1<?php 2 3/** 4 * webtrees: online genealogy 5 * Copyright (C) 2022 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 Fisharebest\Webtrees\Date; 23use Fisharebest\Webtrees\Elements\UnknownElement; 24use Fisharebest\Webtrees\Exceptions\GedcomErrorException; 25use Fisharebest\Webtrees\Family; 26use Fisharebest\Webtrees\Gedcom; 27use Fisharebest\Webtrees\Header; 28use Fisharebest\Webtrees\Individual; 29use Fisharebest\Webtrees\Location; 30use Fisharebest\Webtrees\Media; 31use Fisharebest\Webtrees\Note; 32use Fisharebest\Webtrees\Place; 33use Fisharebest\Webtrees\PlaceLocation; 34use Fisharebest\Webtrees\Registry; 35use Fisharebest\Webtrees\Repository; 36use Fisharebest\Webtrees\Soundex; 37use Fisharebest\Webtrees\Source; 38use Fisharebest\Webtrees\Submission; 39use Fisharebest\Webtrees\Submitter; 40use Fisharebest\Webtrees\Tree; 41use Illuminate\Database\Capsule\Manager as DB; 42use Illuminate\Database\Query\JoinClause; 43 44use function app; 45use function array_chunk; 46use function array_intersect_key; 47use function array_map; 48use function array_unique; 49use function assert; 50use function date; 51use function explode; 52use function max; 53use function mb_substr; 54use function preg_match; 55use function preg_match_all; 56use function preg_replace; 57use function round; 58use function str_contains; 59use function str_replace; 60use function str_starts_with; 61use function strlen; 62use function strtolower; 63use function strtoupper; 64use function strtr; 65use function substr; 66use function trim; 67 68use const PREG_SET_ORDER; 69 70/** 71 * Class GedcomImportService - import GEDCOM data 72 */ 73class GedcomImportService 74{ 75 /** 76 * Tidy up a gedcom record on import, so that we can access it consistently/efficiently. 77 * 78 * @param string $rec 79 * @param Tree $tree 80 * 81 * @return string 82 */ 83 private function reformatRecord(string $rec, Tree $tree): string 84 { 85 $gedcom_service = app(GedcomService::class); 86 assert($gedcom_service instanceof GedcomService); 87 88 // Strip out mac/msdos line endings 89 $rec = preg_replace("/[\r\n]+/", "\n", $rec); 90 91 // Extract lines from the record; lines consist of: level + optional xref + tag + optional data 92 $num_matches = preg_match_all('/^[ \t]*(\d+)[ \t]*(@[^@]*@)?[ \t]*(\w+)[ \t]?(.*)$/m', $rec, $matches, PREG_SET_ORDER); 93 94 // Process the record line-by-line 95 $newrec = ''; 96 foreach ($matches as $n => $match) { 97 [, $level, $xref, $tag, $data] = $match; 98 99 $tag = $gedcom_service->canonicalTag($tag); 100 101 switch ($tag) { 102 case 'AFN': 103 // AFN values are upper case 104 $data = strtoupper($data); 105 break; 106 case 'DATE': 107 // Preserve text from INT dates 108 if (str_contains($data, '(')) { 109 [$date, $text] = explode('(', $data, 2); 110 $text = ' (' . $text; 111 } else { 112 $date = $data; 113 $text = ''; 114 } 115 // Capitals 116 $date = strtoupper($date); 117 // Temporarily add leading/trailing spaces, to allow efficient matching below 118 $date = ' ' . $date . ' '; 119 // Ensure space digits and letters 120 $date = preg_replace('/([A-Z])(\d)/', '$1 $2', $date); 121 $date = preg_replace('/(\d)([A-Z])/', '$1 $2', $date); 122 // Ensure space before/after calendar escapes 123 $date = preg_replace('/@#[^@]+@/', ' $0 ', $date); 124 // "BET." => "BET" 125 $date = preg_replace('/(\w\w)\./', '$1', $date); 126 // "CIR" => "ABT" 127 $date = str_replace(' CIR ', ' ABT ', $date); 128 $date = str_replace(' APX ', ' ABT ', $date); 129 // B.C. => BC (temporarily, to allow easier handling of ".") 