1<?php 2 3/** 4 * webtrees: online genealogy 5 * Copyright (C) 2023 webtrees development team 6 * This program is free software: you can redistribute it and/or modify 7 * it under the terms of the GNU General Public License as published by 8 * the Free Software Foundation, either version 3 of the License, or 9 * (at your option) any later version. 10 * This program is distributed in the hope that it will be useful, 11 * but WITHOUT ANY WARRANTY; without even the implied warranty of 12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 13 * GNU General Public License for more details. 14 * You should have received a copy of the GNU General Public License 15 * along with this program. If not, see <https://www.gnu.org/licenses/>. 16 */ 17 18declare(strict_types=1); 19 20namespace Fisharebest\Webtrees\Services; 21 22use Closure; 23use Fisharebest\Webtrees\Date; 24use Fisharebest\Webtrees\DB; 25use Fisharebest\Webtrees\Family; 26use Fisharebest\Webtrees\Gedcom; 27use Fisharebest\Webtrees\GedcomRecord; 28use Fisharebest\Webtrees\Http\Exceptions\HttpServiceUnavailableException; 29use Fisharebest\Webtrees\I18N; 30use Fisharebest\Webtrees\Individual; 31use Fisharebest\Webtrees\Location; 32use Fisharebest\Webtrees\Media; 33use Fisharebest\Webtrees\Note; 34use Fisharebest\Webtrees\Place; 35use Fisharebest\Webtrees\Registry; 36use Fisharebest\Webtrees\Repository; 37use Fisharebest\Webtrees\SharedNote; 38use Fisharebest\Webtrees\Soundex; 39use Fisharebest\Webtrees\Source; 40use Fisharebest\Webtrees\Submission; 41use Fisharebest\Webtrees\Submitter; 42use Fisharebest\Webtrees\Tree; 43use Illuminate\Database\Query\Builder; 44use Illuminate\Database\Query\Expression; 45use Illuminate\Database\Query\JoinClause; 46use Illuminate\Support\Collection; 47 48use function addcslashes; 49use function array_filter; 50use function array_map; 51use function array_unique; 52use function explode; 53use function implode; 54use function mb_stripos; 55use function preg_match; 56use function preg_quote; 57use function preg_replace; 58use function str_ends_with; 59use function str_starts_with; 60 61use const PHP_INT_MAX; 62 63/** 64 * Search trees for genealogy records. 65 */ 66class SearchService 67{ 68 // Do not attempt to show search results larger than this/ 69 protected const MAX_SEARCH_RESULTS = 5000; 70 71 private TreeService $tree_service; 72 73 /** 74 * @param TreeService $tree_service 75 */ 76 public function __construct( 77 TreeService $tree_service 78 ) { 79 $this->tree_service = $tree_service; 80 } 81 82 /** 83 * @param array<Tree> $trees 84 * @param array<string> $search 85 * 86 * @return Collection<int,Family> 87 */ 88 public function searchFamilies(array $trees, array $search): Collection 89 { 90 $query = DB::table('families'); 91 92 $this->whereTrees($query, 'f_file', $trees); 93 $this->whereSearch($query, 'f_gedcom', $search); 94 95 return $query 96 ->get() 97 ->each($this->rowLimiter()) 98 ->map($this->familyRowMapper()) 99 ->filter(GedcomRecord::accessFilter()) 100 ->filter($this->rawGedcomFilter($search)); 101 } 102 103 /** 104 * Search for families by name. 105 * 106 * @param array<Tree> $trees 107 * @param array<string> $search 108 * @param int $offset 109 * @param int $limit 110 * 111 * @return Collection<int,Family> 112 */ 113 public function searchFamilyNames(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 114 { 115 $query = DB::table('families') 116 ->leftJoin('name AS husb_name', static function (JoinClause $join): void { 117 $join 118 ->on('husb_name.n_file', '=', 'families.f_file') 119 ->on('husb_name.n_id', '=', 'families.f_husb') 120 ->where('husb_name.n_type', '<>', '_MARNM'); 121 }) 122 ->leftJoin('name AS wife_name', static function (JoinClause $join): void { 123 $join 124 ->on('wife_name.n_file', '=', 'families.f_file') 125 ->on('wife_name.n_id', '=', 'families.f_wife') 126 ->where('wife_name.n_type', '<>', '_MARNM'); 127 }); 128 129 $prefix = DB::connection()->getTablePrefix(); 130 $field = new Expression('COALESCE(' . $prefix . "husb_name.n_full, '') || COALESCE(" . $prefix . "wife_name.n_full, '')"); 131 132 $this->whereTrees($query, 'f_file', $trees); 133 $this->whereSearch($query, $field, $search); 134 135 $query 136 ->orderBy('husb_name.n_sort') 137 ->orderBy('wife_name.n_sort') 138 ->select(['families.*', 'husb_name.n_sort', 'wife_name.n_sort']); 139 140 return $this->paginateQuery($query, $this->familyRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 141 } 142 143 /** 144 * @param Place $place 145 * 146 * @return Collection<int,Family> 147 */ 148 public function searchFamiliesInPlace(Place $place): Collection 149 { 150 return DB::table('families') 151 ->join('placelinks', static function (JoinClause $query) { 152 $query 153 ->on('families.f_file', '=', 'placelinks.pl_file') 154 ->on('families.f_id', '=', 'placelinks.pl_gid'); 155 }) 156 ->where('f_file', '=', $place->tree()->id()) 157 ->where('pl_p_id', '=', $place->id()) 158 ->select(['families.