrB, StrH, StrD)) //e.g., 'bough'
|| ((current > 2) && StringAt(workingString, (current - 3), StrB, StrH, StrD)) //e.g., 'broughton'
|| ((current > 3) && StringAt(workingString, (current - 4), StrB, StrH)))
{
current += 2;
break;
}
else
{
//e.g., 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'
if ((current > 2) && (workingString[current - 1] == CharU)
&& StringAt(workingString, (current - 3), StrC, StrG, StrL, StrR, StrT))
{
metaphoneData.Add(StrF);
}
else if ((current > 0) && workingString[current - 1] != CharI)
{
metaphoneData.Add(StrK);
}
current += 2;
break;
}
}
if (workingString[current + 1] == CharN)
{
if ((current == 1) && IsVowel(workingString[0]) && !isSlavoGermanic)
{
metaphoneData.Add(StrKn, StrN);
}
else
//not e.g. 'cagney'
if (!StringAt(workingString, (current + 2), StrEy)
&& (workingString[current + 1] != CharY) && !isSlavoGermanic)
{
metaphoneData.Add(StrN, StrKn);
}
else
{
metaphoneData.Add(StrKn);
}
current += 2;
break;
}
//'tagliaro'
if (StringAt(workingString, (current + 1), StrLi) && !isSlavoGermanic)
{
metaphoneData.Add(StrKl, StrL);
current += 2;
break;
}
//-ges-,-gep-,-gel-, -gie- at beginning
if ((current == 0)
&& ((workingString[current + 1] == CharY)
|| StringAt(workingString, (current + 1), StrEs, StrEp, StrEb, StrEl, StrEy, StrIb, StrIl, StrIn, StrIe, StrEi, StrEr)))
{
metaphoneData.Add(StrK, StrJ);
current += 2;
break;
}
// -ger-, -gy-
if ((StringAt(workingString, (current + 1), StrEr)
|| (workingString[current + 1] == CharY))
&& !StringAt(workingString, 0, StrDanger, StrRanger, StrManger)
&& !StringAt(workingString, (current - 1), StrE, StrI)
&& !StringAt(workingString, (current - 1), StrRgy, StrOgy))
{
metaphoneData.Add(StrK, StrJ);
current += 2;
break;
}
// italian e.g, 'biaggi'
if (StringAt(workingString, (current + 1), StrE, StrI, StrY)
|| StringAt(workingString, (current - 1), StrAggi, StrOggi))
{
//obvious germanic
if ((StringAt(workingString, 0, StrVaNsp, StrVoNsp)
|| StringAt(workingString, 0, StrSch))
|| StringAt(workingString, (current + 1), StrEt))
{
metaphoneData.Add(StrK);
}
else
//always soft if french ending
if (StringAt(workingString, (current + 1), StrIeRsp))
{
metaphoneData.Add(StrJ);
}
else
{
metaphoneData.Add(StrJ, StrK);
}
current += 2;
break;
}
if (workingString[current + 1] == CharG)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrK);
break;
case 'H':
//only keep if first & before vowel or btw. 2 vowels
if (((current == 0) || IsVowel(workingString[current - 1])) && IsVowel(workingString[current + 1]))
{
metaphoneData.Add(StrH);
current += 2;
}
else //also takes care of 'HH'
{
current += 1;
}
break;
case 'J':
//obvious spanish, 'jose', 'san jacinto'
if (StringAt(workingString, current, StrJose) || StringAt(workingString, 0, StrSaNsp))
{
if (((current == 0) && (workingString[current + 4] == ' ')) || StringAt(workingString, 0, StrSaNsp))
{
metaphoneData.Add(StrH);
}
else
{
metaphoneData.Add(StrJ, StrH);
}
current += 1;
break;
}
if ((current == 0) && !StringAt(workingString, current, StrJose))
{
metaphoneData.Add(StrJ, StrA); //Yankelovich/Jankelowicz
}
else
//spanish pron. of e.g. 'bajador'
if (current > 0
&& IsVowel(workingString[current - 1])
&& !isSlavoGermanic && ((workingString[current + 1] == CharA)
|| (workingString[current + 1] == CharO)))
{
metaphoneData.Add(StrJ, StrH);
}
else if (current == last)
{
metaphoneData.Add(StrJ, Sp);
}
else if (!StringAt(workingString, (current + 1), StrL, StrT, StrK, StrS, StrN, StrM, StrB, StrZ)
&& !StringAt(workingString, (current - 1), StrS, StrK, StrL))
{
metaphoneData.Add(StrJ);
}
if (workingString[current + 1] == CharJ) //it could happen!
