feat(agent): 增强多模态链路与工具调用能力

backend/tests/integration/v1/agent/test_sse_flow_live.py

@@ -1,6 +1,8 @@
 from __future__ import annotations
 
+import base64
 import os
+from pathlib import Path
 from uuid import UUID, uuid4
 
 import httpx
@@ -12,6 +14,9 @@ from models.agent_chat_message import AgentChatMessage
 from models.agent_chat_session import AgentChatSession
 
 BASE_URL = os.getenv("AGENT_LIVE_BASE_URL", "http://localhost:5775")
+FIXTURE_IMAGE_PATH = (
+    Path(__file__).resolve().parents[3] / "fixtures" / "images" / "calendar_text_cn.png"
+)
 
 
 async def _live_access_token(client: httpx.AsyncClient) -> str:
@@ -108,6 +113,8 @@ async def test_agent_runs_events_history_live_with_image_input() -> None:
     if os.getenv("AGENT_LIVE_INTEGRATION") != "1":
         pytest.skip("set AGENT_LIVE_INTEGRATION=1 to run live integration test")
 
+    image_data = base64.b64encode(FIXTURE_IMAGE_PATH.read_bytes()).decode("ascii")
+
     async with httpx.AsyncClient(timeout=30.0) as client:
         token = await _live_access_token(client)
         headers = {"Authorization": f"Bearer {token}"}
@@ -128,7 +135,7 @@ async def test_agent_runs_events_history_live_with_image_input() -> None:
                             {"type": "text", "text": "请描述图片里的内容"},
                             {
                                 "type": "binary",
-                                "data": "aGVsbG8=",
+                                "data": image_data,
                                 "mimeType": "image/png",
                             },
                         ],
@@ -142,19 +149,20 @@ async def test_agent_runs_events_history_live_with_image_input() -> None:
         assert run_resp.status_code == 202
 
         events_url = f"{BASE_URL}/api/v1/agent/runs/{thread_id}/events"
-        sse_resp = await client.get(
-            events_url,
-            headers=headers,
-            params={"idle_limit": 150},
-            timeout=60.0,
-        )
-        assert sse_resp.status_code == 200
-        assert sse_resp.headers.get("content-type", "").startswith("text/event-stream")
-        event_names = [
-            line.split(":", 1)[1].strip()
-            for line in sse_resp.text.splitlines()
-            if line.startswith("event:")
-        ]
+        event_names: list[str] = []
+        async with client.stream(
+            "GET", events_url, headers=headers, timeout=90.0
+        ) as sse_resp:
+            assert sse_resp.status_code == 200
+            assert sse_resp.headers.get("content-type", "").startswith(
+                "text/event-stream"
+            )
+            async for line in sse_resp.aiter_lines():
+                if line.startswith("event:"):
+                    event_name = line.split(":", 1)[1].strip()
+                    event_names.append(event_name)
+                    if event_name in {"RUN_FINISHED", "RUN_ERROR"}:
+                        break
 
         assert "RUN_STARTED" in event_names
         assert "RUN_FINISHED" in event_names or "RUN_ERROR" in event_names
@@ -194,7 +202,14 @@ async def test_agent_runs_events_history_live_with_image_input() -> None:
         )
         all_messages = list(rows.scalars().all())
         assert all_messages
-        user_rows = [row for row in all_messages if str(row.role) == "user"]
+        user_rows = [
+            row
+            for row in all_messages
+            if (
+                getattr(row.role, "value", row.role) == "user"
+                or str(getattr(row.role, "value", row.role)) == "user"
+            )
+        ]
         assert user_rows
         metadata = user_rows[0].metadata_json or {}
         attachments = metadata.get("attachments")