*']) 159 ->get() 160 ->each($this->rowLimiter()) 161 ->map($this->familyRowMapper()) 162 ->filter(GedcomRecord::accessFilter()); 163 } 164 165 /** 166 * @param array<Tree> $trees 167 * @param array<string> $search 168 * 169 * @return Collection<int,Individual> 170 */ 171 public function searchIndividuals(array $trees, array $search): Collection 172 { 173 $query = DB::table('individuals'); 174 175 $this->whereTrees($query, 'i_file', $trees); 176 $this->whereSearch($query, 'i_gedcom', $search); 177 178 return $query 179 ->get() 180 ->each($this->rowLimiter()) 181 ->map($this->individualRowMapper()) 182 ->filter(GedcomRecord::accessFilter()) 183 ->filter($this->rawGedcomFilter($search)); 184 } 185 186 /** 187 * Search for individuals by name. 188 * 189 * @param array<Tree> $trees 190 * @param array<string> $search 191 * @param int $offset 192 * @param int $limit 193 * 194 * @return Collection<int,Individual> 195 */ 196 public function searchIndividualNames(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 197 { 198 $query = DB::table('individuals') 199 ->join('name', static function (JoinClause $join): void { 200 $join 201 ->on('name.n_file', '=', 'individuals.i_file') 202 ->on('name.n_id', '=', 'individuals.i_id'); 203 }) 204 ->orderBy('n_sort') 205 ->select(['individuals.*', 'n_sort']); 206 207 $this->whereTrees($query, 'i_file', $trees); 208 $this->whereSearch($query, 'n_full', $search); 209 210 return $this->paginateQuery($query, $this->individualRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 211 } 212 213 /** 214 * @param Place $place 215 * 216 * @return Collection<int,Individual> 217 */ 218 public function searchIndividualsInPlace(Place $place): Collection 219 { 220 return DB::table('individuals') 221 ->join('placelinks', static function (JoinClause $join) { 222 $join 223 ->on('i_file', '=', 'pl_file') 224 ->on('i_id', '=', 'pl_gid'); 225 }) 226 ->where('i_file', '=', $place->tree()->id()) 227 ->where('pl_p_id', '=', $place->id()) 228 ->select(['individuals.*']) 229 ->get() 230 ->each($this->rowLimiter()) 231 ->map($this->individualRowMapper()) 232 ->filter(GedcomRecord::accessFilter()); 233 } 234 235 /** 236 * Search for submissions. 237 * 238 * @param array<Tree> $trees 239 * @param array<string> $search 240 * @param int $offset 241 * @param int $limit 242 * 243 * @return Collection<int,Location> 244 */ 245 public function searchLocations(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 246 { 247 $query = DB::table('other') 248 ->where('o_type', '=', Location::RECORD_TYPE); 249 250 $this->whereTrees($query, 'o_file', $trees); 251 $this->whereSearch($query, 'o_gedcom', $search); 252 253 return $this->paginateQuery($query, $this->locationRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 254 } 255 256 /** 257 * Search for media objects. 258 * 259 * @param array<Tree> $trees 260 * @param array<string> $search 261 * @param int $offset 262 * @param int $limit 263 * 264 * @return Collection<int,Media> 265 */ 266 public function searchMedia(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 267 { 268 $query = DB::table('media'); 269 270 $this->whereTrees($query, 'media.m_file', $trees); 271 $this->whereSearch($query, 'm_gedcom', $search); 272 273 return $this->paginateQuery($query, $this->mediaRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 274 } 275 276 /** 277 * Search for notes. 278 * 279 * @param array<Tree> $trees 280 * @param array<string> $search 281 * @param int $offset 282 * @param int $limit 283 * 284 * @return Collection<int,Note> 285 */ 286 public function searchNotes(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 287 { 288 $query = DB::table('other') 289 ->where('o_type', '=', Note::RECORD_TYPE); 290 291 $this->whereTrees($query, 'o_file', $trees); 292 $this->whereSearch($query, 'o_gedcom', $search); 293 294 return $this->paginateQuery($query, $this->noteRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 295 } 296 297 /** 298 * Search for notes. 299 * 300 * @param array<Tree> $trees 301 * @param array<string> $search 302 * @param int $offset 303 * @param int $limit 304 * 305 * @return Collection<int,SharedNote> 306 */ 307 public function searchSharedNotes(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 308 { 309 $query = DB::table('other') 310 ->where('o_type', '=', SharedNote::RECORD_TYPE); 311 312 $this->whereTrees($query, 'o_file', $trees); 313 $this->whereSearch($query, 'o_gedcom', $search); 314 315 return $this->paginateQuery($query, $this->sharedNoteRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 316 } 317 318 /** 319 * Search for repositories. 320 * 321 * @param array<Tree> $trees 322 * @param array<string> $search 323 * @param int $offset 324 * @param int $limit 325 * 326 * @return Collection<int,Repository> 327 */ 328 public function searchRepositories(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 329 { 330 $query = DB::table('other') 331 ->where('o_type', '=', Repository::RECORD_TYPE); 332 333 $this->whereTrees($query, 'o_file', $trees); 334 $this->whereSearch($query, 'o_gedcom', $search); 335 336 return $this->paginateQuery($query, $this->repositoryRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 337 } 338 339 /** 340 * Search for sources. 