{
current += 2;
}
else
{
current += 1;
}
break;
case CharK:
if (workingString[current + 1] == CharK)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrK);
break;
case CharL:
if (workingString[current + 1] == CharL)
{
//spanish e.g. 'cabrillo', 'gallegos'
if (((current == (input.Length - 3))
&& StringAt(workingString, (current - 1), StrIllo, StrIlla, StrAlle))
|| ((StringAt(workingString, (last - 1), StrAs, StrOs)
|| StringAt(workingString, last, StrA, StrO))
&& StringAt(workingString, (current - 1), StrAlle)))
{
metaphoneData.Add(StrL, Sp);
current += 2;
break;
}
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("L");
break;
case CharM:
if ((StringAt(workingString, (current - 1), StrUmb)
&& (((current + 1) == last)
|| StringAt(workingString, (current + 2), StrEr))) //'dumb','thumb'
|| (workingString[current + 1] == CharM))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("M");
break;
case CharN:
if (workingString[current + 1] == CharN)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrN);
break;
case CharOdash:
current += 1;
metaphoneData.Add(StrN);
break;
case CharP:
if (workingString[current + 1] == CharH)
{
metaphoneData.Add(StrF);
current += 2;
break;
}
//also account for "campbell", "raspberry"
if (StringAt(workingString, (current + 1), StrP, StrB))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrP);
break;
case CharQ:
if (workingString[current + 1] == CharQ)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrK);
break;
case CharR:
//french e.g. 'rogier', but exclude 'hochmeier'
if ((current == last) && !isSlavoGermanic
&& StringAt(workingString, (current - 2), StrIe)
&& !StringAt(workingString, (current - 4), StrMe, StrMa))
{
metaphoneData.Add(string.Empty, StrR);
}
else
{
metaphoneData.Add(StrR);
}
if (workingString[current + 1] == CharR)
{
current += 2;
}
else
{
current += 1;
}
break;
case CharS:
//special cases 'island', 'isle', 'carlisle', 'carlysle'
if (StringAt(workingString, (current - 1), StrIsl, StrYsl))
{
current += 1;
break;
}
//special case 'sugar-'
if ((current == 0) && StringAt(workingString, current, StrSugar))
{
metaphoneData.Add(StrX, StrS);
current += 1;
break;
}
if (StringAt(workingString, current, StrSh))
{
//germanic
if (StringAt(workingString, (current + 1), StrHeim, StrHoek, StrHolm, StrHolz))
{
metaphoneData.Add(StrS);
}
else
{
metaphoneData.Add(StrX);
}
current += 2;
break;
}
//italian & armenian
if (StringAt(workingString, current, StrSio, StrSia) || StringAt(workingString, current, StrSian))
{
if (!isSlavoGermanic)
{
metaphoneData.Add(StrS, StrX);
}
else
{
metaphoneData.Add(StrS);
}
current += 3;
break;
}
//german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'
//also, -sz- in slavic language altho in hungarian it is pronounced 's'
if (((current == 0)
&& StringAt(workingString, (current + 1), StrM, StrN, StrL, StrW))
|| StringAt(workingString, (current + 1), StrZ))
{
metaphoneData.Add(StrS, StrX);
if (StringAt(workingString, (current + 1), StrZ))
{
current += 2;
}
else
{
current += 1;
}
break;
}
if (StringAt(workingString, current, StrSc))
{
//Schlesinger's rule
if (workingString[current + 2] == CharH)
{
//dutch origin, e.g. 'school', 'schooner'
if (StringAt(workingString, (current + 3), StrOo, StrEr, StrEn, StrUy, StrEd, StrEm))
