Merge pull request #150 from j0k3r/dependabot/composer/rector/rector-2.6.7

Merge pull request #149 from j0k3r/dependabot/composer/friendsofphp/php-cs-fixer-3.95.26
build(deps-dev): update rector/rector requirement from 2.6.6 to 2.6.7
2026-09-27 14:36:23 +00:00 · 2026-09-25 07:40:56 +02:00 · 2026-09-25 07:40:47 +02:00 · 2026-09-25 05:34:04 +00:00 · 2026-09-25 05:33:59 +00:00 · 2026-09-18 09:12:38 +02:00
3 changed files with 146 additions and 77 deletions
@@ -30,12 +30,12 @@
        "masterminds/html5": "^2.7"
    },
    "require-dev": {
-        "friendsofphp/php-cs-fixer": "3.95.8",
+        "friendsofphp/php-cs-fixer": "3.95.26",
        "monolog/monolog": "^1.24|^2.1",
        "symfony/phpunit-bridge": "^4.4|^5.3|^6.0|^7.0",
-        "phpstan/phpstan": "2.2.2",
-        "phpstan/phpstan-phpunit": "2.0.16",
-        "rector/rector": "2.4.6"
+        "phpstan/phpstan": "2.2.14",
+        "phpstan/phpstan-phpunit": "2.0.18",
+        "rector/rector": "2.6.7"
    },
    "suggest": {
        "ext-tidy": "Used to clean up given HTML and to avoid problems with bad HTML structure."
@@ -213,7 +213,9 @@ class Readability implements LoggerAwareInterface
     */
    public function init(): bool
    {
-        $this->loadHtml();
+        if (!isset($this->dom)) {
+            $this->loadHtml();
+        }

        if (!isset($this->dom->documentElement)) {
            return false;
@@ -751,6 +753,76 @@ class Readability implements LoggerAwareInterface
        $this->flags &= ~$flag;
    }

+    /**
+     * Load HTML in a DOMDocument.
+     * Apply Pre filters
+     * Cleanup HTML using Tidy (or not).
+     */
+    public function loadHtml(): void
+    {
+        $this->original_html = $this->html;
+
+        $this->logger->debug('Parsing URL: ' . $this->url);
+
+        if ($this->url) {
+            $host = parse_url($this->url, \PHP_URL_HOST);
+            if (null !== $host) {
+                $this->domainRegExp = '/' . strtr((string) preg_replace('/www\d*\./', '', $host), ['.' => '\.']) . '/';
+            }
+        }
+
+        mb_internal_encoding('UTF-8');
+        mb_http_output('UTF-8');
+        mb_regex_encoding('UTF-8');
+
+        // HACK: dirty cleanup to replace some stuff; shouldn't use regexps with HTML but well...
+        if (!$this->flagIsActive(self::FLAG_DISABLE_PREFILTER)) {
+            foreach ($this->pre_filters as $search => $replace) {
+                $this->html = preg_replace($search, $replace, $this->html);
+            }
+            unset($search, $replace);
+        }
+
+        if ('' === trim($this->html)) {
+            $this->html = '<html></html>';
+        }
+
+        /*
+         * Use tidy (if it exists).
+         * This fixes problems with some sites which would otherwise trouble DOMDocument's HTML parsing.
+         * Although sometimes it makes matters worse, which is why there is an option to disable it.
+         */
+        if ($this->useTidy) {
+            $this->logger->debug('Tidying document');
+
+            $tidy = tidy_repair_string($this->html, $this->tidy_config, 'UTF8');
+            if (false !== $tidy && $this->html !== $tidy) {
+                $this->tidied = true;
+                $this->html = $tidy;
+                $this->html = preg_replace('/[\r\n]+/is', "\n", $this->html);
+            }
+            unset($tidy);
+        }
+
+        $this->html = self::entitizeNonAscii((string) $this->html);
+
+        if ('html5lib' === $this->parser || 'html5' === $this->parser) {
+            $this->dom = (new HTML5())->loadHTML($this->html);
+        }
+
+        if ('libxml' === $this->parser) {
+            libxml_use_internal_errors(true);
+
+            $this->dom = new \DOMDocument();
+            $this->dom->preserveWhiteSpace = false;
+            $this->dom->loadHTML($this->html, \LIBXML_NOBLANKS | \LIBXML_COMPACT | \LIBXML_NOERROR);
+
+            libxml_use_internal_errors(false);
+        }
+
+        $this->dom->registerNodeClass(\DOMElement::class, JSLikeHTMLElement::class);
+    }
+
    /**
     * Get the article title as an H1.
     *
@@ -1003,7 +1075,11 @@ class Readability implements LoggerAwareInterface
                    }

                    if ($this->hasSingleTagInsideElement($node, 'p') && $this->getLinkDensity($node) < 0.25) {
-                        $newNode = $node->childNodes->item(0);
+                        // In some cases when tidy is disabled the first item may not be a DOMElement so we apply a filter
+                        $newNode = array_values(array_filter(
+                            iterator_to_array($node->childNodes),
+                            static fn ($childNode) => $childNode instanceof \DOMElement
+                        ))[0];
                        $node->parentNode->replaceChild($newNode, $node);
                        $nodesToScore[] = $newNode;
                    }
@@ -1371,76 +1447,6 @@ class Readability implements LoggerAwareInterface
        }
    }

