fix: Support parsing Photoshop-saved images

Frameright · Feb 1, 2024 · 8ac64a7 · 8ac64a7
1 parent a87b531
commit 8ac64a7
Show file tree

Hide file tree

Showing 4 changed files with 129 additions and 27 deletions.
diff --git a/src/Metadata/Xmp/ImageRegion.php b/src/Metadata/Xmp/ImageRegion.php
@@ -109,54 +109,60 @@ public function __construct($xpath = null, $node = null)
             return;
         }
 
-        $this->id = self::getNodeValue($xpath, 'Iptc4xmpExt:rId', $node);
+        $this->id = self::getChildValueOrAttr($xpath, 'Iptc4xmpExt:rId', $node);
+
+        $xpathNames = "Iptc4xmpExt:Name/rdf:Alt/rdf:li";
         $this->names = self::getNodeValues(
             $xpath,
-            'Iptc4xmpExt:Name/rdf:Alt/rdf:li',
+            // Matches only children and grand-children, not deeper:
+            "./$xpathNames|./*/$xpathNames",
             $node
         );
+
         $this->types = self::getEntityOrConceptValues(
             $xpath,
-            'Iptc4xmpExt:rCtype',
+            './/Iptc4xmpExt:rCtype', // matches any descendant with this name
             $node
         );
         $this->roles = self::getEntityOrConceptValues(
             $xpath,
-            'Iptc4xmpExt:rRole',
+            './/Iptc4xmpExt:rRole', // matches any descendant with this name
             $node
         );
 
-        $this->regionDefinitionId = self::getNodeValue(
+        $this->regionDefinitionId = self::getChildValueOrAttr(
             $xpath,
             'FramerightIdc:RegionDefinitionId',
             $node
         );
 
-        $this->regionName = self::getNodeValue(
+        $this->regionName = self::getChildValueOrAttr(
             $xpath,
             'FramerightIdc:RegionName',
             $node
         );
 
-        $xpathToRb = 'Iptc4xmpExt:RegionBoundary';
-
-        foreach ([
-            'rbShape',
-            'rbUnit',
-            'rbH',
-            'rbW',
-            'rbRx',
-        ] as $property) {
-            $this->$property = self::getNodeValue(
-                $xpath,
-                "$xpathToRb/Iptc4xmpExt:$property",
-                $node
-            );
+        $xpathToRb = './/Iptc4xmpExt:RegionBoundary'; // matches any descendant with this name
+        $rbNode = $xpath->query($xpathToRb, $node)->item(0);
+        if ($rbNode) {
+            foreach ([
+                'rbShape',
+                'rbUnit',
+                'rbH',
+                'rbW',
+                'rbRx',
+            ] as $property) {
+                $this->$property = self::getChildValueOrAttr(
+                    $xpath,
+                    "Iptc4xmpExt:$property",
+                    $rbNode
+                );
+            }
         }
 
         $this->rbXY = new Point(
-            self::getNodeValue($xpath, "$xpathToRb/Iptc4xmpExt:rbX", $node),
-            self::getNodeValue($xpath, "$xpathToRb/Iptc4xmpExt:rbY", $node)
+            self::getChildValueOrAttr($xpath, "Iptc4xmpExt:rbX", $rbNode),
+            self::getChildValueOrAttr($xpath, "Iptc4xmpExt:rbY", $rbNode)
         );
 
         $verticesNodes = $xpath->query(
@@ -167,8 +173,8 @@ public function __construct($xpath = null, $node = null)
             $this->rbVertices = [];
             foreach ($verticesNodes as $verticesNode) {
                 $point = new Point(
-                    self::getNodeValue($xpath, "Iptc4xmpExt:rbX", $verticesNode),
-                    self::getNodeValue($xpath, "Iptc4xmpExt:rbY", $verticesNode)
+                    self::getChildValueOrAttr($xpath, "Iptc4xmpExt:rbX", $verticesNode),
+                    self::getChildValueOrAttr($xpath, "Iptc4xmpExt:rbY", $verticesNode)
                 );
                 array_push($this->rbVertices, $point);
             }
@@ -203,8 +209,10 @@ public function matchesRoleFilter($filter) {
     }
 
     /**
+     * Get the value of the first node matching the given XPath expression.
+     *
      * @param \DOMXPath $xpath
-     * @param string $expression XPath expression leading to one single node.
+     * @param string $expression XPath expression.
      * @param \DOMNode $contextNode
      *
      * @return string|null
@@ -214,6 +222,47 @@ private static function getNodeValue($xpath, $expression, $contextNode) {
         return $node ? $node->nodeValue : null;
     }
 
