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 HEAD')) { 288 $type = 'HEAD'; 289 $xref = 'HEAD'; // For records without an XREF, use the type as a pseudo XREF. 290 } elseif (str_starts_with($gedrec, '0 TRLR')) { 291 $tree->setPreference('imported', '1'); 292 $type = 'TRLR'; 293 $xref = 'TRLR'; // For records without an XREF, use the type as a pseudo XREF. 294 } elseif (str_starts_with($gedrec, '0 _PLAC_DEFN')) { 295 $this->importLegacyPlacDefn($gedrec); 296 297 return; 298 } elseif (str_starts_with($gedrec, '0 _PLAC ')) { 299 $this->importTNGPlac($gedrec); 300 301 return; 302 } else { 303 foreach (Gedcom::CUSTOM_RECORDS_WITHOUT_XREFS as $record_type) { 304 if (preg_match('/^0 ' . $record_type . '\b/', $gedrec) === 1) { 305 return; 306 } 307 } 308 309 throw new GedcomErrorException($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 // The database can only store MFU, and many of the stats queries assume this. 356 $sex = $record->sex(); 357 $sex = $sex === 'M' || $sex === 'F' ? $sex : 'U'; 358 359 DB::table('individuals')->insert([ 360 'i_id' => $xref, 361 'i_file' => $tree_id, 362 'i_rin' => $rin, 363 'i_sex' => $sex, 364 'i_gedcom' => $gedrec, 365 ]); 366 367 // Update the cross-reference/index tables. 368 $this->updatePlaces($xref, $tree, $gedrec); 369 $this->updateDates($xref, $tree_id, $gedrec); 370 $this->updateNames($xref, $tree_id, $record); 371 break; 372 373 case Family::RECORD_TYPE: 374 if (preg_match('/\n1 HUSB @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $match)) { 375 $husb = $match[1]; 376 } else { 377 $husb = ''; 378 } 379 if (preg_match('/\n1 WIFE @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $match)) { 380 $wife = $match[1]; 381 } else { 382 $wife = ''; 383 } 384 $nchi = preg_match_all('/\n1 CHIL @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $match); 385 if (preg_match('/\n1 NCHI (\d+)/', $gedrec, $match)) { 386 $nchi = max($nchi, $match[1]); 387 } 388 389 DB::table('families')->insert([ 390 'f_id' => $xref, 391 'f_file' => $tree_id, 392 'f_husb' => $husb, 393 'f_wife' => $wife, 394 'f_gedcom' => $gedrec, 395 'f_numchil' => $nchi, 396 ]); 397 398 // Update the cross-reference/index tables. 399 $this->updatePlaces($xref, $tree, $gedrec); 400 $this->updateDates($xref, $tree_id, $gedrec); 401 break; 402 403 case Source::RECORD_TYPE: 404 if (preg_match('/\n1 TITL (.+)/', $gedrec, $match)) { 405 $name = $match[1]; 406 } elseif (preg_match('/\n1 ABBR (.+)/', $gedrec, $match)) { 407 $name = $match[1]; 408 } else { 409 $name = $xref; 410 } 411 412 DB::table('sources')->insert([ 413 's_id' => $xref, 414 's_file' => $tree_id, 415 's_name' => mb_substr($name, 0, 255), 416 's_gedcom' => $gedrec, 417 ]); 418 break; 419 420 case Repository::RECORD_TYPE: 421 case Note::RECORD_TYPE: 422 case Submission::RECORD_TYPE: 423 case Submitter::RECORD_TYPE: 424 case Location::RECORD_TYPE: 425 DB::table('other')->insert([ 426 'o_id' => $xref, 427 'o_file' => $tree_id, 428 'o_type' => $type, 429 'o_gedcom' => $gedrec, 430 ]); 431 break; 432 433 case Header::RECORD_TYPE: 434 // Force HEAD records to have a creation date. 435 if (!str_contains($gedrec, "\n1 DATE ")) { 436 $today = strtoupper(date('d M Y')); 437 $gedrec .