{
//'schermerhorn', 'schenker'
if (StringAt(workingString, (current + 3), StrEr, StrEn))
{
metaphoneData.Add(StrX, StrSk);
}
else
{
metaphoneData.Add(StrSk);
}
current += 3;
break;
}
else
{
if ((current == 0) && !IsVowel(workingString[3]) && (workingString[3] != CharW))
{
metaphoneData.Add(StrX, StrS);
}
else
{
metaphoneData.Add(StrX);
}
current += 3;
break;
}
}
if (StringAt(workingString, (current + 2), StrI, StrE, StrY))
{
metaphoneData.Add(StrS);
current += 3;
break;
}
//else
metaphoneData.Add(StrSk);
current += 3;
break;
}
//french e.g. 'resnais', 'artois'
if ((current == last) && StringAt(workingString, (current - 2), StrAi, StrOi))
{
metaphoneData.Add(string.Empty, StrS);
}
else
{
metaphoneData.Add(StrS);
}
if (StringAt(workingString, (current + 1), StrS, StrZ))
{
current += 2;
}
else
{
current += 1;
}
break;
case CharT:
if (StringAt(workingString, current, StrTion))
{
metaphoneData.Add(StrX);
current += 3;
break;
}
if (StringAt(workingString, current, StrTia, StrTch))
{
metaphoneData.Add(StrX);
current += 3;
break;
}
if (StringAt(workingString, current, StrTh) || StringAt(workingString, current, StrTth))
{
//special case 'thomas', 'thames' or germanic
if (StringAt(workingString, (current + 2), StrOm, StrAm)
|| StringAt(workingString, 0, StrVaNsp, StrVoNsp) || StringAt(workingString, 0, StrSch))
{
metaphoneData.Add(StrT);
}
else
{
metaphoneData.Add(StrO, StrT);
}
current += 2;
break;
}
if (StringAt(workingString, (current + 1), StrT, StrD))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrT);
break;
case CharV:
if (workingString[current + 1] == CharV)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrF);
break;
case CharW:
//can also be in middle of word
if (StringAt(workingString, current, StrWr))
{
metaphoneData.Add(StrR);
current += 2;
break;
}
if ((current == 0) && (IsVowel(workingString[current + 1])
|| StringAt(workingString, current, StrWh)))
{
//Wasserman should match Vasserman
if (IsVowel(workingString[current + 1]))
{
metaphoneData.Add(StrA, StrF);
}
else
{
//need Uomo to match Womo
metaphoneData.Add(StrA);
}
}
//Arnow should match Arnoff
if ((current == last && current > 0 && IsVowel(workingString[current - 1]))
|| StringAt(workingString, (current - 1), StrEwski, StrEwsky, StrOwski, StrOwsky)
|| StringAt(workingString, 0, StrSch))
{
metaphoneData.Add(string.Empty, StrF);
current += 1;
break;
}
//polish e.g. 'filipowicz'
if (StringAt(workingString, current, StrWicz, StrWitz))
{
metaphoneData.Add(StrTs, StrFx);
current += 4;
break;
}
//else skip it
current += 1;
break;
case CharX:
//french e.g. breaux
if (!((current == last)
&& (StringAt(workingString, (current - 3), StrIau, StrEau)
|| StringAt(workingString, (current - 2), StrAu, StrOu))))
{
metaphoneData.Add(StrKs);
}
if (StringAt(workingString, (current + 1), StrC, StrX))
{
current += 2;
}
else
{
current += 1;
}
break;
case CharZ:
//chinese pinyin e.g. 'zhao'
if (workingString[current + 1] == CharH)
{
metaphoneData.Add(StrJ);
current += 2;
break;
}
else if (StringAt(workingString, (current + 1), StrZo, StrZi, StrZa)
|| (isSlavoGermanic && ((current > 0) && workingString[current - 1] != CharT)))
{
metaphoneData.Add(StrS, StrTs);
}
else
{