+    /**
+     * Attempts to find a value in the following order:
+     * 1. First direct child node having the given name.
+     * 2. Context node's attribute with this name.
+     * 3. First direct child node's attribute with this name.
+     *
+     * @param \DOMXPath $xpath
+     * @param string $nodeOrAttrName
+     * @param \DOMNode $contextNode
+     *
+     * @return string|null
+     */
+    private static function getChildValueOrAttr($xpath, $nodeOrAttrName, $contextNode) {
+        $childValue = self::getNodeValue($xpath, $nodeOrAttrName, $contextNode);
+        if ($childValue) {
+            return $childValue;
+        }
+
+        $attrNameWithoutNamespace = preg_replace('/^.*:/', '', $nodeOrAttrName);
+        $contextNodeAttrs = $contextNode->attributes;
+        if ($contextNodeAttrs) {
+            $contextNodeAttr = $contextNodeAttrs->getNamedItem($attrNameWithoutNamespace);
+            if ($contextNodeAttr) {
+                return $contextNodeAttr->nodeValue;
+            }
+        }
+
+        $childNodeWithAttr = $xpath->query("*[@$nodeOrAttrName]", $contextNode)->item(0);
+        if (!$childNodeWithAttr) {
+            return null;
+        }
+
+        $childNodeAttrs = $childNodeWithAttr->attributes;
+        if (!$childNodeAttrs) {
+            return null;
+        }
+
+        $childNodeAttr = $childNodeAttrs->getNamedItem($attrNameWithoutNamespace);
+        return $childNodeAttr->nodeValue;
+    }
+
     /**
      * @param \DOMXPath $xpath
      * @param string $expression XPath expression leading to several nodes.

diff --git a/tests/Fixtures/README.md b/tests/Fixtures/README.md
@@ -2,5 +2,7 @@
 
 ## Image sources
 
-* [`IPTC-PhotometadataRef-Std2021.1.jpg`](https://iptc.org/std/photometadata/examples/IPTC-PhotometadataRef-Std2021.1.jpg),
+- [`IPTC-PhotometadataRef-Std2021.1.jpg`](https://iptc.org/std/photometadata/examples/IPTC-PhotometadataRef-Std2021.1.jpg),
   open standard
+- [`skaterphotoshop.jpg`](skaterphotoshop.jpg), authored by
+  [Marina Ekroos](http://marinaekroos.com/)
diff --git a/tests/Fixtures/skaterphotoshop.jpg b/tests/Fixtures/skaterphotoshop.jpg
diff --git a/tests/Metadata/XmpTest.php b/tests/Metadata/XmpTest.php
@@ -378,7 +378,6 @@ public function testGetImageRegionFromFramerightImage()
         $jpeg = JPEG::fromFile(
             __DIR__ . '/../Fixtures/frameright.jpg');
 
-
         $xmp = $jpeg->getXmp();
 
         $expectedRegion = new ImageRegion();
@@ -400,7 +399,59 @@ public function testGetImageRegionFromFramerightImage()
         $this->assertEquals([
             $expectedRegion,
         ], $xmp->getImageRegions());
+    }
 
+    /**
+     * @covers ::getImageRegions
+     *
+     * When opening an image with regions in Photoshop and saving it again, the
+     * RDF/XML structure might change e.g. from
+     *
+     * <Iptc4xmpExt:RegionBoundary rdf:parseType="Resource">
+     *   <Iptc4xmpExt:rbShape>rectangle</Iptc4xmpExt:rbShape>
+     *   <Iptc4xmpExt:rbUnit>relative</Iptc4xmpExt:rbUnit>
+     *   <Iptc4xmpExt:rbW>0.6663040522164808</Iptc4xmpExt:rbW>
+     *   <Iptc4xmpExt:rbH>1</Iptc4xmpExt:rbH>
+     *   <Iptc4xmpExt:rbX>0.25564318738101716</Iptc4xmpExt:rbX>
+     *   <Iptc4xmpExt:rbY>0</Iptc4xmpExt:rbY>
+     * </Iptc4xmpExt:RegionBoundary>
+     *
+     * to
+     *
+     * <Iptc4xmpExt:RegionBoundary Iptc4xmpExt:rbShape="rectangle" Iptc4xmpExt:rbUnit="relative"
+     *     Iptc4xmpExt:rbW="0.6663040522164808" Iptc4xmpExt:rbH="1"
+     *     Iptc4xmpExt:rbX="0.25564318738101716"
+     *     Iptc4xmpExt:rbY="0" />
+     *
+     * This test validates that we can parse what's generated by Photoshop.
+     */
+    public function testGetImageRegionFromImageResavedWithPhotoshop()
+    {
+        $jpeg = JPEG::fromFile(
+            __DIR__ . '/../Fixtures/skaterphotoshop.jpg');
+
+        $xmp = $jpeg->getXmp();
+
+        $expectedFirstRegion = new ImageRegion();
+        $expectedFirstRegion->regionDefinitionId = '6a3d4ec1-7c0c-58db-9645-6185a80efbcd';
+        $expectedFirstRegion->regionName = '1:1 Square';
+        $expectedFirstRegion->id = 'crop-cb6bbb93-539d-401b-bb57-341f461ab2ab';
+        $expectedFirstRegion->names = null;
+        $expectedFirstRegion->types = null;
+        $expectedFirstRegion->roles = [
+            'http://cv.iptc.org/newscodes/imageregionrole/cropping',
+        ];
+        $expectedFirstRegion->rbShape = 'rectangle';
+        $expectedFirstRegion->rbUnit = 'relative';
+        $expectedFirstRegion->rbXY = new Point(0.25564318738101716, 0);
+        $expectedFirstRegion->rbRx = null;
+        $expectedFirstRegion->rbH = '1';
+        $expectedFirstRegion->rbW = '0.6663040522164808';
+
+        $this->assertEquals(
+            $expectedFirstRegion,
+            $xmp->getImageRegions()[0]
+        );
     }
 
     /**