130 $date = str_replace(' B.C. ', ' BC ', $date); 131 // TMG uses "EITHER X OR Y" 132 $date = preg_replace('/^ EITHER (.+) OR (.+)/', ' BET $1 AND $2', $date); 133 // "BET X - Y " => "BET X AND Y" 134 $date = preg_replace('/^(.* BET .+) - (.+)/', '$1 AND $2', $date); 135 $date = preg_replace('/^(.* FROM .+) - (.+)/', '$1 TO $2', $date); 136 // "@#ESC@ FROM X TO Y" => "FROM @#ESC@ X TO @#ESC@ Y" 137 $date = preg_replace('/^ +(@#[^@]+@) +FROM +(.+) +TO +(.+)/', ' FROM $1 $2 TO $1 $3', $date); 138 $date = preg_replace('/^ +(@#[^@]+@) +BET +(.+) +AND +(.+)/', ' BET $1 $2 AND $1 $3', $date); 139 // "@#ESC@ AFT X" => "AFT @#ESC@ X" 140 $date = preg_replace('/^ +(@#[^@]+@) +(FROM|BET|TO|AND|BEF|AFT|CAL|EST|INT|ABT) +(.+)/', ' $2 $1 $3', $date); 141 // Ignore any remaining punctuation, e.g. "14-MAY, 1900" => "14 MAY 1900" 142 // (don't change "/" - it is used in NS/OS dates) 143 $date = preg_replace('/[.,:;-]/', ' ', $date); 144 // BC => B.C. 145 $date = str_replace(' BC ', ' B.C. ', $date); 146 // Append the "INT" text 147 $data = $date . $text; 148 break; 149 case '_FILE': 150 $tag = 'FILE'; 151 break; 152 case 'FORM': 153 // Consistent commas 154 $data = preg_replace('/ *, */', ', ', $data); 155 break; 156 case 'HEAD': 157 // HEAD records don't have an XREF or DATA 158 if ($level === '0') { 159 $xref = ''; 160 $data = ''; 161 } 162 break; 163 case 'NAME': 164 // Tidy up non-printing characters 165 $data = preg_replace('/ +/', ' ', trim($data)); 166 break; 167 case 'PEDI': 168 // PEDI values are lower case 169 $data = strtolower($data); 170 break; 171 case 'PLAC': 172 // Consistent commas 173 $data = preg_replace('/ *[,，،] */u', ', ', $data); 174 // The Master Genealogist stores LAT/LONG data in the PLAC field, e.g. Pennsylvania, USA, 395945N0751013W 175 if (preg_match('/(.*), (\d\d)(\d\d)(\d\d)([NS])(\d\d\d)(\d\d)(\d\d)([EW])$/', $data, $match)) { 176 $data = 177 $match[1] . "\n" . 178 ($level + 1) . " MAP\n" . 179 ($level + 2) . ' LATI ' . ($match[5] . round($match[2] + ($match[3] / 60) + ($match[4] / 3600), 4)) . "\n" . 180 ($level + 2) . ' LONG ' . ($match[9] . round($match[6] + ($match[7] / 60) + ($match[8] / 3600), 4)); 181 } 182 break; 183 case 'RESN': 184 // RESN values are lower case (confidential, privacy, locked, none) 185 $data = strtolower($data); 186 if ($data === 'invisible') { 187 $data = 'confidential'; // From old versions of Legacy. 188 } 189 break; 190 case 'SEX': 191 $data = strtoupper($data); 192 break; 193 case 'STAT': 194 if ($data === 'CANCELLED') { 195 // PhpGedView mis-spells this tag - correct it. 196 $data = 'CANCELED'; 197 } 198 break; 199 case 'TEMP': 200 // Temple codes are upper case 201 $data = strtoupper($data); 202 break; 203 case 'TRLR': 204 // TRLR records don't have an XREF or DATA 205 if ($level === '0') { 206 $xref = ''; 207 $data = ''; 208 } 209 break; 210 } 211 // Suppress "Y", for facts/events with a DATE or PLAC 212 if ($data === 'y') { 213 $data = 'Y'; 214 } 215 if ($level === '1' && $data === 'Y') { 216 for ($i = $n + 1; $i < $num_matches - 1 && $matches[$i][1] !== '1'; ++$i) { 217 if ($matches[$i][3] === 'DATE' || $matches[$i][3] === 'PLAC') { 218 $data = ''; 219 break; 220 } 221 } 222 } 223 // Reassemble components back into a single line 224 switch ($tag) { 225 default: 226 // Remove tabs and multiple/leading/trailing spaces 227 $data = strtr($data, ["\t" => ' ']); 228 $data = trim($data, ' '); 229 while (str_contains($data, ' ')) { 230 $data = strtr($data, [' ' => ' ']); 231 } 232 $newrec .= ($newrec ? "\n" : '') . $level . ' ' . ($level === '0' && $xref ? $xref . ' ' : '') . $tag . ($data === '' && $tag !