fix(route): people charset decoding (#16734) (#16743)

2026-09-01 14:57:14 +08:00 · 2024-09-13 23:10:45 +08:00
parent 810aae3997
commit 487c0469c6
1 changed files with 6 additions and 2 deletions
@@ -32,8 +32,12 @@ async function handler(ctx) {
        responseType: 'buffer',
    });

-    const encoding = site === 'www' ? 'gbk' : 'utf-8';
-    const decodedResponse = iconv.decode(response, encoding);
+    // not seen Content-Type in response headers
+    // try to parse charset from meta tag
+    let decodedResponse = iconv.decode(response, 'utf-8');
+    const parsedCharset = decodedResponse.match(/<meta.*?charset=["']?([^"'>]+)["']?/i);
+    const encoding = parsedCharset ? parsedCharset[1].toLowerCase() : 'utf-8';
+    decodedResponse = encoding === 'utf-8' ? decodedResponse : iconv.decode(response, encoding);
    const $ = load(decodedResponse);

    $('em').remove();