metaphoneData.Add(StrS);
}
if (workingString[current + 1] == CharZ)
{
current += 2;
}
else
{
current += 1;
}
break;
default:
current += 1;
break;
}
}
return metaphoneData.ToString();
}
public static string ToDoubleMetaphone(this string input)
{
MetaphoneData metaphoneData = new MetaphoneData();
int current = 0;
if (input.Length < 1)
{
return(input);
}
int last = input.Length - 1; //zero based index
string workingString = input.ToUpperInvariant() + " ";
bool isSlavoGermanic = (input.IndexOf(CharW) > -1) ||
(input.IndexOf(CharK) > -1) ||
(input.IndexOf(StrCz, StringComparison.OrdinalIgnoreCase) > -1) ||
(input.IndexOf(StrWitz, StringComparison.OrdinalIgnoreCase) > -1);
//skip these when at start of word
if (workingString.StartsWith(StringComparison.OrdinalIgnoreCase, StrGn, StrKn, StrPn, StrWr, StrPs))
{
current += 1;
}
//Initial 'X' is pronounced 'Z' e.g. 'Xavier'
if (workingString[0] == CharX)
{
metaphoneData.Add(StrS); //'Z' maps to 'S'
current += 1;
}
while ((metaphoneData.PrimaryLength < 4) || (metaphoneData.SecondaryLength < 4))
{
if (current >= input.Length)
{
break;
}
switch (workingString[current])
{
case CharA:
case CharE:
case CharI:
case CharO:
case CharU:
case CharY:
if (current == 0)
{
//all init vowels now map to 'A'
metaphoneData.Add("A");
}
current += 1;
break;
case CharB:
//"-mb", e.g", "dumb", already skipped over...
metaphoneData.Add("P");
if (workingString[current + 1] == CharB)
{
current += 2;
}
else
{
current += 1;
}
break;
case CharAdash:
metaphoneData.Add(StrS);
current += 1;
break;
case CharC:
//various germanic
if ((current > 1) &&
!IsVowel(workingString[current - 2]) &&
StringAt(workingString, (current - 1), StrAch) &&
((workingString[current + 2] != CharI) &&
((workingString[current + 2] != CharE) ||
StringAt(workingString, (current - 2), StrBacher, StrMacher))))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
//special case 'caesar'
if ((current == 0) && StringAt(workingString, current, StrCaesar))
{
metaphoneData.Add(StrS);
current += 2;
break;
}
//italian 'chianti'
if (StringAt(workingString, current, StrChia))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
if (StringAt(workingString, current, StrCh))
{
//find 'michael'
if ((current > 0) && StringAt(workingString, current, StrChae))
{
metaphoneData.Add(StrK, StrX);
current += 2;
break;
}
//greek roots e.g. 'chemistry', 'chorus'
if ((current == 0) &&
(StringAt(workingString, (current + 1), StrHarac, StrHaris) ||
StringAt(workingString, (current + 1), StrHor, StrHym, StrHia, StrHem)) &&
!StringAt(workingString, 0, StrChore))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
//germanic, greek, or otherwise 'ch' for 'kh' sound
if ((StringAt(workingString, 0, StrVaNsp, StrVoNsp) ||
StringAt(workingString, 0, StrSch)) || // 'architect but not 'arch', 'orchestra', 'orchid'
StringAt(workingString, (current - 2), StrOrches, StrArchit, StrOrchid) ||
StringAt(workingString, (current + 2), StrT, StrS) ||
((StringAt(workingString, (current - 1), StrA, StrO, StrU, StrE) ||
(current == 0)) && //e.g., 'wachtler', 'wechsler', but not 'tichner'
StringAt(workingString, (current + 2), StrL, StrR, StrN, StrM, StrB, StrH, StrF, StrV, StrW, Sp)))
{
metaphoneData.Add(StrK);
}
else
{
if (current > 0)
