API

DeepAnalyze API Server

🚀 Quick Start

Prerequisites

Start vLLM Model Server:

vllm serve DeepAnalyze-8B --host 0.0.0.0 --port 8000

Starting the Server

cd API
python start_server.py

API Server: http://localhost:8200 (Main API)
File Server: http://localhost:8100 (File downloads)
Health Check: http://localhost:8200/health

The API server will create a new workspace folder in the current directory as the working directory. For each conversation, it will generate a thread subdirectory under this workspace to perform data analysis and generate files.

Quick Test

cd example
python exampleRequest.py          #  requests example
python exampleOpenAI.py    # OpenAI library example

📚 API Usage

1. File Upload

Requests Example:

import requests

with open('data.csv', 'rb') as f:
    files = {'file': ('data.csv', f, 'text/csv')}
    response = requests.post('http://localhost:8200/v1/files', files=files)

file_id = response.json()['id']
print(f"File uploaded: {file_id}")

OpenAI Library Example:

import openai

client = openai.OpenAI(
    base_url="http://localhost:8200/v1",
    api_key="dummy"
)

with open('data.csv', 'rb') as f:
    file_obj = client.files.create(file=f)

print(f"File uploaded: {file_obj.id}")

2. Simple Chat (No Files)

Requests Example:

response = requests.post('http://localhost:8200/v1/chat/completions', json={
    "model": "DeepAnalyze-8B",
    "messages": [
        {"role": "user", "content": "Introduce Python programming language in one sentence"}
    ],
    "temperature": 0.4
})

content = response.json()['choices'][0]['message']['content']
print(content)

OpenAI Library Example:

response = client.chat.completions.create(
    model="DeepAnalyze-8B",
    messages=[
        {"role": "user", "content": "Introduce Python programming language in one sentence"}
    ],
    temperature=0.4
)

print(response.choices[0].message.content)

3. Chat with Files

Requests Example:

response = requests.post('http://localhost:8200/v1/chat/completions', json={
    "model": "DeepAnalyze-8B",
    "messages": [
        {
            "role": "user",
            "content": "Analyze this data file, calculate average salary by department, and generate visualization charts.",
            "file_ids": [file_id]  
        }
    ],
    "temperature": 0.4
})

result = response.json()
content = result['choices'][0]['message']['content']
files = result['choices'][0]['message'].get('files', [])

print(f"Response: {content}")
for file_info in files:
    print(f"Generated file: {file_info['name']} - {file_info['url']}")

OpenAI Library Example:

response = client.chat.completions.create(
    model="DeepAnalyze-8B",
    messages=[
        {
            "role": "user",
            "content": "Analyze this data file, calculate average salary by department, and generate visualization charts.",
            "file_ids": [file_id]  
        }
    ],
    temperature=0.4
)

message = response.choices[0].message
print(f"Response: {message.content}")

# Access generated files (new format)
if hasattr(message, 'files') and message.files:
    for file_info in message.files:
        print(f"Generated file: {file_info['name']} - {file_info['url']}")

4. Multi-turn Conversations with Thread ID

Use thread_id in the latest message to maintain workspace context across multiple requests. This allows files and generated content to be persisted between conversations.

Important: You must maintain the complete conversation history in each request - only add thread_id to the latest message.

Requests Example:

conversation_history = []

# First request - creates new thread and gets thread_id
conversation_history.append({"role": "user", "content": "Create a Python script that calculates Fibonacci numbers"})
response = requests.post('http://localhost:8200/v1/chat/completions', json={
    "model": "DeepAnalyze-8B",
    "messages": conversation_history
})

result = response.json()
thread_id = result['choices'][0]['message']['thread_id']
assistant_response = result['choices'][0]['message']['content']
conversation_history.append({"role": "assistant", "content": assistant_response})

print(f"Thread ID: {thread_id}")

# Second request - includes full history + thread_id in latest message
conversation_history.append({"role": "user", "content": "Now run the script to calculate the first 10 numbers", "thread_id": thread_id})
response = requests.post('http://localhost:8200/v1/chat/completions', json={
    "model": "DeepAnalyze-8B",
    "messages": conversation_history
})

assistant_response = response.json()['choices'][0]['message']['content']
conversation_history.append({"role": "assistant", "content": assistant_response})

