1 <?php

2 namespace TYPO3\CMS\IndexedSearch\Utility;

3 

22 class DoubleMetaPhoneUtility {

23 

24  // properties

28  public $original = '';

29 

33  public $primary = '';

34 

38  public $secondary = '';

39 

43  public $length = 0;

44 

48  public $last = 0;

49 

53  public $current = 0;

54 

55  // methods

56  // TYPO3 specific API to this class. BEGIN

60  public function metaphone($string, $sys_language_uid = 0) {

61  $res = $this->DoubleMetaPhone($string);

62  // debug(array($string,$res['primary']));

63  return $res['primary'];

64  }

65 

66  // TYPO3 specific API to this class. END

67  // Public method

71  public function DoubleMetaPhone($string) {

72  $this->primary = '';

73  $this->secondary = '';

74  $this->current = 0;

75  $this->current = 0;

76  $this->length = strlen($string);

77  $this->last = $this->length - 1;

78  $this->original = $string . ' ';

79  $this->original = strtoupper($this->original);

80  // skip this at beginning of word

81  if ($this->StringAt($this->original, 0, 2, array('GN', 'KN', 'PN', 'WR', 'PS'))) {

82  $this->current++;

83  }

84  // Initial 'X' is pronounced 'Z' e.g. 'Xavier'

85  if ($this->original[0] === 'X') {

86  $this->primary .= 'S';

87  // 'Z' maps to 'S'

88  $this->secondary .= 'S';

89  $this->current++;

90  }

91  // main loop

92  while (strlen($this->primary) < 4 || strlen($this->secondary < 4)) {

93  if ($this->current >= $this->length) {

94  break;

95  }

96  switch (substr($this->original, $this->current, 1)) {

97  case 'A':

98 

99  case 'E':

100 

101  case 'I':

102 

103  case 'O':

104 

105  case 'U':

106 

107  case 'Y':

108  if ($this->current == 0) {

109  // all init vowels now map to 'A'

110  $this->primary .= 'A';

111  $this->secondary .= 'A';

112  }

113  $this->current += 1;

114  break;

115  case 'B':

116  // '-mb', e.g. "dumb", already skipped over ...

117  $this->primary .= 'P';

118  $this->secondary .= 'P';

119  if (substr($this->original, $this->current + 1, 1) == 'B') {

120  $this->current += 2;

121  } else {

122  $this->current += 1;

123  }

124  break;

125  case 'Ç':

126  $this->primary .= 'S';

127  $this->secondary .= 'S';

128  $this->current += 1;

129  break;

130  case 'C':

131  // various gremanic

132  if ($this->current > 1 && !$this->IsVowel($this->original, ($this->current - 2)) && $this->StringAt($this->original, $this->current - 1, 3, array('ACH')) && (substr($this->original, $this->current + 2, 1) != 'I' && (substr($this->original, $this->current + 2, 1) != 'E' || $this->StringAt($this->original, $this->current - 2, 6, array('BACHER', 'MACHER'))))) {

133  $this->primary .= 'K';

134  $this->secondary .= 'K';

135  $this->current += 2;

136  break;

137  }

138  // special case 'caesar'

139  if ($this->current == 0 && $this->StringAt($this->original, $this->current, 6, array('CAESAR'))) {

140  $this->primary .= 'S';

141  $this->secondary .= 'S';

142  $this->current += 2;

143  break;

144  }

145  // italian 'chianti'

146  if ($this->StringAt($this->original, $this->current, 4, array('CHIA'))) {

147  $this->primary .= 'K';

148  $this->secondary .= 'K';

149  $this->current += 2;

150  break;

151  }

152  if ($this->StringAt($this->original, $this->current, 2, array('CH'))) {

153  // find 'michael'

154  if ($this->current > 0 && $this->StringAt($this->original, $this->current, 4, array('CHAE'))) {

155  $this->primary .= 'K';

156  $this->secondary .= 'X';

157  $this->current += 2;

158  break;

159  }

160  // greek roots e.g. 'chemistry', 'chorus'

161  if ($this->current == 0 && ($this->StringAt($this->original, $this->current + 1, 5, array('HARAC', 'HARIS')) || $this->StringAt($this->original, $this->current + 1, 3, array('HOR', 'HYM', 'HIA', 'HEM'))) && !$this->StringAt($this->original, 0, 5, array('CHORE'))) {

162  $this->primary .= 'K';