{
if (StringAt(workingString, 0, StrMc))
{
//e.g., "McHugh"
metaphoneData.Add(StrK);
}
else
{
metaphoneData.Add(StrX, StrK);
}
}
else
{
metaphoneData.Add(StrX);
}
}
current += 2;
break;
}
//e.g, 'czerny'
if (StringAt(workingString, current, StrCz) && !StringAt(workingString, (current - 2), StrWicz))
{
metaphoneData.Add(StrS, StrX);
current += 2;
break;
}
//e.g., 'focaccia'
if (StringAt(workingString, (current + 1), StrCia))
{
metaphoneData.Add(StrX);
current += 3;
break;
}
//double 'C', but not if e.g. 'McClellan'
if (StringAt(workingString, current, StrCc) && !((current == 1) && (workingString[0] == CharM)))
{
//'bellocchio' but not 'bacchus'
if (StringAt(workingString, (current + 2), StrI, StrE, StrH) &&
!StringAt(workingString, (current + 2), StrHu))
{
//'accident', 'accede' 'succeed'
if (((current == 1) && (workingString[current - 1] == CharA)) ||
StringAt(workingString, (current - 1), StrUccee, StrUcces))
{
metaphoneData.Add(StrKs);
}
//'bacci', 'bertucci', other italian
else
{
metaphoneData.Add(StrX);
}
current += 3;
break;
}
else
{
//Pierce's rule
metaphoneData.Add(StrK);
current += 2;
break;
}
}
if (StringAt(workingString, current, StrCk, StrCg, StrCq))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
if (StringAt(workingString, current, StrCi, StrCe, StrCy))
{
//italian vs. english
if (StringAt(workingString, current, StrCio, StrCie, StrCia))
{
metaphoneData.Add(StrS, StrX);
}
else
{
metaphoneData.Add(StrS);
}
current += 2;
break;
}
//else
metaphoneData.Add(StrK);
//name sent in 'mac caffrey', 'mac gregor
if (StringAt(workingString, (current + 1), StrspC, StrspQ, StrspG))
{
current += 3;
}
else if (StringAt(workingString, (current + 1), StrC, StrK, StrQ) &&
!StringAt(workingString, (current + 1), StrCe, StrCi))
{
current += 2;
}
else
{
current += 1;
}
break;
case CharD:
if (StringAt(workingString, current, StrDg))
{
if (StringAt(workingString, (current + 2), StrI, StrE, StrY))
{
//e.g. 'edge'
metaphoneData.Add(StrJ);
current += 3;
break;
}
else
{
//e.g. 'edgar'
metaphoneData.Add(StrTk);
current += 2;
break;
}
}
if (StringAt(workingString, current, StrDt, StrDd))
{
metaphoneData.Add(StrT);
current += 2;
break;
}
//else
metaphoneData.Add(StrT);
current += 1;
break;
case CharF:
if (workingString[current + 1] == CharF)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrF);
break;
case CharG:
if (workingString[current + 1] == CharH)
{
if ((current > 0) && !IsVowel(workingString[current - 1]))
{
metaphoneData.Add(StrK);
current += 2;
break;
}
if (current < 3)
{
//'ghislane', ghiradelli
if (current == 0)
{
if (workingString[current + 2] == CharI)
{
metaphoneData.Add(StrJ);
}
else
{
metaphoneData.Add(StrK);
}
current += 2;
break;
}
}
//Parker's rule (with some further refinements) - e.g., 'hugh'
if (((current > 1) && StringAt(workingString, (current - 2), StrB, StrH, StrD)) || //e.g., 'bough'
((current > 2) && StringAt(workingString, (current - 3), StrB, StrH, StrD)) || //e.g., 'broughton'
((current > 3) && StringAt(workingString, (current - 4), StrB, StrH)))
{
current += 2;
break;
}
else
{
//e.g., 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'
if ((current > 2) && (workingString[current - 1] == CharU) &&
