Include user tags in Beatmapset search

app/Jobs/EsDocument.php

@@ -17,7 +17,7 @@ class EsDocument implements ShouldQueue
 {
     use Dispatchable, InteractsWithQueue, Queueable;
 
-    private array $modelMeta;
+    protected array $modelMeta;
 
     /**
      * Create a new job instance.

app/Jobs/EsDocumentUnique.php

@@ -0,0 +1,20 @@
+<?php
+
+// Copyright (c) ppy Pty Ltd <[email protected]>. Licensed under the GNU Affero General Public License v3.0.
+// See the LICENCE file in the repository root for full licence text.
+
+declare(strict_types=1);
+
+namespace App\Jobs;
+
+use Illuminate\Contracts\Queue\ShouldBeUnique;
+
+class EsDocumentUnique extends EsDocument implements ShouldBeUnique
+{
+    public int $uniqueFor = 600;
+
+    public function uniqueId(): string
+    {
+        return "{$this->modelMeta['class']}-{$this->modelMeta['id']}";
+    }
+}

app/Libraries/Search/BeatmapsetQueryParser.php

@@ -82,6 +82,9 @@ public static function parse(?string $query): array
                 case 'source':
                     $option = static::makeTextOption($op, $m['value']);
                     break;
+                case 'tag':
+                    $option = [static::makeTextOption($op, $m['value'])];
+                    break;
                 case 'title':
                     $option = static::makeTextOption($op, $m['value']);
                     break;

app/Libraries/Search/BeatmapsetSearch.php

@@ -14,6 +14,7 @@
 use App\Models\Beatmapset;
 use App\Models\Follow;
 use App\Models\Solo;
+use App\Models\Tag;
 use App\Models\User;
 use Ds\Set;
 
@@ -60,6 +61,12 @@ public function getQuery()
                     ->should(['term' => ['_id' => ['value' => $this->params->queryString, 'boost' => 100]]])
                     ->should(QueryHelper::queryString($this->params->queryString, $partialMatchFields, 'or', 1 / count($terms)))
                     ->should(QueryHelper::queryString($this->params->queryString, [], 'and'))
+                    ->should([
+                        'nested' => [
+                            'path' => 'beatmaps',
+                            'query' => QueryHelper::queryString($this->params->queryString, ['beatmaps.top_tags'], 'or', 0.5 / count($terms)),
+                        ],
+                    ])
             );
         }
 
@@ -82,6 +89,7 @@ public function getQuery()
         $this->addPlayedFilter($query, $nested);
         $this->addRankFilter($nested);
         $this->addRecommendedFilter($nested);
+        $this->addTagsFilter($nested);
 
         $this->addSimpleFilters($query, $nested);
         $this->addCreatorFilter($query, $nested);
@@ -398,6 +406,25 @@ private function addTextFilter(BoolQuery $query, string $paramField, array $fiel
         $query->must($subQuery);
     }
 
+    private function addTagsFilter(BoolQuery $query): void
+    {
+        if ($this->params->tags === null) {
+            return;
+        }
+
+        $tagSet = new Set(array_map('mb_strtolower', $this->params->tags));
+        $tags = Tag::whereIn('name', $this->params->tags)->limit(10)->pluck('name');
+        $tagSet->remove(...$tags->map(fn ($name) => mb_strtolower($name))->toArray());
+
+        foreach ($tagSet as $tag) {
+            $query->filter(QueryHelper::queryString($tag, ['beatmaps.top_tags'], 'and'));
+        }
+
+        foreach ($tags as $tag) {
+            $query->filter(['term' => ['beatmaps.top_tags.raw' => $tag]]);
+        }
+    }
+
     private function getPlayedBeatmapIds(?array $rank = null)
     {
         $query = Solo\Score

app/Libraries/Search/BeatmapsetSearchParams.php

@@ -45,6 +45,7 @@ class BeatmapsetSearchParams extends SearchParams
     public bool $showSpotlights = false;
     public ?string $source = null;
     public ?string $status = null;
+    public ?array $tags = null;
     public ?string $title = null;
     public ?array $statusRange = null;
     public ?array $totalLength = null;

app/Libraries/Search/BeatmapsetSearchRequestParams.php

@@ -226,6 +226,7 @@ private function parseQuery(): void
             'source' => 'source',
             'stars' => 'difficultyRating',
             'status' => 'statusRange',
+            'tag' => 'tags',
             'title' => 'title',
             'updated' => 'updated',
         ];

app/Models/Beatmap.php

@@ -6,7 +6,7 @@
 namespace App\Models;
 
 use App\Exceptions\InvariantException;
-use App\Jobs\EsDocument;
+use App\Jobs\EsDocumentUnique;
 use App\Libraries\Transactions\AfterCommit;
 use App\Traits\Memoizes;
 use Illuminate\Database\Eloquent\Builder;
@@ -247,7 +247,7 @@ public function afterCommit()
         $beatmapset = $this->beatmapset;
 
         if ($beatmapset !== null) {
-            dispatch(new EsDocument($beatmapset));
+            dispatch(new EsDocumentUnique($beatmapset));
         }
     }
 

app/Models/Beatmapset.php

@@ -10,7 +10,7 @@
 use App\Exceptions\ImageProcessorServiceException;
 use App\Exceptions\InvariantException;
 use App\Jobs\CheckBeatmapsetCovers;
-use App\Jobs\EsDocument;
+use App\Jobs\EsDocumentUnique;
 use App\Jobs\Notifications\BeatmapsetDiscussionLock;
 use App\Jobs\Notifications\BeatmapsetDiscussionUnlock;
 use App\Jobs\Notifications\BeatmapsetDisqualify;
@@ -1508,7 +1508,7 @@ public function refreshCache(bool $resetEligibleMainRulesets = false): void
 