-    /**
-     * Load HTML in a DOMDocument.
-     * Apply Pre filters
-     * Cleanup HTML using Tidy (or not).
-     */
-    private function loadHtml(): void
-    {
-        $this->original_html = $this->html;
-
-        $this->logger->debug('Parsing URL: ' . $this->url);
-
-        if ($this->url) {
-            $host = parse_url($this->url, \PHP_URL_HOST);
-            if (null !== $host) {
-                $this->domainRegExp = '/' . strtr((string) preg_replace('/www\d*\./', '', $host), ['.' => '\.']) . '/';
-            }
-        }
-
-        mb_internal_encoding('UTF-8');
-        mb_http_output('UTF-8');
-        mb_regex_encoding('UTF-8');
-
-        // HACK: dirty cleanup to replace some stuff; shouldn't use regexps with HTML but well...
-        if (!$this->flagIsActive(self::FLAG_DISABLE_PREFILTER)) {
-            foreach ($this->pre_filters as $search => $replace) {
-                $this->html = preg_replace($search, $replace, $this->html);
-            }
-            unset($search, $replace);
-        }
-
-        if ('' === trim($this->html)) {
-            $this->html = '<html></html>';
-        }
-
-        /*
-         * Use tidy (if it exists).
-         * This fixes problems with some sites which would otherwise trouble DOMDocument's HTML parsing.
-         * Although sometimes it makes matters worse, which is why there is an option to disable it.
-         */
-        if ($this->useTidy) {
-            $this->logger->debug('Tidying document');
-
-            $tidy = tidy_repair_string($this->html, $this->tidy_config, 'UTF8');
-            if (false !== $tidy && $this->html !== $tidy) {
-                $this->tidied = true;
-                $this->html = $tidy;
-                $this->html = preg_replace('/[\r\n]+/is', "\n", $this->html);
-            }
-            unset($tidy);
-        }
-
-        $this->html = self::entitizeNonAscii((string) $this->html);
-
-        if ('html5lib' === $this->parser || 'html5' === $this->parser) {
-            $this->dom = (new HTML5())->loadHTML($this->html);
-        }
-
-        if ('libxml' === $this->parser) {
-            libxml_use_internal_errors(true);
-
-            $this->dom = new \DOMDocument();
-            $this->dom->preserveWhiteSpace = false;
-            $this->dom->loadHTML($this->html, \LIBXML_NOBLANKS | \LIBXML_COMPACT | \LIBXML_NOERROR);
-
-            libxml_use_internal_errors(false);
-        }
-
-        $this->dom->registerNodeClass(\DOMElement::class, JSLikeHTMLElement::class);
-    }
-
    private function getAncestors(\DOMElement $node, int $maxDepth = 0): array
    {
        $ancestors = [];
@@ -1483,8 +1489,15 @@ class Readability implements LoggerAwareInterface
        $childNodes = iterator_to_array($node->childNodes);
        $children = array_filter($childNodes, static fn ($childNode) => $childNode instanceof \DOMElement);

+        // array_first() has been added in PHP 8.5, failing back for older versions
+        if (!\function_exists('array_first')) {
+            $firstChild = $children ? $children[array_key_first($children)] : null;
+        } else {
+            $firstChild = array_first($children);
+        }
+
        // There should be exactly 1 element child with given tag
-        if (1 !== \count($children) || $children[0]->nodeName !== $tag) {
+        if (1 !== \count($children) || $firstChild->nodeName !== $tag) {
            return false;
        }

@@ -464,6 +464,62 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
        $this->assertTrue($res);
    }

+    public function dataForHasSingleTagInsideElement(): array
+    {
+        return [
+            'single matching tag, no other content' => [
+                '<div><p>Some text</p></div>', 'p', true,
+            ],
+            'single matching tag with comment' => [
+                '<div><!-- comment --><p>Some text</p></div>', 'p', true,
+            ],
+            'whitespace-only text around single matching tag' => [
+                '<div>   <p>Some text</p>   </div>', 'p', true,
+            ],
+            'two matching tags' => [
+                '<div><p>One</p><p>Two</p></div>', 'p', false,
+            ],
+            'non-whitespace text alongside the single tag' => [
+                '<div>Some text<p>One</p></div>', 'p', false,
+            ],
+            'single tag with a different name' => [
+                '<div><span>One</span></div>', 'p', false,
+            ],
+        ];
+    }
+
+    /**
+     * @dataProvider dataForHasSingleTagInsideElement
+     */
+    public function testHasSingleTagInsideElement(string $html, string $tag, bool $expected): void
+    {
+        $dom = new \DOMDocument();
+        $dom->loadHTML('<html><body>' . $html . '</body></html>');
+        $node = $dom->getElementsByTagName('div')->item(0);
+
+        $readability = $this->getReadability('', '', 'libxml', false);
+
+        $method = new \ReflectionMethod($readability, 'hasSingleTagInsideElement');
+        if (\PHP_VERSION_ID < 80100) {
+            $method->setAccessible(true);
+        }
+
+        $this->assertSame($expected, $method->invoke($readability, $node, $tag));
+    }
+
+    public function testDivToPElementWithComment(): void
+    {
+        $text = str_repeat('Padded real article text to reach decent length for content scoring threshold here. ', 8);
+        $html = '<div><!-- comment --><code>some code snippet</code> ' . $text . '</div>';
+
+        $readability = $this->getReadability($html, 'http://0.0.0.0', 'libxml', false);
+        $res = $readability->init();
+
+        $this->assertTrue($res);
+        $this->assertStringContainsString('some code snippet', $readability->getContent()->getInnerHtml());
+        $this->assertStringContainsString('Padded real article text', $readability->getContent()->getInnerHtml());
+    }
+
    public function testKeepFootnotes(): void
    {
        // from https://www.schreibdichte.de/blog/feed-aggregator-und-spaeter-lesen-dienst-im-team