163  $this->secondary .= 'K';

164  $this->current += 2;

165  break;

166  }

167  // germanic, greek, or otherwise 'ch' for 'kh' sound

168  if ($this->StringAt($this->original, 0, 4, array('VAN ', 'VON ')) || $this->StringAt($this->original, 0, 3, array('SCH')) || $this->StringAt($this->original, $this->current - 2, 6, array('ORCHES', 'ARCHIT', 'ORCHID')) || $this->StringAt($this->original, $this->current + 2, 1, array('T', 'S')) || ($this->StringAt($this->original, $this->current - 1, 1, array('A', 'O', 'U', 'E')) || $this->current == 0) && $this->StringAt($this->original, $this->current + 2, 1, array('L', 'R', 'N', 'M', 'B', 'H', 'F', 'V', 'W', ' '))) {

169  $this->primary .= 'K';

170  $this->secondary .= 'K';

171  } else {

172  if ($this->current > 0) {

173  if ($this->StringAt($this->original, 0, 2, array('MC'))) {

174  // e.g. 'McHugh'

175  $this->primary .= 'K';

176  $this->secondary .= 'K';

177  } else {

178  $this->primary .= 'X';

179  $this->secondary .= 'K';

180  }

181  } else {

182  $this->primary .= 'X';

183  $this->secondary .= 'X';

184  }

185  }

186  $this->current += 2;

187  break;

188  }

189  // e.g. 'czerny'

190  if ($this->StringAt($this->original, $this->current, 2, array('CZ')) && !$this->StringAt($this->original, ($this->current - 2), 4, array('WICZ'))) {

191  $this->primary .= 'S';

192  $this->secondary .= 'X';

193  $this->current += 2;

194  break;

195  }

196  // e.g. 'focaccia'

197  if ($this->StringAt($this->original, $this->current + 1, 3, array('CIA'))) {

198  $this->primary .= 'X';

199  $this->secondary .= 'X';

200  $this->current += 3;

201  break;

202  }

203  // double 'C', but not McClellan'

204  if ($this->StringAt($this->original, $this->current, 2, array('CC')) && !($this->current == 1 && $this->original[0] === 'M')) {

205  // 'bellocchio' but not 'bacchus'

206  if ($this->StringAt($this->original, $this->current + 2, 1, array('I', 'E', 'H')) && !$this->StringAt($this->original, ($this->current + 2), 2, array('HU'))) {

207  // 'accident', 'accede', 'succeed'

208  if ($this->current == 1 && substr($this->original, $this->current - 1, 1) == 'A' || $this->StringAt($this->original, $this->current - 1, 5, array('UCCEE', 'UCCES'))) {

209  $this->primary .= 'KS';

210  $this->secondary .= 'KS';

211  } else {

212  $this->primary .= 'X';

213  $this->secondary .= 'X';

214  }

215  $this->current += 3;

216  break;

217  } else {

218  // Pierce's rule

219  $this->primary .= 'K';

220  $this->secondary .= 'K';

221  $this->current += 2;

222  break;

223  }

224  }

225  if ($this->StringAt($this->original, $this->current, 2, array('CK', 'CG', 'CQ'))) {

226  $this->primary .= 'K';

227  $this->secondary .= 'K';

228  $this->current += 2;

229  break;

230  }

231  if ($this->StringAt($this->original, $this->current, 2, array('CI', 'CE', 'CY'))) {

232  // italian vs. english

233  if ($this->StringAt($this->original, $this->current, 3, array('CIO', 'CIE', 'CIA'))) {

234  $this->primary .= 'S';

235  $this->secondary .= 'X';

236  } else {

237  $this->primary .= 'S';

238  $this->secondary .= 'S';

239  }

240  $this->current += 2;

241  break;

242  }

243  // else

244  $this->primary .= 'K';

245  $this->secondary .= 'K';

246  // name sent in 'mac caffrey', 'mac gregor'

247  if ($this->StringAt($this->original, $this->current + 1, 2, array(' C', ' Q', ' G'))) {

248  $this->current += 3;

249  } else {

250  if ($this->StringAt($this->original, $this->current + 1, 1, array('C', 'K', 'Q')) && !$this->StringAt($this->original, ($this->current + 1), 2, array('CE', 'CI'))) {

251  $this->current += 2;

252  } else {

253  $this->current += 1;

254  }

255  }

256  break;

257  case 'D':