== 'NOTE' ? '' : ' ' . $data); 233 break; 234 case 'NOTE': 235 case 'TEXT': 236 case 'DATA': 237 case 'CONT': 238 $newrec .= ($newrec ? "\n" : '') . $level . ' ' . ($level === '0' && $xref ? $xref . ' ' : '') . $tag . ($data === '' && $tag !== 'NOTE' ? '' : ' ' . $data); 239 break; 240 case 'FILE': 241 // Strip off the user-defined path prefix 242 $GEDCOM_MEDIA_PATH = $tree->getPreference('GEDCOM_MEDIA_PATH'); 243 if ($GEDCOM_MEDIA_PATH !== '' && str_starts_with($data, $GEDCOM_MEDIA_PATH)) { 244 $data = substr($data, strlen($GEDCOM_MEDIA_PATH)); 245 } 246 // convert backslashes in filenames to forward slashes 247 $data = preg_replace("/\\\\/", '/', $data); 248 249 $newrec .= ($newrec ? "\n" : '') . $level . ' ' . ($level === '0' && $xref ? $xref . ' ' : '') . $tag . ($data === '' && $tag !== 'NOTE' ? '' : ' ' . $data); 250 break; 251 case 'CONC': 252 // Merge CONC lines, to simplify access later on. 253 $newrec .= ($tree->getPreference('WORD_WRAPPED_NOTES') ? ' ' : '') . $data; 254 break; 255 } 256 } 257 258 return $newrec; 259 } 260 261 /** 262 * import record into database 263 * this function will parse the given gedcom record and add it to the database 264 * 265 * @param string $gedrec the raw gedcom record to parse 266 * @param Tree $tree import the record into this tree 267 * @param bool $update whether this is an updated record that has been accepted 268 * 269 * @return void 270 * @throws GedcomErrorException 271 */ 272 public function importRecord(string $gedrec, Tree $tree, bool $update): void 273 { 274 $tree_id = $tree->id(); 275 276 // Escaped @ signs (only if importing from file) 277 if (!$update) { 278 $gedrec = str_replace('@@', '@', $gedrec); 279 } 280 281 // Standardise gedcom format 282 $gedrec = $this->reformatRecord($gedrec, $tree); 283 284 // import different types of records 285 if (preg_match('/^0 @(' . Gedcom::REGEX_XREF . ')@ (' . Gedcom::REGEX_TAG . ')/', $gedrec, $match)) { 286 [, $xref, $type] = $match; 287 } elseif (str_starts_with($gedrec, '0 _EVDEF')) { 288 // Created by RootsMagic. We cannot process these records without an XREF. 289 return; 290 } elseif (str_starts_with($gedrec, '0 _EVENT_DEFN')) { 291 // Created by PAF and Legacy. We cannot process these records without an XREF. 292 return; 293 } elseif (str_starts_with($gedrec, '0 _PLAC_DEFN')) { 294 $this->importLegacyPlacDefn($gedrec); 295 296 return; 297 } elseif (str_starts_with($gedrec, '0 _PLAC ')) { 298 $this->importTNGPlac($gedrec); 299 300 return; 301 } elseif (str_starts_with($gedrec, '0 HEAD')) { 302 $type = 'HEAD'; 303 $xref = 'HEAD'; // For records without an XREF, use the type as a pseudo XREF. 304 } elseif (str_starts_with($gedrec, '0 TRLR')) { 305 $tree->setPreference('imported', '1'); 306 $type = 'TRLR'; 307 $xref = 'TRLR'; // For records without an XREF, use the type as a pseudo XREF. 308 } else { 309 throw new GedcomErrorException($gedrec); 310 } 311 312 // Add a _UID 313 if ($tree->getPreference('GENERATE_UIDS') === '1' && !str_contains($gedrec, "\n1 _UID ")) { 314 $element = Registry::elementFactory()->make($type . ':_UID'); 315 if (!$element instanceof UnknownElement) { 316 $gedrec .