Skip to content

API_Endpoints.md

Latest commit

 

History

History
174 lines (131 loc) · 3.68 KB

API_Endpoints.md

File metadata and controls

174 lines (131 loc) · 3.68 KB

✅ 1. WebSocket JSON Payloads (/call/:id)

🎙️ User → Backend (via WebSocket)

Text message:

{
  "type": "text",
  "content": "I've had chest pain for two days."
}

🤖 Backend → Frontend

Response from AI model (to speak via avatar):

{
  "type": "ai",
  "content": "How long have you had the chest pain?",
  "meta": {
    "source": "model 4b",
    "timestamp": "2025-07-04T08:44:32.127Z"
  }
}

✅ 2. Redis Conversation Log Format

Each session is stored as a list under chat:<session_id>, where each item is:

{
  "role": "user" | "ai",
  "content": "The message content",
  "timestamp": "2025-07-04T08:44:32.127Z",
  "s3_doc_url": null | "https://s3.amazonaws.com/bucket/file.pdf"
}

Optional document links (when user uploads a file).

✅ 3. Document Upload API (REST)

POST /upload Frontend sends a document (PDF, image, etc.) to backend:

{
  "session_id": "abc123",
  "user_id": "user001",
  "filename": "blood_test.pdf",
  "content_base64": "<base64-encoded-content>"
}
📤 Response from backend:
{
  "status": "success",
  "s3_url": "https://firebase.storage/bucket/abc123/blood_test.pdf",
  "message": "Document uploaded successfully"
}

Backend saves this document URL into Redis using:

{
  "role": "user",
  "content": "[Document Uploaded: blood_test.pdf]",
  "timestamp": "2025-07-04T08:45:00Z",
  "s3_doc_url": "https://firebase.storage/bucket/abc123/blood_test.pdf"
}

✅ 4. Final Gemini 27B Report Generation Input

When call ends, backend fetches Redis conversation and sends to Gemini 27B:

{
  "conversation": [
    { "role": "user", "content": "I feel dizzy", "timestamp": "..." },
    { "role": "ai", "content": "Have you eaten today?", "timestamp": "..." },
    ...
  ],
  "documents": [
    {
      "name": "blood_test.pdf",
      "url": "https://firebase.storage/bucket/abc123/blood_test.pdf"
    }
  ]
}

🧠 Output from Gemini 27B (Markdown Report)

## Patient Summary

- **Symptoms**: Dizziness, chest pain
- **Onset**: 2 days ago
- **Additional Notes**: Patient uploaded recent blood test.

## Suggested Next Steps

- Consult cardiologist
- Perform ECG and blood panel

✅ 5. Markdown-to-PDF Conversion

No specific JSON needed unless you expose an endpoint like:

{
  "session_id": "abc123",
  "markdown": "<Gemini-generated-markdown>"
}

And backend responds with:

{
  "pdf_url": "https://firebase.storage/bucket/abc123/report.pdf"
}

✅ 6. Final Report Metadata in Firestore for the Doctor Interface

{
  "session_id": "abc123",
  "user_id": "user001",
  "created_at": "2025-07-04T09:00:00Z",
  "markdown": "...",
  "pdf_url": "https://firebase.storage/bucket/abc123/report.pdf"
}

✅ Summary Table of JSON Interfaces

Interface Direction Type Purpose
/call/:id WebSocket ⬅️➡️ JSON Live text/audio interaction
Redis internal JSON Store convo session memory
/upload ➡️ JSON Upload documents
/upload Response ⬅️ JSON Returns S3 URL
Gemini Input ➡️ JSON Full transcript + doc refs
Gemini Output ⬅️ Markdown Health summary report
Final PDF Endpoint (opt.) ➡️ JSON Convert MD to PDF
Firestore internal JSON Archive session + report