feat: Add tool calling support in ollama_dart (#504)

packages/langchain_ollama/lib/src/chat_models/chat_ollama/mappers.dart

@@ -101,7 +101,7 @@ extension ChatResultMapper on GenerateChatCompletionResponse {
     return ChatResult(
       id: id,
       output: AIChatMessage(
-        content: message?.content ?? '',
+        content: message.content,
       ),
       finishReason: _mapFinishReason(doneReason),
       metadata: {

packages/ollama_dart/README.md

@@ -17,17 +17,24 @@ Unofficial Dart client for [Ollama](https://ollama.ai/) API.
 **Supported endpoints:**
 
 - Completions (with streaming support)
-- Chat completions
+- Chat completions (with streaming and tool calling support)
 - Embeddings
 - Models
 - Blobs
+- Version
 
 ## Table of contents
 
 - [Usage](#usage)
   * [Completions](#completions)
+    + [Generate completion](#generate-completion)
+    + [Stream completion](#stream-completion)
   * [Chat completions](#chat-completions)
+    + [Generate chat completion](#generate-chat-completion)
+    + [Stream chat completion](#stream-chat-completion)
+    + [Tool calling](#tool-calling)
   * [Embeddings](#embeddings)
+    + [Generate embedding](#generate-embedding)
   * [Models](#models)
     + [Create model](#create-model)
     + [List models](#list-models)
@@ -54,7 +61,7 @@ Refer to the [documentation](https://github.com/jmorganca/ollama/blob/main/docs/
 
 Given a prompt, the model will generate a response.
 
-**Generate completion:**
+#### Generate completion
 
 ```dart
 final generated = await client.generateCompletion(
@@ -67,7 +74,7 @@ print(generated.response);
 // The sky appears blue because of a phenomenon called Rayleigh scattering...
 ```
 
-**Stream completion:**
+#### Stream completion
 
 ```dart
 final stream = client.generateCompletionStream(
@@ -88,7 +95,7 @@ print(text);
 
 Given a prompt, the model will generate a response in a chat format.
 
-**Generate chat completion:**
+#### Generate chat completion
 
 ```dart
 final res = await client.generateChatCompletion(
@@ -111,7 +118,7 @@ print(res);
 // Message(role: MessageRole.assistant, content: 123456789)
 ```
 
-**Stream chat completion:**
+#### Stream chat completion
 
 ```dart
 final stream = client.generateChatCompletionStream(
@@ -139,11 +146,91 @@ print(text);
 // 123456789
 ```
 
+#### Tool calling
+
+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`](https://ollama.com/library/llama3.1)). 
+
+```dart
+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.
+```
+
 ### Embeddings
 
 Given a prompt, the model will generate an embedding representing the prompt.
 
-**Generate embedding:**
+#### Generate embedding
 
 ```dart
 final generated = await client.generateEmbedding(

packages/ollama_dart/example/ollama_dart_example.dart

@@ -11,6 +11,7 @@ Future<void> main() async {
   await _generateChatCompletion(client);
   await _generateChatCompletionWithHistory(client);
   await _generateChatCompletionStream(client);
+  await _generateChatToolCalling(client);
 
   // Embeddings
   await _generateEmbedding(client);
@@ -86,7 +87,7 @@ Future<void> _generateChatCompletion(final OllamaClient client) async {
       ],
     ),
   );
-  print(generated.message?.content);
+  print(generated.message.content);
 }
 
 Future<void> _generateChatCompletionWithHistory(
@@ -111,7 +112,7 @@ Future<void> _generateChatCompletionWithHistory(
       ],
     ),
   );
-  print(generated.message?.content);
+  print(generated.message.content);
 }
 
 Future<void> _generateChatCompletionStream(final OllamaClient client) async {
@@ -132,11 +133,84 @@ Future<void> _generateChatCompletionStream(final OllamaClient client) async {
   );
   String text = '';
   await for (final res in stream) {
-    text += (res.message?.content ?? '').trim();
+    text += res.message.content.trim();
   }
   print(text);
 }
 
+Future<void> _generateChatToolCalling(final OllamaClient client) async {
+  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: const GenerateChatCompletionRequest(
+      model: 'llama3.1',
+      messages: [userMsg],
+      tools: [tool],
+      keepAlive: 1,
+    ),
+  );
+
+  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,
+        const Message(
+          role: MessageRole.tool,
+          content: toolResult,
+        ),
+      ],
+    ),
+  );
+  print(res2.message.content);
+  // The current weather in Barcelona is 20°C.
+}
+
 Future<void> _generateEmbedding(final OllamaClient client) async {
   final generated = await client.generateEmbedding(
     request: const GenerateEmbeddingRequest(

packages/ollama_dart/lib/src/generated/schema/generate_chat_completion_request.dart

