feat: TS webhook fixtures + parity tests + Postgres lock fix

[patrick-chinchill]

Closes #18, closes #19. 36 fixture tests pass, 10 xfail (real adapter bugs to fix).

[gemini-code-assist]

Code Review

This pull request replaces the two-step lock acquisition in the Postgres state adapter with an atomic upsert to prevent race conditions and adds a comprehensive suite of parity tests for cards, emojis, and fixture replays. Review feedback suggests improving test coverage by including the WhatsApp converter in the aggregate card title test, fully implementing the Discord channel fixture replay, and replacing blocking time.sleep() calls with asyncio.sleep() in asynchronous tests.

[gemini-code-assist]

The test_all_converters_agree_on_title test is a great way to ensure basic parity. However, it seems to be missing a check for the WhatsApp card converter (card_to_whatsapp), even though there's a separate test for it (test_whatsapp_card). Including it in this aggregate test would improve test coverage and ensure it also correctly handles the card title.

    def test_all_converters_agree_on_title(self, card: CardElement):
        """All available converters should include the card title in their output."""
        converters: list[tuple[str, bool, Any]] = [
            ("fallback", True, lambda c: card_to_fallback_text(c)),
        ]
        if _SLACK_CARDS:
            converters.append(("slack", True, lambda c: str(card_to_block_kit(c))))
        if _TEAMS_CARDS:
            converters.append(("teams", True, lambda c: str(card_to_adaptive_card(c))))
        if _GCHAT_CARDS:
            converters.append(("gchat", True, lambda c: str(card_to_google_card(c))))
        if _DISCORD_CARDS:
            converters.append(("discord", True, lambda c: str(card_to_discord_payload(c))))
        if _GITHUB_CARDS:
            converters.append(("github", True, lambda c: card_to_github_markdown(c)))
        if _LINEAR_CARDS:
            converters.append(("linear", True, lambda c: card_to_linear_markdown(c)))
        if _WHATSAPP_CARDS:
            converters.append(("whatsapp", True, lambda c: str(card_to_whatsapp(c))))

        for name, available, converter in converters:
            output = converter(card)
            assert "System Status Report" in output, f"{name} converter should include card title"

[gemini-code-assist]

This test only verifies the structure of the fixture file but doesn't actually test the adapter's handling of the payload. To ensure parity, this test should be implemented fully to process the fixture through adapter.handle_webhook, similar to how other fixtures are tested in this file.

    async def test_discord_channel_fixture(self):
        """channel/discord.json should have parseable gateway events."""
        fixture = load_fixture("channel/discord.json")
        adapter = self._make_adapter(
            application_id=fixture.get("applicationId"),
        )
        mock_chat = _make_mock_chat()
        await adapter.initialize(mock_chat)

        # Mock the Discord API call that creates a thread
        mock_thread_response = {"id": fixture.get("threadChannelId"), "name": "test-thread"}
        with patch.object(adapter, "_create_discord_thread", new_callable=AsyncMock, return_value=mock_thread_response):
            # Gateway events are sent with x-discord-gateway-token header
            gateway_payload = fixture["mention"]
            body = json.dumps(gateway_payload)
            request = _FakeRequest(
                body,
                {
                    "content-type": "application/json",
                    "x-discord-gateway-token": self.BOT_TOKEN,
                },
            )

            result = await adapter.handle_webhook(request)

        assert result["status"] == 200
        assert mock_chat.handle_incoming_message.called

[gemini-code-assist]

Using time.sleep() in an async test blocks the entire event loop, which can lead to issues and is generally considered an anti-pattern. It's better to use await asyncio.sleep() to yield control back to the event loop. You'll need to add import asyncio at the top of the file.

Suggested change

	time.sleep(0.005)
	await asyncio.sleep(0.005)

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 4fac8f60a8

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Make emoji parity tests self-contained

These parity tests depend on a hard-coded external file (/tmp/vercel-chat/packages/chat/src/emoji.ts) and immediately pytest.skip(...) when it is missing, which means the new TestEmojiParity coverage is skipped in normal CI runs where only this repository is checked out (see .github/workflows/test.yml). In practice, regressions in DEFAULT_EMOJI_MAP can pass unnoticed unless a developer manually clones another repo into /tmp, so the test should use a committed fixture/snapshot (or fail explicitly) instead of silently skipping.

Useful? React with 👍 / 👎.