feat: Add tool calling support in ollama_dart

[davidmigloz]

Tool calling allows a model to respond to a given prompt by generating output that matches a user-defined schema, that you can then use to call the tools in your code and return the result back to the model to complete the conversation.

Notes:

• Tool calling requires Ollama 0.2.8 or newer.
• Streaming tool calls is not supported at the moment.
• Not all models support tool calls. Check the Ollama catalogue for models that have the Tools tag (e.g. llama3.1).

const tool = Tool(
  function: ToolFunction(
    name: 'get_current_weather',
    description: 'Get the current weather in a given location',
    parameters: {
      'type': 'object',
      'properties': {
        'location': {
          'type': 'string',
          'description': 'The city and country, e.g. San Francisco, US',
        },
        'unit': {
          'type': 'string',
          'description': 'The unit of temperature to return',
          'enum': ['celsius', 'fahrenheit'],
        },
      },
      'required': ['location'],
    },
  ),
);

const userMsg = Message(
  role: MessageRole.user,
  content: 'What’s the weather like in Barcelona in celsius?',
);

final res1 = await client.generateChatCompletion(
  request: GenerateChatCompletionRequest(
    model: 'llama3.1',
    messages: [userMsg],
    tools: [tool],
  ),
);

print(res1.message.toolCalls);
// [
//   ToolCall(
//     function:
//       ToolCallFunction(
//         name: get_current_weather,
//         arguments: {
//           location: Barcelona, ES,
//           unit: celsius
//         }
//       )
//   )
// ]

// Call your tool here. For this example, we'll just mock the response.
const toolResult = '{"location": "Barcelona, ES", "temperature": 20, "unit": "celsius"}';

// Submit the response of the tool call to the model
final res2 = await client.generateChatCompletion(
  request: GenerateChatCompletionRequest(
    model: 'llama3.1',
    messages: [
      userMsg,
      res1.message,
      Message(
        role: MessageRole.tool,
        content: toolResult,
      ),
    ],
  ),
);
print(res2.message.content);
// The current weather in Barcelona is 20°C.

[davidmigloz]

(@HavenDV tagging you in case you want to regenerate your client)

[HavenDV]

Thank you very much, I'll implement this.
A little later, I planned to implement a periodic poll of the specification and create a new PR/auto-merge/release a new preview version based on this, as I do with the OpenAI specification, but for now your ping helps