Improve decoding JSON

The `Json` class from the crwlr/utils is now used to decode JSON strings. It tries to fix keys without quotes, which is allowed in relaxed JSON. Further, JSON-LD <script> blocks containing an invalid JSON string are ignored and don't lead to an error anymore. Also, you can now pass a PSR-3 LoggerInterface to the `SchemaOrg` class, so it'll log decoding errors.
crwlrsoft · May 17, 2023 · eecf44b · eecf44b
1 parent 0d6b80e
commit eecf44b
Show file tree

Hide file tree

Showing 4 changed files with 145 additions and 5 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -6,5 +6,12 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.2.0] - 2022-05-17
+### Added
+* You can now optionally pass a PSR-3 LoggerInterface to the `SchemaOrg` class, so it'll log decoding errors.
+
+### Fixed
+* The `Json` class from the crwlr/utils is now used to decode JSON strings. It tries to fix keys without quotes, which is allowed in relaxed JSON. Further, JSON-LD <script> blocks containing an invalid JSON string are ignored and don't lead to an error anymore.
+
 ## [0.1.0] - 2022-09-22
 Initial version containing `SchemaOrg` class that finds schema.org JSON-LD objects in HTML documents and converts them to instances of the classes from the spatie schema-org package.
diff --git a/composer.json b/composer.json
@@ -38,7 +38,9 @@
     "require": {
         "php": "^8.0",
         "spatie/schema-org": "~3.11.0",
-        "symfony/dom-crawler": "^6.1"
+        "symfony/dom-crawler": "^6.1",
+        "crwlr/utils": "^1.0",
+        "psr/log": "^2.0|^3.0"
     },
     "require-dev": {
         "pestphp/pest": "^1.22",

diff --git a/src/SchemaOrg.php b/src/SchemaOrg.php
@@ -2,7 +2,10 @@
 
 namespace Crwlr\SchemaOrg;
 
+use Crwlr\Utils\Exceptions\InvalidJsonException;
+use Crwlr\Utils\Json;
 use InvalidArgumentException;
+use Psr\Log\LoggerInterface;
 use Spatie\SchemaOrg\BaseType;
 use Symfony\Component\DomCrawler\Crawler;
 
@@ -12,18 +15,20 @@ class SchemaOrg
 
     private static ?self $singletonInstance = null;
 
-    public function __construct()
+    public function __construct(protected ?LoggerInterface $logger = null)
     {
         $this->types = new TypeList();
     }
 
     /**
      * @return BaseType[]
      */
-    public static function fromHtml(string $html): array
+    public static function fromHtml(string $html, ?LoggerInterface $logger = null): array
     {
         if (!self::$singletonInstance) {
-            self::$singletonInstance = new self();
+            self::$singletonInstance = new self($logger);
+        } elseif ($logger) {
+            self::$singletonInstance->logger = $logger;
         }
 
         return self::$singletonInstance->getFromHtml($html);
@@ -54,7 +59,19 @@ public function getFromHtml(string $html): array
 
     private function getSchemaOrgObjectFromScriptBlock(Crawler $domCrawler): ?BaseType
     {
-        return $this->convertJsonDataToSchemaOrgObject(json_decode($domCrawler->text(), true));
+        try {
+            $jsonData = Json::stringToArray($domCrawler->text());
+        } catch (InvalidJsonException) {
+            $snippetWithReducedSpaces = preg_replace('/\s+/', ' ', $domCrawler->text()) ?? $domCrawler->text();
+
+            $this->logger?->warning(
+                'Failed to parse content of JSON-LD script block as JSON: ' . substr($snippetWithReducedSpaces, 0, 100)
+            );
+
+            return null;
+        }
+
+        return $this->convertJsonDataToSchemaOrgObject($jsonData);
     }
 
     /**

diff --git a/tests/SchemaOrgTest.php b/tests/SchemaOrgTest.php
@@ -3,6 +3,7 @@
 namespace Tests;
 
 use Crwlr\SchemaOrg\SchemaOrg;
+use Psr\Log\LoggerInterface;
 use Spatie\SchemaOrg\Article;
 use Spatie\SchemaOrg\FAQPage;
 use Spatie\SchemaOrg\JobPosting;
@@ -156,3 +157,116 @@
 
     expect($schemaOrgObjects[0]->getProperty('publisher')->getProperty('name'))->toBe('Some Organization, Inc.');
 });
+
+test('there is no error if a json-ld script block contains an invalid JSON string', function () {
+    $html = <<<HTML
+        <!DOCTYPE html>
+        <html lang="de-AT">
+        <head>
+        <title>Some Article</title>
+        </head>
+        <body>
+        <h1>Some Article</h1>
+        <h2>This is some article about something.</h2>
+        <script type="application/ld+json">
+        {
+            "@context": "https:\/\/schema.org",
+            "@type": "Article",
+            name: Some Article,
+            url: https://de.example.org/articles/some,
+        ]
+        </script>
+        </body>
+        </html>
+        HTML;
+
+    $schemaOrgObjects = SchemaOrg::fromHtml($html);
+
+    expect($schemaOrgObjects)->toBeEmpty();
+});
+
+test('you can pass it a PSR-3 LoggerInterface and it will log an error message for invalid JSON string', function () {
+    $scriptBlockContent = <<<INVALIDJSON
+        {
+            "@context": "https:\/\/schema.org",
+            "@type": "Article",
+            name: Some Article,
+            url: https://de.example.org/articles/some,
+        ]
+        INVALIDJSON;
+
+    $html = <<<HTML
+        <!DOCTYPE html>
+        <html lang="de-AT">
+        <head>
+        <title>Some Article</title>
+        </head>
+        <body>
+        <h1>Some Article</h1>
+        <h2>This is some article about something.</h2>
+        <script type="application/ld+json">{$scriptBlockContent}</script>
+        </body>
+        </html>
+        HTML;
+
+    $logger = new class () implements LoggerInterface {
+        /**
+         * @var array<array<string|string[]>>
+         */
+        public array $messages = [];
+
+        public function emergency(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('emergency', $message, $context);
+        }
+
+        public function alert(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('alert', $message, $context);
+        }
+
+        public function critical(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('critical', $message, $context);
+        }
+
+        public function error(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('error', $message, $context);
+        }
+
+        public function warning(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('warning', $message, $context);
+        }
+
+        public function notice(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('notice', $message, $context);
+        }
+
+        public function info(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('info', $message, $context);
+        }
+
+        public function debug(\Stringable|string $message, array $context = []): void
+        {
+            $this->log('debug', $message, $context);
+        }
+
+        public function log($level, \Stringable|string $message, array $context = []): void
+        {
+            $this->messages[] = ['level' => $level, 'message' => $message, 'context' => $context];
+        }
+    };
+
+    SchemaOrg::fromHtml($html, $logger);
+
+    expect($logger->messages[0])->toBe([
+        'level' => 'warning',
+        'message' => 'Failed to parse content of JSON-LD script block as JSON: { "@context": "https:\/\/schema.org", ' .
+            '"@type": "Article", name: Some Article, url: https://de.exampl',
+        'context' => [],
+    ]);
+});