chore(webview): generalize dialog bridge into SyncServer

[pavelfeldman]

Summary

• Generalize the WebView DialogBridge into a reusable SyncServer (JSON in / JSON out).
• Each handler is registered under an unguessable guid path, so the endpoint can't be discovered by page-side scripts just from the randomly bound port.
• WVPage now owns registering/unregistering its own dialog handler and keeps the dialog-specific request interpretation.

[github-actions]

Test results for "tests 1"

1 failed
❌ [installation tests] › playwright-cdn.spec.ts:43 › playwright cdn failover should work (https://cdn.playwright.dev) @package-installations-ubuntu-latest

3 flaky

⚠️ [chromium-library] › library/video.spec.ts:736 › screencast › should work with video+trace `@chromium-ubuntu-22.04-arm-node20`
⚠️ [chromium-library] › library/chromium/chromium.spec.ts:371 › should produce network events, routing, and annotations for Service Worker `@realtime-time-library-chromium-linux`
⚠️ [chromium-library] › library/video.spec.ts:699 › screencast › should capture full viewport on hidpi `@chromium-ubuntu-22.04-node20`

49965 passed, 1188 skipped

---

Merge workflow run.

[github-actions]

Test results for "MCP"

3 failed
❌ [chrome] › mcp/annotate.spec.ts:57 › should capture multiple screenshots in one annotation @mcp-macos-latest-chrome
❌ [chrome] › mcp/cli-navigation.spec.ts:43 › tab-new with url @mcp-macos-latest-chrome
❌ [chromium] › mcp/annotate.spec.ts:173 › user-initiated annotate downloads zip with feedback.md @mcp-macos-latest-chromium

7757 passed, 1249 skipped

---

Merge workflow run.

[github-actions]

Hi, I'm the Playwright bot and I took a look at the failing CI.

🟢 The one failure is unrelated to this PR — CI is clear

The single red test is playwright-cdn.spec.ts:43 › playwright cdn failover should work on package-installations-ubuntu-latest. It's a CDN download failure (npx playwright install exited 1 while pulling Chrome from cdn.playwright.dev), which this WebKit WebView SyncServer refactor can't touch.

Details

This PR only refactors packages/playwright-core/src/server/webkit/webview/ (dialog bridge → SyncServer). None of the failing surfaces exercise that code.

Pre-existing flake / infra

• [installation tests] › playwright-cdn.spec.ts:43 › playwright cdn failover should work (https://cdn.playwright.dev) — network/infra, not a code failure. The error is npx playwright install exiting 1 mid-download of the Chrome build from the CDN. Across the test-results DB this test's final verdict flips: failed 2 of 420 runs, passed 418, and the other failure was on PR feat(webkit): roll to r2327 #41684 (sha 922cd4cf) — a different PR, so this test breaks where this branch can't be responsible. Nothing in the WebView SyncServer change reaches the install/CDN path.

Also seen (already green)

• 3 flaky, all self-rescued on retry and unrelated to WebView: library/video.spec.ts:736, library/chromium/chromium.spec.ts:371 (Service Worker), library/video.spec.ts:699.

_{Triaged by the Playwright bot - agent run}

[github-actions]

Hi, I'm the Playwright bot and I took a look at the failing CI.

🟢 All failures are pre-existing flakes / infra — this PR is clear

This PR only refactors the WebKit WebView dialog bridge into SyncServer (packages/playwright-core/src/server/webkit/webview/). None of the failing surfaces — a CDN install test and three MCP tests on macOS Chrome/Chromium — go anywhere near that code, and every one of them is a known flake that fails on unrelated SHAs.

Details

Pre-existing flake / infra

• [installation tests] › playwright-cdn.spec.ts:43 › playwright cdn failover should work — network/infra, npx playwright install exiting 1 mid-download from cdn.playwright.dev. Across the DB it flips failed 2 of 420 runs, passed 418, and the other failure was on PR feat(webkit): roll to r2327 #41684 (sha 922cd4cf) — a different PR. The install/CDN path doesn't touch the WebView change.

• [chrome] › mcp/annotate.spec.ts:57 › should capture multiple screenshots in one annotation — bimodal flake: failed 7 of 302 chrome runs, and fails on 4 other SHAs (dbb805e6, a89eab73, 9c551255, 90cbbbb9), not this branch.

• [chrome] › mcp/cli-navigation.spec.ts:43 › tab-new with url — flake: failed 1 of 302 chrome runs, also failed on sha c51e3f58, a different branch.

• [chromium] › mcp/annotate.spec.ts:173 › user-initiated annotate downloads zip with feedback.md — one of the flakiest of the group: final verdict flips 12/297 (firefox), 11/304 (webkit), 2/301 (chromium), failing across 6 distinct SHAs unrelated to this PR.

The MCP tools exercise Chrome/Chromium/CDP; a WebKit WebView SyncServer refactor can't reach them.

Also seen (already green)

• 3 flaky, all self-rescued on retry and unrelated to WebView: library/video.spec.ts:736, library/chromium/chromium.spec.ts:371 (Service Worker), library/video.spec.ts:699.

_{Triaged by the Playwright bot - agent run}