341 * 342 * @param array<Tree> $trees 343 * @param array<string> $search 344 * @param int $offset 345 * @param int $limit 346 * 347 * @return Collection<int,Source> 348 */ 349 public function searchSources(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 350 { 351 $query = DB::table('sources'); 352 353 $this->whereTrees($query, 's_file', $trees); 354 $this->whereSearch($query, 's_gedcom', $search); 355 356 return $this->paginateQuery($query, $this->sourceRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 357 } 358 359 /** 360 * Search for sources by name. 361 * 362 * @param array<Tree> $trees 363 * @param array<string> $search 364 * @param int $offset 365 * @param int $limit 366 * 367 * @return Collection<int,Source> 368 */ 369 public function searchSourcesByName(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 370 { 371 $query = DB::table('sources') 372 ->orderBy('s_name'); 373 374 $this->whereTrees($query, 's_file', $trees); 375 $this->whereSearch($query, 's_name', $search); 376 377 return $this->paginateQuery($query, $this->sourceRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 378 } 379 380 /** 381 * Search for sources. 382 * 383 * @param array<Tree> $trees 384 * @param array<string> $search 385 * @param int $offset 386 * @param int $limit 387 * 388 * @return Collection<int,string> 389 */ 390 public function searchSurnames(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 391 { 392 $query = DB::table('name'); 393 394 $this->whereTrees($query, 'n_file', $trees); 395 $this->whereSearch($query, 'n_surname', $search); 396 397 return $query 398 ->groupBy(['n_surname']) 399 ->orderBy('n_surname') 400 ->skip($offset) 401 ->take($limit) 402 ->pluck('n_surname'); 403 } 404 405 /** 406 * Search for submissions. 407 * 408 * @param array<Tree> $trees 409 * @param array<string> $search 410 * @param int $offset 411 * @param int $limit 412 * 413 * @return Collection<int,Submission> 414 */ 415 public function searchSubmissions(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 416 { 417 $query = DB::table('other') 418 ->where('o_type', '=', Submission::RECORD_TYPE); 419 420 $this->whereTrees($query, 'o_file', $trees); 421 $this->whereSearch($query, 'o_gedcom', $search); 422 423 return $this->paginateQuery($query, $this->submissionRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 424 } 425 426 /** 427 * Search for submitters. 428 * 429 * @param array<Tree> $trees 430 * @param array<string> $search 431 * @param int $offset 432 * @param int $limit 433 * 434 * @return Collection<int,Submitter> 435 */ 436 public function searchSubmitters(array $trees, array $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 437 { 438 $query = DB::table('other') 439 ->where('o_type', '=', Submitter::RECORD_TYPE); 440 441 $this->whereTrees($query, 'o_file', $trees); 442 $this->whereSearch($query, 'o_gedcom', $search); 443 444 return $this->paginateQuery($query, $this->submitterRowMapper(), GedcomRecord::accessFilter(), $offset, $limit); 445 } 446 447 /** 448 * Search for places. 449 * 450 * @param Tree $tree 451 * @param string $search 452 * @param int $offset 453 * @param int $limit 454 * 455 * @return Collection<int,Place> 456 */ 457 public function searchPlaces(Tree $tree, string $search, int $offset = 0, int $limit = PHP_INT_MAX): Collection 458 { 459 $query = DB::table('places AS p0') 460 ->where('p0.p_file', '=', $tree->id()) 461 ->leftJoin('places AS p1', 'p1.p_id', '=', 'p0.p_parent_id') 462 ->leftJoin('places AS p2', 'p2.p_id', '=', 'p1.p_parent_id') 463 ->leftJoin('places AS p3', 'p3.p_id', '=', 'p2.p_parent_id') 464 ->leftJoin('places AS p4', 'p4.p_id', '=', 'p3.p_parent_id') 465 ->leftJoin('places AS p5', 'p5.p_id', '=', 'p4.p_parent_id') 466 ->leftJoin('places AS p6', 'p6.p_id', '=', 'p5.p_parent_id') 467 ->leftJoin('places AS p7', 'p7.p_id', '=', 'p6.p_parent_id') 468 ->leftJoin('places AS p8', 'p8.p_id', '=', 'p7.p_parent_id') 469 ->orderBy('p0.p_place') 470 ->orderBy('p1.p_place') 471 ->orderBy('p2.p_place') 472 ->orderBy('p3.p_place') 473 ->orderBy('p4.p_place') 474 ->orderBy('p5.p_place') 475 ->orderBy('p6.p_place') 476 ->orderBy('p7.p_place') 477 ->orderBy('p8.p_place') 478 ->select([ 479 'p0.p_place AS place0', 480 'p1.p_place AS place1', 481 'p2.p_place AS place2', 482 'p3.p_place AS place3', 483 'p4.p_place AS place4', 484 'p5.p_place AS place5', 485 'p6.p_place AS place6', 486 'p7.p_place AS place7', 487 'p8.p_place AS place8', 488 ]); 489 490 // Filter each level of the hierarchy. 491 foreach (explode(',', $search, 9) as $level => $string) { 492 $query->where('p' . $level . '.p_place', $this->iLike(), '%' . addcslashes($string, '\\%_') . '%'); 493 } 494 495 $row_mapper = static function (object $row) use ($tree): Place { 496 $place = implode(', ', array_filter((array) $row)); 497 498 return new Place($place, $tree); 499 }; 500 501 $filter = static function (): bool { 502 return true; 503 }; 504 505 return $this->paginateQuery($query, $row_mapper, $filter, $offset, $limit); 506 } 507 508 /** 509 * @param Tree $tree 510 * @param array<string,string> $fields 511 * @param array<string,string> $modifiers 512 * 513 * @return Collection<int,Individual> 514 */ 515 public function searchIndividualsAdvanced(Tree $tree, array $fields, array $modifiers): Collection 516 { 517 $fields = array_filter($fields, static fn (string $x): bool => $x !== ''); 518 519 $query = DB::table('individuals') 520 ->where('i_file', '=', $tree->id()) 521 ->select(['individuals.