258  if ($this->StringAt($this->original, $this->current, 2, array('DG'))) {

259  if ($this->StringAt($this->original, $this->current + 2, 1, array('I', 'E', 'Y'))) {

260  // e.g. 'edge'

261  $this->primary .= 'J';

262  $this->secondary .= 'J';

263  $this->current += 3;

264  break;

265  } else {

266  // e.g. 'edgar'

267  $this->primary .= 'TK';

268  $this->secondary .= 'TK';

269  $this->current += 2;

270  break;

271  }

272  }

273  if ($this->StringAt($this->original, $this->current, 2, array('DT', 'DD'))) {

274  $this->primary .= 'T';

275  $this->secondary .= 'T';

276  $this->current += 2;

277  break;

278  }

279  // else

280  $this->primary .= 'T';

281  $this->secondary .= 'T';

282  $this->current += 1;

283  break;

284  case 'F':

285  if (substr($this->original, $this->current + 1, 1) == 'F') {

286  $this->current += 2;

287  } else {

288  $this->current += 1;

289  }

290  $this->primary .= 'F';

291  $this->secondary .= 'F';

292  break;

293  case 'G':

294  if (substr($this->original, $this->current + 1, 1) == 'H') {

295  if ($this->current > 0 && !$this->IsVowel($this->original, ($this->current - 1))) {

296  $this->primary .= 'K';

297  $this->secondary .= 'K';

298  $this->current += 2;

299  break;

300  }

301  if ($this->current < 3) {

302  // 'ghislane', 'ghiradelli'

303  if ($this->current == 0) {

304  if (substr($this->original, $this->current + 2, 1) == 'I') {

305  $this->primary .= 'J';

306  $this->secondary .= 'J';

307  } else {

308  $this->primary .= 'K';

309  $this->secondary .= 'K';

310  }

311  $this->current += 2;

312  break;

313  }

314  }

315  // Parker's rule (with some further refinements) - e.g. 'hugh'

316  if ($this->current > 1 && $this->StringAt($this->original, $this->current - 2, 1, array('B', 'H', 'D')) || $this->current > 2 && $this->StringAt($this->original, $this->current - 3, 1, array('B', 'H', 'D')) || $this->current > 3 && $this->StringAt($this->original, $this->current - 4, 1, array('B', 'H'))) {

317  $this->current += 2;

318  break;

319  } else {

320  // e.g. 'laugh', 'McLaughlin', 'cough', 'gough', 'rough', 'tough'

321  if ($this->current > 2 && substr($this->original, $this->current - 1, 1) == 'U' && $this->StringAt($this->original, $this->current - 3, 1, array('C', 'G', 'L', 'R', 'T'))) {

322  $this->primary .= 'F';

323  $this->secondary .= 'F';

324  } elseif ($this->current > 0 && substr($this->original, $this->current - 1, 1) != 'I') {

325  $this->primary .= 'K';

326  $this->secondary .= 'K';

327  }

328  $this->current += 2;

329  break;

330  }

331  }

332  if (substr($this->original, $this->current + 1, 1) == 'N') {

333  if ($this->current == 1 && $this->IsVowel($this->original, 0) && !$this->SlavoGermanic($this->original)) {

334  $this->primary .= 'KN';

335  $this->secondary .= 'N';

336  } else {

337  // not e.g. 'cagney'

338  if (!$this->StringAt($this->original, ($this->current + 2), 2, array('EY')) && substr($this->original, $this->current + 1) != 'Y' && !$this->SlavoGermanic($this->original)) {

339  $this->primary .= 'N';

340  $this->secondary .= 'KN';

341  } else {

342  $this->primary .= 'KN';

343  $this->secondary .= 'KN';

344  }

345  }

346  $this->current += 2;

347  break;

348  }

349  // 'tagliaro'

350  if ($this->StringAt($this->original, $this->current + 1, 2, array('LI')) && !$this->SlavoGermanic($this->original)) {

351  $this->primary .= 'KL';

352  $this->secondary .= 'L';

353  $this->current += 2;

354  break;

355  }

356  // -ges-, -gep-, -gel- at beginning

357  if ($this->current == 0 && (substr($this->original, $this->current + 1, 1) == 'Y' || $this->StringAt($this->original, $this->current + 1, 2, array(

358  'ES',

359  'EP',

360  'EB',

361  'EL',

362  'EY',

363  'IB',

364  'IL',

365  'IN',

366  'IE',

367  'EI',

368  'ER'

