test: Add system prompts to ollama tests (#3415)

Author: sjrl

integrations/ollama/tests/test_chat_generator.py

@@ -1509,7 +1509,10 @@ def test_live_run_with_repeated_tool_calls(self, tools, streaming_callback):
         component = OllamaChatGenerator(model="qwen3:0.6b", tools=tools, streaming_callback=streaming_callback)
         tool_invoker = ToolInvoker(tools=tools)
 
-        messages = [ChatMessage.from_user("What is the weather in Paris and London?")]
+        messages = [
+            ChatMessage.from_system("Use the tools to answer the question."),
+            ChatMessage.from_user("What is the weather in Paris and London?"),
+        ]
         response = component.run(messages)
 
         assert len(response["replies"]) == 1
@@ -1538,9 +1541,12 @@ def test_live_run_with_tools_and_format(self, tools):
             "required": ["capital", "population"],
         }
         chat_generator = OllamaChatGenerator(model="qwen3:0.6b", tools=tools, response_format=response_format)
-        message = ChatMessage.from_user("What's the weather in Paris?")
+        messages = [
+            ChatMessage.from_system("Use the tools to answer the question."),
+            ChatMessage.from_user("What's the weather in Paris?"),
+        ]
 
-        result = chat_generator.run([message])
+        result = chat_generator.run(messages)
 
         assert isinstance(result, dict)
         assert isinstance(result["replies"], list)
@@ -1569,8 +1575,11 @@ def population(city: Annotated[str, "The city to get the population for"]) -> st
         mixed_tools = [tools[0], population_toolset]
         component = OllamaChatGenerator(model="qwen3:0.6b", tools=mixed_tools, streaming_callback=streaming_callback)
 
-        message = ChatMessage.from_user("What is the weather and population in Paris?")
-        response = component.run([message])
+        messages = [
+            ChatMessage.from_system("Use the tools to answer the question."),
+            ChatMessage.from_user("What is the weather and population in Paris?"),
+        ]
+        response = component.run(messages)
 
         assert len(response["replies"]) == 1
         message = response["replies"][0]
@@ -1619,7 +1628,10 @@ async def async_callback(chunk: StreamingChunk) -> None:
     async def test_run_async_with_tools(self, tools):
         """Test async with tool calls."""
         chat_generator = OllamaChatGenerator(model="qwen3:0.6b", tools=tools)
-        messages = [ChatMessage.from_user("What's the weather in Paris?")]
+        messages = [
+            ChatMessage.from_system("Use the tools to answer the question."),
+            ChatMessage.from_user("What's the weather in Paris?"),
+        ]
 
         response = await chat_generator.run_async(messages)
 
@@ -1652,7 +1664,10 @@ async def callback(_chunk: StreamingChunk) -> None:
             chunks_received = True
 
         chat_generator = OllamaChatGenerator(model="qwen3:0.6b", tools=tools, streaming_callback=callback)
-        messages = [ChatMessage.from_user("What's the weather in Berlin?")]
+        messages = [
+            ChatMessage.from_system("Use the tools to answer the question."),
+            ChatMessage.from_user("What's the weather in Berlin?"),
+        ]
 
         response = await chat_generator.run_async(messages)