Merge pull request #1282 from nextcloud/fix/1279-json-with-array

Make parser more generic to handle more websites

CHANGELOG.md

@@ -26,6 +26,8 @@
   [1122](https://github.com/nextcloud/cookbook/pull/1122) @MarcelRobitaille
 - Add filter to prevent special chars in folder names
   [#1268](https://github.com/nextcloud/cookbook/pull/1268) @christianlupus
+- Fix bug with websites that provide array of schema entries
+  [#1282](https://github.com/nextcloud/cookbook/pull/1282) @christianlupus
 
 ### Maintenance
 - Use the pre-built database images for MySQL and PostgreSQL tests

lib/Helper/HTMLParser/HttpJsonLdParser.php

@@ -70,15 +70,18 @@ private function parseJsonLdElement(\DOMNode $node): array {
 		// Look for an array of recipes
 		$this->mapArray($json);
 
-		// Ensure the type of the object is never an array
-		$this->checkForArrayType($json);
+		if ($this->jsonService->isSchemaObject($json, 'Recipe', true, false)) {
+			// Ensure the type of the object is never an array
+			$this->checkForArrayType($json);
 
-		if ($this->jsonService->isSchemaObject($json, 'Recipe')) {
 			// We found our recipe
 			return $json;
 		} else {
-			throw new HtmlParsingException($this->l->t('No recipe was found.'));
+			// Continue with other approaches
 		}
+
+		//
+		throw new HtmlParsingException($this->l->t('No recipe was found.'));
 	}
 
 	/**
@@ -156,7 +159,7 @@ private function searchForRecipeInArray(array $arr): ?array {
 		// Iterate through all objects in the array ...
 		foreach ($arr as $item) {
 			// ... looking for a recipe
-			if ($this->jsonService->isSchemaObject($item, 'Recipe')) {
+			if ($this->jsonService->isSchemaObject($item, 'Recipe', true, false)) {
 				// We found a recipe in the array, use it
 				return $item;
 			}
@@ -174,12 +177,8 @@ private function searchForRecipeInArray(array $arr): ?array {
 	 * @param array $json The JSON object to parse
 	 */
 	private function checkForArrayType(array &$json) {
-		if (! $this->jsonService->isSchemaObject($json)) {
-			return;
-		}
-
 		if (is_array($json['@type'])) {
-			$json['@type'] = $json['@type'][0];
+			$json['@type'] = 'Recipe';
 		}
 	}
 }

lib/Service/JsonService.php

@@ -17,9 +17,10 @@ class JsonService {
 	 * @param mixed $obj The object to check
 	 * @param string $type The type to check for. If null or '' no type check is performed
 	 * @param bool $checkContext If true, check for a present context entry
+	 * @param bool $uniqueType If false, also accept JSON objects that contain multiple types as @type.
 	 * @return bool true, if $obj is an object and optionally satisfies the type check
 	 */
-	public function isSchemaObject($obj, string $type = null, bool $checkContext = true): bool {
+	public function isSchemaObject($obj, string $type = null, bool $checkContext = true, bool $uniqueType = true): bool {
 		if (! is_array($obj)) {
 			// Objects must bve encoded as arrays in JSON
 			return false;
@@ -44,6 +45,21 @@ public function isSchemaObject($obj, string $type = null, bool $checkContext = t
 			return true;
 		}
 
+		if (is_array($obj['@type'])) {
+			if ($uniqueType) {
+				if (count($obj['@type']) === 1 && $obj['@type'][0] === $type) {
+					return true;
+				}
+				return false;
+			}
+
+			$foundTypes = array_filter($obj['@type'], function ($x) use ($type) {
+				return trim($x) === $type;
+			});
+
+			return count($foundTypes) > 0;
+		}
+
 		// Check if type matches
 		return (strcmp($obj['@type'], $type) === 0);
 	}

tests/Unit/Helper/HTMLParser/HttpJsonLdParserTest.php

@@ -27,6 +27,7 @@ public function dataProvider(): array {
 			'caseG' => ['caseG.html', true, 'caseG.json'],
 			'caseH' => ['caseH.html', true, 'caseH.json'],
 			'caseI' => ['caseI.html', true, 'caseI.json'],
+			'caseJ' => ['caseJ.html', true, 'caseJ.json'],
 		];
 	}