Release new client based on OpenAPI Specification Update - 2024-11-19 (#102)

* Improve open api spec

* The generated code based on OpenAPI spec doesn't have errors, and the dart example works.

* Add moderation example + update readme

* Bump version to 1.1.0 in pubspec.yaml

* Rename batch related functions for better generated code naming

* Add batch inference feature and example usage

* Update CHANGELOG

* Update list models section in readme

---------

Co-authored-by: gawi151 <[email protected]>
Co-authored-by: Marcin Gruchala <[email protected]>

Author: 3 people

CHANGELOG.md

@@ -1,5 +1,13 @@
 # Changelog
 
+## 1.1.0
+
+- Generate new client code based on [Mistral OAS](https://github.com/mistralai/platform-docs-public/commit/89abfa4879279981adde99e23c0cda8153da6eef#diff-d910ba2ef878f7db0223a966b81c8b3f3b65027bb39e4431bb05140171eece39R2722)
+- New feature: Moderation ([docs](https://docs.mistral.ai/capabilities/guardrailing/)). Check [example](example/mistralai_client_moderation_example.dart) for usage.
+- New feature: Batch Inference ([docs](https://docs.mistral.ai/capabilities/batch/)). Check [example](example/mistral_client_batch_inference_example.dart) for usage.
+- New function: `downloadFile` to download files from Mistral AI
+- BREAKING CHANGE: Changes to `ModelCard` to support `BaseModelCard` and `FTModelCard`
+
 ## 1.0.0
 
 - Support newest [Mistral OAS](https://github.com/mistralai/platform-docs-public/commit/2f2868013495c8b332c987be3c8c945204a544d4)

README.md

@@ -27,8 +27,8 @@ final client = MistralAIClient(apiKey: 'your api key here');
 
 ```dart
 final modelsResult = await client.listModels();
-final models = modelsResult.data.map((e) => e.id).toList();
-print(models.join(', '));
+final models = modelsResult.data?.map((e) => e.id).toList();
+print(models?.join(', '));
 ```
 
 ### Chat
@@ -102,6 +102,12 @@ Check examples:
 
 - [Agents example](example/agents_example.dart)
 
+### Moderation
+
+Check examples:
+
+- [Moderation example](example/mistralai_client_moderation_example.dart)
+
 ## Resources
 
 You can check the [official Mistral AI docs](https://docs.mistral.ai/).

example/mistral_client_batch_inference_example.dart

@@ -0,0 +1,49 @@
+// ignore_for_file: avoid_print
+
+import 'dart:io';
+
+import 'package:http/http.dart';
+import 'package:mistralai_client_dart/mistralai_client_dart.dart';
+
+import 'api_key.dart';
+
+void main() async {
+  final client = MistralAIClient(apiKey: mistralApiKey);
+
+  // Create a new file
+  final file = File('example/fine-tuning-file.jsonl');
+  final createdFile = await client.uploadFile(
+    request: [
+      MultipartFile.fromBytes(
+        'file',
+        file.readAsBytesSync(),
+        filename: file.path,
+      ),
+    ],
+  );
+  print('Created file: ${createdFile.id} \n');
+  print(createdFile);
+
+  final createdBatchJob = await client.createBatchJob(
+    request: BatchJobIn(
+      model: 'mistral-small-latest',
+      endpoint: ApiEndpoint.v1ChatCompletions,
+      inputFiles: [createdFile.id],
+      metadata: {'job_type': 'testing'},
+    ),
+  );
+  print('Created job: ${createdBatchJob.id}');
+  print(createdBatchJob);
+
+  final retrieveJob = await client.getBatchJob(jobId: createdBatchJob.id);
+  print('Retrieved job: ${retrieveJob.id}');
+  print(retrieveJob);
+
+  final listJobs = await client.getBatchJobs();
+  print('List of jobs:');
+  print(listJobs);
+
+  final cancelJob = await client.cancelBatchJob(jobId: createdBatchJob.id);
+  print('Cancelled job: ${cancelJob.id}');
+  print(cancelJob);
+}

example/mistralai_client_function_calling_dart_example.dart

@@ -89,8 +89,9 @@ void main() async {
 
   final messages = <dynamic>[
     const UserMessage(
-      content:
-          UserMessageContent.string("What's the status of my transaction?"),
+      content: UserMessageContent.string(
+        "What's the status of my transaction?",
+      ),
     ),
   ];
 

example/mistralai_client_moderation_example.dart

@@ -0,0 +1,56 @@
+// ignore_for_file: avoid_print
+
+import 'package:mistralai_client_dart/mistralai_client_dart.dart';
+
+import 'api_key.dart';
+
+Future<void> main() async {
+  final client = MistralAIClient(apiKey: mistralApiKey);
+
+  const financialMessage = 'Invest in our revolutionary cryptocurrency today'
+      ' and secure a guaranteed 300% return within the'
+      ' first week! This once-in-a-lifetime opportunity is'
+      ' backed by cutting-edge technology and an elite team of investors.'
+      ' Sign up now—no prior knowledge or experience required! Act fast,'
+      ' as spots are limited!';
+
+  // raw text moderation
+  const rawTextClassificationRequest = ClassificationRequest(
+    input: ClassificationRequestInput.string(
+      financialMessage,
+    ),
+    model: 'mistral-moderation-latest',
+  );
+
+  final rawTextModerationResponse =
+      await client.moderations(request: rawTextClassificationRequest);
+  final classificationObject = rawTextModerationResponse.results?.first;
+
+  print(classificationObject);
+
+  // chat text moderation
+  const conversationClassificationRequest = ChatClassificationRequest(
+    input: Input.array(
+      [
+        [
+          AssistantMessage(
+            content: AssistantMessageContent.string(
+              financialMessage,
+            ),
+          ),
+          UserMessage(
+            content: UserMessageContent.string(
+              'How can I Invest in your cryptocurrency?',
+            ),
+          ),
+        ],
+      ],
+    ),
+    model: 'mistral-moderation-latest',
+  );
+
+  final conversationModerationResponse =
+      await client.chatModerations(request: conversationClassificationRequest);
+
+  print(conversationModerationResponse);
+}