Optimize metadata parsing

src/Annotation.php

@@ -68,15 +68,16 @@ public static function validate(array $data): void
 
     public function getLabel(array $categories): Label
     {
-        $categoryIndex = array_search($this->category_id, array_column($categories, 'id'));
-        $category = $categories[$categoryIndex];
+        $category = $categories[$this->category_id];
+
         return new Label(id: $category->id, name: $category->name);
     }
 
     public function getLabelAndUsers(array $categories): array
     {
         $cocoUser = Coco::getCocoUser();
         $label = $this->getLabel($categories);
+
         return [new LabelAndUser(label: $label, user: $cocoUser)];
     }
 
@@ -85,6 +86,11 @@ public function getPoints(): array
         if ($this->getShape()->id === Shape::circleId()) {
             return $this->getCirclePoints();
         }
+
+        if ($this->getShape()->id === Shape::rectangleId()) {
+            return array_slice($this->segmentation, 0, 8);
+        }
+
         return $this->segmentation;
     }
 
@@ -204,27 +210,22 @@ function euclidean_distance($x1, $y1, $x2, $y2)
 
     public function isRectangleShape(): bool
     {
-        if (count($this->segmentation) !== 8) {
+        // Some rectangles have an additional last point equalling the first point,
+        // similar to polygons.
+        $segmentationCount = count($this->segmentation);
+        if ($segmentationCount !== 8 && $segmentationCount !== 10) {
             return false;
         }
 
-        // Toleranz für Gleitkomma-Vergleiche
+        // Tolerance for float comparison.
         $tolerance = 0.01;
 
-        // Punkte (x1, y1), (x2, y2), (x3, y3), (x4, y4)
         list($x1, $y1, $x2, $y2, $x3, $y3, $x4, $y4) = $this->segmentation;
 
-        // Berechne die Seitenlängen
-        $d1 = $this->euclidean_distance($x1, $y1, $x2, $y2); // Distanz zwischen P1 und P2
-        $d2 = $this->euclidean_distance($x2, $y2, $x3, $y3); // Distanz zwischen P2 und P3
-        $d3 = $this->euclidean_distance($x3, $y3, $x4, $y4); // Distanz zwischen P3 und P4
-        $d4 = $this->euclidean_distance($x4, $y4, $x1, $y1); // Distanz zwischen P4 und P1
-
-        // Berechne die Diagonalen
-        $diag1 = $this->euclidean_distance($x1, $y1, $x3, $y3); // Diagonale P1 -> P3
-        $diag2 = $this->euclidean_distance($x2, $y2, $x4, $y4); // Diagonale P2 -> P4
+        $diag1 = $this->euclidean_distance($x1, $y1, $x3, $y3);
+        $diag2 = $this->euclidean_distance($x2, $y2, $x4, $y4);
 
-        // Prüfen, ob gegenüberliegende Seiten gleich lang sind und Diagonalen gleich lang sind
-        return abs($d1 - $d3) < $tolerance && abs($d2 - $d4) < $tolerance && abs($diag1 - $diag2) < $tolerance;
+        // If diagonals are equal we have a rectangle.
+        return abs($diag1 - $diag2) < $tolerance;
     }
 }

src/Coco.php

@@ -51,10 +51,12 @@ public static function create(array $data): self
         }, $data['annotations']);
 
         // Create the Category objects
-        $instance->categories = array_map(function ($categoryData) {
+        $categories = array_map(function ($categoryData) {
             return Category::create($categoryData);
         }, $data['categories']);
 