= "\n1 _UID " . $element->default($tree); 317 } 318 } 319 320 // If the user has downloaded their GEDCOM data (containing media objects) and edited it 321 // using an application which does not support (and deletes) media objects, then add them 322 // back in. 323 if ($tree->getPreference('keep_media')) { 324 $old_linked_media = DB::table('link') 325 ->where('l_from', '=', $xref) 326 ->where('l_file', '=', $tree_id) 327 ->where('l_type', '=', 'OBJE') 328 ->pluck('l_to'); 329 330 // Delete these links - so that we do not insert them again in updateLinks() 331 DB::table('link') 332 ->where('l_from', '=', $xref) 333 ->where('l_file', '=', $tree_id) 334 ->where('l_type', '=', 'OBJE') 335 ->delete(); 336 337 foreach ($old_linked_media as $media_id) { 338 $gedrec .= "\n1 OBJE @" . $media_id . '@'; 339 } 340 } 341 342 // Convert inline media into media objects 343 $gedrec = $this->convertInlineMedia($tree, $gedrec); 344 345 switch ($type) { 346 case Individual::RECORD_TYPE: 347 $record = Registry::individualFactory()->new($xref, $gedrec, null, $tree); 348 349 if (preg_match('/\n1 RIN (.+)/', $gedrec, $match)) { 350 $rin = $match[1]; 351 } else { 352 $rin = $xref; 353 } 354 355 DB::table('individuals')->insert([ 356 'i_id' => $xref, 357 'i_file' => $tree_id, 358 'i_rin' => $rin, 359 'i_sex' => $record->sex(), 360 'i_gedcom' => $gedrec, 361 ]); 362 363 // Update the cross-reference/index tables. 364 $this->updatePlaces($xref, $tree, $gedrec); 365 $this->updateDates($xref, $tree_id, $gedrec); 366 $this->updateNames($xref, $tree_id, $record); 367 break; 368 369 case Family::RECORD_TYPE: 370 if (preg_match('/\n1 HUSB @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $match)) { 371 $husb = $match[1]; 372 } else { 373 $husb = ''; 374 } 375 if (preg_match('/\n1 WIFE @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $match)) { 376 $wife = $match[1]; 377 } else { 378 $wife = ''; 379 } 380 $nchi = preg_match_all('/\n1 CHIL @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $match); 381 if (preg_match('/\n1 NCHI (\d+)/', $gedrec, $match)) { 382 $nchi = max($nchi, $match[1]); 383 } 384 385 DB::table('families')->insert([ 386 'f_id' => $xref, 387 'f_file' => $tree_id, 388 'f_husb' => $husb, 389 'f_wife' => $wife, 390 'f_gedcom' => $gedrec, 391 'f_numchil' => $nchi, 392 ]); 393 394 // Update the cross-reference/index tables. 395 $this->updatePlaces($xref, $tree, $gedrec); 396 $this->updateDates($xref, $tree_id, $gedrec); 397 break; 398 399 case Source::RECORD_TYPE: 400 if (preg_match('/\n1 TITL (.+)/', $gedrec, $match)) { 401 $name = $match[1]; 402 } elseif (preg_match('/\n1 ABBR (.+)/', $gedrec, $match)) { 403 $name = $match[1]; 404 } else { 405 $name = $xref; 406 } 407 408 DB::table('sources')->insert([ 409 's_id' => $xref, 410 's_file' => $tree_id, 411 's_name' => mb_substr($name, 0, 255), 412 's_gedcom' => $gedrec, 413 ]); 414 break; 415 416 case Repository::RECORD_TYPE: 417 case Note::RECORD_TYPE: 418 case Submission::RECORD_TYPE: 419 case Submitter::RECORD_TYPE: 420 case Location::RECORD_TYPE: 421 DB::table('other')->insert([ 422 'o_id' => $xref, 423 'o_file' => $tree_id, 424 'o_type' => $type, 425 'o_gedcom' => $gedrec, 426 ]); 427 break; 428 429 case Header::RECORD_TYPE: 430 // Force HEAD records to have a creation date. 431 if (!str_contains($gedrec, "\n1 DATE ")) { 432 $today = strtoupper(date('d M Y')); 433 $gedrec .= "\n1 DATE " . $today; 434 } 435 436 DB::table('other')->insert([ 437 'o_id' => $xref, 438 'o_file' => $tree_id, 439 'o_type' => Header::RECORD_TYPE, 440 'o_gedcom' => $gedrec, 441 ]); 442 break; 443 444 445 case Media::RECORD_TYPE: 446 $record = Registry::mediaFactory()->new($xref, $gedrec, null, $tree); 447 448 DB::table('media')->insert([ 449 'm_id' => $xref, 450 'm_file' => $tree_id, 451 'm_gedcom' => $gedrec, 452 ]); 453 454 foreach ($record->mediaFiles() as $media_file) { 455 DB::table('media_file')->insert([ 456 'm_id' => $xref, 457 'm_file' => $tree_id, 458 'multimedia_file_refn' => mb_substr($media_file->filename(), 0, 248), 459 'multimedia_format' => mb_substr($media_file->format(), 0, 4), 460 'source_media_type' => mb_substr($media_file->type(), 0, 15), 461 'descriptive_title' => mb_substr($media_file->title(), 0, 248), 462 ]); 463 } 464 break; 465 466 default: // Custom record types. 