fix: support for chat.models override base configuration (#60)

Author: zhengxs2018

src/base/common/language-models/providers/TongyiProvider.ts

@@ -79,14 +79,25 @@ export class TongyiLanguageModelProvider implements ILanguageModelProvider {
 	private _newLLM(options: { [name: string]: any }) {
 		const config = this.configService;
 
+		const {
+			model,
+			temperature,
+			maxTokens,
+			topP,
+			apiKey = config.get<string>('tongyi.apiKey'),
+			enableSearch = config.get('tongyi.enableSearch', true),
+			clientOptions = {},
+		} = options;
+
 		return new ChatAlibabaTongyi({
-			alibabaApiKey: config.get<string>('tongyi.apiKey'),
+			alibabaApiKey: apiKey,
 			streaming: true,
-			model: this._resolveChatModel(options.model),
-			temperature: options.temperature,
-			maxTokens: options.maxTokens,
-			topP: options.topP,
-			enableSearch: config.get('tongyi.enableSearch'),
+			model: this._resolveChatModel(model),
+			temperature: temperature,
+			maxTokens: maxTokens,
+			topP: topP,
+			enableSearch: enableSearch,
+			...clientOptions,
 		});
 	}
 

src/base/common/language-models/providers/WenxinProvider.ts

@@ -79,14 +79,25 @@ export class WenxinLanguageModelProvider implements ILanguageModelProvider {
 	private _newLLM(options: { [name: string]: any }) {
 		const config = this.configService;
 
+		const {
+			model,
+			temperature,
+			penaltyScore,
+			topP,
+			apiKey = config.get<string>('qianfan.apiKey'),
+			secretKey = config.get('qianfan.secretKey'),
+			clientOptions = {},
+		} = options;
+
 		return new ChatBaiduWenxin({
-			baiduApiKey: config.get<string>('qianfan.apiKey'),
-			baiduSecretKey: config.get<string>('qianfan.secretKey'),
+			baiduApiKey: apiKey,
+			baiduSecretKey: secretKey,
 			streaming: true,
-			model: this._resolveChatModel(options.model),
-			temperature: options.temperature,
-			topP: options.topP,
-			penaltyScore: options.penaltyScore,
+			model: this._resolveChatModel(model),
+			temperature: temperature,
+			topP: topP,
+			penaltyScore: penaltyScore,
+			...clientOptions,
 		});
 	}
 

src/base/common/language-models/providers/anthropicProvider.ts

@@ -78,15 +78,25 @@ export class AnthropicLanguageModelProvider implements ILanguageModelProvider {
 
 	private _newLLM(options: { [name: string]: any }) {
 		const config = this.configService;
+		const {
+			model,
+			baseURL = config.get('anthropic.baseURL'),
+			apiKey = config.get('anthropic.apiKey'),
+			temperature,
+			maxTokens,
+			topP,
+			clientOptions = {}
+		} = options;
 
 		return new ChatAnthropic({
-			anthropicApiKey: config.get<string>('anthropic.apiKey'),
-			anthropicApiUrl: config.get<string>('anthropic.baseURL'),
+			anthropicApiKey: apiKey,
+			anthropicApiUrl: baseURL,
 			streaming: true,
-			model: this._resolveChatModel(options.model),
-			temperature: options.temperature,
-			maxTokens: options.maxTokens,
-			topP: options.topP,
+			temperature,
+			maxTokens,
+			topP,
+			model: this._resolveChatModel(model),
+			...clientOptions,
 		});
 	}
 

src/base/common/language-models/providers/openaiProvider.ts

@@ -178,24 +178,29 @@ export class OpenAILanguageModelProvider implements ILanguageModelProvider {
 	}
 
 	private _newLLM(options: { [name: string]: any }) {
-		const { baseURL, apiKey, project, organization } = options;
-		const configService = this.configService;
-
-		const apiType = configService.get<'openai' | 'azure'>('openai.apiType');
+		const config = this.configService;
+		const {
+			baseURL = config.get('openai.baseURL'),
+			apiKey = config.get('openai.apiKey'),
+			project = config.get('openai.project'),
+			organization = config.get('openai.organization'),
+			deployment = organization,
+			apiType = config.get<'openai' | 'azure'>('openai.apiType'),
+		} = options;
 
 		if (apiType === 'azure') {
 			return new AzureOpenAI({
-				baseURL: baseURL || configService.get('openai.baseURL'),
-				apiKey: apiKey || configService.get('openai.apiKey'),
-				deployment: organization || configService.get('openai.organization'),
+				baseURL: baseURL,
+				apiKey: apiKey,
+				deployment: deployment,
 			});
 		}
 
 		return new OpenAI({
-			baseURL: configService.get('openai.baseURL'),
-			project: project || configService.get('openai.project'),
-			apiKey: apiKey || configService.get('openai.apiKey'),
-			organization: organization || configService.get('openai.organization'),
+			baseURL: baseURL,
+			project: project,
+			apiKey: apiKey,
+			organization: organization,
 		});
 	}
 

src/editor/views/chat/continue/continueViewProvider.ts

@@ -235,17 +235,7 @@ export class ContinueViewProvider extends AbstractWebviewViewProvider implements
 
 	private async listModels(): Promise<IChatModelResource[]> {
 		const resources = this.configService.getConfig<IChatModelResource[]>('chat.models');
-		if (!resources) {
-			return [];
-		}
-
-		return resources.map(res => {
-			return {
-				title: res.title,
-				provider: res.provider,
-				model: res.model,
-			};
-		});
+		return resources ?? [];
 	}
 
 	readonly slashCommandsMap: Record<string, string> = {
@@ -324,16 +314,15 @@ export class ContinueViewProvider extends AbstractWebviewViewProvider implements
 			const models = await this.listModels();
 
 			const title = event.data.title;
-			const metadata = models.find(m => m.title === title);
+			const resource = models.find(m => m.title === title);
 
 			const completionOptions = event.data.completionOptions;
 
 			await this.lm.chat(
 				mapToChatMessages(event.data.messages),
 				{
+					...resource,
 					...completionOptions,
-					provider: metadata?.provider,
-					model: metadata?.model,
 				},
 				{
 					report(fragment) {