perf(search): denormalize public compound catalog onto molecules

app/Actions/Project/ArchiveProject.php

@@ -3,7 +3,7 @@
 namespace App\Actions\Project;
 
 use App\Models\Project;
-use App\Support\Public\PublicMoleculeAggregates;
+use App\Support\Public\PublicMoleculeCatalogIndexer;
 
 class ArchiveProject
 {
@@ -36,7 +36,7 @@ public function toggleArchive($project)
         }
         $project->save();
 
-        PublicMoleculeAggregates::forgetPublicCatalogTotalCache();
+        app(PublicMoleculeCatalogIndexer::class)->refreshForProject($project);
     }
 
     /**

app/Actions/Project/PublishProject.php

@@ -4,7 +4,7 @@
 
 use App\Jobs\ProcessMetadataExtractionBagitGenerationJob;
 use App\Models\Project;
-use App\Support\Public\PublicMoleculeAggregates;
+use App\Support\Public\PublicMoleculeCatalogIndexer;
 
 class PublishProject
 {
@@ -32,6 +32,6 @@ public function publish($project)
             }
         }
 
-        PublicMoleculeAggregates::forgetPublicCatalogTotalCache();
+        app(PublicMoleculeCatalogIndexer::class)->refreshForProject($project);
     }
 }

app/Actions/Project/UnPublishProject.php

@@ -3,7 +3,7 @@
 namespace App\Actions\Project;
 
 use App\Models\Project;
-use App\Support\Public\PublicMoleculeAggregates;
+use App\Support\Public\PublicMoleculeCatalogIndexer;
 
 class UnPublishProject
 {
@@ -29,6 +29,6 @@ public function unPublish($project)
             }
         }
 
-        PublicMoleculeAggregates::forgetPublicCatalogTotalCache();
+        app(PublicMoleculeCatalogIndexer::class)->refreshForProject($project);
     }
 }

app/Actions/Study/PublishStudy.php

@@ -5,7 +5,7 @@
 use App\Jobs\ProcessMetadataExtractionBagitGenerationJob;
 use App\Models\Study;
 use App\Services\ChemotionRepositoryTrackerService;
-use App\Support\Public\PublicMoleculeAggregates;
+use App\Support\Public\PublicMoleculeCatalogIndexer;
 use Illuminate\Support\Facades\Log;
 
 class PublishStudy
@@ -26,6 +26,7 @@ public function publish($study)
             $dataset->save();
         }
 
+        app(PublicMoleculeCatalogIndexer::class)->refreshForStudy($study);
         if ($study->is_public && $study->has_nmrium && filled($study->download_url)) {
             ProcessMetadataExtractionBagitGenerationJob::dispatch($study->id);
         }

app/Console/Commands/IndexPublicMoleculeCatalogCommand.php

@@ -0,0 +1,30 @@
+<?php
+
+namespace App\Console\Commands;
+
+use App\Support\Public\PublicMoleculeCatalogIndexer;
+use Illuminate\Console\Attributes\Description;
+use Illuminate\Console\Attributes\Signature;
+use Illuminate\Console\Command;
+
+#[Signature('nmrxiv:index-public-molecule-catalog {--molecule= : Single molecule id} {--chunk=100 : Molecules per enrichment chunk}')]
+#[Description('Denormalize public-catalog membership and compound-card badges onto molecules')]
+class IndexPublicMoleculeCatalogCommand extends Command
+{
+    public function handle(PublicMoleculeCatalogIndexer $indexer): int
+    {
+        $chunk = max(1, (int) $this->option('chunk'));
+        $moleculeOption = $this->option('molecule');
+        $moleculeIds = filled($moleculeOption) ? [(int) $moleculeOption] : null;
+
+        $result = $indexer->refresh($moleculeIds, $chunk);
+
+        $this->info(sprintf(
+            'Indexed public molecule catalog (%s in catalog, %s cleared).',
+            number_format($result['indexed']),
+            number_format($result['cleared']),
+        ));
+
+        return self::SUCCESS;
+    }
+}

app/Http/Controllers/API/SearchController.php

@@ -764,7 +764,7 @@ public function search(Request $request, ?string $smiles = null)
 
             $queryType = strtolower($queryType);
 
-            $publicSpectraExists = PublicMoleculeAggregates::hasPublicSpectraExistsSql('molecules.id');
+            $publicSpectraFilter = PublicMoleculeAggregates::hasPublicSpectraColumnSql();
             $orderByRecentSql = $sort === 'recent' ? 'ORDER BY molecules.created_at DESC' : '';
 