# Third request - continue with same thread, full history maintained
conversation_history.append({"role": "user", "content": "List all files in the current workspace", "thread_id": thread_id})
response = requests.post('http://localhost:8200/v1/chat/completions', json={
    "model": "DeepAnalyze-8B",
    "messages": conversation_history
})

OpenAI Library Example:

messages = []

# First request
messages.append({"role": "user", "content": "Create a data analysis script"})
response = client.chat.completions.create(
    model="DeepAnalyze-8B",
    messages=messages
)

# Check for thread_id using hasattr()
thread_id = None
if hasattr(response.choices[0].message, 'thread_id'):
    thread_id = response.choices[0].message.thread_id

messages.append({"role": "assistant", "content": response.choices[0].message.content})

# Continue conversation - include full history + thread_id in latest message
messages.append({"role": "user", "content": "Run the analysis script", "thread_id": thread_id})
response = client.chat.completions.create(
    model="DeepAnalyze-8B",
    messages=messages
)

if hasattr(response.choices[0].message, 'thread_id'):
    print(f"Thread ID: {response.choices[0].message.thread_id}")

messages.append({"role": "assistant", "content": response.choices[0].message.content})

Key Points:

thread_id is returned in the response message
Include thread_id in the latest user message of your conversation history
You should send the complete conversation history in each request
Files created in previous requests remain available in the thread workspace
The workspace persists across requests with the same thread_id

5. Streaming Chat with Files

Requests Example:

response = requests.post('http://localhost:8200/v1/chat/completions', json={
    "model": "DeepAnalyze-8B",
    "messages": [
        {
            "role": "user",
            "content": "Analyze this data and generate trend charts.",
            "file_ids": [file_id]
        }
    ],
    "stream": True
}, stream=True)

for line in response.iter_lines():
    if line:
        line_str = line.decode('utf-8')
        if line_str.startswith('data: '):
            data_str = line_str[6:]
            if data_str == '[DONE]':
                break
            chunk = json.loads(data_str)
            if 'choices' in chunk and chunk['choices']:
                delta = chunk['choices'][0].get('delta', {})
                if 'content' in delta:
                    print(delta['content'], end='', flush=True)
                if 'thread_id' in delta:
                    print(f"\nThread ID: {delta['thread_id']}")

OpenAI Library Example:

stream = client.chat.completions.create(
    model="DeepAnalyze-8B",
    messages=[
        {
            "role": "user",
            "content": "Analyze this data and generate trend charts.",
            "file_ids": [file_id]
        }
    ],
    stream=True
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end='', flush=True)
    if hasattr(chunk, 'generated_files') and chunk.generated_files:
        collected_files.extend(chunk.generated_files)
    if hasattr(chunk.choices[0].delta, 'thread_id'):
        print(f"\nThread ID: {chunk.choices[0].delta.thread_id}")

📋 API Reference

Files API

POST /v1/files

Upload a file for analysis.

Request:

POST /v1/files
Content-Type: multipart/form-data

file: [binary file data]

Response:

{
  "id": "file-abc123...",
  "object": "file",
  "bytes": 1024,
  "created_at": 1704067200,
  "filename": "data.csv"
}

GET /v1/files

List all uploaded files.

Request:

GET /v1/files

Response:

{
  "object": "list",
  "data": [
    {
      "id": "file-abc123...",
      "object": "file",
      "bytes": 1024,
      "created_at": 1704067200,
      "filename": "data.csv"
    }
  ]
}

GET /v1/files/{file_id}/content

Download file content.

Request:

GET /v1/files/{file_id}/content

Response: Binary file content

DELETE /v1/files/{file_id}

Delete a file.

Request:

DELETE /v1/files/{file_id}

Response:

{
  "id": "file-abc123...",
  "object": "file",
  "deleted": true
}

Chat Completions API

POST /v1/chat/completions

Extended chat completion with file support.

