Merge pull request #24 from Chinchill-AI/coverage-and-docs

test: coverage improvements + doc updates

Author: patrick-chinchill

.coverage

@@ -1,5 +1,15 @@
 # Changelog
 
+## 0.0.1a7 (2026-04-03)
+
+Fixture replay tests and coverage hardening.
+
+- All fixture replay tests pass (46 tests, 28 JSON fixtures copied from TS SDK)
+- GChat: fix float startIndex in annotation parsing
+- Postgres: atomic lock upsert to eliminate TOCTOU race
+- Coverage tests added for 4 low-coverage modules
+- Documentation updates (TESTING.md, ARCHITECTURE.md)
+
 ## 0.0.1a6 (2026-04-07)
 
 Systematic port fidelity scan — 10 more bugs fixed.

CHANGELOG.md

@@ -2,7 +2,7 @@
 
 Multi-platform async chat SDK for Python. Port of [Vercel Chat](https://github.com/vercel/chat).
 
-> **Status: Alpha (0.0.1a6)** — API may change. Not yet tested in production.
+> **Status: Alpha (0.0.1a7)** — API may change. Not yet tested in production.
 
 ## Why chat-sdk?
 

README.md

@@ -228,15 +228,15 @@ def assert_no_camel_case_keys(d, path="", *, skip_raw=True):
 
 This test suite is a regression guard. Without it, any refactoring of adapter dispatch code could reintroduce camelCase keys without any test failure. The test runs on every CI build and takes < 1 second.
 
-## Plans: Webhook JSON Fixtures
+## Webhook JSON Fixtures (Done)
 
-The TS SDK repository contains recorded webhook JSON payloads used for testing. We plan to copy these fixtures into `tests/fixtures/` to enable:
+28 webhook JSON fixtures were copied from the TS SDK repository into `tests/fixtures/` and 46 replay tests now pass against them. The fixtures enable:
 
-1. **Cross-language parity testing**: Verify that the Python adapter produces the same normalized `Message` objects as the TS adapter for identical webhook payloads.
+1. **Cross-language parity testing**: The Python adapter produces the same normalized `Message` objects as the TS adapter for identical webhook payloads.
 2. **Platform version regression**: When platforms change their webhook format, the fixture files make it obvious what changed.
-3. **Replay tests without mocks**: Drive the full adapter pipeline with real payloads instead of hand-constructed dicts.
+3. **Replay tests without mocks**: The full adapter pipeline is driven with real payloads instead of hand-constructed dicts.
 
-The fixture format will be:
+The fixture layout:
 ```
 tests/fixtures/
   slack/

docs/TESTING.md

@@ -1,6 +1,6 @@
 [project]
 name = "chat-sdk"
-version = "0.0.1a6"
+version = "0.0.1a7"
 description = "Multi-platform async chat SDK for Python — port of Vercel Chat"
 readme = "README.md"
 license = {text = "MIT"}
@@ -34,7 +34,8 @@ discord = ["pynacl>=1.5", "aiohttp>=3.9"]
 teams = ["aiohttp>=3.9"]
 telegram = ["aiohttp>=3.9"]
 whatsapp = ["aiohttp>=3.9"]
-google-chat = ["aiohttp>=3.9", "google-auth>=2.0", "pyjwt>=2.8"]
+google-chat = ["aiohttp>=3.9",
+    "pyjwt[crypto]>=2.8", "google-auth>=2.0", "pyjwt>=2.8"]
 linear = ["aiohttp>=3.9"]
 all = [
     "slack-sdk>=3.27.0",
@@ -44,6 +45,7 @@ all = [
     "cryptography>=42.0",
     "pynacl>=1.5",
     "aiohttp>=3.9",
+    "pyjwt[crypto]>=2.8",
     "google-auth>=2.0",
 ]
 
@@ -70,5 +72,7 @@ dev = [
     "pytest>=8.0",
     "pytest-asyncio>=0.23.0",
     "pytest-cov>=5.0",
+    "aiohttp>=3.9",
+    "pyjwt[crypto]>=2.8",
     "ruff>=0.4.0",
 ]