feat: Add DeepSeek web (cookie-based) account support

[Zepp-Hanzj]

feat: Add DeepSeek web (cookie-based) account support

Add support for DeepSeek accounts authenticated via browser cookies and tokens, enabling free API access through the DeepSeek web interface.

What's New

• DeepSeek Cookie Auth: Create accounts using browser Authorization token + Cookie
• PoW Challenge Solver: Keccak-based proof-of-work solver (DeepSeekHashV1) for anti-bot bypass
• SSE Stream Converter: Translates DeepSeek's SSE format to OpenAI Chat Completions format
• OpenAI-Compatible Gateway: DeepSeek requests go through the existing /v1/chat/completions endpoint
• Multi-Model Support: Supports deepseek-chat, deepseek-v4-pro, deepseek-coder, deepseek-reasoner, deepseek-r1
• Frontend UI: Account creation modal with Cookie/Token input, platform badge, and color theme

Architecture

Client → /v1/chat/completions → OpenAI Gateway → DeepSeek Handler
    → Create Session → Solve PoW → Call Completion → Convert SSE → Response

Files Changed

Category	Files	Summary
DeepSeek core	5 new files	PoW solver, HTTP client, SSE converter, constants, gateway handler
Backend routing	9 modified	Platform constants, gateway routing, account/group validation, service layer
Frontend	8 modified	Account modals, types, colors, i18n, badge, groups view
Total	22 files	+1343 / -29 lines

Testing

All models verified through Sub2API proxy in both stream and non-stream modes:

Model	Non-stream	Stream
deepseek-chat	✅	✅
deepseek-v4-pro	✅	✅
deepseek-coder	✅	✅
deepseek-reasoner	✅	✅
deepseek-r1	✅	✅

How to Use

1. Go to Admin → Accounts → Create Account
2. Select DeepSeek as platform type
3. Enter Authorization token (from browser DevTools → Network → chat.deepseek.com request headers)
4. Enter Cookie (from browser DevTools → Application → Cookies)
5. Optionally specify a Model ID (defaults to deepseek-chat)
6. Use the standard /v1/chat/completions endpoint with the DeepSeek account API key

Notes

• DeepSeek PoW challenges are solved client-side to bypass rate limiting
• The SSE stream converter handles DeepSeek's JSON Patch format and converts to OpenAI streaming format
• Browser request headers are mimicked to avoid WAF (429) blocks
• Token refresh requires manual re-entry (same as re-authenticating in browser)

[github-actions]

All contributors have signed the CLA. ✅
_{Posted by the CLA Assistant Lite bot.}

[Zepp-Hanzj]

Thank you for your contribution! Before we can merge this PR, we need you to sign our Contributor License Agreement (CLA).

To sign, please reply with the following comment:

I have read the CLA Document and I hereby sign the CLA

You only need to sign once — it will be valid for all your future contributions to this project.

I have read the CLA Document and I hereby sign the CLA

You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.

I have read the CLA Document and I hereby sign the CLA

[Zepp-Hanzj]

recheck

[Zepp-Hanzj]

I have read the CLA Document and I hereby sign the CLA

[Zepp-Hanzj]

Update: DeepSeek Web Search Support

Added search_enabled: true to the DeepSeek completion payload. This enables the model to perform web searches for real-time information, matching the behavior of the DeepSeek web interface.

Before: "我没有联网获取实时数据的能力"
After: Real-time weather, news, etc. with citations `[citation:N]

[Wei-Shaw]

不考虑合并。