Add OpenAI integration with CLI for generating Mermaid diagrams.

[BaoLiqi]

#5 #4

I tested my fork with the model 'deepseek/deepseek-chat' on OpenRouter.
It seems that the generated diagram is heavily influenced by the contents of the README file, as my changes did not appear in the output.
To confirm this, I removed my local README file, tested again, and obtained the following diagram, which looks improved.

flowchart TD
    subgraph "API Gateway"
        NGINX["Nginx"]:::gateway
        click NGINX "nginx/api.conf"
        click NGINX "nginx/setup_nginx.sh"
    end

    subgraph "Application Server"
        APP["Application Server (main.py)"]:::app
        click APP "app/main.py"
    end

    subgraph "Router Layer"
        MODIFY["Modify Router (modify.py)"]:::router
        GENERATE["Generate Router (generate.py)"]:::router
        click MODIFY "app/routers/modify.py"
        click GENERATE "app/routers/generate.py"
    end

    subgraph "Service Layer"
        GITHUB["GitHub Service (github_service.py)"]:::service
        CLAUDE["Claude Service (claude_service.py)"]:::service
        OPENAI["OpenAI Service (openai_service.py)"]:::service
        click GITHUB "app/services/github_service.py"
        click CLAUDE "app/services/claude_service.py"
        click OPENAI "app/services/openai_service.py"
    end

    subgraph "Rate Limiter"
        LIMITER["Rate Limiter (limiter.py)"]:::limiter
        click LIMITER "app/core/limiter.py"
    end

    subgraph "External Services"
        GITHUB_API["GitHub API"]:::external
        CLAUDE_API["Claude API"]:::external
        OPENAI_API["OpenAI API"]:::external
    end

    subgraph "Deployment"
        DOCKER["Dockerfile"]:::deploy
        DEPLOY["deploy.sh"]:::deploy
        click DOCKER "Dockerfile"
        click DEPLOY "deploy.sh"
    end

    Client --> NGINX
    NGINX --> APP
    APP --> MODIFY
    APP --> GENERATE
    MODIFY --> GITHUB
    MODIFY --> CLAUDE
    MODIFY --> OPENAI
    GENERATE --> GITHUB
    GENERATE --> CLAUDE
    GENERATE --> OPENAI
    GITHUB --> GITHUB_API
    CLAUDE --> CLAUDE_API
    OPENAI --> OPENAI_API
    LIMITER --> APP
    DOCKER --> NGINX
    DEPLOY --> NGINX

    classDef gateway fill:#f9f,stroke:#333,stroke-width:2px
    classDef app fill:#bbf,stroke:#333,stroke-width:2px
    classDef router fill:#fbb,stroke:#333,stroke-width:2px
    classDef service fill:#bfb,stroke:#333,stroke-width:2px
    classDef limiter fill:#ffb,stroke:#333,stroke-width:2px
    classDef external fill:#ccc,stroke:#333,stroke-width:2px
    classDef deploy fill:#bff,stroke:#333,stroke-width:2px

Loading

[vercel]

@BaoLiqi is attempting to deploy a commit to the Ahmed Khaleel's projects Team on Vercel.

A member of the Team first needs to authorize it.

[winterjung]

I think it would be an idea to respect the local .gitignore contents as well.

[BaoLiqi]

Thank you for the feedback! 👍 I’ve gone ahead and updated the code to respect the local .gitignore contents.

[peng456]

I used the model 'deepseek/deepseek-chat' and found it unstable. I'm not sure why. After you removed the README file, did the results become stable?

[BaoLiqi]

I used the model 'deepseek/deepseek-chat' and found it unstable. I'm not sure why. After you removed the README file, did the results become stable?

No, it did not. Occasionally, deepseek-chat even produced incorrect Mermaid syntax. I employed this model to test my code because it is cost-effective. For better stability, you can opt for more reliable models.

[ahmedkhaleel2004]

this is great work but I've just changed a lot including optimizing the app to be used with the much cheaper, faster, smarter OpenAI o3-mini (from openrouter since I don't have the tier 3 key). I think a CLI would be a nice to have but probably has to be reimplemented all over again now