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      * Can take a callback which is used for reporting progress.
  55      * Callback receives three arguments:
  56      * - An instance of the model being processed
  57      * - The number that have been processed so far.
  58      * - The total number of that model to be processed.
  59      *
  60      * @param callable(Entity, int, int)|null $progressCallback
  61      */
  62     public function indexAllEntities(?callable $progressCallback = null)
  63     {
  64         SearchTerm::query()->truncate();
  65
  66         foreach ($this->entityProvider->all() as $entityModel) {
  67             $selectFields = ['id', 'name', $entityModel->textField];
  68             $total = $entityModel->newQuery()->withTrashed()->count();
  69             $chunkSize = 250;
  70             $processed = 0;
  71
  72             $chunkCallback = function (Collection $entities) use ($progressCallback, &$processed, $total, $chunkSize, $entityModel) {
  73                 $this->indexEntities($entities->all());
  74                 $processed = min($processed + $chunkSize, $total);
  75
  76                 if (is_callable($progressCallback)) {
  77                     $progressCallback($entityModel, $processed, $total);
  78                 }
  79             };
  80
  81             $entityModel->newQuery()
  82                 ->select($selectFields)
  83                 ->chunk($chunkSize, $chunkCallback);
  84         }
  85     }
  86
  87     /**
  88      * Delete related Entity search terms.
  89      */
  90     public function deleteEntityTerms(Entity $entity)
  91     {
  92         $entity->searchTerms()->delete();
  93     }
  94
  95     /**
  96      * Create a scored term array from the given text.
  97      *
  98      * @returns array{term: string, score: float}
  99      */
 100     protected function generateTermArrayFromText(string $text, int $scoreAdjustment = 1): array
 101     {
 102         $tokenMap = []; // {TextToken => OccurrenceCount}
 103         $splitChars = " \n\t.,!?:;()[]{}<>`'\"";
 104         $token = strtok($text, $splitChars);
 105
 106         while ($token !== false) {
 107             if (!isset($tokenMap[$token])) {
 108                 $tokenMap[$token] = 0;
 109             }
 110             $tokenMap[$token]++;
 111             $token = strtok($splitChars);
 112         }
 113
 114         $terms = [];
 115         foreach ($tokenMap as $token => $count) {
 116             $terms[] = [
 117                 'term'  => $token,
 118                 'score' => $count * $scoreAdjustment,
 119             ];
 120         }
 121
 122         return $terms;
 123     }
 124
 125     /**
 126      * For the given entity, Generate an array of term data details.
 127      * Is the raw term data, not instances of SearchTerm models.
 128      *
 129      * @returns array{term: string, score: float}[]
 130      */
 131     protected function entityToTermDataArray(Entity $entity): array
 132     {
 133         $nameTerms = $this->generateTermArrayFromText($entity->name, 40 * $entity->searchFactor);
 134         $bodyTerms = $this->generateTermArrayFromText($entity->getText(), 1 * $entity->searchFactor);
 135         $termData = array_merge($nameTerms, $bodyTerms);
 136
 137         foreach ($termData as $index => $term) {
 138             $termData[$index]['entity_type'] = $entity->getMorphClass();
 139             $termData[$index]['entity_id'] = $entity->id;
 140         }
 141
 142         return $termData;
 143     }
 144 }