             $ids = [];
@@ -774,7 +774,7 @@ public function search(Request $request, ?string $smiles = null)
                 ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                     [
                         'from' => 'FROM molecules',
-                        'where' => "WHERE identifier IS NOT NULL AND (smiles LIKE ? OR absolute_smiles LIKE ? OR canonical_smiles LIKE ?) AND {$publicSpectraExists}",
+                        'where' => "WHERE identifier IS NOT NULL AND (smiles LIKE ? OR absolute_smiles LIKE ? OR canonical_smiles LIKE ?) AND {$publicSpectraFilter}",
                         'order' => $orderByRecentSql,
                     ],
                     ['%'.$query.'%', '%'.$query.'%', '%'.$query.'%'],
@@ -786,7 +786,7 @@ public function search(Request $request, ?string $smiles = null)
                     ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                         [
                             'from' => 'FROM mols INNER JOIN molecules ON molecules.id = mols.id',
-                            'where' => "WHERE m@>? AND molecules.identifier IS NOT NULL AND {$publicSpectraExists}",
+                            'where' => "WHERE m@>? AND molecules.identifier IS NOT NULL AND {$publicSpectraFilter}",
                             'id' => 'mols.id',
                             'order' => $orderByRecentSql,
                         ],
@@ -803,7 +803,7 @@ public function search(Request $request, ?string $smiles = null)
                 ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                     [
                         'from' => 'FROM molecules',
-                        'where' => "WHERE identifier IS NOT NULL AND (inchi LIKE ? OR standard_inchi LIKE ?) AND {$publicSpectraExists}",
+                        'where' => "WHERE identifier IS NOT NULL AND (inchi LIKE ? OR standard_inchi LIKE ?) AND {$publicSpectraFilter}",
                         'order' => $orderByRecentSql,
                     ],
                     ['%'.$query.'%', '%'.$query.'%'],
@@ -814,7 +814,7 @@ public function search(Request $request, ?string $smiles = null)
                 ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                     [
                         'from' => 'FROM molecules',
-                        'where' => "WHERE identifier IS NOT NULL AND (inchi_key LIKE ? OR standard_inchi_key LIKE ?) AND {$publicSpectraExists}",
+                        'where' => "WHERE identifier IS NOT NULL AND (inchi_key LIKE ? OR standard_inchi_key LIKE ?) AND {$publicSpectraFilter}",
                         'order' => $orderByRecentSql,
                     ],
                     ['%'.$query.'%', '%'.$query.'%'],
@@ -826,7 +826,7 @@ public function search(Request $request, ?string $smiles = null)
                     ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                         [
                             'from' => 'FROM mols INNER JOIN molecules ON molecules.id = mols.id',
-                            'where' => "WHERE m@=? AND molecules.identifier IS NOT NULL AND {$publicSpectraExists}",
+                            'where' => "WHERE m@=? AND molecules.identifier IS NOT NULL AND {$publicSpectraFilter}",
                             'id' => 'mols.id',
                             'order' => $orderByRecentSql,
                         ],
@@ -844,7 +844,7 @@ public function search(Request $request, ?string $smiles = null)
                     ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                         [
                             'from' => 'FROM fps INNER JOIN molecules ON molecules.id = fps.id',
-                            'where' => "WHERE mfp2%morganbv_fp(?) AND molecules.identifier IS NOT NULL AND {$publicSpectraExists}",
+                            'where' => "WHERE mfp2%morganbv_fp(?) AND molecules.identifier IS NOT NULL AND {$publicSpectraFilter}",
                             'id' => 'fps.id',
                             'order' => $orderByRecentSql,
                         ],
@@ -884,7 +884,7 @@ public function search(Request $request, ?string $smiles = null)
                 ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                     [
                         'from' => 'FROM molecules',
-                        'where' => "WHERE identifier IS NOT NULL AND (name::TEXT ILIKE ? OR iupac_name ILIKE ? OR synonyms::TEXT ILIKE ? OR identifier::TEXT ILIKE ?) AND {$publicSpectraExists}",
+                        'where' => "WHERE identifier IS NOT NULL AND (name::TEXT ILIKE ? OR iupac_name ILIKE ? OR synonyms::TEXT ILIKE ? OR identifier::TEXT ILIKE ?) AND {$publicSpectraFilter}",
                         'order' => $orderByRecentSql,
                     ],
                     ['%'.$query.'%', '%'.$query.'%', '%'.$query.'%', '%'.$query.'%'],
@@ -1135,7 +1135,7 @@ private function buildSecureFilterQuery(string $query, array $filterMap, int $li
             }
 
             $whereClause = implode(' OR ', $whereConditions);
-            $publicSpectraFilter = PublicMoleculeAggregates::hasPublicSpectraExistsSql('molecules.id');
+            $publicSpectraFilter = PublicMoleculeAggregates::hasPublicSpectraColumnSql();
 
             ['ids' => $ids, 'total' => $count] = PublicMoleculeAggregates::paginateIds(
                 [

app/Models/Molecule.php

@@ -38,6 +38,10 @@ protected function casts(): array
     {
         return [
             'workspace_experiment_type_counts' => 'array',
+            'has_public_spectra' => 'boolean',
+            'public_samples_count' => 'integer',
+            'public_experiment_type_counts' => 'array',
+            'public_catalog_indexed_at' => 'datetime',
         ];
     }
 

app/Support/Public/PublicMoleculeAggregates.php

@@ -21,8 +21,17 @@ final class PublicMoleculeAggregates
 
     private const PUBLIC_CATALOG_TOTAL_CACHE_SECONDS = 300;
 