*']) 522 ->distinct(); 523 524 // Join the following tables 525 $father_name = false; 526 $mother_name = false; 527 $spouse_family = false; 528 $indi_name = false; 529 $indi_dates = []; 530 $fam_dates = []; 531 $indi_plac = false; 532 $fam_plac = false; 533 534 foreach ($fields as $field_name => $field_value) { 535 if (str_starts_with($field_name, 'FATHER:NAME')) { 536 $father_name = true; 537 } elseif (str_starts_with($field_name, 'MOTHER:NAME')) { 538 $mother_name = true; 539 } elseif (str_starts_with($field_name, 'INDI:NAME:GIVN')) { 540 $indi_name = true; 541 } elseif (str_starts_with($field_name, 'INDI:NAME:SURN')) { 542 $indi_name = true; 543 } elseif (str_starts_with($field_name, 'FAM:')) { 544 $spouse_family = true; 545 if (str_ends_with($field_name, ':DATE')) { 546 $fam_dates[] = explode(':', $field_name)[1]; 547 } elseif (str_ends_with($field_name, ':PLAC')) { 548 $fam_plac = true; 549 } 550 } elseif (str_starts_with($field_name, 'INDI:')) { 551 if (str_ends_with($field_name, ':DATE')) { 552 $indi_dates[] = explode(':', $field_name)[1]; 553 } elseif (str_ends_with($field_name, ':PLAC')) { 554 $indi_plac = true; 555 } 556 } 557 } 558 559 if ($father_name || $mother_name) { 560 $query->join('link AS l1', static function (JoinClause $join): void { 561 $join 562 ->on('l1.l_file', '=', 'individuals.i_file') 563 ->on('l1.l_from', '=', 'individuals.i_id') 564 ->where('l1.l_type', '=', 'FAMC'); 565 }); 566 567 if ($father_name) { 568 $query->join('link AS l2', static function (JoinClause $join): void { 569 $join 570 ->on('l2.l_file', '=', 'l1.l_file') 571 ->on('l2.l_from', '=', 'l1.l_to') 572 ->where('l2.l_type', '=', 'HUSB'); 573 }); 574 $query->join('name AS father_name', static function (JoinClause $join): void { 575 $join 576 ->on('father_name.n_file', '=', 'l2.l_file') 577 ->on('father_name.n_id', '=', 'l2.l_to'); 578 }); 579 } 580 581 if ($mother_name) { 582 $query->join('link AS l3', static function (JoinClause $join): void { 583 $join 584 ->on('l3.l_file', '=', 'l1.l_file') 585 ->on('l3.l_from', '=', 'l1.l_to') 586 ->where('l3.l_type', '=', 'WIFE'); 587 }); 588 $query->join('name AS mother_name', static function (JoinClause $join): void { 589 $join 590 ->on('mother_name.n_file', '=', 'l3.l_file') 591 ->on('mother_name.n_id', '=', 'l3.l_to'); 592 }); 593 } 594 } 595 596 if ($spouse_family) { 597 $query->join('link AS l4', static function (JoinClause $join): void { 598 $join 599 ->on('l4.l_file', '=', 'individuals.i_file') 600 ->on('l4.l_from', '=', 'individuals.i_id') 601 ->where('l4.l_type', '=', 'FAMS'); 602 }); 603 $query->join('families AS spouse_families', static function (JoinClause $join): void { 604 $join 605 ->on('spouse_families.f_file', '=', 'l4.l_file') 606 ->on('spouse_families.f_id', '=', 'l4.l_to'); 607 }); 608 } 609 610 if ($indi_name) { 611 $query->join('name AS individual_name', static function (JoinClause $join): void { 612 $join 613 ->on('individual_name.n_file', '=', 'individuals.i_file') 614 ->on('individual_name.n_id', '=', 'individuals.i_id'); 615 }); 616 } 617 618 foreach (array_unique($indi_dates) as $indi_date) { 619 $query->join('dates AS date_' . $indi_date, static function (JoinClause $join) use ($indi_date): void { 620 $join 621 ->on('date_' . $indi_date . '.d_file', '=', 'individuals.i_file') 622 ->on('date_' . $indi_date . '.d_gid', '=', 'individuals.i_id'); 623 }); 624 } 625 626 foreach (array_unique($fam_dates) as $fam_date) { 627 $query->join('dates AS date_' . $fam_date, static function (JoinClause $join) use ($fam_date): void { 628 $join 629 ->on('date_' . $fam_date . '.d_file', '=', 'spouse_families.f_file') 630 ->on('date_' . $fam_date . '.d_gid', '=', 'spouse_families.f_id'); 631 }); 632 } 633 634 if ($indi_plac) { 635 $query->join('placelinks AS individual_placelinks', static function (JoinClause $join): void { 636 $join 637 ->on('individual_placelinks.pl_file', '=', 'individuals.i_file') 638 ->on('individual_placelinks.pl_gid', '=', 'individuals.i_id'); 639 }); 640 $query->join('places AS individual_places', static function (JoinClause $join): void { 641 $join 642 ->on('individual_places.p_file', '=', 'individual_placelinks.pl_file') 643 ->on('individual_places.p_id', '=', 'individual_placelinks.pl_p_id'); 644 }); 645 } 646 647 if ($fam_plac) { 648 $query->join('placelinks AS family_placelinks', static function (JoinClause $join): void { 649 $join 650 ->on('family_placelinks.pl_file', '=', 'spouse_families.f_file') 651 ->on('family_placelinks.pl_gid', '=', 'spouse_families.f_id'); 652 }); 653 $query->join('places AS family_places', static function (JoinClause $join): void { 654 $join 655 ->on('family_places.p_file', '=', 'family_placelinks.pl_file') 656 ->on('family_places.p_id', '=', 'family_placelinks.pl_p_id'); 657 }); 658 } 659 660 foreach ($fields as $field_name => $field_value) { 661 $parts = explode(':', $field_name . ':::'); 662 if (str_starts_with($field_name, 'INDI:NAME:')) { 663 switch ($field_name) { 664 case 'INDI:NAME:GIVN': 665 switch ($modifiers[$field_name]) { 666 case 'EXACT': 667 $query->where('individual_name.n_givn', '=', $field_value); 668 break; 669 case 'BEGINS': 670 $query->where('individual_name.n_givn', $this->iLike(), $field_value . '%'); 671 break; 672 case 'CONTAINS': 673 $query->where('individual_name.n_givn', $this->iLike(), '%' . $field_value . '%'); 674 break; 675 case 'SDX_STD': 676 $sdx = Soundex::russell($field_value); 677 if ($sdx !== '') { 678 $this->wherePhonetic($query, 'individual_name.n_soundex_givn_std', $sdx); 679 } else { 680 // No phonetic content? Use a substring match 681 $query->where('individual_name.n_givn', $this->iLike(), '%' . $field_value . '%'); 682 } 683 break; 684 case 'SDX': // SDX uses DM by default. 