= "\n1 DATE " . $today; 438 } 439 440 DB::table('other')->insert([ 441 'o_id' => $xref, 442 'o_file' => $tree_id, 443 'o_type' => Header::RECORD_TYPE, 444 'o_gedcom' => $gedrec, 445 ]); 446 break; 447 448 449 case Media::RECORD_TYPE: 450 $record = Registry::mediaFactory()->new($xref, $gedrec, null, $tree); 451 452 DB::table('media')->insert([ 453 'm_id' => $xref, 454 'm_file' => $tree_id, 455 'm_gedcom' => $gedrec, 456 ]); 457 458 foreach ($record->mediaFiles() as $media_file) { 459 DB::table('media_file')->insert([ 460 'm_id' => $xref, 461 'm_file' => $tree_id, 462 'multimedia_file_refn' => mb_substr($media_file->filename(), 0, 248), 463 'multimedia_format' => mb_substr($media_file->format(), 0, 4), 464 'source_media_type' => mb_substr($media_file->type(), 0, 15), 465 'descriptive_title' => mb_substr($media_file->title(), 0, 248), 466 ]); 467 } 468 break; 469 470 default: // Custom record types. 471 DB::table('other')->insert([ 472 'o_id' => $xref, 473 'o_file' => $tree_id, 474 'o_type' => mb_substr($type, 0, 15), 475 'o_gedcom' => $gedrec, 476 ]); 477 break; 478 } 479 480 // Update the cross-reference/index tables. 481 $this->updateLinks($xref, $tree_id, $gedrec); 482 } 483 484 /** 485 * Legacy Family Tree software generates _PLAC_DEFN records containing LAT/LONG values 486 * 487 * @param string $gedcom 488 */ 489 private function importLegacyPlacDefn(string $gedcom): void 490 { 491 $gedcom_service = new GedcomService(); 492 493 if (preg_match('/\n1 PLAC (.+)/', $gedcom, $match)) { 494 $place_name = $match[1]; 495 } else { 496 return; 497 } 498 499 if (preg_match('/\n3 LATI ([NS].+)/', $gedcom, $match)) { 500 $latitude = $gedcom_service->readLatitude($match[1]); 501 } else { 502 return; 503 } 504 505 if (preg_match('/\n3 LONG ([EW].+)/', $gedcom, $match)) { 506 $longitude = $gedcom_service->readLongitude($match[1]); 507 } else { 508 return; 509 } 510 511 $location = new PlaceLocation($place_name); 512 513 if ($location->latitude() === null && $location->longitude() === null) { 514 DB::table('place_location') 515 ->where('id', '=', $location->id()) 516 ->update([ 517 'latitude' => $latitude, 518 'longitude' => $longitude, 519 ]); 520 } 521 } 522 523 /** 524 * Legacy Family Tree software generates _PLAC records containing LAT/LONG values 525 * 526 * @param string $gedcom 527 */ 528 private function importTNGPlac(string $gedcom): void 529 { 530 if (preg_match('/^0 _PLAC (.+)/', $gedcom, $match)) { 531 $place_name = $match[1]; 532 } else { 533 return; 534 } 535 536 if (preg_match('/\n2 LATI (.+)/', $gedcom, $match)) { 537 $latitude = (float) $match[1]; 538 } else { 539 return; 540 } 541 542 if (preg_match('/\n2 LONG (.+)/', $gedcom, $match)) { 543 $longitude = (float) $match[1]; 544 } else { 545 return; 546 } 547 548 $location = new PlaceLocation($place_name); 549 550 if ($location->latitude() === null && $location->longitude() === null) { 551 DB::table('place_location') 552 ->where('id', '=', $location->id()) 553 ->update([ 554 'latitude' => $latitude, 555 'longitude' => $longitude, 556 ]); 557 } 558 } 559 560 /** 561 * Extract all level 2 places from the given record and insert them into the places table 562 * 563 * @param string $xref 564 * @param Tree $tree 565 * @param string $gedrec 566 * 567 * @return void 568 */ 569 public function updatePlaces(string $xref, Tree $tree, string $gedrec): void 570 { 571 // Insert all new rows together 572 $rows = []; 573 574 preg_match_all('/\n2 PLAC (.+)/', $gedrec, $matches); 575 576 $places = array_unique($matches[1]); 577 578 foreach ($places as $place_name) { 579 $place = new Place($place_name, $tree); 580 581 // Calling Place::id() will create the entry in the database, if it doesn't already exist. 582 while ($place->id() !== 0) { 583 $rows[] = [ 584 'pl_p_id' => $place->id(), 585 'pl_gid' => $xref, 586 'pl_file' => $tree->id(), 587 ]; 588 589 $place = $place->parent(); 590 } 591 } 592 593 // array_unique doesn't work with arrays of arrays 594 $rows = array_intersect_key($rows, array_unique(array_map('serialize', $rows))); 595 596 // PDO has a limit of 65535 placeholders, and each row requires 3 placeholders. 