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     /**
  14      * @var EntityProvider
  15      */
  16     protected $entityProvider;
  17
  18     public function __construct(EntityProvider $entityProvider)
  19     {
  20         $this->entityProvider = $entityProvider;
  21     }
  22
  23     /**
  24      * Index the given entity.
  25      */
  26     public function indexEntity(Entity $entity)
  27     {
  28         $this->deleteEntityTerms($entity);
  29         $terms = $this->entityToTermDataArray($entity);
  30         SearchTerm::query()->insert($terms);
  31     }
  32
  33     /**
  34      * Index multiple Entities at once.
  35      *
  36      * @param Entity[] $entities
  37      */
  38     public function indexEntities(array $entities)
  39     {
  40         $terms = [];
  41         foreach ($entities as $entity) {
  42             $entityTerms = $this->entityToTermDataArray($entity);
  43             array_push($terms, ...$entityTerms);
  44         }
  45
  46         $chunkedTerms = array_chunk($terms, 500);
  47         foreach ($chunkedTerms as $termChunk) {
  48             SearchTerm::query()->insert($termChunk);
  49         }
  50     }
  51
  52     /**
  53      * Delete and re-index the terms for all entities in the system.
  54      */
  55     public function indexAllEntities()
  56     {
  57         SearchTerm::query()->truncate();
  58
  59         foreach ($this->entityProvider->all() as $entityModel) {
  60             $selectFields = ['id', 'name', $entityModel->textField];
  61             $entityModel->newQuery()
  62                 ->withTrashed()
  63                 ->select($selectFields)
  64                 ->chunk(1000, function (Collection $entities) {
  65                     $this->indexEntities($entities->all());
  66                 });
  67         }
  68     }
  69
  70     /**
  71      * Delete related Entity search terms.
  72      */
  73     public function deleteEntityTerms(Entity $entity)
  74     {
  75         $entity->searchTerms()->delete();
  76     }
  77
  78     /**
  79      * Create a scored term array from the given text.
  80      *
  81      * @returns array{term: string, score: float}
  82      */
  83     protected function generateTermArrayFromText(string $text, int $scoreAdjustment = 1): array
  84     {
  85         $tokenMap = []; // {TextToken => OccurrenceCount}
  86         $splitChars = " \n\t.,!?:;()[]{}<>`'\"";
  87         $token = strtok($text, $splitChars);
  88
  89         while ($token !== false) {
  90             if (!isset($tokenMap[$token])) {
  91                 $tokenMap[$token] = 0;
  92             }
  93             $tokenMap[$token]++;
  94             $token = strtok($splitChars);
  95         }
  96
  97         $terms = [];
  98         foreach ($tokenMap as $token => $count) {
  99             $terms[] = [
 100                 'term'  => $token,
 101                 'score' => $count * $scoreAdjustment,
 102             ];
 103         }
 104
 105         return $terms;
 106     }
 107
 108     /**
 109      * For the given entity, Generate an array of term data details.
 110      * Is the raw term data, not instances of SearchTerm models.
 111      *
 112      * @returns array{term: string, score: float}[]
 113      */
 114     protected function entityToTermDataArray(Entity $entity): array
 115     {
 116         $nameTerms = $this->generateTermArrayFromText($entity->name, 40 * $entity->searchFactor);
 117         $bodyTerms = $this->generateTermArrayFromText($entity->getText(), 1 * $entity->searchFactor);
 118         $termData = array_merge($nameTerms, $bodyTerms);
 119
 120         foreach ($termData as $index => $term) {
 121             $termData[$index]['entity_type'] = $entity->getMorphClass();
 122             $termData[$index]['entity_id'] = $entity->id;
 123         }
 124
 125         return $termData;
 126     }
 127 }