fix(har): embed content that is not valid UTF-8 as base64

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

@@ -14,6 +14,8 @@
  * limitations under the License.
  */
 
+import { isUtf8 } from 'buffer';
+
 import mime from 'mime';
 import { base64ByteLength } from '@isomorphic/base64';
 import { ManualPromise } from '@isomorphic/manualPromise';
@@ -268,7 +270,7 @@ export class HarTracer {
       content.mimeType = contentType;
     // Keep the size as -1 "not available" when we did not actually read the body.
     if (event.body) {
-      this._storeResponseContent(event.body, content, 'other');
+      this._storeResponseContent(event.body, content);
       if (!this._options.omitSizes)
         harEntry.response.bodySize = event.body.length;
     }
@@ -380,7 +382,7 @@ export class HarTracer {
 
       const content = harEntry.response.content;
       compressionCalculationBarrier?.setDecodedBodySize(buffer.length);
-      this._storeResponseContent(buffer, content, request.resourceType());
+      this._storeResponseContent(buffer, content);
     }).catch(() => {
       compressionCalculationBarrier?.setDecodedBodySize(0);
     }).then(() => {
@@ -545,14 +547,14 @@ export class HarTracer {
       this._delegate.onEntryStarted(harEntry);
   }
 
-  private _storeResponseContent(buffer: Buffer, content: har.Content, resourceType: string) {
+  private _storeResponseContent(buffer: Buffer, content: har.Content) {
     if (!this._options.omitSizes)
       content.size = buffer.length;
 
     if (this._options.content === 'embed') {
       // 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') {
+      if (isTextualMimeType(content.mimeType) && isUtf8(buffer)) {
         content.text = buffer.toString();
       } else {
         content.text = buffer.toString('base64');

tests/library/har.spec.ts

@@ -368,6 +368,24 @@ it('HA-021 should include content @smoke', async ({ contextFactory, server }, te
   expect(log.entries[2].response.content.compression).toBe(0);
 });
 
+it('HA-113 should store binary content with a textual mime type as base64', {
+  annotation: { type: 'issue', description: 'https://github.com/microsoft/playwright/issues/43493' },
+}, async ({ contextFactory, server, browserName }, testInfo) => {
+  it.fixme(browserName === 'webkit', 'WebKit decodes the body of a textual resource before it gets to us');
+  const body = Buffer.from([0x80, 0xc3, 0x28, 0xff, 0x41]);
+  server.setRoute('/binary.js', (req, res) => {
+    res.setHeader('Content-Type', 'application/javascript');
+    res.end(body);
+  });
+  const { page, getLog } = await pageWithHar(contextFactory, testInfo);
+  await page.goto(server.EMPTY_PAGE);
+  await page.evaluate(() => fetch('/binary.js').then(r => r.arrayBuffer()));
+  const log = await getLog();
+  const content = log.entries.find(entry => entry.request.url.endsWith('/binary.js'))!.response.content;
+  expect(content.encoding).toBe('base64');
+  expect(Buffer.from(content.text!, 'base64')).toEqual(body);
+});
+
 it('HA-022 should use attach mode for zip extension', async ({ contextFactory, server }, testInfo) => {
   const { page, getZip } = await pageWithHar(contextFactory, testInfo, { outputPath: 'test.har.zip' });
   await page.goto(server.PREFIX + '/har.html');

tests/registry.md

@@ -26,7 +26,7 @@ subsystem, add a row to the Subsystems table. Then update this file:
 
 ## Subsystems
 
-98 subsystems, 4933 tests (5330 with loop copies).
+98 subsystems, 4934 tests (5331 with loop copies).
 
 | Code | Subsystem | Files | Tests |
 |---|---|---|---|
@@ -78,7 +78,7 @@ subsystem, add a row to the Subsystems table. Then update this file:
 | `FR` | Frame | `page/{frame-evaluate, frame-frame-element, frame-goto, frame-hierarchy}` | 36 |
 | `GE` | Geolocation | `library/geolocation` | 9 |
 | `GN` | Locator and selector generator | `library/{locator-generator, selector-generator}` | 104 |
-| `HA` | HAR | `library/{har, browsercontext-har, har-websocket}` | 112 |
+| `HA` | HAR | `library/{har, browsercontext-har, har-websocket}` | 113 |
 | `HE` | Headful | `library/headful` | 16 |
 | `HT` | Hit target | `library/hit-target` | 22 |
 | `IF` | Input files | `page/page-set-input-files` | 21 |
@@ -3157,6 +3157,7 @@ subsystem, add a row to the Subsystems table. Then update this file:
 | `HA-110` | `library/har-websocket.spec.ts` | should still allow routeWebSocket to fully mock the connection when capturing HAR |
 | `HA-111` | `library/har-websocket.spec.ts` | should still allow routeWebSocket to modify messages when capturing HAR |
 | `HA-112` | `library/har-websocket.spec.ts` | should respect PLAYWRIGHT\_HAR\_NO\_WEBSOCKET\_FRAMES |
+| `HA-113` | `library/har.spec.ts` | should store binary content with a textual mime type as base64 |
 
 ### HE — Headful