+    /**
+     * Indexed public-catalog membership written by PublicMoleculeCatalogIndexer.
+     */
+    public static function hasPublicSpectraColumnSql(string $table = 'molecules'): string
+    {
+        return "{$table}.has_public_spectra = true";
+    }
+
     /**
      * Correlated EXISTS: molecule has ≥1 spectrum in the public catalog.
+     * Used by the catalog indexer, not by request-path search.
      */
     public static function hasPublicSpectraExistsSql(string $moleculeIdColumn = 'molecules.id'): string
     {
@@ -133,12 +142,12 @@ public static function paginatePublicCatalog(
         int $offset,
         bool $orderByRecent = true,
     ): array {
-        $exists = self::hasPublicSpectraExistsSql('molecules.id');
+        $public = self::hasPublicSpectraColumnSql();
 
         return self::paginateIds(
             [
                 'from' => 'FROM molecules',
-                'where' => "WHERE molecules.identifier IS NOT NULL AND {$exists}",
+                'where' => "WHERE molecules.identifier IS NOT NULL AND {$public}",
                 'id' => 'molecules.id',
                 'order' => $orderByRecent ? 'ORDER BY molecules.created_at DESC' : '',
             ],
@@ -192,11 +201,11 @@ public static function forgetPublicCatalogTotalCache(): void
      */
     public static function publicCatalogTotal(): int
     {
-        $exists = self::hasPublicSpectraExistsSql('molecules.id');
+        $public = self::hasPublicSpectraColumnSql();
 
         return self::countIds(
             'FROM molecules',
-            "WHERE molecules.identifier IS NOT NULL AND {$exists}",
+            "WHERE molecules.identifier IS NOT NULL AND {$public}",
             'molecules.id',
             [],
             self::PUBLIC_CATALOG_TOTAL_CACHE_KEY,
@@ -211,7 +220,7 @@ public static function scopePublicCatalog(Builder $query): Builder
     {
         return $query
             ->whereNotNull('identifier')
-            ->whereRaw(self::hasPublicSpectraExistsSql($query->getModel()->getTable().'.id'));
+            ->where('has_public_spectra', true);
     }
 
     /**
@@ -224,19 +233,9 @@ public static function enrich(array $molecules): array
             return $molecules;
         }
 
-        $ids = array_values(array_unique(array_map(
-            fn ($molecule) => $molecule instanceof Molecule ? (int) $molecule->id : (int) $molecule->id,
-            $molecules,
-        )));
-
-        $sampleCounts = self::sampleCountsByMoleculeId($ids);
-        $experimentCounts = (new MoleculeExperimentTypeCounts)->forPublicCatalog($ids);
-
         foreach ($molecules as $molecule) {
-            $id = $molecule instanceof Molecule ? (int) $molecule->id : (int) $molecule->id;
-
-            $samples = $sampleCounts[$id] ?? 0;
-            $experiments = $experimentCounts[$id] ?? [];
+            $samples = self::publicSamplesCountFromRow($molecule);
+            $experiments = self::publicExperimentTypeCountsFromRow($molecule);
 
             if ($molecule instanceof Molecule) {
                 $molecule->setAttribute('workspace_samples_count', $samples);
@@ -250,6 +249,20 @@ public static function enrich(array $molecules): array
         return $molecules;
     }
 
+    /**
+     * Sample and experiment-type badge payload for the catalog indexer.
+     *
+     * @param  list<int>  $moleculeIds
+     * @return array{sample_counts: array<int, int>, experiment_counts: array<int, array<string, int>>}
+     */
+    public static function catalogCardPayload(array $moleculeIds): array
+    {
+        return [
+            'sample_counts' => self::sampleCountsByMoleculeId($moleculeIds),
+            'experiment_counts' => (new MoleculeExperimentTypeCounts)->forPublicCatalog($moleculeIds),
+        ];
+    }
+
     /**
      * @param  list<mixed>  $bindings
      */
@@ -277,6 +290,29 @@ private static function countIds(
         return (int) Cache::remember($totalCacheKey, $totalCacheSeconds, $resolve);
     }
 
+    private static function publicSamplesCountFromRow(object|Molecule $molecule): int
+    {
+        return (int) ($molecule->public_samples_count ?? 0);
+    }
+
+    /**
+     * @return array<string, int>
+     */
+    private static function publicExperimentTypeCountsFromRow(object|Molecule $molecule): array
+    {
+        $value = $molecule->public_experiment_type_counts ?? [];
+
+        if (is_string($value)) {
+            $value = json_decode($value, true);
+        }
+
+        if (is_object($value)) {
+            $value = json_decode(json_encode($value), true);
+        }
+
+        return is_array($value) ? $value : [];
+    }
+
     /**
      * @param  array<int, int>  $moleculeIds
      * @return array<int, int>