     public function afterCommit()
     {
-        dispatch(new EsDocument($this));
+        dispatch(new EsDocumentUnique($this));
     }
 
     public function notificationCover()

app/Models/Traits/Es/BaseDbIndexable.php

@@ -82,12 +82,13 @@ public function esRouting()
 
     public function esDeleteDocument(array $options = [])
     {
-        $document = array_merge([
+        $document = [
             'index' => static::esIndexName(),
             'routing' => $this->esRouting(),
             'id' => $this->getEsId(),
             'client' => ['ignore' => 404],
-        ], $options);
+            ...$options,
+        ];
 
         return Es::getClient()->delete($document);
     }
@@ -98,12 +99,13 @@ public function esIndexDocument(array $options = [])
             return $this->esDeleteDocument($options);
         }
 
-        $document = array_merge([
+        $document = [
             'index' => static::esIndexName(),
             'routing' => $this->esRouting(),
             'id' => $this->getEsId(),
             'body' => $this->toEsJson(),
-        ], $options);
+            ...$options,
+        ];
 
         return Es::getClient()->index($document);
     }

app/Models/Traits/Es/BeatmapsetSearch.php

@@ -31,18 +31,30 @@ public static function esSchemaFile()
         return config_path('schemas/beatmapsets.json');
     }
 
+    private static function esBeatmapTags(Beatmap $beatmap): array
+    {
+        $tags = app('tags');
+
+        return array_reject_null(
+            array_map(
+                fn ($tagId) => $tags->get($tagId['tag_id'])?->name,
+                $beatmap->topTagIds()
+            )
+        );
+    }
+
     public function esShouldIndex()
     {
         return !$this->trashed() && !present($this->download_disabled_url);
     }
 
     public function toEsJson()
     {
-        return array_merge(
-            $this->esBeatmapsetValues(),
-            ['beatmaps' => $this->esBeatmapsValues()],
-            ['difficulties' => $this->esDifficultiesValues()]
-        );
+        return [
+            ...$this->esBeatmapsetValues(),
+            'beatmaps' => $this->esBeatmapsValues(),
+            'difficulties' => $this->esDifficultiesValues(),
+        ];
     }
 
     private function esBeatmapsetValues()
@@ -78,12 +90,17 @@ private function esBeatmapsValues()
         foreach ($this->beatmaps as $beatmap) {
             $beatmapValues = [];
             foreach ($mappings as $field => $mapping) {
-                $beatmapValues[$field] = $beatmap->$field;
+                $value = match ($field) {
+                    'top_tags' => $this->esBeatmapTags($beatmap),
+                    // TODO: remove adding $beatmap->user_id once everything else also populated beatmap_owners by default.
+                    // Duplicate user_id in the array should be fine for now since the field isn't scored for querying.
+                    'user_id' => $beatmap->beatmapOwners->pluck('user_id')->add($beatmap->user_id),
+                    default => $beatmap->$field,
+                };
+
+                $beatmapValues[$field] = $value;
             }
 
-            // TODO: remove adding $beatmap->user_id once everything else also populated beatmap_owners by default.
-            // Duplicate user_id in the array should be fine for now since the field isn't scored for querying.
-            $beatmapValues['user_id'] = $beatmap->beatmapOwners->pluck('user_id')->add($beatmap->user_id);
             $values[] = $beatmapValues;
 
             if ($beatmap->playmode === Beatmap::MODES['osu']) {
@@ -96,11 +113,16 @@ private function esBeatmapsValues()
                     $convert->playmode = $modeInt;
                     $convert->convert = true;
                     $convertValues = [];
-                    foreach ($mappings as $field => $mapping) {
-                        $convertValues[$field] = $convert->$field;
+                    foreach ($mappings as $field => $_mapping) {
+                        $convertValues[$field] = match ($field) {
+                            // just add a copy for converts too.
+                            'top_tags',
+                            'user_id' => $beatmapValues[$field],
+
+                            default => $convert->$field,
+                        };
                     }
 
-                    $convertValues['user_id'] = $beatmapValues['user_id']; // just add a copy for converts too.
                     $values[] = $convertValues;
                 }
             }

config/schemas/beatmapsets.json

@@ -84,6 +84,14 @@
           "playmode": {
             "type": "byte"
           },
+          "top_tags": {
+            "type": "text",
+            "fields": {
+              "raw": {
+                "type": "keyword"
+              }
+            }
+          },
           "total_length": {
             "type": "long"
           },

tests/Libraries/Search/BeatmapsetQueryParserTest.php

@@ -46,6 +46,9 @@ public static function queryDataProvider()
             ['ranked>="2020-07-21 12:30:30 +09:00"', ['keywords' => null, 'options' => ['ranked' => ['gte' => static::parseTime('2020-07-21 03:30:30')]]]],
             ['ranked="2020-07-21 12:30:30 +09:00"', ['keywords' => null, 'options' => ['ranked' => ['gte' => static::parseTime('2020-07-21 03:30:30'), 'lt' => static::parseTime('2020-07-21 03:30:31')]]]],
             ['ranked="invalid date format"', ['keywords' => 'ranked="invalid date format"', 'options' => []]],
+            ['tag=hello', ['keywords' => null, 'options' => ['tag' => ['hello']]]],
+            ['tag=hello tag=world', ['keywords' => null, 'options' => ['tag' => ['hello', 'world']]]],
+            ['tag="hello world"', ['keywords' => null, 'options' => ['tag' => ['hello world']]]],
 
             // multiple options
             ['artist=hello creator:world', ['keywords' => null, 'options' => ['artist' => 'hello', 'creator' => 'world']]],