369  )))) {

370  $this->primary .= 'K';

371  $this->secondary .= 'J';

372  $this->current += 2;

373  break;

374  }

375  // -ger-, -gy-

376  if (($this->StringAt($this->original, $this->current + 1, 2, array('ER')) || substr($this->original, $this->current + 1, 1) == 'Y') && !$this->StringAt($this->original, 0, 6, array('DANGER', 'RANGER', 'MANGER')) && !$this->StringAt($this->original, ($this->current - 1), 1, array('E', 'I')) && !$this->StringAt($this->original, ($this->current - 1), 3, array('RGY', 'OGY'))) {

377  $this->primary .= 'K';

378  $this->secondary .= 'J';

379  $this->current += 2;

380  break;

381  }

382  // italian e.g. 'biaggi'

383  if ($this->StringAt($this->original, $this->current + 1, 1, array('E', 'I', 'Y')) || $this->StringAt($this->original, $this->current - 1, 4, array('AGGI', 'OGGI'))) {

384  // obvious germanic

385  if ($this->StringAt($this->original, 0, 4, array('VAN ', 'VON ')) || $this->StringAt($this->original, 0, 3, array('SCH')) || $this->StringAt($this->original, $this->current + 1, 2, array('ET'))) {

386  $this->primary .= 'K';

387  $this->secondary .= 'K';

388  } else {

389  // always soft if french ending

390  if ($this->StringAt($this->original, $this->current + 1, 4, array('IER '))) {

391  $this->primary .= 'J';

392  $this->secondary .= 'J';

393  } else {

394  $this->primary .= 'J';

395  $this->secondary .= 'K';

396  }

397  }

398  $this->current += 2;

399  break;

400  }

401  if (substr($this->original, $this->current + 1, 1) == 'G') {

402  $this->current += 2;

403  } else {

404  $this->current += 1;

405  }

406  $this->primary .= 'K';

407  $this->secondary .= 'K';

408  break;

409  case 'H':

410  // only keep if first & before vowel or btw. 2 vowels

411  if (($this->current == 0 || $this->IsVowel($this->original, $this->current - 1)) && $this->IsVowel($this->original, $this->current + 1)) {

412  $this->primary .= 'H';

413  $this->secondary .= 'H';

414  $this->current += 2;

415  } else {

416  $this->current += 1;

417  }

418  break;

419  case 'J':

420  // obvious spanish, 'jose', 'san jacinto'

421  if ($this->StringAt($this->original, $this->current, 4, array('JOSE')) || $this->StringAt($this->original, 0, 4, array('SAN '))) {

422  if ($this->current == 0 && substr($this->original, $this->current + 4, 1) == ' ' || $this->StringAt($this->original, 0, 4, array('SAN '))) {

423  $this->primary .= 'H';

424  $this->secondary .= 'H';

425  } else {

426  $this->primary .= 'J';

427  $this->secondary .= 'H';

428  }

429  $this->current += 1;

430  break;

431  }

432  if ($this->current == 0 && !$this->StringAt($this->original, $this->current, 4, array('JOSE'))) {

433  $this->primary .= 'J';

434  // Yankelovich/Jankelowicz

435  $this->secondary .= 'A';

436  } else {

437  // spanish pron. of .e.g. 'bajador'

438  if ($this->IsVowel($this->original, $this->current - 1) && !$this->SlavoGermanic($this->original) && (substr($this->original, $this->current + 1, 1) == 'A' || substr($this->original, $this->current + 1, 1) == 'O')) {

439  $this->primary .= 'J';

440  $this->secondary .= 'H';

441  } else {

442  if ($this->current == $this->last) {

443  $this->primary .= 'J';

444  $this->secondary .= '';

445  } else {

446  if (!$this->StringAt($this->original, ($this->current + 1), 1, array('L', 'T', 'K', 'S', 'N', 'M', 'B', 'Z')) && !$this->StringAt($this->original, ($this->current - 1), 1, array('S', 'K', 'L'))) {

447  $this->primary .= 'J';

448  $this->secondary .= 'J';

449  }

450  }

451  }

452  }

453  if (substr($this->original, $this->current + 1, 1) == 'J') {

454  // it could happen

455  $this->current += 2;

456  } else {

457  $this->current += 1;

458  }

459  break;

460  case 'K':

461  if (substr($this->original, $this->current + 1, 1) == 'K') {

462  $this->current += 2;