467 DB::table('other')->insert([ 468 'o_id' => $xref, 469 'o_file' => $tree_id, 470 'o_type' => mb_substr($type, 0, 15), 471 'o_gedcom' => $gedrec, 472 ]); 473 break; 474 } 475 476 // Update the cross-reference/index tables. 477 $this->updateLinks($xref, $tree_id, $gedrec); 478 } 479 480 /** 481 * Legacy Family Tree software generates _PLAC_DEFN records containing LAT/LONG values 482 * 483 * @param string $gedcom 484 */ 485 private function importLegacyPlacDefn(string $gedcom): void 486 { 487 $gedcom_service = new GedcomService(); 488 489 if (preg_match('/\n1 PLAC (.+)/', $gedcom, $match)) { 490 $place_name = $match[1]; 491 } else { 492 return; 493 } 494 495 if (preg_match('/\n3 LATI ([NS].+)/', $gedcom, $match)) { 496 $latitude = $gedcom_service->readLatitude($match[1]); 497 } else { 498 return; 499 } 500 501 if (preg_match('/\n3 LONG ([EW].+)/', $gedcom, $match)) { 502 $longitude = $gedcom_service->readLongitude($match[1]); 503 } else { 504 return; 505 } 506 507 $location = new PlaceLocation($place_name); 508 509 if ($location->latitude() === null && $location->longitude() === null) { 510 DB::table('place_location') 511 ->where('id', '=', $location->id()) 512 ->update([ 513 'latitude' => $latitude, 514 'longitude' => $longitude, 515 ]); 516 } 517 } 518 519 /** 520 * Legacy Family Tree software generates _PLAC records containing LAT/LONG values 521 * 522 * @param string $gedcom 523 */ 524 private function importTNGPlac(string $gedcom): void 525 { 526 if (preg_match('/^0 _PLAC (.+)/', $gedcom, $match)) { 527 $place_name = $match[1]; 528 } else { 529 return; 530 } 531 532 if (preg_match('/\n2 LATI (.+)/', $gedcom, $match)) { 533 $latitude = (float) $match[1]; 534 } else { 535 return; 536 } 537 538 if (preg_match('/\n2 LONG (.+)/', $gedcom, $match)) { 539 $longitude = (float) $match[1]; 540 } else { 541 return; 542 } 543 544 $location = new PlaceLocation($place_name); 545 546 if ($location->latitude() === null && $location->longitude() === null) { 547 DB::table('place_location') 548 ->where('id', '=', $location->id()) 549 ->update([ 550 'latitude' => $latitude, 551 'longitude' => $longitude, 552 ]); 553 } 554 } 555 556 /** 557 * Extract all level 2 places from the given record and insert them into the places table 558 * 559 * @param string $xref 560 * @param Tree $tree 561 * @param string $gedrec 562 * 563 * @return void 564 */ 565 public function updatePlaces(string $xref, Tree $tree, string $gedrec): void 566 { 567 // Insert all new rows together 568 $rows = []; 569 570 preg_match_all('/\n2 PLAC (.+)/', $gedrec, $matches); 571 572 $places = array_unique($matches[1]); 573 574 foreach ($places as $place_name) { 575 $place = new Place($place_name, $tree); 576 577 // Calling Place::id() will create the entry in the database, if it doesn't already exist. 578 while ($place->id() !== 0) { 579 $rows[] = [ 580 'pl_p_id' => $place->id(), 581 'pl_gid' => $xref, 582 'pl_file' => $tree->id(), 583 ]; 584 585 $place = $place->parent(); 586 } 587 } 588 589 // array_unique doesn't work with arrays of arrays 590 $rows = array_intersect_key($rows, array_unique(array_map('serialize', $rows))); 591 592 // PDO has a limit of 65535 placeholders, and each row requires 3 placeholders. 