From 81ab6b3fde101d7801f0159d26908cd09b894858 Mon Sep 17 00:00:00 2001
From: musou1500 <musou1500@gmail.com>
Date: Mon, 13 Dec 2021 07:54:00 +0900
Subject: [PATCH] fix(tracing): overwrite attr value specifies charset other
 than utf-8 (#10848)

---
 .../trace/recorder/snapshotterInjected.ts     | 22 ++++++++++++++++++-
 tests/snapshotter.spec.ts                     | 14 ++++++++++++
 2 files changed, 35 insertions(+), 1 deletion(-)

diff --git a/packages/playwright-core/src/server/trace/recorder/snapshotterInjected.ts b/packages/playwright-core/src/server/trace/recorder/snapshotterInjected.ts
index 64c4972cfa..e3eeb64dfe 100644
--- a/packages/playwright-core/src/server/trace/recorder/snapshotterInjected.ts
+++ b/packages/playwright-core/src/server/trace/recorder/snapshotterInjected.ts
@@ -196,6 +196,24 @@ export function frameSnapshotStreamer(snapshotStreamer: string) {
       visitNode(this._fakeBase);
     }
 
+    private __sanitizeMetaAttribute(name: string, value: string, httpEquiv: string) {
+      if (name === 'charset')
+        return 'utf-8';
+
+      if (httpEquiv.toLowerCase() !== 'content-type' || name !== 'content')
+        return value;
+
+      const [type, ...params] = value.split(';');
+      if (type !== 'text/html' || params.length <= 0)
+        return value;
+
+      const charsetParamIdx = params.findIndex(param => param.trim().startsWith('charset='));
+      if (charsetParamIdx > -1)
+        params[charsetParamIdx] = 'charset=utf-8';
+
+      return `${type}; ${params.join('; ')}`;
+    }
+
     private _sanitizeUrl(url: string): string {
       if (url.startsWith('javascript:'))
         return '';
@@ -420,7 +438,9 @@ export function frameSnapshotStreamer(snapshotStreamer: string) {
             if (nodeName === 'IFRAME' && (name === 'src' || name === 'sandbox'))
               continue;
             let value = element.attributes[i].value;
-            if (name === 'src' && (nodeName === 'IMG'))
+            if (nodeName === 'META')
+              value = this.__sanitizeMetaAttribute(name, value, (node as HTMLMetaElement).httpEquiv);
+            else if (name === 'src' && (nodeName === 'IMG'))
               value = this._sanitizeUrl(value);
             else if (name === 'srcset' && (nodeName === 'IMG'))
               value = this._sanitizeSrcSet(value);
diff --git a/tests/snapshotter.spec.ts b/tests/snapshotter.spec.ts
index 485cd13c25..9b8a16e0b7 100644
--- a/tests/snapshotter.spec.ts
+++ b/tests/snapshotter.spec.ts
@@ -103,6 +103,20 @@ it.describe('snapshots', () => {
     expect(distillSnapshot(snapshot)).toBe('<!DOCTYPE foo>hi');
   });
 
+  it('should replace meta charset attr that specifies charset', async ({ page, server, toImpl, snapshotter }) => {
+    await page.goto(server.EMPTY_PAGE);
+    await page.setContent('<meta charset="shift-jis" />');
+    const snapshot = await snapshotter.captureSnapshot(toImpl(page), 'snapshot');
+    expect(distillSnapshot(snapshot)).toBe('<META charset="utf-8">');
+  });
+
+  it('should replace meta content attr that specifies charset', async ({ page, server, toImpl, snapshotter }) => {
+    await page.goto(server.EMPTY_PAGE);
+    await page.setContent('<meta http-equiv="Content-Type" content="text/html; charset=Shift_JIS">');
+    const snapshot = await snapshotter.captureSnapshot(toImpl(page), 'snapshot');
+    expect(distillSnapshot(snapshot)).toBe('<META http-equiv="Content-Type" content="text/html; charset=utf-8">');
+  });
+
   it('should respect subresource CSSOM change', async ({ page, server, toImpl, snapshotter }) => {
     await page.goto(server.EMPTY_PAGE);
     await page.route('**/style.css', route => {