fix: gateway proxy reads delta.message and detects truncated SSE

Standardize the stream delta on LangChain's own message dict (delta.message)
instead of a bespoke {text, tool_calls} shape, and raise when the stream ends
without the [DONE] sentinel so mid-stream truncation is no longer silent.

EvoScientist/llm/gateway_proxy.py

@@ -122,11 +122,13 @@ class GatewayProxyChatModel(BaseChatModel):
                 json=payload,
             ) as response:
                 response.raise_for_status()
+                saw_done = False
                 async for line in response.aiter_lines():
                     if not line.startswith("data:"):
                         continue
                     data = line[len("data:"):].strip()
                     if data == "[DONE]":
+                        saw_done = True
                         break
                     chunk = json.loads(data)
                     message = _chunk_to_message(chunk)
@@ -134,13 +136,18 @@ class GatewayProxyChatModel(BaseChatModel):
                         message=message,
                         generation_info=chunk.get("generation_info"),
                     )
+                if not saw_done:
+                    raise RuntimeError("AI4SCI_MODEL_STREAM_INCOMPLETE")
 
 
 def _chunk_to_message(chunk: dict[str, Any]) -> BaseMessage:
     delta = chunk.get("delta") or {}
-    parsed = messages_from_dict(
-        [delta.get("message") or {"type": "AIMessageChunk", "data": {"content": delta.get("text", "")}}]
-    )
+    message_dict = delta.get("message")
+    if message_dict is None:
+        raise RuntimeError("AI4SCI_MODEL_STREAM_DELTA_INVALID")
+    parsed = messages_from_dict([message_dict])
+    if len(parsed) != 1:
+        raise RuntimeError("AI4SCI_MODEL_STREAM_DELTA_INVALID")
     return parsed[0]
 
 

tests/test_gateway_proxy.py

@@ -1,3 +1,5 @@
+import json
+
 import httpx
 import pytest
 from langchain_core.messages import HumanMessage
@@ -48,9 +50,10 @@ async def test_astream_yields_chunks_from_sse(monkeypatch):
         run_id="run-1",
         envelope_signature="sig",
     )
+    msg = {"type": "AIMessageChunk", "data": {"content": "hello"}}
     lines = [
-        'data: {"delta": {"text": "hello"}}\n',
-        'data: {"delta": {"text": " world"}}\n',
+        f'data: {json.dumps({"delta": {"message": msg}})}\n',
+        'data: {"delta": {"message": {"type": "AIMessageChunk", "data": {"content": " world"}}}}\n',
         "data: [DONE]\n",
     ]
     fake = _FakeClient(lines)
@@ -61,3 +64,53 @@ async def test_astream_yields_chunks_from_sse(monkeypatch):
     assert fake.sent["stream"] is True
     assert len(chunks) == 2
     assert chunks[0].message.content == "hello"
+
+
+@pytest.mark.anyio
+async def test_astream_roundtrips_streaming_tool_call_chunks(monkeypatch):
+    model = GatewayProxyChatModel(
+        gateway_url="http://gw",
+        run_id="run-1",
+        envelope_signature="sig",
+    )
+    msg = {
+        "type": "AIMessageChunk",
+        "data": {
+            "content": "",
+            "tool_call_chunks": [
+                {
+                    "name": "read_file",
+                    "args": '{"path":',
+                    "id": "call-1",
+                    "index": 0,
+                    "type": "tool_call_chunk",
+                }
+            ],
+        },
+    }
+    lines = [f'data: {json.dumps({"delta": {"message": msg}})}\n', "data: [DONE]\n"]
+    fake = _FakeClient(lines)
+    monkeypatch.setattr(httpx, "AsyncClient", lambda **kw: fake)
+
+    chunks = [c async for c in model._astream([HumanMessage(content="hi")])]
+
+    assert len(chunks) == 1
+    assert chunks[0].message.tool_call_chunks[0]["name"] == "read_file"
+    assert chunks[0].message.tool_call_chunks[0]["id"] == "call-1"
+
+
+@pytest.mark.anyio
+async def test_astream_raises_on_missing_done(monkeypatch):
+    model = GatewayProxyChatModel(
+        gateway_url="http://gw",
+        run_id="run-1",
+        envelope_signature="sig",
+    )
+    lines = [
+        'data: {"delta": {"message": {"type": "AIMessageChunk", "data": {"content": "hi"}}}}\n'
+    ]
+    fake = _FakeClient(lines)
+    monkeypatch.setattr(httpx, "AsyncClient", lambda **kw: fake)
+
+    with pytest.raises(RuntimeError, match="AI4SCI_MODEL_STREAM_INCOMPLETE"):
+        _ = [c async for c in model._astream([HumanMessage(content="hi")])]