Artifact: support downloading artifacts with CJK characters in their name (#2341)

* Artifact: support downloading artifacts with CJK characters in their name * Fix some linting/PR comments * One more linting fix
2026-03-11 09:30:15 -04:00
parent 6fd292ebdd
commit 21229dc09e
4 changed files with 135 additions and 5 deletions
@@ -977,5 +977,125 @@ describe('download-artifact', () => {
      )
      expect(fs.existsSync(maliciousPath)).toBe(false)
    })
+
+    it('should correctly handle Content-Disposition with filename* parameter (RFC 5987)', async () => {
+      const rawFileContent = 'content with rfc5987 encoding'
+      const expectedFileName = '报告-土-x.txt'
+      const asciiFileName = '__-_-x.txt'
+
+      const mockGetRfc5987File = jest.fn(() => {
+        const message = new http.IncomingMessage(new net.Socket())
+        message.statusCode = 200
+        message.headers['content-type'] = 'text/plain'
+        // Server sends both: filename with _ fallbacks, filename* with UTF-8 encoding
+        message.headers['content-disposition'] =
+          `attachment; filename="${asciiFileName}"; filename*=UTF-8''${encodeURIComponent(expectedFileName)}`
+        message.push(Buffer.from(rawFileContent, 'utf8'))
+        message.push(null)
+        return {
+          message
+        }
+      })
+
+      const mockHttpClient = (HttpClient as jest.Mock).mockImplementation(
+        () => {
+          return {
+            get: mockGetRfc5987File
+          }
+        }
+      )
+
+      await streamExtractExternal(
+        fixtures.blobStorageUrl,
+        fixtures.workspaceDir
+      )
+
+      expect(mockHttpClient).toHaveBeenCalledWith(getUserAgentString())
+      const savedFilePath = path.join(fixtures.workspaceDir, expectedFileName)
+      expect(fs.existsSync(savedFilePath)).toBe(true)
+      expect(fs.readFileSync(savedFilePath, 'utf8')).toBe(rawFileContent)
+    })
+
+    it('should handle zip artifacts with Chinese characters in the artifact name', async () => {
+      // Simulate Azure Blob Storage URL with rscd containing Chinese filename
+      const chineseArtifactName = 'probe-土-x'
+      const asciiArtifactName = 'probe-_-x'
+      const blobUrlWithChineseName = `https://blob-storage.local/artifact.zip?rscd=${encodeURIComponent(`attachment; filename="${asciiArtifactName}.zip"; filename*=UTF-8''${encodeURIComponent(`${chineseArtifactName}.zip`)}`)}&rsct=application%2Fzip&sig=abc123`
+
+      const mockGetZip = jest.fn(() => {
+        const message = new http.IncomingMessage(new net.Socket())
+        message.statusCode = 200
+        message.headers['content-type'] = 'application/zip'
+        message.headers['content-disposition'] =
+          `attachment; filename="${asciiArtifactName}.zip"; filename*=UTF-8''${encodeURIComponent(`${chineseArtifactName}.zip`)}`
+        message.push(fs.readFileSync(fixtures.exampleArtifact.path))
+        message.push(null)
+        return {
+          message
+        }
+      })
+
+      const mockHttpClient = (HttpClient as jest.Mock).mockImplementation(
+        () => {
+          return {
+            get: mockGetZip
+          }
+        }
+      )
+
+      await streamExtractExternal(blobUrlWithChineseName, fixtures.workspaceDir)
+
+      expect(mockHttpClient).toHaveBeenCalledWith(getUserAgentString())
+      // Zip should be extracted normally regardless of Chinese artifact name
+      await expectExtractedArchive(fixtures.workspaceDir)
+    })
+
+    it.each([
+      ['土', '_'], // U+571F - known to cause 400 errors
+      ['日', '_'], // U+65E5 - reported to work fine
+      ['中文测试', '____'], // multiple Chinese characters
+      ['文件-2026年', '__-2026_'], // mixed Chinese and numbers
+      ['データ', '___'], // Japanese katakana
+      ['테스트', '___'] // Korean characters
+    ])(
+      'should prefer filename* over filename for non-ASCII character %s (%s)',
+      async (chars, asciiReplacement) => {
+        const rawFileContent = `content for ${chars}`
+        const expectedFileName = `artifact-${chars}.txt`
+        const asciiFileName = `artifact-${asciiReplacement}.txt`
+
+        const mockGetFile = jest.fn(() => {
+          const message = new http.IncomingMessage(new net.Socket())
+          message.statusCode = 200
+          message.headers['content-type'] = 'text/plain'
+          // Server sends filename with _ replacing non-ASCII, filename* with proper encoding
+          message.headers['content-disposition'] =
+            `attachment; filename="${asciiFileName}"; filename*=UTF-8''${encodeURIComponent(expectedFileName)}`
+          message.push(Buffer.from(rawFileContent, 'utf8'))
+          message.push(null)
+          return {
+            message
+          }
+        })
+
+        const mockHttpClient = (HttpClient as jest.Mock).mockImplementation(
+          () => {
+            return {
+              get: mockGetFile
+            }
+          }
+        )
+
+        await streamExtractExternal(
+          fixtures.blobStorageUrl,
+          fixtures.workspaceDir
+        )
+
+        expect(mockHttpClient).toHaveBeenCalledWith(getUserAgentString())
+        const savedFilePath = path.join(fixtures.workspaceDir, expectedFileName)
+        expect(fs.existsSync(savedFilePath)).toBe(true)
+        expect(fs.readFileSync(savedFilePath, 'utf8')).toBe(rawFileContent)
+      }
+    )
  })
 })