StringAt(workingString, (current - 3), StrC, StrG, StrL, StrR, StrT))
{
metaphoneData.Add(StrF);
}
else if ((current > 0) && workingString[current - 1] != CharI)
{
metaphoneData.Add(StrK);
}
current += 2;
break;
}
}
if (workingString[current + 1] == CharN)
{
if ((current == 1) && IsVowel(workingString[0]) && !isSlavoGermanic)
{
metaphoneData.Add(StrKn, StrN);
}
else
//not e.g. 'cagney'
if (!StringAt(workingString, (current + 2), StrEy) &&
(workingString[current + 1] != CharY) && !isSlavoGermanic)
{
metaphoneData.Add(StrN, StrKn);
}
else
{
metaphoneData.Add(StrKn);
}
current += 2;
break;
}
//'tagliaro'
if (StringAt(workingString, (current + 1), StrLi) && !isSlavoGermanic)
{
metaphoneData.Add(StrKl, StrL);
current += 2;
break;
}
//-ges-,-gep-,-gel-, -gie- at beginning
if ((current == 0) &&
((workingString[current + 1] == CharY) ||
StringAt(workingString, (current + 1), StrEs, StrEp, StrEb, StrEl, StrEy, StrIb, StrIl, StrIn, StrIe, StrEi, StrEr)))
{
metaphoneData.Add(StrK, StrJ);
current += 2;
break;
}
// -ger-, -gy-
if ((StringAt(workingString, (current + 1), StrEr) ||
(workingString[current + 1] == CharY)) &&
!StringAt(workingString, 0, StrDanger, StrRanger, StrManger) &&
!StringAt(workingString, (current - 1), StrE, StrI) &&
!StringAt(workingString, (current - 1), StrRgy, StrOgy))
{
metaphoneData.Add(StrK, StrJ);
current += 2;
break;
}
// italian e.g, 'biaggi'
if (StringAt(workingString, (current + 1), StrE, StrI, StrY) ||
StringAt(workingString, (current - 1), StrAggi, StrOggi))
{
//obvious germanic
if ((StringAt(workingString, 0, StrVaNsp, StrVoNsp) ||
StringAt(workingString, 0, StrSch)) ||
StringAt(workingString, (current + 1), StrEt))
{
metaphoneData.Add(StrK);
}
else
//always soft if french ending
if (StringAt(workingString, (current + 1), StrIeRsp))
{
metaphoneData.Add(StrJ);
}
else
{
metaphoneData.Add(StrJ, StrK);
}
current += 2;
break;
}
if (workingString[current + 1] == CharG)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrK);
break;
case 'H':
//only keep if first & before vowel or btw. 2 vowels
if (((current == 0) || IsVowel(workingString[current - 1])) && IsVowel(workingString[current + 1]))
{
metaphoneData.Add(StrH);
current += 2;
}
else //also takes care of 'HH'
{
current += 1;
}
break;
case 'J':
//obvious spanish, 'jose', 'san jacinto'
if (StringAt(workingString, current, StrJose) || StringAt(workingString, 0, StrSaNsp))
{
if (((current == 0) && (workingString[current + 4] == ' ')) || StringAt(workingString, 0, StrSaNsp))
{
metaphoneData.Add(StrH);
}
else
{
metaphoneData.Add(StrJ, StrH);
}
current += 1;
break;
}
if ((current == 0) && !StringAt(workingString, current, StrJose))
{
metaphoneData.Add(StrJ, StrA); //Yankelovich/Jankelowicz
}
else
//spanish pron. of e.g. 'bajador'
if (current > 0 &&
IsVowel(workingString[current - 1]) &&
!isSlavoGermanic && ((workingString[current + 1] == CharA) ||
(workingString[current + 1] == CharO)))
{
metaphoneData.Add(StrJ, StrH);
}
else if (current == last)
{
metaphoneData.Add(StrJ, Sp);
}
else if (!StringAt(workingString, (current + 1), StrL, StrT, StrK, StrS, StrN, StrM, StrB, StrZ) &&
!StringAt(workingString, (current - 1), StrS, StrK, StrL))
{
metaphoneData.Add(StrJ);
}
if (workingString[current + 1] == CharJ) //it could happen!