597 foreach (array_chunk($rows, 20000) as $chunk) { 598 DB::table('placelinks')->insert($chunk); 599 } 600 } 601 602 /** 603 * Extract all the dates from the given record and insert them into the database. 604 * 605 * @param string $xref 606 * @param int $ged_id 607 * @param string $gedrec 608 * 609 * @return void 610 */ 611 private function updateDates(string $xref, int $ged_id, string $gedrec): void 612 { 613 // Insert all new rows together 614 $rows = []; 615 616 preg_match_all("/\n1 (\w+).*(?:\n[2-9].*)*\n2 DATE (.+)(?:\n[2-9].*)*/", $gedrec, $matches, PREG_SET_ORDER); 617 618 foreach ($matches as $match) { 619 $fact = $match[1]; 620 $date = new Date($match[2]); 621 $rows[] = [ 622 'd_day' => $date->minimumDate()->day, 623 'd_month' => $date->minimumDate()->format('%O'), 624 'd_mon' => $date->minimumDate()->month, 625 'd_year' => $date->minimumDate()->year, 626 'd_julianday1' => $date->minimumDate()->minimumJulianDay(), 627 'd_julianday2' => $date->minimumDate()->maximumJulianDay(), 628 'd_fact' => $fact, 629 'd_gid' => $xref, 630 'd_file' => $ged_id, 631 'd_type' => $date->minimumDate()->format('%@'), 632 ]; 633 634 $rows[] = [ 635 'd_day' => $date->maximumDate()->day, 636 'd_month' => $date->maximumDate()->format('%O'), 637 'd_mon' => $date->maximumDate()->month, 638 'd_year' => $date->maximumDate()->year, 639 'd_julianday1' => $date->maximumDate()->minimumJulianDay(), 640 'd_julianday2' => $date->maximumDate()->maximumJulianDay(), 641 'd_fact' => $fact, 642 'd_gid' => $xref, 643 'd_file' => $ged_id, 644 'd_type' => $date->minimumDate()->format('%@'), 645 ]; 646 } 647 648 // array_unique doesn't work with arrays of arrays 649 $rows = array_intersect_key($rows, array_unique(array_map('serialize', $rows))); 650 651 DB::table('dates')->insert($rows); 652 } 653 654 /** 655 * Extract all the links from the given record and insert them into the database 656 * 657 * @param string $xref 658 * @param int $ged_id 659 * @param string $gedrec 660 * 661 * @return void 662 */ 663 private function updateLinks(string $xref, int $ged_id, string $gedrec): void 664 { 665 // Insert all new rows together 666 $rows = []; 667 668 preg_match_all('/\n\d+ (' . Gedcom::REGEX_TAG . ') @(' . Gedcom::REGEX_XREF . ')@/', $gedrec, $matches, PREG_SET_ORDER); 669 670 foreach ($matches as $match) { 671 // Some applications (e.g. GenoPro) create links longer than 15 characters. 672 $link = mb_substr($match[1], 0, 15); 673 674 // Take care of "duplicates" that differ on case/collation, e.g. "SOUR @S1@" and "SOUR @s1@" 675 $rows[$link . strtoupper($match[2])] = [ 676 'l_from' => $xref, 677 'l_to' => $match[2], 678 'l_type' => $link, 679 'l_file' => $ged_id, 680 ]; 681 } 682 683 DB::table('link')->insert($rows); 684 } 685 686 /** 687 * Extract all the names from the given record and insert them into the database. 688 * 689 * @param string $xref 690 * @param int $ged_id 691 * @param Individual $record 692 * 693 * @return void 694 */ 695 private function updateNames(string $xref, int $ged_id, Individual $record): void 696 { 697 // Insert all new rows together 698 $rows = []; 699 700 foreach ($record->getAllNames() as $n => $name) { 701 if ($name['givn'] === Individual::PRAENOMEN_NESCIO) { 702 $soundex_givn_std = null; 703 $soundex_givn_dm = null; 704 } else { 705 $soundex_givn_std = Soundex::russell($name['givn']); 706 $soundex_givn_dm = Soundex::daitchMokotoff($name['givn']); 707 } 708 709 if ($name['surn'] === Individual::NOMEN_NESCIO) { 710 $soundex_surn_std = null; 711 $soundex_surn_dm = null; 712 } else { 713 $soundex_surn_std = Soundex::russell($name['surname']); 714 $soundex_surn_dm = Soundex::daitchMokotoff($name['surname']); 715 } 716 717 $rows[] = [ 718 'n_file' => $ged_id, 719 'n_id' => $xref, 720 'n_num' => $n, 721 'n_type' => $name['type'], 722 'n_sort' => mb_substr($name['sort'], 0, 255), 723 'n_full' => mb_substr($name['fullNN'], 0, 255), 724 'n_surname' => mb_substr($name['surname'], 0, 255), 725 'n_surn' => mb_substr($name['surn'], 0, 255), 726 'n_givn' => mb_substr($name['givn'], 0, 255), 727 'n_soundex_givn_std' => $soundex_givn_std, 728 'n_soundex_surn_std' => $soundex_surn_std, 729 'n_soundex_givn_dm' => $soundex_givn_dm, 730 'n_soundex_surn_dm' => $soundex_surn_dm, 731 ]; 732 } 733 734 DB::table('name')->insert($rows); 735 } 736 737 /** 738 * Extract inline media data, and convert to media objects. 739 * 740 * @param Tree $tree 741 * @param string $gedcom 742 * 743 * @return string 744 */ 745 private function convertInlineMedia(Tree $tree, string $gedcom): string 746 { 747 while (preg_match('/\n1 OBJE(?:\n[2-9].+)+/', $gedcom, $match)) { 748 $xref = $this->createMediaObject($match[0], $tree); 749 $gedcom = strtr($gedcom, [$match[0] => "\n1 OBJE @" . $xref . '@']); 750 } 751 while (preg_match('/\n2 OBJE(?:\n[3-9].+)+/', $gedcom, $match)) { 752 $xref = $this->createMediaObject($match[0], $tree); 753 $gedcom = strtr($gedcom, [$match[0] => "\n2 OBJE @" . $xref . '@']); 754 } 755 while (preg_match('/\n3 OBJE(?:\n[4-9].+)+/', $gedcom, $match)) { 756 $xref = $this->createMediaObject($match[0], $tree); 757 $gedcom = strtr($gedcom, [$match[0] => "\n3 OBJE @" . $xref . '@']); 758 } 759 760 return $gedcom; 761 } 762 763 /** 764 * Create a new media object, from inline media data. 765 * 766 * GEDCOM 5.5.1 specifies: +1 FILE / +2 FORM / +3 MEDI / +1 TITL 767 * GEDCOM 5.5 specifies: +1 FILE / +1 FORM / +1 TITL 768 * GEDCOM 5.5.1 says that GEDCOM 5.5 specifies: +1 FILE / +1 FORM / +2 MEDI 769 * 770 * Legacy generates: +1 FORM / +1 FILE / +1 TITL / +1 _SCBK / +1 _PRIM / +1 _TYPE / +1 NOTE 771 * RootsMagic generates: +1 FILE / +1 FORM / +1 TITL 772 * 773 * @param string $gedcom 774 * @param Tree $tree 775 * 776 * @return string 777 */ 778 private function createMediaObject(string $gedcom, Tree $tree): string 779 { 780 preg_match('/\n\d FILE (.+)/', $gedcom, $match); 781 $file = $match[1] ?? ''; 782 783 preg_match('/\n\d TITL (.+)/', $gedcom, $match); 784 $title = $match[1] ?? ''; 785 786 preg_match('/\n\d FORM (.+)/', $gedcom, $match); 787 $format = $match[1] ?? ''; 788 789 preg_match('/\n\d MEDI (.+)/', $gedcom, $match); 790 $media = $match[1] ?? ''; 791 792 preg_match('/\n\d _SCBK (.+)/', $gedcom, $match); 793 $scrapbook = $match[1] ?? ''; 794 795 preg_match('/\n\d _PRIM (.+)/', $gedcom, $match); 796 $primary = $match[1] ?? ''; 797 798 preg_match('/\n\d _TYPE (.+)/', $gedcom, $match); 799 if ($media === '') { 800 // Legacy uses _TYPE instead of MEDI 801 $media = $match[1] ?? ''; 802 $type = ''; 803 } else { 804 $type = $match[1] ?? ''; 805 } 806 807 preg_match_all('/\n\d NOTE (.+(?:\n\d CONT.*)*)/', $gedcom, $matches); 808 $notes = $matches[1] ?? []; 809 810 // Have we already created a media object with the same title/filename? 811 $xref = DB::table('media_file') 812 ->where('m_file', '=', $tree->id()) 813 ->where('descriptive_title', '=', mb_substr($title, 0, 248)) 814 ->where('multimedia_file_refn', '=', mb_substr($file, 0, 248)) 815 ->value('m_id'); 816 817 if ($xref === null) { 818 $xref = Registry::xrefFactory()->make(Media::RECORD_TYPE); 819 820 // convert to a media-object 821 $gedcom = '0 @' . $xref . "@ OBJE\n1 FILE " . $file; 822 823 if ($format !