feat(har): store textual content without base64 encoding (#14772)

packages/playwright-core/src/server/har/harTracer.ts

@@ -191,7 +191,7 @@ export class HarTracer {
     const contentType = event.headers['content-type'];
     if (contentType)
       content.mimeType = contentType;
-    this._storeResponseContent(event.body, content);
+    this._storeResponseContent(event.body, content, 'other');
 
     if (this._started)
       this._delegate.onEntryFinished(harEntry);
@@ -261,7 +261,7 @@ export class HarTracer {
 
       const content = harEntry.response.content;
       compressionCalculationBarrier.setDecodedBodySize(buffer.length);
-      this._storeResponseContent(buffer, content);
+      this._storeResponseContent(buffer, content, request.resourceType());
     }).catch(() => {
       compressionCalculationBarrier.setDecodedBodySize(0);
     }).then(() => {
@@ -296,15 +296,21 @@ export class HarTracer {
       this._delegate.onEntryFinished(harEntry);
   }
 
-  private _storeResponseContent(buffer: Buffer | undefined, content: har.Content) {
+  private _storeResponseContent(buffer: Buffer | undefined, content: har.Content, resourceType: string) {
     if (!buffer) {
       content.size = 0;
       return;
     }
     content.size = buffer.length;
     if (this._options.content === 'embedded') {
+      // Sometimes, we can receive a font/media file with textual mime type. Browser
+      // still interprets them correctly, but the 'content-type' header is obviously wrong.
+      if (isTextualMimeType(content.mimeType) && resourceType !== 'font') {
+        content.text = buffer.toString();
+      } else {
         content.text = buffer.toString('base64');
         content.encoding = 'base64';
+      }
     } else if (this._options.content === 'sha1') {
       content._sha1 = calculateSha1(buffer) + '.' + (mime.getExtension(content.mimeType) || 'dat');
       if (this._started)
@@ -529,3 +535,7 @@ function parseCookie(c: string): har.Cookie {
   }
   return cookie;
 }
+
+function isTextualMimeType(mimeType: string) {
+  return !!mimeType.match(/^(text\/.*?|application\/(json|(x-)?javascript|xml.*?|ecmascript)|image\/svg(\+xml)?|application\/.*?(\+json|\+xml))(;\s*charset=.*)?$/);
+}

tests/library/har.spec.ts

@@ -248,21 +248,26 @@ it('should include secure set-cookies', async ({ contextFactory, httpsServer },
 it('should include content @smoke', async ({ contextFactory, server }, testInfo) => {
   const { page, getLog } = await pageWithHar(contextFactory, testInfo);
   await page.goto(server.PREFIX + '/har.html');
+  await page.evaluate(() => fetch('/pptr.png').then(r => r.arrayBuffer()));
   const log = await getLog();
 
-  expect(log.entries[0].response.httpVersion).toBe('HTTP/1.1');
+  expect(log.entries[0].response.content.encoding).toBe(undefined);
-  expect(log.entries[0].response.content.encoding).toBe('base64');
   expect(log.entries[0].response.content.mimeType).toBe('text/html; charset=utf-8');
-  expect(Buffer.from(log.entries[0].response.content.text, 'base64').toString()).toContain('HAR Page');
+  expect(log.entries[0].response.content.text).toContain('HAR Page');
   expect(log.entries[0].response.content.size).toBeGreaterThanOrEqual(96);
   expect(log.entries[0].response.content.compression).toBe(0);
 
-  expect(log.entries[1].response.httpVersion).toBe('HTTP/1.1');
+  expect(log.entries[1].response.content.encoding).toBe(undefined);
-  expect(log.entries[1].response.content.encoding).toBe('base64');
   expect(log.entries[1].response.content.mimeType).toBe('text/css; charset=utf-8');
-  expect(Buffer.from(log.entries[1].response.content.text, 'base64').toString()).toContain('pink');
+  expect(log.entries[1].response.content.text).toContain('pink');
   expect(log.entries[1].response.content.size).toBeGreaterThanOrEqual(37);
   expect(log.entries[1].response.content.compression).toBe(0);
+
+  expect(log.entries[2].response.content.encoding).toBe('base64');
+  expect(log.entries[2].response.content.mimeType).toBe('image/png');
+  expect(Buffer.from(log.entries[2].response.content.text, 'base64').byteLength).toBeGreaterThan(0);
+  expect(log.entries[2].response.content.size).toBeGreaterThanOrEqual(6000);
+  expect(log.entries[2].response.content.compression).toBe(0);
 });
 
 it('should filter by glob', async ({ contextFactory, server }, testInfo) => {
@@ -580,7 +585,7 @@ it('should contain http2 for http2 requests', async ({ contextFactory, browserNa
   const log = await getLog();
   expect(log.entries[0].request.httpVersion).toBe('HTTP/2.0');
   expect(log.entries[0].response.httpVersion).toBe('HTTP/2.0');
-  expect(Buffer.from(log.entries[0].response.content.text, 'base64').toString()).toBe('<h1>Hello World</h1>');
+  expect(log.entries[0].response.content.text).toBe('<h1>Hello World</h1>');
   server.close();
 });
 
@@ -734,7 +739,7 @@ it('should include API request', async ({ contextFactory, server }, testInfo) =>
   expect(entry.response.status).toBe(200);
   expect(entry.response.headers.find(h => h.name.toLowerCase() === 'content-type')?.value).toContain('application/json');
   expect(entry.response.content.size).toBe(15);
-  expect(entry.response.content.text).toBe(responseBody.toString('base64'));
+  expect(entry.response.content.text).toBe(responseBody.toString());
 });
 
 it('should not hang on resources served from cache', async ({ contextFactory, server, browserName }, testInfo) => {