Fix discarding `html[lang]`

`DOMDocument::loadHTML` will parse HTML documents as ISO-8859-1 if there is no `meta[charset]` tag. This means that UTF-8-encoded HTML fragments such as those coming from JSON-LD `articleBody` field would be parsed with incorrect encoding. In f14428e4c0, we tried to resolve it by putting `meta[charset]` tag at the start of the HTML fragment. Unfortunately, it turns out that causes parser to auto-insert a `html` element, losing the attributes of the original `html` tag. Let’s try to insert the `meta[charset]` tag into the proper place in the HTML document. We do not need to use the same trick with `JSLikeHTMLElement::__set`. That expects smaller HTML fragments, not `html` documents, so creating `html` and `head` elements will not be a problem. (cherry picked from commit efbbc86df9) Had to strip type hints since we still target PHP 5.6.
1 year ago · f1c6297e3c
parent 5afefcff34
commit f1c6297e3c
2 changed files with 84 additions and 1 deletions
--- a/src/Readability.php
+++ b/src/Readability.php
@ -1435,7 +1435,7 @@ class Readability implements LoggerAwareInterface
            unset($tidy);
        }

-        $this->html = '<meta charset="utf-8">' . (string) $this->html;
+        $this->html = self::ensureMetaCharset((string) $this->html);

        if ('html5lib' === $this->parser || 'html5' === $this->parser) {
            $this->dom = (new HTML5())->loadHTML($this->html);
@ -1453,4 +1453,45 @@ class Readability implements LoggerAwareInterface

        $this->dom->registerNodeClass('DOMElement', 'Readability\JSLikeHTMLElement');
    }
+
+    /**
+     * Tries to insert `meta[charset]` tag into the proper place in the passed HTML document.
+     *
+     * `DOMDocument::loadHTML` will parse HTML documents as ISO-8859-1 if there is no `meta[charset]` tag.
+     * This means that UTF-8-encoded HTML fragments such as those coming from JSON-LD `articleBody` field would be parsed with incorrect encoding.
+     * Unfortunately, we cannot just put the tag at the start of the HTML fragment, since that would cause parser to auto-insert a `html` element, losing the attributes of the original `html` tag.
+     *
+     * @param string $html UTF-8 encoded document
+     */
+    private static function ensureMetaCharset($html)
+    {
+        $charsetTag = '<meta charset="utf-8">';
+
+        // Only look at first 1024 bytes since, according to HTML5 specification,
+        // that’s where <meta> elements declaring a character encoding must be located.
+        // https://developer.mozilla.org/en-US/docs/Web/HTML/Element/meta#charset
+        $start = substr($html, 0, 1000);
+
+        if (1 === preg_match('/<meta[^>]+charset/i', $start)) {
+            // <meta> tag is already present, no need for modification.
+            return $html;
+        }
+
+        if (1 === preg_match('/<head[^>]*>/i', $start)) {
+            // <head> tag was located, <meta> tags go there.
+            $html = preg_replace('/<head[^>]*>/i', '$0' . $charsetTag, $html, 1);
+
+            return $html;
+        }
+
+        if (1 === preg_match('/<html[^>]*>/i', $start)) {
+            // <html> tag was located, let’s put it inside and have parser create <head>.
+            $html = preg_replace('/<html[^>]*>/i', '$0' . $charsetTag, $html, 1);
+
+            return $html;
+        }
+
+        // Fallback – just plop the <meta> at the start of the fragment.
+        return $charsetTag . $html;
+    }
 }
--- a/tests/ReadabilityTest.php
+++ b/tests/ReadabilityTest.php
@ -486,6 +486,48 @@ class ReadabilityTest extends \PHPUnit\Framework\TestCase
        $this->assertStringContainsString('<a href="alice-I.html">Down the Rabbit-Hole</a>', $readability->getContent()->getInnerHtml());
    }

+    /**
+     * @return array<string, array{0: string, 1: string, 2?: bool}>
+     */
+    public function dataForHtmlLang()
+    {
+        return [
+            'meta' => [
+                '<html lang="fr"><head><meta charset="utf-8"></head><body><article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article></body></html>',
+                'fr',
+            ],
+            'head' => [
+                '<html lang="fr"><head><title>Foo</title></head><body><article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article></body></html>',
+                'fr',
+            ],
+            'headless' => [
+                '<html lang="fr"><body><article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article></body></html>',
+                'fr',
+                // tidy would add <head> tag.
+                false,
+            ],
+            'fragment' => [
+                '<article>' . str_repeat('<p>This is the awesome content :)</p>', 7) . '</article>',
+                '',
+                // tidy would add <html>.
+                false,
+            ],
+        ];
+    }
+
+    /**
+     * @dataProvider dataForHtmlLang
+     */
+    public function testHtmlLang($html, $lang, $useTidy = true)
+    {
+        $readability = $this->getReadability($html, 'http://0.0.0.0', 'libxml', $useTidy);
+        $res = $readability->init();
+
+        $this->assertTrue($res);
+        $this->assertInstanceOf(\DOMDocument::class, $readability->dom);
+        $this->assertSame($lang, $readability->dom->documentElement->getAttribute('lang'));
+    }
+
    private function getReadability($html, $url = null, $parser = 'libxml', $useTidy = true)
    {
        $readability = new Readability($html, $url, $parser, $useTidy);