== '') { 824 $gedcom .= "\n2 FORM " . $format; 825 826 if ($media !== '') { 827 $gedcom .= "\n3 TYPE " . $media; 828 } 829 } 830 831 if ($title !== '') { 832 $gedcom .= "\n3 TITL " . $title; 833 } 834 835 if ($scrapbook !== '') { 836 $gedcom .= "\n1 _SCBK " . $scrapbook; 837 } 838 839 if ($primary !== '') { 840 $gedcom .= "\n1 _PRIM " . $primary; 841 } 842 843 if ($type !== '') { 844 $gedcom .= "\n1 _TYPE " . $type; 845 } 846 847 foreach ($notes as $note) { 848 $gedcom .= "\n1 NOTE " . strtr($note, ["\n3" => "\n2", "\n4" => "\n2", "\n5" => "\n2"]); 849 } 850 851 DB::table('media')->insert([ 852 'm_id' => $xref, 853 'm_file' => $tree->id(), 854 'm_gedcom' => $gedcom, 855 ]); 856 857 DB::table('media_file')->insert([ 858 'm_id' => $xref, 859 'm_file' => $tree->id(), 860 'multimedia_file_refn' => mb_substr($file, 0, 248), 861 'multimedia_format' => mb_substr($format, 0, 4), 862 'source_media_type' => mb_substr($media, 0, 15), 863 'descriptive_title' => mb_substr($title, 0, 248), 864 ]); 865 } 866 867 return $xref; 868 } 869 870 /** 871 * update a record in the database 872 * 873 * @param string $gedrec 874 * @param Tree $tree 875 * @param bool $delete 876 * 877 * @return void 878 * @throws GedcomErrorException 879 */ 880 public function updateRecord(string $gedrec, Tree $tree, bool $delete): void 881 { 882 if (preg_match('/^0 @(' . Gedcom::REGEX_XREF . ')@ (' . Gedcom::REGEX_TAG . ')/', $gedrec, $match)) { 883 [, $gid, $type] = $match; 884 } elseif (preg_match('/^0 (HEAD)(?:\n|$)/', $gedrec, $match)) { 885 // The HEAD record has no XREF. Any others? 886 $gid = $match[1]; 887 $type = $match[1]; 888 } else { 889 throw new GedcomErrorException($gedrec); 890 } 891 892 // Place links 893 DB::table('placelinks') 894 ->where('pl_gid', '=', $gid) 895 ->where('pl_file', '=', $tree->id()) 896 ->delete(); 897 898 // Orphaned places. If we're deleting "Westminster, London, England", 899 // then we may also need to delete "London, England" and "England". 900 do { 901 $affected = DB::table('places') 902 ->leftJoin('placelinks', function (JoinClause $join): void { 903 $join 904 ->on('p_id', '=', 'pl_p_id') 905 ->on('p_file', '=', 'pl_file'); 906 }) 907 ->whereNull('pl_p_id') 908 ->delete(); 909 } while ($affected > 0); 910 911 DB::table('dates') 912 ->where('d_gid', '=', $gid) 913 ->where('d_file', '=', $tree->id()) 914 ->delete(); 915 916 DB::table('name') 917 ->where('n_id', '=', $gid) 918 ->where('n_file', '=', $tree->id()) 919 ->delete(); 920 921 DB::table('link') 922 ->where('l_from', '=', $gid) 923 ->where('l_file', '=', $tree->id()) 924 ->delete(); 925 926 switch ($type) { 927 case Individual::RECORD_TYPE: 928 DB::table('individuals') 929 ->where('i_id', '=', $gid) 930 ->where('i_file', '=', $tree->id()) 931 ->delete(); 932 break; 933 934 case Family::RECORD_TYPE: 935 DB::table('families') 936 ->where('f_id', '=', $gid) 937 ->where('f_file', '=', $tree->id()) 938 ->delete(); 939 break; 940 941 case Source::RECORD_TYPE: 942 DB::table('sources') 943 ->where('s_id', '=', $gid) 944 ->where('s_file', '=', $tree->id()) 945 ->delete(); 946 break; 947 948 case Media::RECORD_TYPE: 949 DB::table('media_file') 950 ->where('m_id', '=', $gid) 951 ->where('m_file', '=', $tree->id()) 952 ->delete(); 953 954 DB::table('media') 955 ->where('m_id', '=', $gid) 956 ->where('m_file', '=', $tree->id()) 957 ->delete(); 958 break; 959 960 default: 961 DB::table('other') 962 ->where('o_id', '=', $gid) 963 ->where('o_file', '=', $tree->id()) 964 ->delete(); 965 break; 966 } 967 968 if (!$delete) { 969 $this->importRecord($gedrec, $tree, true); 970 } 971 } 972} 973