From 60249e0bc2396c8ae3a4f37b4aacf592570e17cb Mon Sep 17 00:00:00 2001
From: Sergio M <sergiomorenoalbert@gmail.com>
Date: Sat, 18 May 2019 16:15:16 +0200
Subject: [PATCH] fix: Page.setContent working with unicode strings (#4433)

Fix `page.setContent` with unicode strings that exceeds the range of a 8-bit byte.

This patch implements decoding part of the [MDN's solution #4 for the "unicode problem"](https://developer.mozilla.org/en-US/docs/Web/API/WindowBase64/Base64_encoding_and_decoding#Solution_4_%E2%80%93_escaping_the_string_before_encoding_it).

Since we rely on node.js buffer to encode into base64, we don't have troubles with base64 encoding, so it is left as-is.

Fixes #4424
---
 experimental/puppeteer-firefox/lib/DOMWorld.js |  2 +-
 lib/DOMWorld.js                                |  2 +-
 test/page.spec.js                              | 12 ++++++++++++
 3 files changed, 14 insertions(+), 2 deletions(-)

diff --git a/experimental/puppeteer-firefox/lib/DOMWorld.js b/experimental/puppeteer-firefox/lib/DOMWorld.js
index 8d50da02..12cfc67d 100644
--- a/experimental/puppeteer-firefox/lib/DOMWorld.js
+++ b/experimental/puppeteer-firefox/lib/DOMWorld.js
@@ -134,7 +134,7 @@ class DOMWorld {
   async setContent(html) {
     await this.evaluate(base64html => {
       document.open();
-      document.write(atob(base64html));
+      document.write(decodeURIComponent(atob(base64html).split('').map(c => '%' + ('00' + c.charCodeAt(0).toString(16)).slice(-2)).join('')));
       document.close();
     }, Buffer.from(html).toString('base64'));
   }
diff --git a/lib/DOMWorld.js b/lib/DOMWorld.js
index 503df690..0a0210fe 100644
--- a/lib/DOMWorld.js
+++ b/lib/DOMWorld.js
@@ -205,7 +205,7 @@ class DOMWorld {
     // lifecycle event. @see https://crrev.com/608658
     await this.evaluate(base64html => {
       document.open();
-      document.write(atob(base64html));
+      document.write(decodeURIComponent(atob(base64html).split('').map(c => '%' + ('00' + c.charCodeAt(0).toString(16)).slice(-2)).join('')));
       document.close();
     }, Buffer.from(html).toString('base64'));
     const watcher = new LifecycleWatcher(this._frameManager, this._frame, waitUntil, timeout);
diff --git a/test/page.spec.js b/test/page.spec.js
index cf78ce1c..bc2a065d 100644
--- a/test/page.spec.js
+++ b/test/page.spec.js
@@ -760,6 +760,18 @@ module.exports.addTests = function({testRunner, expect, headless, puppeteer, CHR
       await page.setContent('<div>hello world</div>' + '\x7F');
       expect(await page.$eval('div', div => div.textContent)).toBe('hello world');
     });
+    it('should work with accents', async({page, server}) => {
+      await page.setContent('<div>aberración</div>');
+      expect(await page.$eval('div', div => div.textContent)).toBe('aberración');
+    });
+    it('should work with emojis', async({page, server}) => {
+      await page.setContent('<div>🐥</div>');
+      expect(await page.$eval('div', div => div.textContent)).toBe('🐥');
+    });
+    it('should work with newline', async({page, server}) => {
+      await page.setContent('<div>\n</div>');
+      expect(await page.$eval('div', div => div.textContent)).toBe('\n');
+    });
   });
 
   describe_fails_ffox('Page.setBypassCSP', function() {