Use context prefix for json-ld schema (see #1457)

Description ----------- As discussed in #1236 (comment) Commits ------- fa39c59 Use context prefix for json-ld schema a43de85 Remove contao:pageId 498fd07 Fix the merge conflict 46c8f34 Merge branch '4.9' into fix/google-schema-validation
contao · Mar 2, 2020 · 19ff24f · 19ff24f
1 parent 0059dd4
commit 19ff24f
Show file tree

Hide file tree

Showing 3 changed files with 102 additions and 16 deletions.
diff --git a/core-bundle/src/Resources/contao/pages/PageRegular.php b/core-bundle/src/Resources/contao/pages/PageRegular.php
@@ -825,12 +825,12 @@ protected function createFooterScripts($objLayout, $objPage = null)
 
 			$meta = array
 			(
-				'@context' => 'https://schema.contao.org/',
-				'@type' => 'RegularPage',
-				'noSearch' => $noSearch,
-				'protected' => (bool) $objPage->protected,
-				'groups' => array_map('intval', array_filter((array) $objPage->groups)),
-				'fePreview' => System::getContainer()->get('contao.security.token_checker')->isPreviewMode()
+				'@context' => array('contao' => 'https://schema.contao.org/'),
+				'@type' => 'contao:RegularPage',
+				'contao:noSearch' => $noSearch,
+				'contao:protected' => (bool) $objPage->protected,
+				'contao:groups' => array_map('intval', array_filter((array) $objPage->groups)),
+				'contao:fePreview' => System::getContainer()->get('contao.security.token_checker')->isPreviewMode()
 			);
 
 			$strScripts .= '<script type="application/ld+json">' . json_encode($meta) . '</script>';

diff --git a/core-bundle/src/Search/Document.php b/core-bundle/src/Search/Document.php
@@ -142,17 +142,87 @@ private function filterJsonLd(array $jsonLds, string $context = '', string $type
         $matching = [];
 
         foreach ($jsonLds as $data) {
-            if ('' !== $context && (!isset($data['@context']) || $data['@context'] !== $context)) {
-                continue;
-            }
+            $data = $this->expandJsonLdContexts($data);
 
-            if ('' !== $type && (!isset($data['@type']) || $data['@type'] !== $type)) {
+            if ('' !== $type && (!isset($data['@type']) || $data['@type'] !== $context.$type)) {
                 continue;
             }
 
-            $matching[] = $data;
+            if (\count($filtered = $this->filterJsonLdContexts($data, [$context]))) {
+                $matching[] = $filtered;
+            }
         }
 
         return $matching;
     }
+
+    private function expandJsonLdContexts(array $data): array
+    {
+        if (empty($data['@context'])) {
+            return $data;
+        }
+
+        if (\is_string($data['@context'])) {
+            foreach ($data as $key => $value) {
+                if ('@type' === $key) {
+                    $data[$key] = $data['@context'].$value;
+                    continue;
+                }
+
+                if ('@' !== $key[0]) {
+                    unset($data[$key]);
+                    $data[$data['@context'].$key] = $value;
+                }
+            }
+
+            return $data;
+        }
+
+        if (\is_array($data['@context'])) {
+            foreach ($data['@context'] as $prefix => $context) {
+                if (isset($data['@type']) && 0 === strncmp($data['@type'], $prefix.':', \strlen($prefix) + 1)) {
+                    $data['@type'] = substr($data['@type'], \strlen($prefix) + 1);
+                }
+
+                foreach ($data as $key => $value) {
+                    if (0 === strncmp($prefix.':', $key, \strlen($prefix) + 1)) {
+                        unset($data[$key]);
+                        $data[$context.substr($key, \strlen($prefix) + 1)] = $value;
+                    }
+                }
+            }
+
+            return $data;
+        }
+
+        throw new \RuntimeException('Unable to expand JSON-LD data');
+    }
+
+    private function filterJsonLdContexts(array $data, array $contexts): array
+    {
+        $newData = [];
+        $found = false;
+
+        foreach ($data as $key => $value) {
+            foreach ($contexts as $context) {
+                if ('@type' === $key) {
+                    $newData[$key] = $value;
+
+                    if (0 === strncmp($value, $context, \strlen($context))) {
+                        $newData[$key] = substr($value, \strlen($context));
+                        $found = true;
+                        break;
+                    }
+                }
+
+                if (0 === strncmp($context, $key, \strlen($context))) {
+                    $newData[substr($key, \strlen($context))] = $value;
+                    $found = true;
+                    break;
+                }
+            }
+        }
+
+        return $found ? $newData : [];
+    }
 }
diff --git a/core-bundle/tests/Search/DocumentTest.php b/core-bundle/tests/Search/DocumentTest.php
@@ -50,7 +50,7 @@ public function testExtractsTheJsdonLdScript(string $body, array $expectedJsonLd
         $this->assertSame('https://example.com', (string) $document->getUri());
         $this->assertSame(200, $document->getStatusCode());
         $this->assertSame(['content-type' => ['text/html']], $document->getHeaders());
-        $this->assertSame($expectedJsonLds, $document->extractJsonLdScripts());
+        $this->assertSame($expectedJsonLds, $document->extractJsonLdScripts('https://contao.org/'));
     }
 
     public function documentProvider(): \Generator
@@ -64,7 +64,26 @@ public function documentProvider(): \Generator
             '<html><body><script type="application/ld+json">{"@context":"https:\/\/contao.org\/","@type":"PageMetaData","foobar":true}</script></body></html>',
             [
                 [
-                    '@context' => 'https://contao.org/',
+                    '@type' => 'PageMetaData',
+                    'foobar' => true,
+                ],
+            ],
+        ];
+
+        yield 'Test with one valid json ld element without context' => [
+            '<html><body><script type="application/ld+json">{"@type":"https:\/\/contao.org\/PageMetaData","https:\/\/contao.org\/foobar":true}</script></body></html>',
+            [
+                [
+                    '@type' => 'PageMetaData',
+                    'foobar' => true,
+                ],
+            ],
+        ];
+
+        yield 'Test with one valid json ld element with context prefix' => [
+            '<html><body><script type="application/ld+json">{"@context":{"contao":"https:\/\/contao.org\/"},"@type":"contao:PageMetaData","contao:foobar":true}</script></body></html>',
+            [
+                [
                     '@type' => 'PageMetaData',
                     'foobar' => true,
                 ],
@@ -75,12 +94,10 @@ public function documentProvider(): \Generator
             '<html><body><script type="application/ld+json">{"@context":"https:\/\/contao.org\/","@type":"PageMetaData","foobar":true}</script><script type="application/ld+json">{"@context":"https:\/\/contao.org\/","@type":"PageMetaData","foobar":false}</script></body></html>',
             [
                 [
-                    '@context' => 'https://contao.org/',
                     '@type' => 'PageMetaData',
                     'foobar' => true,
                 ],
                 [
-                    '@context' => 'https://contao.org/',
                     '@type' => 'PageMetaData',
                     'foobar' => false,
                 ],
@@ -91,7 +108,6 @@ public function documentProvider(): \Generator
             '<html><body><script type="application/ld+json">{"@context":"https:\/\/contao.org\/","@type":"PageMetaData","foobar":true}</script><script type="application/ld+json">{"@context":"https:\/\/contao.org\/", ...</script></body></html>',
             [
                 [
-                    '@context' => 'https://contao.org/',
                     '@type' => 'PageMetaData',
                     'foobar' => true,
                 ],