{
current += 2;
}
else
{
current += 1;
}
break;
case CharK:
if (workingString[current + 1] == CharK)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrK);
break;
case CharL:
if (workingString[current + 1] == CharL)
{
//spanish e.g. 'cabrillo', 'gallegos'
if (((current == (input.Length - 3)) &&
StringAt(workingString, (current - 1), StrIllo, StrIlla, StrAlle)) ||
((StringAt(workingString, (last - 1), StrAs, StrOs) ||
StringAt(workingString, last, StrA, StrO)) &&
StringAt(workingString, (current - 1), StrAlle)))
{
metaphoneData.Add(StrL, Sp);
current += 2;
break;
}
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("L");
break;
case CharM:
if ((StringAt(workingString, (current - 1), StrUmb) &&
(((current + 1) == last) ||
StringAt(workingString, (current + 2), StrEr))) || //'dumb','thumb'
(workingString[current + 1] == CharM))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add("M");
break;
case CharN:
if (workingString[current + 1] == CharN)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrN);
break;
case CharOdash:
current += 1;
metaphoneData.Add(StrN);
break;
case CharP:
if (workingString[current + 1] == CharH)
{
metaphoneData.Add(StrF);
current += 2;
break;
}
//also account for "campbell", "raspberry"
if (StringAt(workingString, (current + 1), StrP, StrB))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrP);
break;
case CharQ:
if (workingString[current + 1] == CharQ)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrK);
break;
case CharR:
//french e.g. 'rogier', but exclude 'hochmeier'
if ((current == last) && !isSlavoGermanic &&
StringAt(workingString, (current - 2), StrIe) &&
!StringAt(workingString, (current - 4), StrMe, StrMa))
{
metaphoneData.Add(string.Empty, StrR);
}
else
{
metaphoneData.Add(StrR);
}
if (workingString[current + 1] == CharR)
{
current += 2;
}
else
{
current += 1;
}
break;
case CharS:
//special cases 'island', 'isle', 'carlisle', 'carlysle'
if (StringAt(workingString, (current - 1), StrIsl, StrYsl))
{
current += 1;
break;
}
//special case 'sugar-'
if ((current == 0) && StringAt(workingString, current, StrSugar))
{
metaphoneData.Add(StrX, StrS);
current += 1;
break;
}
if (StringAt(workingString, current, StrSh))
{
//germanic
if (StringAt(workingString, (current + 1), StrHeim, StrHoek, StrHolm, StrHolz))
{
metaphoneData.Add(StrS);
}
else
{
metaphoneData.Add(StrX);
}
current += 2;
break;
}
//italian & armenian
if (StringAt(workingString, current, StrSio, StrSia) || StringAt(workingString, current, StrSian))
{
if (!isSlavoGermanic)
{
metaphoneData.Add(StrS, StrX);
}
else
{
metaphoneData.Add(StrS);
}
current += 3;
break;
}
//german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'
//also, -sz- in slavic language altho in hungarian it is pronounced 's'
if (((current == 0) &&
StringAt(workingString, (current + 1), StrM, StrN, StrL, StrW)) ||
StringAt(workingString, (current + 1), StrZ))
{
metaphoneData.Add(StrS, StrX);
if (StringAt(workingString, (current + 1), StrZ))
{
current += 2;
}
else
{
current += 1;
}
break;
}
if (StringAt(workingString, current, StrSc))
{
//Schlesinger's rule
if (workingString[current + 2] == CharH)
{
//dutch origin, e.g. 'school', 'schooner'
if (StringAt(workingString, (current + 3), StrOo, StrEr, StrEn, StrUy, StrEd, StrEm))
{
//'schermerhorn', 'schenker'
