fix(accent): address review cancellation edge cases

Author: torrid-fish

api/accent/preprocess.py

@@ -310,10 +310,11 @@ def cap_chunk_length(chunk: str, token_boundaries: set[int] | None = None) -> li
     r"hz|db|nm|m|g|l|w|v|a|s|h)"
 )
 NUMERIC_UNIT_RE = re.compile(rf"^{_NUMERIC_UNIT_BODY}$", re.IGNORECASE)
-# Do not retain a unit-looking substring inside an ASCII identifier (for
-# example Model10mph or abc53mmdef) when hiding English from OpenJTalk.
+# Do not retain a unit-looking substring inside an ASCII identifier or compound
+# (for example Model10mph, abc53mmdef, or RTX-4090m) when hiding English from
+# OpenJTalk. Dots and hyphens are identifier bridges throughout the pipeline.
 _NUMERIC_UNIT_TEXT_RE = re.compile(
-    rf"(?<![A-Za-z0-9_]){_NUMERIC_UNIT_BODY}(?![A-Za-z0-9_])", re.IGNORECASE
+    rf"(?<![A-Za-z0-9_.-]){_NUMERIC_UNIT_BODY}(?![A-Za-z0-9_.-])", re.IGNORECASE
 )
 
 

api/accent/routes.py

@@ -106,17 +106,18 @@ async def _mark_accent_until_disconnect(
 
     processing = asyncio.create_task(_mark_accent(request))
     disconnected = asyncio.create_task(_wait_for_disconnect(raw_request.receive))
-    done, _pending = await asyncio.wait(
-        {processing, disconnected}, return_when=asyncio.FIRST_COMPLETED
-    )
-    if processing in done:
-        disconnected.cancel()
-        await asyncio.gather(disconnected, return_exceptions=True)
-        return await processing
+    try:
+        done, _pending = await asyncio.wait(
+            {processing, disconnected}, return_when=asyncio.FIRST_COMPLETED
+        )
+        if processing in done:
+            return await processing
 
-    processing.cancel()
-    await asyncio.gather(processing, return_exceptions=True)
-    raise asyncio.CancelledError
+        raise asyncio.CancelledError
+    finally:
+        processing.cancel()
+        disconnected.cancel()
+        await asyncio.gather(processing, disconnected, return_exceptions=True)
 
 
 async def _wait_for_disconnect(receive: Receive) -> None:

tests/test_preprocess_regressions.py

@@ -14,7 +14,9 @@
 
 def test_clean_hidden_english_keeps_only_standalone_numeric_units() -> None:
     assert clean_hidden_english("Model10mph猫 abc53mmdef猫") == "10猫 53猫"
+    assert clean_hidden_english("RTX-4090m猫 53mm-pro猫") == "-4090猫 53-猫"
     assert clean_hidden_english("53mm猫 3kgの荷物") == "53mm猫 3kgの荷物"
+    assert clean_hidden_english("-53mm猫") == "-53mm猫"
 
 
 def _word(surface: str) -> WordAccentResult:

tests/test_request_limits.py

@@ -135,15 +135,33 @@ async def scenario() -> None:
         limiter = threading.BoundedSemaphore(1)
         monkeypatch.setattr(routes, "_REQUEST_LIMITER", limiter)
         monkeypatch.setattr(routes, "ACCENT_REQUEST_TIMEOUT_SECONDS", 0.01)
+        processing_cancelled = asyncio.Event()
+        receive_cancelled = asyncio.Event()
 
         async def never_finishes(_request: Request) -> None:
-            await asyncio.Event().wait()
+            try:
+                await asyncio.Event().wait()
+            except asyncio.CancelledError:
+                processing_cancelled.set()
+                raise
+
+        async def never_disconnects() -> Message:
+            try:
+                await asyncio.Event().wait()
+            except asyncio.CancelledError:
+                receive_cancelled.set()
+                raise
 
         monkeypatch.setattr(routes, "_mark_accent", never_finishes)
+        raw_request = cast(
+            StarletteRequest, SimpleNamespace(receive=never_disconnects)
+        )
 
         with pytest.raises(HTTPException) as caught:
-            await routes.mark_accent(Request(text="猫"), _raw_request())
+            await routes.mark_accent(Request(text="猫"), raw_request)
         assert caught.value.status_code == 504
+        assert processing_cancelled.is_set()
+        assert receive_cancelled.is_set()
         assert limiter.acquire(blocking=False)
         limiter.release()