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

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