685 case 'SDX_DM': 686 $sdx = Soundex::daitchMokotoff($field_value); 687 if ($sdx !== '') { 688 $this->wherePhonetic($query, 'individual_name.n_soundex_givn_dm', $sdx); 689 } else { 690 // No phonetic content? Use a substring match 691 $query->where('individual_name.n_givn', $this->iLike(), '%' . $field_value . '%'); 692 } 693 break; 694 } 695 unset($fields[$field_name]); 696 break; 697 case 'INDI:NAME:SURN': 698 switch ($modifiers[$field_name]) { 699 case 'EXACT': 700 $query->where(function (Builder $query) use ($field_value): void { 701 $query 702 ->where('individual_name.n_surn', '=', $field_value) 703 ->orWhere('individual_name.n_surname', '=', $field_value); 704 }); 705 break; 706 case 'BEGINS': 707 $query->where(function (Builder $query) use ($field_value): void { 708 $query 709 ->where('individual_name.n_surn', $this->iLike(), $field_value . '%') 710 ->orWhere('individual_name.n_surname', $this->iLike(), $field_value . '%'); 711 }); 712 break; 713 case 'CONTAINS': 714 $query->where(function (Builder $query) use ($field_value): void { 715 $query 716 ->where('individual_name.n_surn', $this->iLike(), '%' . $field_value . '%') 717 ->orWhere('individual_name.n_surname', $this->iLike(), '%' . $field_value . '%'); 718 }); 719 break; 720 case 'SDX_STD': 721 $sdx = Soundex::russell($field_value); 722 if ($sdx !== '') { 723 $this->wherePhonetic($query, 'individual_name.n_soundex_surn_std', $sdx); 724 } else { 725 // No phonetic content? Use a substring match 726 $query->where(function (Builder $query) use ($field_value): void { 727 $query 728 ->where('individual_name.n_surn', $this->iLike(), '%' . $field_value . '%') 729 ->orWhere('individual_name.n_surname', $this->iLike(), '%' . $field_value . '%'); 730 }); 731 } 732 break; 733 case 'SDX': // SDX uses DM by default. 734 case 'SDX_DM': 735 $sdx = Soundex::daitchMokotoff($field_value); 736 if ($sdx !== '') { 737 $this->wherePhonetic($query, 'individual_name.n_soundex_surn_dm', $sdx); 738 } else { 739 // No phonetic content? Use a substring match 740 $query->where(function (Builder $query) use ($field_value): void { 741 $query 742 ->where('individual_name.n_surn', $this->iLike(), '%' . $field_value . '%') 743 ->orWhere('individual_name.n_surname', $this->iLike(), '%' . $field_value . '%'); 744 }); 745 } 746 break; 747 } 748 unset($fields[$field_name]); 749 break; 750 case 'INDI:NAME:NICK': 751 case 'INDI:NAME:_MARNM': 752 case 'INDI:NAME:_HEB': 753 case 'INDI:NAME:_AKA': 754 $like = "%\n1 NAME%\n2 " . $parts[2] . ' %' . preg_quote($field_value, '/') . '%'; 755 $query->where('individuals.i_gedcom', $this->iLike(), $like); 756 break; 757 } 758 } elseif (str_starts_with($field_name, 'INDI:') && str_ends_with($field_name, ':DATE')) { 759 $date = new Date($field_value); 760 if ($date->isOK()) { 761 $delta = 365 * (int) ($modifiers[$field_name] ?? 0); 762 $query 763 ->where('date_' . $parts[1] . '.d_fact', '=', $parts[1]) 764 ->where('date_' . $parts[1] . '.d_julianday1', '>=', $date->minimumJulianDay() - $delta) 765 ->where('date_' . $parts[1] . '.d_julianday2', '<=', $date->maximumJulianDay() + $delta); 766 } 767 unset($fields[$field_name]); 768 } elseif (str_starts_with($field_name, 'FAM:') && str_ends_with($field_name, ':DATE')) { 769 $date = new Date($field_value); 770 if ($date->isOK()) { 771 $delta = 365 * (int) ($modifiers[$field_name] ?? 0); 772 $query 773 ->where('date_' . $parts[1] . '.d_fact', '=', $parts[1]) 774 ->where('date_' . $parts[1] . '.d_julianday1', '>=', $date->minimumJulianDay() - $delta) 775 ->where('date_' . $parts[1] . '.d_julianday2', '<=', $date->maximumJulianDay() + $delta); 776 } 777 unset($fields[$field_name]); 778 } elseif (str_starts_with($field_name, 'INDI:') && str_ends_with($field_name, ':PLAC')) { 779 // SQL can only link a place to a person/family, not to an event. 780 $query->where('individual_places.p_id', '=', $field_value); 781 } elseif (str_starts_with($field_name, 'FAM:') && str_ends_with($field_name, ':PLAC')) { 782 // SQL can only link a place to a person/family, not to an event. 783 $query->where('family_places.p_id', '=', $field_value); 784 } elseif (str_starts_with($field_name, 'MOTHER:NAME:') || str_starts_with($field_name, 'FATHER:NAME:')) { 785 $table = str_starts_with($field_name, 'FATHER:NAME:') ? 'father_name' : 'mother_name'; 786 switch ($parts[2]) { 787 case 'GIVN': 788 switch ($modifiers[$field_name]) { 789 case 'EXACT': 790 $query->where($table . '.n_givn', '=', $field_value); 791 break; 792 case 'BEGINS': 793 $query->where($table . '.n_givn', $this->iLike(), $field_value . '%'); 794 break; 795 case 'CONTAINS': 796 $query->where($table . '.n_givn', $this->iLike(), '%' . $field_value . '%'); 797 break; 798 case 'SDX_STD': 799 $sdx = Soundex::russell($field_value); 800 if ($sdx !== '') { 801 $this->wherePhonetic($query, $table . '.n_soundex_givn_std', $sdx); 802 } else { 803 // No phonetic content? Use a substring match 804 $query->where($table . '.n_givn', $this->iLike(), '%' . $field_value . '%'); 805 } 806 break; 807 case 'SDX': // SDX uses DM by default. 808 case 'SDX_DM': 809 $sdx = Soundex::daitchMokotoff($field_value); 810 if ($sdx !