if (StringAt(workingString, (current + 3), StrEr, StrEn))
{
metaphoneData.Add(StrX, StrSk);
}
else
{
metaphoneData.Add(StrSk);
}
current += 3;
break;
}
else
{
if ((current == 0) && !IsVowel(workingString[3]) && (workingString[3] != CharW))
{
metaphoneData.Add(StrX, StrS);
}
else
{
metaphoneData.Add(StrX);
}
current += 3;
break;
}
}
if (StringAt(workingString, (current + 2), StrI, StrE, StrY))
{
metaphoneData.Add(StrS);
current += 3;
break;
}
//else
metaphoneData.Add(StrSk);
current += 3;
break;
}
//french e.g. 'resnais', 'artois'
if ((current == last) && StringAt(workingString, (current - 2), StrAi, StrOi))
{
metaphoneData.Add(string.Empty, StrS);
}
else
{
metaphoneData.Add(StrS);
}
if (StringAt(workingString, (current + 1), StrS, StrZ))
{
current += 2;
}
else
{
current += 1;
}
break;
case CharT:
if (StringAt(workingString, current, StrTion))
{
metaphoneData.Add(StrX);
current += 3;
break;
}
if (StringAt(workingString, current, StrTia, StrTch))
{
metaphoneData.Add(StrX);
current += 3;
break;
}
if (StringAt(workingString, current, StrTh) || StringAt(workingString, current, StrTth))
{
//special case 'thomas', 'thames' or germanic
if (StringAt(workingString, (current + 2), StrOm, StrAm) ||
StringAt(workingString, 0, StrVaNsp, StrVoNsp) || StringAt(workingString, 0, StrSch))
{
metaphoneData.Add(StrT);
}
else
{
metaphoneData.Add(StrO, StrT);
}
current += 2;
break;
}
if (StringAt(workingString, (current + 1), StrT, StrD))
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrT);
break;
case CharV:
if (workingString[current + 1] == CharV)
{
current += 2;
}
else
{
current += 1;
}
metaphoneData.Add(StrF);
break;
case CharW:
//can also be in middle of word
if (StringAt(workingString, current, StrWr))
{
metaphoneData.Add(StrR);
current += 2;
break;
}
if ((current == 0) && (IsVowel(workingString[current + 1]) ||
StringAt(workingString, current, StrWh)))
{
//Wasserman should match Vasserman
if (IsVowel(workingString[current + 1]))
{
metaphoneData.Add(StrA, StrF);
}
else
{
//need Uomo to match Womo
metaphoneData.Add(StrA);
}
}
//Arnow should match Arnoff
if ((current == last && current > 0 && IsVowel(workingString[current - 1])) ||
StringAt(workingString, (current - 1), StrEwski, StrEwsky, StrOwski, StrOwsky) ||
StringAt(workingString, 0, StrSch))
{
metaphoneData.Add(string.Empty, StrF);
current += 1;
break;
}
//polish e.g. 'filipowicz'
if (StringAt(workingString, current, StrWicz, StrWitz))
{
metaphoneData.Add(StrTs, StrFx);
current += 4;
break;
}
//else skip it
current += 1;
break;
case CharX:
//french e.g. breaux
if (!((current == last) &&
(StringAt(workingString, (current - 3), StrIau, StrEau) ||
StringAt(workingString, (current - 2), StrAu, StrOu))))
{
metaphoneData.Add(StrKs);
}
if (StringAt(workingString, (current + 1), StrC, StrX))
{
current += 2;
}
else
{
current += 1;
}
break;
case CharZ:
//chinese pinyin e.g. 'zhao'
if (workingString[current + 1] == CharH)
{
metaphoneData.Add(StrJ);
current += 2;
break;
}
else if (StringAt(workingString, (current + 1), StrZo, StrZi, StrZa) ||
(isSlavoGermanic && ((current > 0) && workingString[current - 1] != CharT)))
{
metaphoneData.Add(StrS, StrTs);
}
else
{
metaphoneData.Add(StrS);
}
if (workingString[current + 1] == CharZ)
{
current += 2;
}
else
{
current += 1;
}
break;
default:
current += 1;
break;
}
}
return(metaphoneData.ToString());
}
