Issue with bolt.diy release 0.0.7

[Ikaroyo]

There was an error processing your request: Custom error: Invalid JSON response

using bolt.diy v0.0.7
it seems new bolt slice messages.

still work with v0.0.6

[Ikaroyo]

fixed: by changing the code in

bolt.dit\app\lib\modules\llm\providers\openai-like.ts

to:

import { BaseProvider, getOpenAILikeModel } from '~/lib/modules/llm/base-provider';
import type { ModelInfo } from '~/lib/modules/llm/types';
import type { IProviderSetting } from '~/types/model';
import type { LanguageModelV1 } from 'ai';
import { createOpenAI } from '@ai-sdk/openai';
import { logger } from '~/utils/logger';

export default class OpenAILikeProvider extends BaseProvider {
  name = 'OpenAILike';
  getApiKeyLink = undefined;

  config = {
    baseUrlKey: 'OPENAI_LIKE_API_BASE_URL',
    apiTokenKey: 'OPENAI_LIKE_API_KEY',
  };

  staticModels: ModelInfo[] = [];

  async getDynamicModels(
    apiKeys?: Record<string, string>,
    settings?: IProviderSetting,
    serverEnv: Record<string, string> = {},
  ): Promise<ModelInfo[]> {
    const { baseUrl, apiKey } = this.getProviderBaseUrlAndKey({
      apiKeys,
      providerSettings: settings,
      serverEnv,
      defaultBaseUrlKey: 'OPENAI_LIKE_API_BASE_URL',
      defaultApiTokenKey: 'OPENAI_LIKE_API_KEY',
    });

    if (!baseUrl || !apiKey) {
      return [];
    }

    const response = await fetch(`${baseUrl}/models`, {
      headers: {
        Authorization: `Bearer ${apiKey}`,
      },
    });

    const res = (await response.json()) as any;

    return res.data.map((model: any) => ({
      name: model.id,
      label: model.id,
      provider: this.name,
      maxTokenAllowed: 8000,
    }));
  }

  getModelInstance(options: {
    model: string;
    serverEnv: Env;
    apiKeys?: Record<string, string>;
    providerSettings?: Record<string, IProviderSetting>;
  }): LanguageModelV1 {
    const { model, serverEnv, apiKeys, providerSettings } = options;

    const { baseUrl, apiKey } = this.getProviderBaseUrlAndKey({
      apiKeys,
      providerSettings: providerSettings?.[this.name],
      serverEnv: serverEnv as any,
      defaultBaseUrlKey: 'OPENAI_LIKE_API_BASE_URL',
      defaultApiTokenKey: 'OPENAI_LIKE_API_KEY',
    });

    if (!baseUrl || !apiKey) {
      throw new Error(`Missing configuration for ${this.name} provider`);
    }
    
    const isClaudeModel = model.toLowerCase().includes('claude');
    
    const openai = createOpenAI({
      baseURL: baseUrl,
      apiKey,
      fetch: async (url, options) => {
        const response = await fetch(url, options);
        
        // Only process JSON responses
        const contentType = response.headers.get('content-type');
        if (!contentType || !contentType.includes('application/json')) {
          return response;
        }
        
        // Clone the response to avoid consuming it
        const clonedResponse = response.clone();
        
        try {
          // Get the response body
          const originalBody = await clonedResponse.json() as { choices?: Array<any> };
          
          // Check if this is a Claude response that needs fixing
          if (isClaudeModel && 
              originalBody.choices && 
              Array.isArray(originalBody.choices) &&
              originalBody.choices.length > 0) {
            
            // Add missing index field to each choice
            const transformedBody = {
              ...originalBody,
              choices: originalBody.choices.map((choice, index) => ({
                ...choice,
                index
              }))
            };
            
            logger.debug('OpenAILike: Transformed Claude response for compatibility');
            
            // Return a new response with the transformed body
            return new Response(JSON.stringify(transformedBody), {
              status: response.status,
              statusText: response.statusText,
              headers: response.headers
            });
          }
        } catch (error) {
          logger.error('Error transforming response:', error);
        }
        
        // Return the original response if no transformation needed or if an error occurred
        return response;
      }
    });

    return openai(model);
  }
}

[ericc-ch]

Seems like GitHub Copilot API isn't responding with 100% OpenAI compatible response format.

I think the best way to do it for now is I'll add a few test cases and modify the response where needed.

[ericc-ch]

Hi just wanted to let you know I'll stop working on this for now

Please look at https://github.com/ericc-ch/copilot-api?tab=readme-ov-file#warning

[ericc-ch]

Hey, I want sincerely apologize for my knee-jerk reaction of just immediately archiving the repo. I realize that was an overreaction and not the best action to make. I should have investigated the rate limit issue further and waited for more information from GitHub before making the repository inaccessible. The repository is now unarchived and open again.

For now, other people seem to be getting the same rate limit issue. Even the one using official LM API:

• VS Code LM API - New Models (Claude 3.7, o3-mini, o1) don't create/edit files properly RooCodeInc/Roo-Code#1187
• VS Code LM API - Copilot - Claude 3.7 Sonnet - "Model is not supported for this request." RooCodeInc/Roo-Code#1203
• https://github.com/jjleng/copilot-more?tab=readme-ov-file#must-read

I still will not be working on this for a while until things become clearer though