463  } else {

464  $this->current += 1;

465  }

466  $this->primary .= 'K';

467  $this->secondary .= 'K';

468  break;

469  case 'L':

470  if (substr($this->original, $this->current + 1, 1) == 'L') {

471  // spanish e.g. 'cabrillo', 'gallegos'

472  if ($this->current == $this->length - 3 && $this->StringAt($this->original, $this->current - 1, 4, array('ILLO', 'ILLA', 'ALLE')) || ($this->StringAt($this->original, $this->last - 1, 2, array('AS', 'OS')) || $this->StringAt($this->original, $this->last, 1, array('A', 'O'))) && $this->StringAt($this->original, $this->current - 1, 4, array('ALLE'))) {

473  $this->primary .= 'L';

474  $this->secondary .= '';

475  $this->current += 2;

476  break;

477  }

478  $this->current += 2;

479  } else {

480  $this->current += 1;

481  }

482  $this->primary .= 'L';

483  $this->secondary .= 'L';

484  break;

485  case 'M':

486  if ($this->StringAt($this->original, $this->current - 1, 3, array('UMB')) && ($this->current + 1 == $this->last || $this->StringAt($this->original, $this->current + 2, 2, array('ER'))) || substr($this->original, $this->current + 1, 1) == 'M') {

487  $this->current += 2;

488  } else {

489  $this->current += 1;

490  }

491  $this->primary .= 'M';

492  $this->secondary .= 'M';

493  break;

494  case 'N':

495  if (substr($this->original, $this->current + 1, 1) == 'N') {

496  $this->current += 2;

497  } else {

498  $this->current += 1;

499  }

500  $this->primary .= 'N';

501  $this->secondary .= 'N';

502  break;

503  case 'Ñ':

504  $this->current += 1;

505  $this->primary .= 'N';

506  $this->secondary .= 'N';

507  break;

508  case 'P':

509  if (substr($this->original, $this->current + 1, 1) == 'H') {

510  $this->current += 2;

511  $this->primary .= 'F';

512  $this->secondary .= 'F';

513  break;

514  }

515  // also account for "campbell" and "raspberry"

516  if ($this->StringAt($this->original, $this->current + 1, 1, array('P', 'B'))) {

517  $this->current += 2;

518  } else {

519  $this->current += 1;

520  }

521  $this->primary .= 'P';

522  $this->secondary .= 'P';

523  break;

524  case 'Q':

525  if (substr($this->original, $this->current + 1, 1) == 'Q') {

526  $this->current += 2;

527  } else {

528  $this->current += 1;

529  }

530  $this->primary .= 'K';

531  $this->secondary .= 'K';

532  break;

533  case 'R':

534  // french e.g. 'rogier', but exclude 'hochmeier'

535  if ($this->current == $this->last && !$this->SlavoGermanic($this->original) && $this->StringAt($this->original, $this->current - 2, 2, array('IE')) && !$this->StringAt($this->original, ($this->current - 4), 2, array('ME', 'MA'))) {

536  $this->primary .= '';

537  $this->secondary .= 'R';

538  } else {

539  $this->primary .= 'R';

540  $this->secondary .= 'R';

541  }

542  if (substr($this->original, $this->current + 1, 1) == 'R') {

543  $this->current += 2;

544  } else {

545  $this->current += 1;

546  }

547  break;

548  case 'S':

549  // special cases 'island', 'isle', 'carlisle', 'carlysle'

550  if ($this->StringAt($this->original, $this->current - 1, 3, array('ISL', 'YSL'))) {

551  $this->current += 1;

552  break;

553  }

554  // special case 'sugar-'

555  if ($this->current == 0 && $this->StringAt($this->original, $this->current, 5, array('SUGAR'))) {

556  $this->primary .= 'X';

557  $this->secondary .= 'S';

558  $this->current += 1;

559  break;

560  }

561  if ($this->StringAt($this->original, $this->current, 2, array('SH'))) {

562  // germanic

563  if ($this->StringAt($this->original, $this->current + 1, 4, array('HEIM', 'HOEK', 'HOLM', 'HOLZ'))) {

564  $this->primary .= 'S';

565  $this->secondary .= 'S';

566  } else {

567  $this->primary .= 'X';

568  $this->secondary .= 'X';

569  }

570  $this->current += 2;

571  break;