@@ -47,6 +47,9 @@ class GenerateChatCompletionRequest with _$GenerateChatCompletionRequest {
     /// - If set to 0, the model will be unloaded immediately once finished.
     /// - If not set, the model will stay loaded for 5 minutes by default
     @JsonKey(name: 'keep_alive', includeIfNull: false) int? keepAlive,
+
+    /// A list of tools the model may call.
+    @JsonKey(includeIfNull: false) List<Tool>? tools,
   }) = _GenerateChatCompletionRequest;
 
   /// Object construction from a JSON representation
@@ -60,7 +63,8 @@ class GenerateChatCompletionRequest with _$GenerateChatCompletionRequest {
     'format',
     'options',
     'stream',
-    'keep_alive'
+    'keep_alive',
+    'tools'
   ];
 
   /// Perform validations on the schema property values
@@ -77,6 +81,7 @@ class GenerateChatCompletionRequest with _$GenerateChatCompletionRequest {
       'options': options,
       'stream': stream,
       'keep_alive': keepAlive,
+      'tools': tools,
     };
   }
 }

packages/ollama_dart/lib/src/generated/schema/generate_chat_completion_response.dart

@@ -16,18 +16,18 @@ class GenerateChatCompletionResponse with _$GenerateChatCompletionResponse {
   /// Factory constructor for GenerateChatCompletionResponse
   const factory GenerateChatCompletionResponse({
     /// A message in the chat endpoint
-    @JsonKey(includeIfNull: false) Message? message,
+    required Message message,
 
     /// The model name.
     ///
     /// Model names follow a `model:tag` format. Some examples are `orca-mini:3b-q4_1` and `llama3:70b`. The tag is optional and, if not provided, will default to `latest`. The tag is used to identify a specific version.
-    @JsonKey(includeIfNull: false) String? model,
+    required String model,
 
     /// Date on which a model was created.
-    @JsonKey(name: 'created_at', includeIfNull: false) String? createdAt,
+    @JsonKey(name: 'created_at') required String createdAt,
 
     /// Whether the response has completed.
-    @JsonKey(includeIfNull: false) bool? done,
+    required bool done,
 
     /// Reason why the model is done generating a response.
     @JsonKey(

packages/ollama_dart/lib/src/generated/schema/message.dart

@@ -23,14 +23,23 @@ class Message with _$Message {
 
     /// (optional) a list of Base64-encoded images to include in the message (for multimodal models such as llava)
     @JsonKey(includeIfNull: false) List<String>? images,
+
+    /// A list of tools the model wants to call.
+    @JsonKey(name: 'tool_calls', includeIfNull: false)
+    List<ToolCall>? toolCalls,
   }) = _Message;
 
   /// Object construction from a JSON representation
   factory Message.fromJson(Map<String, dynamic> json) =>
       _$MessageFromJson(json);
 
   /// List of all property names of schema
-  static const List<String> propertyNames = ['role', 'content', 'images'];
+  static const List<String> propertyNames = [
+    'role',
+    'content',
+    'images',
+    'tool_calls'
+  ];
 
   /// Perform validations on the schema property values
   String? validateSchema() {
@@ -43,6 +52,7 @@ class Message with _$Message {
       'role': role,
       'content': content,
       'images': images,
+      'tool_calls': toolCalls,
     };
   }
 }
@@ -59,4 +69,6 @@ enum MessageRole {
   user,
   @JsonValue('assistant')
   assistant,
+  @JsonValue('tool')
+  tool,
 }

packages/ollama_dart/lib/src/generated/schema/schema.dart

@@ -19,6 +19,12 @@ part 'generate_chat_completion_request.dart';
 part 'generate_chat_completion_response.dart';
 part 'done_reason.dart';
 part 'message.dart';
+part 'tool.dart';
+part 'tool_function.dart';
+part 'tool_function_params.dart';
+part 'tool_call.dart';
+part 'tool_call_function.dart';
+part 'tool_call_function_args.dart';
 part 'generate_embedding_request.dart';
 part 'generate_embedding_response.dart';
 part 'create_model_request.dart';