Request:

{
  "model": "DeepAnalyze-8B",
  "messages": [
    {
      "role": "user",
      "content": "分析这个数据文件",
      "file_ids": ["file-abc123"],     // OpenAI compatible: file_ids in messages
      "thread_id": "thread-xyz789..."  // Optional: thread_id in latest message for workspace persistence
    }
  ],
  "file_ids": ["file-def456"],         // Optional: file_ids parameter (backward compatibility)
  "temperature": 0.4,
  "stream": false
}

Response (Non-Streaming):

{
  "id": "chatcmpl-xyz789...",
  "object": "chat.completion",
  "created": 1704067200,
  "model": "DeepAnalyze-8B",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "分析结果...",
        "thread_id": "thread-abc123...", // Thread ID for workspace persistence
        "files": [                      // New format: files in message
          {
            "name": "chart.png",
            "url": "http://localhost:8100/thread-123/generated/chart.png"
          }
        ]
      },
      "finish_reason": "stop"
    }
  ],
  "generated_files": [                // Backward compatibility: generated_files field
    {
      "name": "chart.png",
      "url": "http://localhost:8100/thread-123/generated/chart.png"
    }
  ],
  "attached_files": ["file-abc123"]   // Input files
}

Response (Streaming):

data: {"id": "chatcmpl-xyz789...", "object": "chat.completion.chunk", "choices": [{"delta": {"content": "分析"}}]}
data: {"id": "chatcmpl-xyz789...", "object": "chat.completion.chunk", "choices": [{"delta": {"files": [{"name":"chart.png","url":"..."}], "thread_id": "thread-abc123..."}, "finish_reason": "stop"}]}
data: [DONE]

Health Check API

GET /health

Check API server status.

Request:

GET /health

Response:

{
  "status": "healthy",
  "timestamp": 1704067200
}

🏗️ Architecture

Multi-Port Design

Port 8000: vLLM model server (external)
Port 8200: Main API server (FastAPI)
Port 8100: File HTTP server for downloads

🔧 Configuration

Environment Variables

# API Configuration
API_BASE = "http://localhost:8000/v1"  # vLLM endpoint
MODEL_PATH = "DeepAnalyze-8B"          # Model name
WORKSPACE_BASE_DIR = "workspace"       # File storage
HTTP_SERVER_PORT = 8100               # File server port

# Model Settings
DEFAULT_TEMPERATURE = 0.4            # Default sampling temperature
MAX_NEW_TOKENS = 32768               # Maximum response tokens
STOP_TOKEN_IDS = [32000, 32007]      # Special token IDs

🛠️ Examples

The example/ directory contains comprehensive examples:

exampleRequest.py - Simple requests example
exampleOpenAI.py - OpenAI library example

Name		Name	Last commit message	Last commit date
parent directory ..
example		example
README.md		README.md
README_ZH.md		README_ZH.md
__init__.py		__init__.py
admin_api.py		admin_api.py
chat_api.py		chat_api.py
config.py		config.py
file_api.py		file_api.py
main.py		main.py
models.py		models.py
models_api.py		models_api.py
start_server.py		start_server.py
storage.py		storage.py
utils.py		utils.py

Navigation Menu

Uh oh!

README.md

DeepAnalyze API Server

🚀 Quick Start

Prerequisites

Starting the Server

Quick Test

📚 API Usage

1. File Upload

2. Simple Chat (No Files)

3. Chat with Files

4. Multi-turn Conversations with Thread ID

5. Streaming Chat with Files

📋 API Reference

Files API

POST /v1/files

GET /v1/files

GET /v1/files/{file_id}/content

DELETE /v1/files/{file_id}

Chat Completions API

POST /v1/chat/completions

Health Check API

GET /health

🏗️ Architecture

Multi-Port Design

🔧 Configuration

Environment Variables

🛠️ Examples

Uh oh!

FilesExpand file tree

API

Directory actions

More options

Directory actions

More options

Latest commit

History

API

Folders and files

parent directory

README.md

DeepAnalyze API Server

🚀 Quick Start

Prerequisites

Starting the Server

Quick Test

📚 API Usage

1. File Upload

2. Simple Chat (No Files)

3. Chat with Files

4. Multi-turn Conversations with Thread ID

5. Streaming Chat with Files

📋 API Reference

Files API

POST /v1/files

GET /v1/files

GET /v1/files/{file_id}/content

DELETE /v1/files/{file_id}

Chat Completions API

POST /v1/chat/completions

Health Check API

GET /health

🏗️ Architecture

Multi-Port Design

🔧 Configuration

Environment Variables

🛠️ Examples