== '') { 811 $this->wherePhonetic($query, $table . '.n_soundex_givn_dm', $sdx); 812 } else { 813 // No phonetic content? Use a substring match 814 $query->where($table . '.n_givn', $this->iLike(), '%' . $field_value . '%'); 815 } 816 break; 817 } 818 break; 819 case 'SURN': 820 switch ($modifiers[$field_name]) { 821 case 'EXACT': 822 $query->where($table . '.n_surn', '=', $field_value); 823 break; 824 case 'BEGINS': 825 $query->where($table . '.n_surn', $this->iLike(), $field_value . '%'); 826 break; 827 case 'CONTAINS': 828 $query->where($table . '.n_surn', $this->iLike(), '%' . $field_value . '%'); 829 break; 830 case 'SDX_STD': 831 $sdx = Soundex::russell($field_value); 832 if ($sdx !== '') { 833 $this->wherePhonetic($query, $table . '.n_soundex_surn_std', $sdx); 834 } else { 835 // No phonetic content? Use a substring match 836 $query->where($table . '.n_surn', $this->iLike(), '%' . $field_value . '%'); 837 } 838 break; 839 case 'SDX': // SDX uses DM by default. 840 case 'SDX_DM': 841 $sdx = Soundex::daitchMokotoff($field_value); 842 if ($sdx !== '') { 843 $this->wherePhonetic($query, $table . '.n_soundex_surn_dm', $sdx); 844 } else { 845 // No phonetic content? Use a substring match 846 $query->where($table . '.n_surn', $this->iLike(), '%' . $field_value . '%'); 847 } 848 break; 849 } 850 break; 851 } 852 unset($fields[$field_name]); 853 } elseif (str_starts_with($field_name, 'FAM:')) { 854 // e.g. searches for occupation, religion, note, etc. 855 // Initial matching only. Need PHP to apply filter. 856 $query->where('spouse_families.f_gedcom', $this->iLike(), "%\n1 " . $parts[1] . ' %' . $field_value . '%'); 857 } elseif (str_starts_with($field_name, 'INDI:') && str_ends_with($field_name, ':TYPE')) { 858 // Initial matching only. Need PHP to apply filter. 859 $query->where('individuals.i_gedcom', $this->iLike(), "%\n1 " . $parts[1] . "%\n2 TYPE %" . $field_value . '%'); 860 } elseif (str_starts_with($field_name, 'INDI:')) { 861 // e.g. searches for occupation, religion, note, etc. 862 // Initial matching only. Need PHP to apply filter. 863 $query->where('individuals.i_gedcom', $this->iLike(), "%\n1 " . $parts[1] . '%' . $parts[2] . '%' . $field_value . '%'); 864 } 865 } 866 867 return $query 868 ->get() 869 ->each($this->rowLimiter()) 870 ->map($this->individualRowMapper()) 871 ->filter(GedcomRecord::accessFilter()) 872 ->filter(static function (Individual $individual) use ($fields, $tree): bool { 873 // Check for searches which were only partially matched by SQL 874 foreach ($fields as $field_name => $field_value) { 875 $parts = explode(':', $field_name . '::::'); 876 877 if (str_starts_with($field_name, 'INDI:NAME:') && $field_name !== 'INDI:NAME:GIVN' && $field_name !== 'INDI:NAME:SURN') { 878 $regex = '/\n1 NAME.*(?:\n2.*)*\n2 ' . $parts[2] . ' .*' . preg_quote($field_value, '/') . '/i'; 879 880 if (preg_match($regex, $individual->gedcom()) === 1) { 881 continue; 882 } 883 884 return false; 885 } 886 887 if (str_starts_with($field_name, 'INDI:') && str_ends_with($field_name, ':PLAC')) { 888 $place = Place::find((int) $field_value, $tree); 889 890 foreach ($individual->facts([$parts[1]]) as $fact) { 891 if ($fact->place()->gedcomName() === $place->gedcomName() || str_ends_with($fact->place()->gedcomName(), ', ' . $place->gedcomName())) { 892 continue 2; 893 } 894 } 895 return false; 896 } 897 898 if (str_starts_with($field_name, 'FAM:') && str_ends_with($field_name, ':PLAC')) { 899 $place = Place::find((int) $field_value, $tree); 900 901 foreach ($individual->spouseFamilies() as $family) { 902 foreach ($family->facts([$parts[1]]) as $fact) { 903 if ($fact->place()->gedcomName() === $place->gedcomName() || str_ends_with($fact->place()->gedcomName(), ', ' . $place->gedcomName())) { 904 continue 3; 905 } 906 } 907 } 908 return false; 909 } 910 911 $regex = '/' . preg_quote($field_value, '/') . '/i'; 912 913 if ($field_name === 'INDI:FACT:TYPE' || $field_name === 'INDI:EVEN:TYPE' || $field_name === 'INDI:CHAN:_WT_USER') { 914 foreach ($individual->facts([$parts[1]]) as $fact) { 915 if (preg_match($regex, $fact->attribute($parts[2])) === 1) { 916 continue 2; 917 } 918 } 919 920 return false; 921 } 922 923 if (str_starts_with($field_name, 'INDI:')) { 924 foreach ($individual->facts([$parts[1]]) as $fact) { 925 if (preg_match($regex, $fact->value()) === 1) { 926 continue 2; 927 } 928 } 929 930 return false; 931 } 932 933 if (str_starts_with($field_name, 'FAM:')) { 934 foreach ($individual->spouseFamilies() as $family) { 935 foreach ($family->facts([$parts[1]]) as $fact) { 936 if (preg_match($regex, $fact->value()) === 1) { 937 continue 3; 938 } 939 } 940 } 941 return false; 942 } 943 } 944 945 return true; 946 }); 947 } 948 949 /** 950 * @param string $soundex 951 * @param string $lastname 952 * @param string $firstname 953 * @param string $place 954 * @param array<Tree> $search_trees 955 * 956 * @return Collection<int,Individual> 957 */ 958 public function searchIndividualsPhonetic(string $soundex, string $lastname, string $firstname, string $place, array $search_trees): Collection 959 { 960 switch ($soundex) { 961 default: 962 case 'Russell': 963 $givn_sdx = Soundex::russell($firstname); 964 $surn_sdx = Soundex::russell($lastname); 965 $plac_sdx = Soundex::russell($place); 966 $givn_field = 'n_soundex_givn_std'; 967 $surn_field = 'n_soundex_surn_std'; 968 $plac_field = 'p_std_soundex'; 969 break; 970 case 'DaitchM': 971 $givn_sdx = Soundex::daitchMokotoff($firstname); 972 $surn_sdx = Soundex::daitchMokotoff($lastname); 973 $plac_sdx = Soundex::daitchMokotoff($place); 974 $givn_field = 'n_soundex_givn_dm'; 975 $surn_field = 'n_soundex_surn_dm'; 976 $plac_field = 'p_dm_soundex'; 977 break; 978 } 979 980 // Nothing to search for? Return nothing. 