593 foreach (array_chunk($rows, 20000) as $chunk) { 594 DB::table('placelinks')->insert($chunk); 595 } 596 } 597 598 /** 599 * Extract all the dates from the given record and insert them into the database. 600 * 601 * @param string $xref 602 * @param int $ged_id 603 * @param string $gedrec 604 * 605 * @return void 606 */ 607 private function updateDates(string $xref, int $ged_id, string $gedrec): void 608 { 609 // Insert all new rows together 610 $rows = []; 611 612 preg_match_all("/\n1 (\w+).*(?:\n[2-9].*)*\n2 DATE (.+)(?:\n[2-9].*)*/", $gedrec, $matches, PREG_SET_ORDER); 613 614 foreach ($matches as $match) { 615 $fact = $match[1]; 616 $date = new Date($match[2]); 617 $rows[] = [ 618 'd_day' => $date->minimumDate()->day, 619 'd_month' => $date->minimumDate()->format('%O'), 620 'd_mon' => $date->minimumDate()->month, 621 'd_year' => $date->minimumDate()->year, 622 'd_julianday1' => $date->minimumDate()->minimumJulianDay(), 623 'd_julianday2' => $date->minimumDate()->maximumJulianDay(), 624 'd_fact' => $fact, 625 'd_gid' => $xref, 626 'd_file' => $ged_id, 627 'd_type' => $date->minimumDate()->format('%@'), 628 ]; 629 630 $rows[] = [ 631 'd_day' => $date->maximumDate()->day, 632 'd_month' => $date->maximumDate()->format('%O'), 633 'd_mon' => $date->maximumDate()->month, 634 'd_year' => $date->maximumDate()->year, 635 'd_julianday1' => $date->maximumDate()->minimumJulianDay(), 636 'd_julianday2' => $date->maximumDate()->maximumJulianDay(), 637 'd_fact' => $fact, 638 'd_gid' => $xref, 639 'd_file' => $ged_id, 640 'd_type' => $date->minimumDate()->format('%@'), 641 ]; 642 } 643 644 // array_unique doesn't work with arrays of arrays 645 $rows = array_intersect_key($rows, array_unique(array_map('serialize', $rows))); 646 647 DB::table('dates')->insert($rows); 648 } 649 650 /** 651 * Extract all the links from the given record and insert them into the database 652 * 653 * @param string $xref 654 * @param int $ged_id 655 * @param string $gedrec 656 * 657 * @return void 658 */ 659 private function updateLinks(string $xref, int $ged_id, string $gedrec): void 660 { 661 // Insert all new rows together 662 $rows = []; 663 664 preg_match_all('/\n\d+ (' . Gedcom::REGEX_TAG . ') @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $matches, PREG_SET_ORDER); 665 666 foreach ($matches as $match) { 667 // Take care of "duplicates" that differ on case/collation, e.g. "SOUR @S1@" and "SOUR @s1@" 668 $rows[$match[1] . strtoupper($match[2])] = [ 669 'l_from' => $xref, 670 'l_to' => $match[2], 671 'l_type' => $match[1], 672 'l_file' => $ged_id, 673 ]; 674 } 675 676 DB::table('link')->insert($rows); 677 } 678 679 /** 680 * Extract all the names from the given record and insert them into the database. 681 * 682 * @param string $xref 683 * @param int $ged_id 684 * @param Individual $record 685 * 686 * @return void 687 */ 688 private function updateNames(string $xref, int $ged_id, Individual $record): void 689 { 690 // Insert all new rows together 691 $rows = []; 692 693 foreach ($record->getAllNames() as $n => $name) { 694 if ($name['givn'] === Individual::PRAENOMEN_NESCIO) { 695 $soundex_givn_std = null; 696 $soundex_givn_dm = null; 697 } else { 698 $soundex_givn_std = Soundex::russell($name['givn']); 699 $soundex_givn_dm = Soundex::daitchMokotoff($name['givn']); 700 } 701 702 if ($name['surn'] === Individual::NOMEN_NESCIO) { 703 $soundex_surn_std = null; 704 $soundex_surn_dm = null; 705 } else { 706 $soundex_surn_std = Soundex::russell($name['surname']); 707 $soundex_surn_dm = Soundex::daitchMokotoff($name['surname']); 708 } 709 710 $rows[] = [ 711 'n_file' => $ged_id, 712 'n_id' => $xref, 713 'n_num' => $n, 714 'n_type' => $name['type'], 715 'n_sort' => mb_substr($name['sort'], 0, 255), 716 'n_full' => mb_substr($name['fullNN'], 0, 255), 717 'n_surname' => mb_substr($name['surname'], 0, 255), 718 'n_surn' => mb_substr($name['surn'], 0, 255), 719 'n_givn' => mb_substr($name['givn'], 0, 255), 720 'n_soundex_givn_std' => $soundex_givn_std, 721 'n_soundex_surn_std' => $soundex_surn_std, 722 'n_soundex_givn_dm' => $soundex_givn_dm, 723 'n_soundex_surn_dm' => $soundex_surn_dm, 724 ]; 725 } 726 727 DB::table('name')->insert($rows); 728 } 729 730 /** 731 * Extract inline media data, and convert to media objects. 732 * 733 * @param Tree $tree 734 * @param string $gedcom 735 * 736 * @return string 737 */ 738 private function convertInlineMedia(Tree $tree, string $gedcom): string 739 { 740 while (preg_match('/\n1 OBJE(?:\n[2-9].+)+/', $gedcom, $match)) { 741 $xref = $this->createMediaObject($match[0], $tree); 742 $gedcom = strtr($gedcom, [$match[0] => "\n1 OBJE @" . $xref . '@']); 743 } 744 while (preg_match('/\n2 OBJE(?:\n[3-9].+)+/', $gedcom, $match)) { 745 $xref = $this->createMediaObject($match[0], $tree); 746 $gedcom = strtr($gedcom, [$match[0] => "\n2 OBJE @" . $xref . '@']); 747 } 748 while (preg_match('/\n3 OBJE(?:\n[4-9].+)+/', $gedcom, $match)) { 749 $xref = $this->createMediaObject($match[0], $tree); 750 $gedcom = strtr($gedcom, [$match[0] => "\n3 OBJE @" . $xref . '@']); 751 } 752 753 return $gedcom; 754 } 755 756 /** 757 * Create a new media object, from inline media data. 758 * 759 * GEDCOM 5.5.1 specifies: +1 FILE / +2 FORM / +3 MEDI / +1 TITL 760 * GEDCOM 5.5 specifies: +1 FILE / +1 FORM / +1 TITL 761 * GEDCOM 5.5.1 says that GEDCOM 5.5 specifies: +1 FILE / +1 FORM / +2 MEDI 762 * 763 * Legacy generates: +1 FORM / +1 FILE / +1 TITL / +1 _SCBK / +1 _PRIM / +1 _TYPE / +1 NOTE 764 * RootsMagic generates: +1 FILE / +1 FORM / +1 TITL 765 * 766 * @param string $gedcom 767 * @param Tree $tree 768 * 769 * @return string 770 */ 771 private function createMediaObject(string $gedcom, Tree $tree): string 772 { 773 preg_match('/\n\d FILE (.+)/', $gedcom, $match); 774 $file = $match[1] ?? ''; 775 776 preg_match('/\n\d TITL (.+)/', $gedcom, $match); 777 $title = $match[1] ?? ''; 778 779 preg_match('/\n\d FORM (.+)/', $gedcom, $match); 780 $format = $match[1] ?? ''; 781 782 preg_match('/\n\d MEDI (.+)/', $gedcom, $match); 783 $media = $match[1] ?? ''; 784 785 preg_match('/\n\d _SCBK (.+)/', $gedcom, $match); 786 $scrapbook = $match[1] ?? ''; 787 788 preg_match('/\n\d _PRIM (.+)/', $gedcom, $match); 789 $primary = $match[1] ?? ''; 790 791 preg_match('/\n\d _TYPE (.+)/', $gedcom, $match); 792 if ($media === '') { 793 // Legacy uses _TYPE instead of MEDI 794 $media = $match[1] ?? ''; 795 $type = ''; 796 } else { 797 $type = $match[1] ?? ''; 798 } 799 800 preg_match_all('/\n\d NOTE (.+(?:\n\d CONT.*)*)/', $gedcom, $matches); 801 $notes = $matches[1] ?? []; 802 803 // Have we already created a media object with the same title/filename? 804 $xref = DB::table('media_file') 805 ->where('m_file', '=', $tree->id()) 806 ->where('descriptive_title', '=', mb_substr($title, 0, 248)) 807 ->where('multimedia_file_refn', '=', mb_substr($file, 0, 248)) 808 ->value('m_id'); 809 810 if ($xref === null) { 811 $xref = Registry::xrefFactory()->make(Media::RECORD_TYPE); 812 813 // convert to a media-object 814 $gedcom = '0 @' . $xref . "@ OBJE\n1 FILE " . $file; 815 816 if ($format !== '') { 817 $gedcom .