Increasing compatibility with JsonLD Parser

lib/Helper/Filter/JSON/FixInstructionsFilter.php

@@ -98,6 +98,11 @@ public function apply(array &$json): bool {
 					continue;
 				}
 
+				if ($this->jsonService->isSchemaObject($value, 'HowToTip', false)) {
+					$instructions[$key] = [$this->extractHowToStep($value)];
+					continue;
+				}
+
 				if ($this->jsonService->isSchemaObject($value, 'HowToSection', false)) {
 					$newInstructions = $this->flattenHowToSection($value);
 					$instructions[$key] = $newInstructions;

lib/Helper/HTMLParser/HttpJsonLdParser.php

@@ -26,7 +26,7 @@ public function __construct(IL10N $l10n, JsonService $jsonService) {
 	public function parse(\DOMDocument $document, ?string $url): array {
 		$xpath = new \DOMXPath($document);
 
-		$json_ld_elements = $xpath->query("//*[@type='application/ld+json']");
+		$json_ld_elements = $xpath->query("//*[@type='application/ld+json'] | //script");
 
 		foreach ($json_ld_elements as $json_ld_element) {
 			if (!$json_ld_element || !$json_ld_element->nodeValue) {
@@ -58,7 +58,12 @@ private function parseJsonLdElement(\DOMNode $node): array {
 		$json = json_decode($string, true);
 
 		if ($json === null) {
-			throw new HtmlParsingException($this->l->t('JSON cannot be decoded.'));
+			$extractedJson = $this->extractNextJsJson($string);
+			if ($extractedJson === null) {
+				throw new HtmlParsingException($this->l->t('JSON cannot be decoded.'));
+			}
+
+			$json = json_decode($extractedJson, true);
 		}
 
 		if ($json === false || $json === true || !is_array($json)) {
@@ -85,6 +90,34 @@ private function parseJsonLdElement(\DOMNode $node): array {
 		throw new HtmlParsingException($this->l->t('No recipe was found.'));
 	}
 
+	/**
+	 * Try to extract escaped JSON from a Next.js flight payload.
+	 *
+	 * Example:
+	 * self.__next_f.push([1,"{\"@context\":\"https://schema.org\",...}"])
+	 *
+	 * @param string $rawContent
+	 * @return string|null
+	 */
+	private function extractNextJsJson(string $rawContent): ?string {
+		if (strpos($rawContent, 'self.__next_f.push(') === false) {
+			return null;
+		}
+
+		$matches = [];
+		$matched = preg_match('/self\.__next_f\.push\(\[\s*\d+\s*,\s*"((?:\\\\.|[^"\\\\])*)"/s', $rawContent, $matches);
+		if ($matched !== 1 || !isset($matches[1])) {
+			return null;
+		}
+
+		$decoded = stripcslashes($matches[1]);
+		if ($decoded === '') {
+			return null;
+		}
+
+		return $decoded;
+	}
+
 	/**
 	 * Fix any JSON issues before trying to decode it
 	 *

tests/Unit/Helper/HTMLParser/HttpJsonLdParserTest.php

@@ -28,6 +28,7 @@ public static function dataProvider(): array {
 			'caseJ' => ['caseJ.html', true, 'caseJ.json'],
 			//'caseK' => ['caseK.html', true, 'caseK.json'],
 			'caseL' => ['caseL.html', true, 'caseL.json'],
+			'caseM' => ['caseM.html', true, 'caseM.json'],
 		];
 	}
 
@@ -70,6 +71,7 @@ public function testHTMLFile($file, $valid, $jsonFile): void {
 		$parser = new HttpJsonLdParser($l, $jsonService);
 
 		$content = file_get_contents(__DIR__ . "/res_JsonLd/$file");
+		$content = $this->normalizeLineEndings($content);
 
 		$document = new \DOMDocument();
 		$document->loadHTML($content);
@@ -78,6 +80,7 @@ public function testHTMLFile($file, $valid, $jsonFile): void {
 			$res = $parser->parse($document, 'http://example.com');
 
 			$jsonDest = file_get_contents(__DIR__ . "/res_JsonLd/$jsonFile");
+			$jsonDest = $this->normalizeLineEndings($jsonDest);
 			$expected = json_decode($jsonDest, true);
 
 			$this->assertEquals($expected, $res);
@@ -86,4 +89,8 @@ public function testHTMLFile($file, $valid, $jsonFile): void {
 			$this->assertFalse($valid);
 		}
 	}
+
+	function normalizeLineEndings(string $text): string {
+		return str_replace(["\r\n", "\r"], "\n", $text);
+	}
 }

tests/Unit/Helper/HTMLParser/HttpMicrodataParserTest.php

@@ -63,15 +63,41 @@ public function testHTMLFile($filename, $valid, $jsonFile, $skipped = false): vo
 		$parser = new HttpMicrodataParser($l);
 
 		$content = file_get_contents(__DIR__ . "/res_Microdata/$filename");
+		$content = $this->normalizeLineEndings($content);
 
 		$document = new \DOMDocument();
 		$document->loadHTML($content);
 
 		try {
 			$res = $parser->parse($document, 'http://example.com');
+			if (isset($res['recipeIngredient'])) {
+				$res['recipeIngredient'] = array_map(
+					fn($line) => str_replace(["\r\n", "\r"], "\n", $line),
+					$res['recipeIngredient']
+				);
+			}
+			if (isset($res['recipeInstructions'])) {
+				$res['recipeInstructions'] = array_map(
+					fn($line) => str_replace(["\r\n", "\r"], "\n", $line),
+					$res['recipeInstructions']
+				);
+			}
 
 			$jsonDest = file_get_contents(__DIR__ . "/res_Microdata/$jsonFile");
+			$jsonDest = $this->normalizeLineEndings($jsonDest);
 			$expected = json_decode($jsonDest, true);
+			if (isset($expected['recipeIngredient'])) {
+				$expected['recipeIngredient'] = array_map(
+					fn($line) => str_replace(["\r\n", "\r"], "\n", $line),
+					$expected['recipeIngredient']
+				);
+			}
+			if (isset($expected['recipeInstructions'])) {
+				$expected['recipeInstructions'] = array_map(
+					fn($line) => str_replace(["\r\n", "\r"], "\n", $line),
+					$expected['recipeInstructions']
+				);
+			}
 
 			// $this->markTestSkipped();
 
@@ -168,4 +194,8 @@ private function finishTest($parser, $content, $jsonFile): void {
 			$this->assertFalse(true);
 		}
 	}
+
+	function normalizeLineEndings(string $text): string {
+		return str_replace(["\r\n", "\r"], "\n", $text);
+	}
 }