mb_convert_encodingを用いたHTMLエンティティ化が非推奨になったので、別実装で代替

shikorism · Nov 23, 2024 · 5c8e008 · 5c8e008
1 parent dfc4c9f
commit 5c8e008
Show file tree

Hide file tree

Showing 7 changed files with 28 additions and 7 deletions.
diff --git a/app/MetadataResolver/ActivityPubResolver.php b/app/MetadataResolver/ActivityPubResolver.php
@@ -2,6 +2,7 @@
 
 namespace App\MetadataResolver;
 
+use App\Facades\Formatter;
 use GuzzleHttp\Exception\TransferException;
 use Illuminate\Support\Facades\Log;
 use Psr\Http\Message\ResponseInterface;
@@ -81,7 +82,7 @@ private function html2text(string $html): string
             return '';
         }
 
-        $html = mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8');
+        $html = Formatter::htmlEntities($html, 'UTF-8');
         $html = preg_replace('~<br\s*/?\s*>|</p>\s*<p[^>]*>~i', "\n", $html);
         $dom = new \DOMDocument();
         $dom->loadHTML($html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);

diff --git a/app/MetadataResolver/DLsiteResolver.php b/app/MetadataResolver/DLsiteResolver.php
@@ -2,6 +2,7 @@
 
 namespace App\MetadataResolver;
 
+use App\Facades\Formatter;
 use GuzzleHttp\Client;
 
 class DLsiteResolver implements Resolver
@@ -29,7 +30,7 @@ public function __construct(Client $client, OGPResolver $ogpResolver)
     public function extractTags(string $html): array
     {
         $dom = new \DOMDocument();
-        @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
+        @$dom->loadHTML(Formatter::htmlEntities($html, 'UTF-8'));
         $xpath = new \DOMXPath($dom);
 
         $genreNode = $xpath->query("//div[@class='main_genre'][1]");
@@ -79,7 +80,7 @@ public function resolve(string $url): Metadata
         $metadata = $this->ogpResolver->parse($res->getBody());
 
         $dom = new \DOMDocument();
-        @$dom->loadHTML(mb_convert_encoding($res->getBody(), 'HTML-ENTITIES', 'UTF-8'));
+        @$dom->loadHTML(Formatter::htmlEntities($res->getBody(), 'UTF-8'));
         $xpath = new \DOMXPath($dom);
 
         // OGPタイトルから[]に囲まれているmakerを取得する

diff --git a/app/MetadataResolver/MelonbooksResolver.php b/app/MetadataResolver/MelonbooksResolver.php
@@ -2,6 +2,7 @@
 
 namespace App\MetadataResolver;
 
+use App\Facades\Formatter;
 use GuzzleHttp\Client;
 use GuzzleHttp\Cookie\CookieJar;
 
@@ -38,7 +39,7 @@ public function resolve(string $url): Metadata
         $metadata = $this->ogpResolver->parse($res->getBody());
 
         $dom = new \DOMDocument();
-        @$dom->loadHTML(mb_convert_encoding($res->getBody(), 'HTML-ENTITIES', 'UTF-8'));
+        @$dom->loadHTML(Formatter::htmlEntities($res->getBody(), 'UTF-8'));
         $xpath = new \DOMXPath($dom);
         $descriptionNodelist = $xpath->query('//div[contains(@class, "item-detail")]/*[contains(@class, "page-headline") and contains(text(), "作品詳細")]/following-sibling::div[1]');
         $specialDescriptionNodelist = $xpath->query('//div[contains(@class, "item-detail")]/*[contains(@class, "page-headline") and contains(text(), "スタッフのオススメポイント")]/following-sibling::div[1]');

diff --git a/app/MetadataResolver/NarouResolver.php b/app/MetadataResolver/NarouResolver.php
@@ -2,6 +2,7 @@
 
 namespace App\MetadataResolver;
 
+use App\Facades\Formatter;
 use GuzzleHttp\Client;
 use GuzzleHttp\Cookie\CookieJar;
 
@@ -35,7 +36,7 @@ public function resolve(string $url): Metadata
 
         // 一見旧式のDOMDocumentを使っているように見えるがこれは罠で、なろうのHTMLはDOMCrawlerだとパースに失敗する
         $dom = new \DOMDocument();
-        @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'ASCII,JIS,UTF-8,eucJP-win,SJIS-win'));
+        @$dom->loadHTML(Formatter::htmlEntities($html, 'ASCII,JIS,UTF-8,eucJP-win,SJIS-win'));
         $xpath = new \DOMXPath($dom);
 
         $metadata = $this->ogpResolver->parse($html);

diff --git a/app/MetadataResolver/OGPResolver.php b/app/MetadataResolver/OGPResolver.php
@@ -2,6 +2,7 @@
 
 namespace App\MetadataResolver;
 
+use App\Facades\Formatter;
 use GuzzleHttp\Client;
 use GuzzleHttp\Cookie\CookieJar;
 use GuzzleHttp\RequestOptions;
@@ -30,7 +31,7 @@ public function parse(string $html, ?OGPParsePriority $priority = null): Metadat
         }
 
         $dom = new \DOMDocument();
-        @$dom->loadHTML(mb_convert_encoding($html, 'HTML-ENTITIES', 'ASCII,JIS,UTF-8,eucJP-win,SJIS-win'));
+        @$dom->loadHTML(Formatter::htmlEntities($html, 'ASCII,JIS,UTF-8,eucJP-win,SJIS-win'));
         $xpath = new \DOMXPath($dom);
 
         $metadata = new Metadata();

diff --git a/app/MetadataResolver/ToranoanaResolver.php b/app/MetadataResolver/ToranoanaResolver.php
@@ -2,6 +2,7 @@
 
 namespace App\MetadataResolver;
 
+use App\Facades\Formatter;
 use GuzzleHttp\Client;
 use GuzzleHttp\Cookie\CookieJar;
 
@@ -28,7 +29,7 @@ public function resolve(string $url): Metadata
         $metadata = $this->ogpResolver->parse($res->getBody());
 
         $dom = new \DOMDocument();
-        @$dom->loadHTML(mb_convert_encoding($res->getBody(), 'HTML-ENTITIES', 'UTF-8'));
+        @$dom->loadHTML(Formatter::htmlEntities($res->getBody(), 'UTF-8'));
         $xpath = new \DOMXPath($dom);
         $imgNode = $xpath->query('//*[@id="preview"]//img')->item(0);
         if ($imgNode !== null) {

diff --git a/app/Utilities/Formatter.php b/app/Utilities/Formatter.php
@@ -150,4 +150,19 @@ public function sanitizeLike(string $value): string
     {
         return preg_replace('/[%_]/', '\\\\$0', $value);
     }
+
+    /**
+     * PHP 8.1までの `mb_convert_encoding($input, 'HTML-ENTITIES', $from)` 相当のHTMLエンティティ化処理を行います。
+     * @param string $input エンコードする文字列
+     * @param string|null $fromEncoding `$input` の文字コード (nullの場合はdefault_charsetに準ずる、通常はUTF-8)
+     * @return string エンティティ化された文字列
+     */
+    public function htmlEntities(string $input, string $fromEncoding = null): string
+    {
+        // 非Unicode文字列は上手く処理できないので、UTF-8に正規化する
+        $input = mb_convert_encoding($input, 'UTF-8', $fromEncoding);
+
+        // 参考: https://github.com/php/php-src/pull/7177#issuecomment-1317296767
+        return mb_encode_numericentity($input, [0x80, 0x10fffff, 0, 0x1fffff]);
+    }
 }