572  }

573  // italian & armenian

574  if ($this->StringAt($this->original, $this->current, 3, array('SIO', 'SIA')) || $this->StringAt($this->original, $this->current, 4, array('SIAN'))) {

575  if (!$this->SlavoGermanic($this->original)) {

576  $this->primary .= 'S';

577  $this->secondary .= 'X';

578  } else {

579  $this->primary .= 'S';

580  $this->secondary .= 'S';

581  }

582  $this->current += 3;

583  break;

584  }

585  // german & anglicisations, e.g. 'smith' match 'schmidt', 'snider' match 'schneider'

586  // also, -sz- in slavic language altho in hungarian it is pronounced 's'

587  if ($this->current == 0 && $this->StringAt($this->original, $this->current + 1, 1, array('M', 'N', 'L', 'W')) || $this->StringAt($this->original, $this->current + 1, 1, array('Z'))) {

588  $this->primary .= 'S';

589  $this->secondary .= 'X';

590  if ($this->StringAt($this->original, $this->current + 1, 1, array('Z'))) {

591  $this->current += 2;

592  } else {

593  $this->current += 1;

594  }

595  break;

596  }

597  if ($this->StringAt($this->original, $this->current, 2, array('SC'))) {

598  // Schlesinger's rule

599  if (substr($this->original, $this->current + 2, 1) == 'H') {

600  // dutch origin, e.g. 'school', 'schooner'

601  if ($this->StringAt($this->original, $this->current + 3, 2, array('OO', 'ER', 'EN', 'UY', 'ED', 'EM'))) {

602  // 'schermerhorn', 'schenker'

603  if ($this->StringAt($this->original, $this->current + 3, 2, array('ER', 'EN'))) {

604  $this->primary .= 'X';

605  $this->secondary .= 'SK';

606  } else {

607  $this->primary .= 'SK';

608  $this->secondary .= 'SK';

609  }

610  $this->current += 3;

611  break;

612  } else {

613  if ($this->current == 0 && !$this->IsVowel($this->original, 3) && substr($this->original, $this->current + 3, 1) != 'W') {

614  $this->primary .= 'X';

615  $this->secondary .= 'S';

616  } else {

617  $this->primary .= 'X';

618  $this->secondary .= 'X';

619  }

620  $this->current += 3;

621  break;

622  }

623  }

624  if ($this->StringAt($this->original, $this->current + 2, 1, array('I', 'E', 'Y'))) {

625  $this->primary .= 'S';

626  $this->secondary .= 'S';

627  $this->current += 3;

628  break;

629  }

630  // else

631  $this->primary .= 'SK';

632  $this->secondary .= 'SK';

633  $this->current += 3;

634  break;

635  }

636  // french e.g. 'resnais', 'artois'

637  if ($this->current == $this->last && $this->StringAt($this->original, $this->current - 2, 2, array('AI', 'OI'))) {

638  $this->primary .= '';

639  $this->secondary .= 'S';

640  } else {

641  $this->primary .= 'S';

642  $this->secondary .= 'S';

643  }

644  if ($this->StringAt($this->original, $this->current + 1, 1, array('S', 'Z'))) {

645  $this->current += 2;

646  } else {

647  $this->current += 1;

648  }

649  break;

650  case 'T':

651  if ($this->StringAt($this->original, $this->current, 4, array('TION'))) {

652  $this->primary .= 'X';

653  $this->secondary .= 'X';

654  $this->current += 3;

655  break;

656  }

657  if ($this->StringAt($this->original, $this->current, 3, array('TIA', 'TCH'))) {

658  $this->primary .= 'X';

659  $this->secondary .= 'X';

660  $this->current += 3;

661  break;

662  }

663  if ($this->StringAt($this->original, $this->current, 2, array('TH')) || $this->StringAt($this->original, $this->current, 3, array('TTH'))) {

664  // special case 'thomas', 'thames' or germanic

665  if ($this->StringAt($this->original, $this->current + 2, 2, array('OM', 'AM')) || $this->StringAt($this->original, 0, 4, array('VAN ', 'VON ')) || $this->StringAt($this->original, 0, 3, array('SCH'))) {

666  $this->primary .= 'T';

667  $this->secondary .= 'T';

668  } else {

669  $this->primary .= '0';

670  $this->secondary .= 'T';

671  }

672  $this->current += 2;

673  break;

674  }

675  if ($this->StringAt($this->original, $this->current + 1, 1, array('T', 'D'))) {

676  $this->current += 2;