981 if ($givn_sdx === '' && $surn_sdx === '' && $plac_sdx === '') { 982 return new Collection(); 983 } 984 985 $query = DB::table('individuals') 986 ->select(['individuals.*']) 987 ->distinct(); 988 989 $this->whereTrees($query, 'i_file', $search_trees); 990 991 if ($plac_sdx !== '') { 992 $query->join('placelinks', static function (JoinClause $join): void { 993 $join 994 ->on('placelinks.pl_file', '=', 'individuals.i_file') 995 ->on('placelinks.pl_gid', '=', 'individuals.i_id'); 996 }); 997 $query->join('places', static function (JoinClause $join): void { 998 $join 999 ->on('places.p_file', '=', 'placelinks.pl_file') 1000 ->on('places.p_id', '=', 'placelinks.pl_p_id'); 1001 }); 1002 1003 $this->wherePhonetic($query, $plac_field, $plac_sdx); 1004 } 1005 1006 if ($givn_sdx !== '' || $surn_sdx !== '') { 1007 $query->join('name', static function (JoinClause $join): void { 1008 $join 1009 ->on('name.n_file', '=', 'individuals.i_file') 1010 ->on('name.n_id', '=', 'individuals.i_id'); 1011 }); 1012 1013 $this->wherePhonetic($query, $givn_field, $givn_sdx); 1014 $this->wherePhonetic($query, $surn_field, $surn_sdx); 1015 } 1016 1017 return $query 1018 ->get() 1019 ->each($this->rowLimiter()) 1020 ->map($this->individualRowMapper()) 1021 ->filter(GedcomRecord::accessFilter()); 1022 } 1023 1024 /** 1025 * Paginate a search query. 1026 * 1027 * @param Builder $query Searches the database for the desired records. 1028 * @param Closure $row_mapper Converts a row from the query into a record. 1029 * @param Closure $row_filter 1030 * @param int $offset Skip this many rows. 1031 * @param int $limit Take this many rows. 1032 * 1033 * @return Collection<int,mixed> 1034 */ 1035 private function paginateQuery(Builder $query, Closure $row_mapper, Closure $row_filter, int $offset, int $limit): Collection 1036 { 1037 $collection = new Collection(); 1038 1039 foreach ($query->cursor() as $row) { 1040 $record = $row_mapper($row); 1041 // searchIndividualNames() and searchFamilyNames() can return duplicate rows, 1042 // where individuals have multiple names - and we need to sort results by name. 1043 if ($collection->containsStrict($record)) { 1044 continue; 1045 } 1046 // If the object has a method "canShow()", then use it to filter for privacy. 1047 if ($row_filter($record)) { 1048 if ($offset > 0) { 1049 $offset--; 1050 } else { 1051 if ($limit > 0) { 1052 $collection->push($record); 1053 } 1054 1055 $limit--; 1056 1057 if ($limit === 0) { 1058 break; 1059 } 1060 } 1061 } 1062 } 1063 1064 1065 return $collection; 1066 } 1067 1068 /** 1069 * Apply search filters to a SQL query column. Apply collation rules to MySQL. 1070 * 1071 * @param Builder $query 1072 * @param Expression|string $column 1073 * @param array<string> $search_terms 1074 */ 1075 private function whereSearch(Builder $query, Expression|string $column, array $search_terms): void 1076 { 1077 foreach ($search_terms as $search_term) { 1078 $query->where($column, $this->iLike(), '%' . addcslashes($search_term, '\\%_') . '%'); 1079 } 1080 } 1081 1082 /** 1083 * Apply soundex search filters to a SQL query column. 1084 * 1085 * @param Builder $query 1086 * @param Expression|string $field 1087 * @param string $soundex 1088 */ 1089 private function wherePhonetic(Builder $query, $field, string $soundex): void 1090 { 1091 if ($soundex !== '') { 1092 $query->where(function (Builder $query) use ($soundex, $field): void { 1093 foreach (explode(':', $soundex) as $sdx) { 1094 $query->orWhere($field, $this->iLike(), '%' . $sdx . '%'); 1095 } 1096 }); 1097 } 1098 } 1099 1100 /** 1101 * @param Builder $query 1102 * @param string $tree_id_field 1103 * @param array<Tree> $trees 1104 */ 1105 private function whereTrees(Builder $query, string $tree_id_field, array $trees): void 1106 { 1107 $tree_ids = array_map(static function (Tree $tree): int { 1108 return $tree->id(); 1109 }, $trees); 1110 1111 $query->whereIn($tree_id_field, $tree_ids); 1112 } 1113 1114 /** 1115 * Find the media object that uses a particular media file. 1116 * 1117 * @param string $file 1118 * 1119 * @return array<Media> 1120 */ 1121 public function findMediaObjectsForMediaFile(string $file): array 1122 { 1123 return DB::table('media') 1124 ->join('media_file', static function (JoinClause $join): void { 1125 $join 1126 ->on('media_file.m_file', '=', 'media.m_file') 1127 ->on('media_file.m_id', '=', 'media.m_id'); 1128 }) 1129 ->join('gedcom_setting', 'media.m_file', '=', 'gedcom_setting.gedcom_id') 1130 ->where(new Expression('setting_value || multimedia_file_refn'), '=', $file) 1131 ->select(['media.*']) 1132 ->distinct() 1133 ->get() 1134 ->map($this->mediaRowMapper()) 1135 ->all(); 1136 } 1137 1138 /** 1139 * A closure to filter records by privacy-filtered GEDCOM data. 1140 * 1141 * @param array<string> $search_terms 1142 * 1143 * @return Closure(GedcomRecord):bool 1144 */ 1145 private function rawGedcomFilter(array $search_terms): Closure 1146 { 1147 return static function (GedcomRecord $record) use ($search_terms): bool { 1148 // Ignore non-genealogy fields 1149 $gedcom = preg_replace('/\n\d (?:_UID|_WT_USER) .