+        $instance->categories = array_combine(array_map(fn ($c) => $c->id, $categories), $categories);
+
         // validate the data consistency
         $instance->validateCategoriesInData();
         $instance->validateImagesInData();

src/CocoParser.php

@@ -3,12 +3,13 @@
 namespace Biigle\Modules\MetadataCoco;
 
 use Biigle\MediaType;
-use Biigle\Services\MetadataParsing\MetadataParser;
-use Biigle\Services\MetadataParsing\VolumeMetadata;
-use Biigle\Services\MetadataParsing\ImageAnnotation;
-use Biigle\Services\MetadataParsing\ImageMetadata;
 use Biigle\Modules\MetadataCoco\Coco;
 use Biigle\Modules\MetadataCoco\Image;
+use Biigle\Services\MetadataParsing\ImageAnnotation;
+use Biigle\Services\MetadataParsing\ImageMetadata;
+use Biigle\Services\MetadataParsing\MetadataParser;
+use Biigle\Services\MetadataParsing\VolumeMetadata;
+use Illuminate\Support\Collection;
 
 class CocoParser extends MetadataParser
 {
@@ -74,32 +75,32 @@ public function getMetadata(): VolumeMetadata
             handle: null,
         );
 
+        $annotations = collect($coco->annotations)->groupBy('image_id');
+        $categories = $coco->categories;
+
         foreach ($coco->images as $image) {
             $imageMetaData = new ImageMetadata(
                 name: $image->file_name
             );
 
-            $this->processImageAnnotations($image, $imageMetaData);
+            if ($annotations->has($image->id)) {
+                $imageMetaData->annotations = $this->processImageAnnotations($annotations->get($image->id), $categories);
+            }
 
             $metadata->addFile($imageMetaData);
         }
 
         return $metadata;
     }
 
-    private function processImageAnnotations(Image $image, ImageMetadata $imageMetaData)
+    private function processImageAnnotations(Collection $annotations, array $categories)
     {
-        $annotations = array_filter($this->getCoco()->annotations, function ($annotation) use ($image) {
-            return $annotation->image_id === $image->id;
-        });
-
-        foreach ($annotations as $annotation) {
-            $metaDataAnnotation = new ImageAnnotation(
+        return $annotations->map(function ($annotation) use ($categories) {
+            return new ImageAnnotation(
                 shape: $annotation->getShape(),
                 points: $annotation->getPoints(),
-                labels: $annotation->getLabelAndUsers($this->getCoco()->categories),
+                labels: $annotation->getLabelAndUsers($categories),
             );
-            $imageMetaData->addAnnotation($metaDataAnnotation);
-        }
+        })->toArray();
     }
 }

tests/CocoParserTest.php

@@ -163,6 +163,37 @@ public function testIsRectangleShape()
         $this->assertSame($rectangleAnnotation->getPoints(), $rectangleAnnotation->segmentation);
     }
 
+    public function testIsRectangleShapeWithMorePoints()
+    {
+        $rectangleAnnotation = Annotation::create([
+            'id' => 1,
+            'image_id' => 1,
+            'category_id' => 1,
+            'bbox' => null,
+            'segmentation' => [
+                [
+                    1853.22,
+                    596.22,
+
+                    1776.16,
+                    799.04,
+
+                    1597.21,
+                    731.04,
+
+                    1674.27,
+                    528.23,
+
+                    1853.22,
+                    596.22,
+                ]
+            ],
+        ]);
+        $this->assertTrue($rectangleAnnotation->isRectangleShape());
+        $this->assertSame($rectangleAnnotation->getShape(), Shape::rectangle());
+        $this->assertSame($rectangleAnnotation->getPoints(), array_slice($rectangleAnnotation->segmentation, 0, 8));
+    }
+
     public function testIsCircleShape()
     {
         $circleAnnotation = Annotation::create([
@@ -328,7 +359,7 @@ public function testIsPolygonShape()
             'image_id' => 1,
             'category_id' => 1,
             'bbox' => null,
-            'segmentation' => [[1, 1, 2, 2, 3, 3, 4, 4, 1, 1]]
+            'segmentation' => [[1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 1, 1]]
         ]);
         $this->assertSame($polygonAnnotation->getShape(), Shape::polygon());
     }