677  } else {

678  $this->current += 1;

679  }

680  $this->primary .= 'T';

681  $this->secondary .= 'T';

682  break;

683  case 'V':

684  if (substr($this->original, $this->current + 1, 1) == 'V') {

685  $this->current += 2;

686  } else {

687  $this->current += 1;

688  }

689  $this->primary .= 'F';

690  $this->secondary .= 'F';

691  break;

692  case 'W':

693  // can also be in middle of word

694  if ($this->StringAt($this->original, $this->current, 2, array('WR'))) {

695  $this->primary .= 'R';

696  $this->secondary .= 'R';

697  $this->current += 2;

698  break;

699  }

700  if ($this->current == 0 && ($this->IsVowel($this->original, $this->current + 1) || $this->StringAt($this->original, $this->current, 2, array('WH')))) {

701  // Wasserman should match Vasserman

702  if ($this->IsVowel($this->original, $this->current + 1)) {

703  $this->primary .= 'A';

704  $this->secondary .= 'F';

705  } else {

706  // need Uomo to match Womo

707  $this->primary .= 'A';

708  $this->secondary .= 'A';

709  }

710  }

711  // Arnow should match Arnoff

712  if ($this->current == $this->last && $this->IsVowel($this->original, $this->current - 1) || $this->StringAt($this->original, $this->current - 1, 5, array('EWSKI', 'EWSKY', 'OWSKI', 'OWSKY')) || $this->StringAt($this->original, 0, 3, array('SCH'))) {

713  $this->primary .= '';

714  $this->secondary .= 'F';

715  $this->current += 1;

716  break;

717  }

718  // polish e.g. 'filipowicz'

719  if ($this->StringAt($this->original, $this->current, 4, array('WICZ', 'WITZ'))) {

720  $this->primary .= 'TS';

721  $this->secondary .= 'FX';

722  $this->current += 4;

723  break;

724  }

725  // else skip it

726  $this->current += 1;

727  break;

728  case 'X':

729  // french e.g. breaux

730  if (!($this->current == $this->last && ($this->StringAt($this->original, $this->current - 3, 3, array('IAU', 'EAU')) || $this->StringAt($this->original, $this->current - 2, 2, array('AU', 'OU'))))) {

731  $this->primary .= 'KS';

732  $this->secondary .= 'KS';

733  }

734  if ($this->StringAt($this->original, $this->current + 1, 1, array('C', 'X'))) {

735  $this->current += 2;

736  } else {

737  $this->current += 1;

738  }

739  break;

740  case 'Z':

741  // chinese pinyin e.g. 'zhao'

742  if (substr($this->original, $this->current + 1, 1) == 'H') {

743  $this->primary .= 'J';

744  $this->secondary .= 'J';

745  $this->current += 2;

746  break;

747  } elseif ($this->StringAt($this->original, $this->current + 1, 2, array('ZO', 'ZI', 'ZA')) || $this->SlavoGermanic($this->original) && ($this->current > 0 && substr($this->original, $this->current - 1, 1) != 'T')) {

748  $this->primary .= 'S';

749  $this->secondary .= 'TS';

750  } else {

751  $this->primary .= 'S';

752  $this->secondary .= 'S';

753  }

754  if (substr($this->original, $this->current + 1, 1) == 'Z') {

755  $this->current += 2;

756  } else {

757  $this->current += 1;

758  }

759  break;

760  default:

761  $this->current += 1;

762  }

763  }

764  // end while

765  $this->primary = substr($this->primary, 0, 4);

766  $this->secondary = substr($this->secondary, 0, 4);

767  $result['primary'] = $this->primary;

768  $result['secondary'] = $this->secondary;

769  return $result;

770  }

771 

772  // end of function MetaPhone

773  // Private methods

777  public function StringAt($string, $start, $length, $list) {

778  if ($start < 0 || $start >= strlen($string)) {

779  return 0;

780  }

781  $listCount = count($list);

782  for ($i = 0; $i < $listCount; $i++) {

783  if ($list[$i] == substr($string, $start, $length)) {

784  return 1;

785  }

786  }

787  return 0;

788  }

789 

798  public function IsVowel($string, $pos) {

799  return preg_match('/[AEIOUY]/', substr($string, $pos, 1));

800  }

801 

809  public function SlavoGermanic($string) {

810  return preg_match('/W|K|CZ|WITZ/', $string);

811  }

812 

813 }