BookStack Code Mirror - bookstack/blob - app/Entities/Tools/SearchIndex.php

   1 <?php
   2
   3 namespace BookStack\Entities\Tools;
   4
   5 use BookStack\Entities\EntityProvider;
   6 use BookStack\Entities\Models\Entity;
   7 use BookStack\Entities\Models\SearchTerm;
   8 use Illuminate\Support\Collection;
   9
  10 class SearchIndex
  11 {
  12     /**
  13      * @var SearchTerm
  14      */
  15     protected $searchTerm;
  16
  17     /**
  18      * @var EntityProvider
  19      */
  20     protected $entityProvider;
  21
  22     public function __construct(SearchTerm $searchTerm, EntityProvider $entityProvider)
  23     {
  24         $this->searchTerm = $searchTerm;
  25         $this->entityProvider = $entityProvider;
  26     }
  27
  28     /**
  29      * Index the given entity.
  30      */
  31     public function indexEntity(Entity $entity)
  32     {
  33         $this->deleteEntityTerms($entity);
  34         $nameTerms = $this->generateTermArrayFromText($entity->name, 5 * $entity->searchFactor);
  35         $bodyTerms = $this->generateTermArrayFromText($entity->getText(), 1 * $entity->searchFactor);
  36         $terms = array_merge($nameTerms, $bodyTerms);
  37         foreach ($terms as $index => $term) {
  38             $terms[$index]['entity_type'] = $entity->getMorphClass();
  39             $terms[$index]['entity_id'] = $entity->id;
  40         }
  41         $this->searchTerm->newQuery()->insert($terms);
  42     }
  43
  44     /**
  45      * Index multiple Entities at once.
  46      *
  47      * @param Entity[] $entities
  48      */
  49     protected function indexEntities(array $entities)
  50     {
  51         $terms = [];
  52         foreach ($entities as $entity) {
  53             $nameTerms = $this->generateTermArrayFromText($entity->name, 5 * $entity->searchFactor);
  54             $bodyTerms = $this->generateTermArrayFromText($entity->getText(), 1 * $entity->searchFactor);
  55             foreach (array_merge($nameTerms, $bodyTerms) as $term) {
  56                 $term['entity_id'] = $entity->id;
  57                 $term['entity_type'] = $entity->getMorphClass();
  58                 $terms[] = $term;
  59             }
  60         }
  61
  62         $chunkedTerms = array_chunk($terms, 500);
  63         foreach ($chunkedTerms as $termChunk) {
  64             $this->searchTerm->newQuery()->insert($termChunk);
  65         }
  66     }
  67
  68     /**
  69      * Delete and re-index the terms for all entities in the system.
  70      */
  71     public function indexAllEntities()
  72     {
  73         $this->searchTerm->newQuery()->truncate();
  74
  75         foreach ($this->entityProvider->all() as $entityModel) {
  76             $selectFields = ['id', 'name', $entityModel->textField];
  77             $entityModel->newQuery()
  78                 ->withTrashed()
  79                 ->select($selectFields)
  80                 ->chunk(1000, function (Collection $entities) {
  81                     $this->indexEntities($entities->all());
  82                 });
  83         }
  84     }
  85
  86     /**
  87      * Delete related Entity search terms.
  88      */
  89     public function deleteEntityTerms(Entity $entity)
  90     {
  91         $entity->searchTerms()->delete();
  92     }
  93
  94     /**
  95      * Create a scored term array from the given text.
  96      */
  97     protected function generateTermArrayFromText(string $text, int $scoreAdjustment = 1): array
  98     {
  99         $tokenMap = []; // {TextToken => OccurrenceCount}
 100         $splitChars = " \n\t.,!?:;()[]{}<>`'\"";
 101         $token = strtok($text, $splitChars);
 102
 103         while ($token !== false) {
 104             if (!isset($tokenMap[$token])) {
 105                 $tokenMap[$token] = 0;
 106             }
 107             $tokenMap[$token]++;
 108             $token = strtok($splitChars);
 109         }
 110
 111         $terms = [];
 112         foreach ($tokenMap as $token => $count) {
 113             $terms[] = [
 114                 'term'  => $token,
 115                 'score' => $count * $scoreAdjustment,
 116             ];
 117         }
 118
 119         return $terms;
 120     }
 121 }