feat(rag): retry a failed LLM request before the first token lands — BOR_LLM_RETRIES/BOR_LLM_RETRY_DELAY with a live 'retrying' status

tests/unit/test_sse_events.py

@@ -4,7 +4,12 @@ from __future__ import annotations
 import json
 
 from app.api.chat import sse_event
-from app.schemas import ChatErrorEvent, ChatThinkingEvent, ChatToolEvent
+from app.schemas import (
+    ChatErrorEvent,
+    ChatRetryEvent,
+    ChatThinkingEvent,
+    ChatToolEvent,
+)
 
 
 def _payload(frame: str) -> dict:
@@ -100,3 +105,23 @@ def test_tool_event_shape_is_type_name_argument_only() -> None:
     dumped = ChatToolEvent(name="read_document", argument="S/p.md").model_dump()
     assert set(dumped.keys()) == {"type", "name", "argument"}
     assert dumped["type"] == "tool"  # default — call sites never spell it out
+
+
+def test_retry_frame_serializes_exactly() -> None:
+    """Phase 67: the ``retry`` frame is exactly ``{type: "retry",
+    attempt: int, max_attempts: int}`` — a transient status the client's
+    readSSE handler branches on ("Communication interrupted — retrying
+    (n of N)…"), never an error. ``attempt`` is the attempt the server is
+    about to try next (1-based); the contract the frontend branch and the
+    E2E suite key off, locked here ahead of the JS implementation."""
+    frame = sse_event(ChatRetryEvent(attempt=2, max_attempts=4).model_dump())
+    assert frame == 'data: {"type": "retry", "attempt": 2, "max_attempts": 4}\n\n'
+    assert _payload(frame) == {"type": "retry", "attempt": 2, "max_attempts": 4}
+
+
+def test_retry_event_shape_is_type_attempt_max_attempts_only() -> None:
+    dumped = ChatRetryEvent(attempt=2, max_attempts=4).model_dump()
+    assert set(dumped.keys()) == {"type", "attempt", "max_attempts"}
+    assert dumped["type"] == "retry"  # default — call sites never spell it out
+    assert isinstance(dumped["attempt"], int)
+    assert isinstance(dumped["max_attempts"], int)