= "\n2 FORM " . $format; 818 819 if ($media !== '') { 820 $gedcom .= "\n3 TYPE " . $media; 821 } 822 } 823 824 if ($title !== '') { 825 $gedcom .= "\n3 TITL " . $title; 826 } 827 828 if ($scrapbook !== '') { 829 $gedcom .= "\n1 _SCBK " . $scrapbook; 830 } 831 832 if ($primary !== '') { 833 $gedcom .= "\n1 _PRIM " . $primary; 834 } 835 836 if ($type !== '') { 837 $gedcom .= "\n1 _TYPE " . $type; 838 } 839 840 foreach ($notes as $note) { 841 $gedcom .= "\n1 NOTE " . strtr($note, ["\n3" => "\n2", "\n4" => "\n2", "\n5" => "\n2"]); 842 } 843 844 DB::table('media')->insert([ 845 'm_id' => $xref, 846 'm_file' => $tree->id(), 847 'm_gedcom' => $gedcom, 848 ]); 849 850 DB::table('media_file')->insert([ 851 'm_id' => $xref, 852 'm_file' => $tree->id(), 853 'multimedia_file_refn' => mb_substr($file, 0, 248), 854 'multimedia_format' => mb_substr($format, 0, 4), 855 'source_media_type' => mb_substr($media, 0, 15), 856 'descriptive_title' => mb_substr($title, 0, 248), 857 ]); 858 } 859 860 return $xref; 861 } 862 863 /** 864 * update a record in the database 865 * 866 * @param string $gedrec 867 * @param Tree $tree 868 * @param bool $delete 869 * 870 * @return void 871 * @throws GedcomErrorException 872 */ 873 public function updateRecord(string $gedrec, Tree $tree, bool $delete): void 874 { 875 if (preg_match('/^0 @(' . Gedcom::REGEX_XREF . ')@ (' . Gedcom::REGEX_TAG . ')/', $gedrec, $match)) { 876 [, $gid, $type] = $match; 877 } elseif (preg_match('/^0 (HEAD)(?:\n|$)/', $gedrec, $match)) { 878 // The HEAD record has no XREF. Any others? 879 $gid = $match[1]; 880 $type = $match[1]; 881 } else { 882 throw new GedcomErrorException($gedrec); 883 } 884 885 // Place links 886 DB::table('placelinks') 887 ->where('pl_gid', '=', $gid) 888 ->where('pl_file', '=', $tree->id()) 889 ->delete(); 890 891 // Orphaned places. If we're deleting "Westminster, London, England", 892 // then we may also need to delete "London, England" and "England". 893 do { 894 $affected = DB::table('places') 895 ->leftJoin('placelinks', function (JoinClause $join): void { 896 $join 897 ->on('p_id', '=', 'pl_p_id') 898 ->on('p_file', '=', 'pl_file'); 899 }) 900 ->whereNull('pl_p_id') 901 ->delete(); 902 } while ($affected > 0); 903 904 DB::table('dates') 905 ->where('d_gid', '=', $gid) 906 ->where('d_file', '=', $tree->id()) 907 ->delete(); 908 909 DB::table('name') 910 ->where('n_id', '=', $gid) 911 ->where('n_file', '=', $tree->id()) 912 ->delete(); 913 914 DB::table('link') 915 ->where('l_from', '=', $gid) 916 ->where('l_file', '=', $tree->id()) 917 ->delete(); 918 919 switch ($type) { 920 case Individual::RECORD_TYPE: 921 DB::table('individuals') 922 ->where('i_id', '=', $gid) 923 ->where('i_file', '=', $tree->id()) 924 ->delete(); 925 break; 926 927 case Family::RECORD_TYPE: 928 DB::table('families') 929 ->where('f_id', '=', $gid) 930 ->where('f_file', '=', $tree->id()) 931 ->delete(); 932 break; 933 934 case Source::RECORD_TYPE: 935 DB::table('sources') 936 ->where('s_id', '=', $gid) 937 ->where('s_file', '=', $tree->id()) 938 ->delete(); 939 break; 940 941 case Media::RECORD_TYPE: 942 DB::table('media_file') 943 ->where('m_id', '=', $gid) 944 ->where('m_file', '=', $tree->id()) 945 ->delete(); 946 947 DB::table('media') 948 ->where('m_id', '=', $gid) 949 ->where('m_file', '=', $tree->id()) 950 ->delete(); 951 break; 952 953 default: 954 DB::table('other') 955 ->where('o_id', '=', $gid) 956 ->where('o_file', '=', $tree->id()) 957 ->delete(); 958 break; 959 } 960 961 if (!$delete) { 962 $this->importRecord($gedrec, $tree, true); 963 } 964 } 965} 966