feat(engine): add DeepSeekEngine (#446)

Add DeepSeekEngine to support DeepSeek API.  This
includes a new DeepSeekConfig interface and updates to
the engine selection logic.

feat(README.md, src/commands/config.ts): Add DeepSeek support

Adds support for the DeepSeek AI provider.  Updates the README,
config validation, and model list to include DeepSeek.  This
allows users to utilize DeepSeek models with the OpenCommit
tool.

fix(deepseek.ts): update DeepSeek API base URL to include version number v1

refactor(deepseek.ts): improve DeepSeekEngine constructor

The DeepSeekEngine constructor is refactored to use the
spread syntax for better readability and maintainability
when merging config parameters.  The baseURL is now
explicitly set within the constructor.

fix(README.md): remove Groq from the list of supported AI providers

refactor(deepseek.ts): rename interface DeepseekConfig to DeepSeekEngineeekConfig and fix typo

Revert "refactor(deepseek.ts): rename interface DeepseekConfig to DeepSeekEngineeekConfig and fix typo"

This reverts commit f492367d3885fa97cd685feca889f93d6c465b2f.

refactor(deepseek.ts): Rename DeepseekConfig to DeepSeekConfig for consistency

✨ feat(engine): add DeepSeekEngine to support DeepSeek API

♻️ refactor(engine): improve OpenAiEngine and create a
    new DeepSeekEngine class to handle DeepSeek API
    requests.  The DeepSeekEngine class inherits from
    OpenAiEngine and overrides the
    generateCommitMessage method to use the DeepSeek
    API.  This change improves code organization and
    maintainability.

🐛 Fix: Correct DeepSeekEngine import and class name

The import path and class name for DeepSeekEngine were
incorrect, causing a runtime error. This commit corrects
the import path and class name to `DeepseekEngine` to
resolve the issue.

Revert "🐛 Fix: Correct DeepSeekEngine import and class name"

This reverts commit 738fd36c434d9df9c3a458b1e8230c974bd2a76e.

🐛 Fix: Correct DeepSeekEngine import and class name

The import path and class name for DeepSeekEngine were corrected to match the actual file and class name.  This fixes a runtime error.

Restore ./out directory to master state

src/commands/config.ts

@@ -128,6 +128,10 @@ export const MODEL_LIST = {
     'mistral-embed',
     'mistral-moderation-2411',
     'mistral-moderation-latest',
+  ],
+  deepseek : [
+	  'deepseek-chat',
+	  'deepseek-reasoner',
   ]
 };
 
@@ -145,6 +149,8 @@ const getDefaultModel = (provider: string | undefined): string => {
       return MODEL_LIST.groq[0];
     case 'mistral':
       return MODEL_LIST.mistral[0];
+    case 'deepseek':
+	return MODEL_LIST.deepseek[0];
     default:
       return MODEL_LIST.openai[0];
   }
@@ -184,7 +190,7 @@ export const configValidators = {
     validateConfig(
       'OCO_API_KEY',
       value,
-      'You need to provide the OCO_API_KEY when OCO_AI_PROVIDER set to "openai" (default) or "ollama" or "mlx" or "azure" or "gemini" or "flowise" or "anthropic". Run `oco config set OCO_API_KEY=your_key OCO_AI_PROVIDER=openai`'
+      'You need to provide the OCO_API_KEY when OCO_AI_PROVIDER set to "openai" (default) or "ollama" or "mlx" or "azure" or "gemini" or "flowise" or "anthropic" or "deepseek". Run `oco config set OCO_API_KEY=your_key OCO_AI_PROVIDER=openai`'
     );
 
     return value;
@@ -307,9 +313,10 @@ export const configValidators = {
         'azure',
         'test',
         'flowise',
-        'groq'
+        'groq',
+	'deepseek'
       ].includes(value) || value.startsWith('ollama'),
-      `${value} is not supported yet, use 'ollama', 'mlx', 'anthropic', 'azure', 'gemini', 'flowise', 'mistral' or 'openai' (default)`
+      `${value} is not supported yet, use 'ollama', 'mlx', 'anthropic', 'azure', 'gemini', 'flowise', 'mistral', 'deepseek' or 'openai' (default)`
     );
 
     return value;
@@ -356,7 +363,8 @@ export enum OCO_AI_PROVIDER_ENUM {
   FLOWISE = 'flowise',
   GROQ = 'groq',
   MISTRAL = 'mistral',
-  MLX = 'mlx'
+  MLX = 'mlx',
+  DEEPSEEK = 'deepseek'
 }
 
 export type ConfigType = {

src/engine/deepseek.ts

@@ -0,0 +1,60 @@
+import axios from 'axios';
+import { OpenAI } from 'openai';
+import { GenerateCommitMessageErrorEnum } from '../generateCommitMessageFromGitDiff';
+import { tokenCount } from '../utils/tokenCount';
+import { OpenAiEngine, OpenAiConfig } from './openAI';
+
+export interface DeepseekConfig extends OpenAiConfig {}
+
+export class DeepseekEngine extends OpenAiEngine {
+  constructor(config: DeepseekConfig) {
+    // Call OpenAIEngine constructor with forced Deepseek baseURL
+    super({
+      ...config,
+      baseURL: 'https://api.deepseek.com/v1'
+    });
+  }
+
+  // Identical method from OpenAiEngine, re-implemented here
+  public generateCommitMessage = async (
+    messages: Array<OpenAI.Chat.Completions.ChatCompletionMessageParam>
+  ): Promise<string | null> => {
+    const params = {
+      model: this.config.model,
+      messages,
+      temperature: 0,
+      top_p: 0.1,
+      max_tokens: this.config.maxTokensOutput
+    };
+
+    try {
+      const REQUEST_TOKENS = messages
+        .map((msg) => tokenCount(msg.content as string) + 4)
+        .reduce((a, b) => a + b, 0);
+
+      if (
+        REQUEST_TOKENS >
+        this.config.maxTokensInput - this.config.maxTokensOutput
+      )
+        throw new Error(GenerateCommitMessageErrorEnum.tooMuchTokens);
+
+      const completion = await this.client.chat.completions.create(params);
+
+      const message = completion.choices[0].message;
+
+      return message?.content;
+    } catch (error) {
+      const err = error as Error;
+      if (
+        axios.isAxiosError<{ error?: { message: string } }>(error) &&
+        error.response?.status === 401
+      ) {
+        const openAiError = error.response.data.error;
+
+        if (openAiError) throw new Error(openAiError.message);
+      }
+
+      throw err;
+    }
+  };
+}

src/utils/engine.ts

@@ -10,6 +10,7 @@ import { MistralAiEngine } from '../engine/mistral';
 import { TestAi, TestMockType } from '../engine/testAi';
 import { GroqEngine } from '../engine/groq';
 import { MLXEngine } from '../engine/mlx';
+import { DeepseekEngine } from '../engine/deepseek';
 
 export function getEngine(): AiEngine {
   const config = getConfig();
@@ -51,6 +52,9 @@ export function getEngine(): AiEngine {
     case OCO_AI_PROVIDER_ENUM.MLX:
       return new MLXEngine(DEFAULT_CONFIG);
 
+    case OCO_AI_PROVIDER_ENUM.DEEPSEEK:
+      return new DeepseekEngine(DEFAULT_CONFIG);
+
     default:
       return new OpenAiEngine(DEFAULT_CONFIG);
   }