fix(oci): address Bugbot review — deduplicate Streamer and add V1 stream-start

1. Remove duplicate Streamer class (manually_maintained/streaming.py)
   and import from aws_client.py instead. Both were identical
   SyncByteStream wrappers.

2. Emit stream-start event with generation_id at the beginning of V1
   streams, matching the standard Cohere V1 streaming chat format.
   Consumers relying on stream-start for state initialization will now
   receive it before text-generation events.

Updated test_v1_stream_wrapper_preserves_finish_reason to verify
stream-start is emitted first.

Author: fede-kamel

src/cohere/manually_maintained/streaming.py

@@ -11,8 +11,8 @@
 import requests
 from .client import Client, ClientEnvironment
 from .client_v2 import ClientV2
+from .aws_client import Streamer
 from .manually_maintained.lazy_oci_deps import lazy_oci
-from .manually_maintained.streaming import Streamer
 from httpx import URL, ByteStream
 
 
@@ -1038,16 +1038,22 @@ def _transform_v2_event(oci_event: typing.Dict[str, typing.Any]) -> typing.Itera
                 final_usage = _usage_from_oci(oci_event.get("usage"))
             yield _emit_v2_event(cohere_event)
 
-    def _transform_v1_event(oci_event: typing.Dict[str, typing.Any]) -> bytes:
-        nonlocal full_v1_text, final_v1_finish_reason
+    def _transform_v1_event(oci_event: typing.Dict[str, typing.Any]) -> typing.Iterator[bytes]:
+        nonlocal emitted_start, full_v1_text, final_v1_finish_reason
+        if not emitted_start:
+            yield _emit_v1_event({
+                "event_type": "stream-start",
+                "generation_id": generation_id,
+                "is_finished": False,
+            })
+            emitted_start = True
         event = transform_stream_event(endpoint, oci_event, is_v2=False)
         if isinstance(event, dict):
             if event.get("event_type") == "text-generation" and event.get("text"):
                 full_v1_text += typing.cast(str, event["text"])
             if "finishReason" in oci_event:
                 final_v1_finish_reason = oci_event.get("finishReason", final_v1_finish_reason)
-            return _emit_v1_event(event)
-        return b""
+            yield _emit_v1_event(event)
 
     def _process_line(line: str) -> typing.Iterator[bytes]:
         if not line.startswith("data: "):
@@ -1091,7 +1097,8 @@ def _process_line(line: str) -> typing.Iterator[bytes]:
                 for event_bytes in _transform_v2_event(oci_event):
                     yield event_bytes
             else:
-                yield _transform_v1_event(oci_event)
+                for event_bytes in _transform_v1_event(oci_event):
+                    yield event_bytes
         except Exception as exc:
             raise RuntimeError(f"OCI stream event transformation failed for endpoint '{endpoint}': {exc}") from exc
 

src/cohere/oci_client.py

@@ -800,9 +800,13 @@ def test_v1_stream_wrapper_preserves_finish_reason(self):
             for raw in transform_oci_stream_wrapper(iter(chunks), "chat", is_v2=False)
         ]
 
-        self.assertEqual(events[2]["event_type"], "stream-end")
-        self.assertEqual(events[2]["finish_reason"], "MAX_TOKENS")
-        self.assertEqual(events[2]["response"]["text"], "Hello world")
+        # First event should be stream-start with generation_id
+        self.assertEqual(events[0]["event_type"], "stream-start")
+        self.assertIn("generation_id", events[0])
+
+        self.assertEqual(events[3]["event_type"], "stream-end")
+        self.assertEqual(events[3]["finish_reason"], "MAX_TOKENS")
+        self.assertEqual(events[3]["response"]["text"], "Hello world")
 
     def test_transform_chat_request_tool_message_fields(self):
         """Test tool message fields are converted to OCI names."""