*/', '', $record->gedcom()); 1150 1151 // Ignore matches in links 1152 $gedcom = preg_replace('/\n\d ' . Gedcom::REGEX_TAG . '( @' . Gedcom::REGEX_XREF . '@)?/', '', $gedcom); 1153 1154 // Re-apply the filtering 1155 foreach ($search_terms as $search_term) { 1156 if (mb_stripos($gedcom, $search_term) === false) { 1157 return false; 1158 } 1159 } 1160 1161 return true; 1162 }; 1163 } 1164 1165 /** 1166 * Searching for short or common text can give more results than the system can process. 1167 * 1168 * @param int $limit 1169 * 1170 * @return Closure():void 1171 */ 1172 private function rowLimiter(int $limit = self::MAX_SEARCH_RESULTS): Closure 1173 { 1174 return static function () use ($limit): void { 1175 static $n = 0; 1176 1177 if (++$n > $limit) { 1178 $message = I18N::translate('The search returned too many results.'); 1179 1180 throw new HttpServiceUnavailableException($message); 1181 } 1182 }; 1183 } 1184 1185 /** 1186 * Convert a row from any tree in the families table into a family object. 1187 * 1188 * @return Closure(object):Family 1189 */ 1190 private function familyRowMapper(): Closure 1191 { 1192 return function (object $row): Family { 1193 $tree = $this->tree_service->find((int) $row->f_file); 1194 1195 return Registry::familyFactory()->mapper($tree)($row); 1196 }; 1197 } 1198 1199 /** 1200 * Convert a row from any tree in the individuals table into an individual object. 1201 * 1202 * @return Closure(object):Individual 1203 */ 1204 private function individualRowMapper(): Closure 1205 { 1206 return function (object $row): Individual { 1207 $tree = $this->tree_service->find((int) $row->i_file); 1208 1209 return Registry::individualFactory()->mapper($tree)($row); 1210 }; 1211 } 1212 1213 /** 1214 * Convert a row from any tree in the media table into a location object. 1215 * 1216 * @return Closure(object):Location 1217 */ 1218 private function locationRowMapper(): Closure 1219 { 1220 return function (object $row): Location { 1221 $tree = $this->tree_service->find((int) $row->o_file); 1222 1223 return Registry::locationFactory()->mapper($tree)($row); 1224 }; 1225 } 1226 1227 /** 1228 * Convert a row from any tree in the media table into an media object. 1229 * 1230 * @return Closure(object):Media 1231 */ 1232 private function mediaRowMapper(): Closure 1233 { 1234 return function (object $row): Media { 1235 $tree = $this->tree_service->find((int) $row->m_file); 1236 1237 return Registry::mediaFactory()->mapper($tree)($row); 1238 }; 1239 } 1240 1241 /** 1242 * Convert a row from any tree in the other table into a note object. 1243 * 1244 * @return Closure:Note 1245 */ 1246 private function noteRowMapper(): Closure 1247 { 1248 return function (object $row): Note { 1249 $tree = $this->tree_service->find((int) $row->o_file); 1250 1251 return Registry::noteFactory()->mapper($tree)($row); 1252 }; 1253 } 1254 1255 /** 1256 * Convert a row from any tree in the other table into a repository object. 1257 * 1258 * @return Closure:Repository 1259 */ 1260 private function repositoryRowMapper(): Closure 1261 { 1262 return function (object $row): Repository { 1263 $tree = $this->tree_service->find((int) $row->o_file); 1264 1265 return Registry::repositoryFactory()->mapper($tree)($row); 1266 }; 1267 } 1268 1269 /** 1270 * Convert a row from any tree in the other table into a note object. 1271 * 1272 * @return Closure(object):SharedNote 1273 */ 1274 private function sharedNoteRowMapper(): Closure 1275 { 1276 return function (object $row): Note { 1277 $tree = $this->tree_service->find((int) $row->o_file); 1278 1279 return Registry::sharedNoteFactory()->mapper($tree)($row); 1280 }; 1281 } 1282 1283 /** 1284 * Convert a row from any tree in the sources table into a source object. 1285 * 1286 * @return Closure:Source 1287 */ 1288 private function sourceRowMapper(): Closure 1289 { 1290 return function (object $row): Source { 1291 $tree = $this->tree_service->find((int) $row->s_file); 1292 1293 return Registry::sourceFactory()->mapper($tree)($row); 1294 }; 1295 } 1296 1297 /** 1298 * Convert a row from any tree in the other table into a submission object. 1299 * 1300 * @return Closure(object):Submission 1301 */ 1302 private function submissionRowMapper(): Closure 1303 { 1304 return function (object $row): Submission { 1305 $tree = $this->tree_service->find((int) $row->o_file); 1306 1307 return Registry::submissionFactory()->mapper($tree)($row); 1308 }; 1309 } 1310 1311 /** 1312 * Convert a row from any tree in the other table into a submitter object. 1313 * 1314 * @return Closure(object):Submitter 1315 */ 1316 private function submitterRowMapper(): Closure 1317 { 1318 return function (object $row): Submitter { 1319 $tree = $this->tree_service->find((int) $row->o_file); 1320 1321 return Registry::submitterFactory()->mapper($tree)($row); 1322 }; 1323 } 1324 1325 /** 1326 * @internal - a better solution would support other RDBMS, probably by using collations. 1327 */ 1328 private function iLike(): string 1329 { 1330 if (DB::connection()->getDriverName() === 'pgsql') { 1331 return 'ILIKE'; 1332 } 1333 1334 return 'LIKE'; 1335 } 1336} 1337