BettaFish Micro-Sentiment System - LionCC API Complete Deployment Guide

Categories: Tech 2025-11-07

Author: Lion CC (lioncc.ai) WeChat: HSQBJ088888888

Last Updated: 2025-11-07 System Version: v1.0.1 Deployment Method: Docker Update Notes: Added Gemini 2.5 Pro model configuration, updated recommended model list

💡 About LionCC API Service

This guide is tested and verified based on LionCC API (VibeCodingAPI.ai) platform, fully compatible with all LLM models in BettaFish Micro-Sentiment System (Claude, GPT-4, Gemini, etc.).

🎁 Special Offer: Register LionCC API through codecodex.ai community and get $20 testing credits (limited to first 1000 users)

Add Business Assistant on WeChat

Scan to add business assistant, note: BettaFish

Get $20 API testing credits + technical support

Close

📋 Table of Contents

1. System Architecture Overview
2. Third-Party API Configuration
3. Docker Deployment Notes
4. Common Issues & Solutions
5. 🚀 Quick Start Guide ⭐️ ← Recommended Starting Point

🏗️ System Architecture Overview

Core Components

BettaFish Micro-Sentiment System
├── InsightEngine     - Private database mining (Claude 3.5 Sonnet)
├── MediaEngine       - Multimodal content analysis (Gemini Pro)
├── QueryEngine       - Web search reasoning (DeepSeek Chat)
├── ForumEngine       - Agent collaboration forum
└── ReportEngine      - Intelligent report generation (Gemini 2.5 Pro)

Tech Stack

• Backend: Python 3.11 + Flask + LangGraph
• Database: MySQL 8.0
• Containers: Docker + Docker Network
• Frontend: Streamlit (single engine) + HTML/JavaScript (main app)

🔑 Third-Party API Configuration

1. LLM API Configuration (LionCC API)

📍 API Provider

• Service Name: LionCC API
• Platform: VibeCodingAPI.ai
• Website: https://vibecodingapi.ai
• Features: OpenAI-compatible interface, optimized for BettaFish Micro-Sentiment System, supports all mainstream LLM models

🔐 API Key Configuration

# All engines use the same API Key
INSIGHT_ENGINE_API_KEY=sk-YOUR_API_KEY_HERE
MEDIA_ENGINE_API_KEY=sk-YOUR_API_KEY_HERE
QUERY_ENGINE_API_KEY=sk-YOUR_API_KEY_HERE
REPORT_ENGINE_API_KEY=sk-YOUR_API_KEY_HERE
FORUM_HOST_API_KEY=sk-YOUR_API_KEY_HERE
MINDSPIDER_API_KEY=sk-YOUR_API_KEY_HERE
KEYWORD_OPTIMIZER_API_KEY=sk-YOUR_API_KEY_HERE

🎯 Recommended Model Configuration

⚠️ Tested Model List

✅ Available Models (tested and verified through LionCC API):

- gpt-4o              ✅ (Latest GPT-4 Optimized)
- gpt-4               ✅
- gpt-4-turbo         ✅
- gpt-3.5-turbo       ✅
- claude-3-5-sonnet-20241022  ✅ (Recommended)
- claude-3-opus-20240229      ✅
- gemini-pro          ✅
- gemini-2.5-pro      ✅ (Latest version, outstanding performance, recommended for reports)
- deepseek-chat       ✅ (Recommended, cost-effective)
- deepseek-reasoner   ⚠️ (Unstable, fails under high load, not recommended)

❌ Unavailable Models:

- kimi-k2-0711-preview    ❌ (HTTP 500)
- qwen-plus               ❌ (HTTP 503)
- qwen-turbo              ❌ (HTTP 503)
- gemini-2.0-flash-exp    ❌ (HTTP 500)

🆕 Gemini 2.5 Pro New Features

Recommended Scenario: ReportEngine (report generation)

Advantages:

• ✅ Latest generation Gemini model with significant performance improvements
• ✅ Long context support (1M+ tokens), ideal for large-scale report generation
• ✅ Optimized for multiple languages, excellent Chinese and English generation quality
• ✅ Strong structured output capabilities, suitable for HTML/Markdown format reports
• ✅ Better cost-performance ratio than GPT-4o

Notes:

• ⚠️ Slightly higher token consumption compared to Gemini Pro
• ⚠️ Some API providers may not yet support this model
• ⚠️ Test availability before large-scale use

Test Command:

# Test Gemini 2.5 Pro availability
curl -X POST "https://vibecodingapi.ai/v1/chat/completions" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-2.5-pro",
    "messages": [{"role": "user", "content": "test"}],
    "max_tokens": 100
  }'

💡 Model Switching Recommendations

If a specific model is unavailable, switch according to these priorities:

Text Analysis Tasks:

claude-3-5-sonnet > gpt-4o > gpt-4-turbo > deepseek-chat

Reasoning Tasks:

deepseek-chat > gpt-4-turbo > claude-3-opus

Report Generation:

gemini-2.5-pro > gpt-4o > gpt-4 > claude-3-5-sonnet

Cost Optimization:

deepseek-chat > gpt-3.5-turbo > gemini-pro

2. Bocha API Configuration (Multimodal Search)

📍 API Information

• Name: Bocha AI
• Website: https://api.bochaai.com or https://open.bochaai.com
• Function: Multimodal web search (images, videos, web pages)
• Used by: MediaEngine

🔐 Configuration Points

# Bocha API Configuration
BOCHA_BASE_URL=https://api.bochaai.com
BOCHA_WEB_SEARCH_API_KEY=sk-YOUR_BOCHA_KEY_HERE

⚠️ Key Notes

1. API Endpoint Selection

Bocha offers two API packages:

2. Code Configuration

By default, BettaFish uses /v1/ai-search. If your package is Web Search, modify:

# Edit file: MediaEngine/tools/search.py
# Line 94
BASE_URL = "https://api.bochaai.com/v1/web-search"  # Change to web-search

3. Quota Check

Login to Bocha dashboard to check:

• Account balance
• Package type (Web Search / AI Search)
• Remaining calls

4. Common Errors

{"message": "You do not have enough money or package quota", "code": "403"}

Solutions:

• Recharge account
• Confirm using correct API endpoint
• Check if API Key is correct

3. Tavily API Configuration (Web Search)

📍 API Information

• Name: Tavily
• Website: https://www.tavily.com
• Function: Text web search
• Used by: InsightEngine, QueryEngine

🔐 Configuration

TAVILY_API_KEY=tvly-YOUR_KEY_HERE

✅ Features

• Free Quota: Usually includes trial credits
• Stability: High
• Speed: Fast
• Languages: Supports Chinese and English

💡 Test Command

curl -X POST 'https://api.tavily.com/search' \
  -H 'Content-Type: application/json' \
  -d '{
    "api_key": "YOUR_KEY",
    "query": "test",
    "max_results": 1
  }'

🐳 Docker Deployment Notes

1. Docker Image Building

📋 Basic Configuration

FROM python:3.11-slim

# Key dependencies
RUN apt-get update && apt-get install -y \
    build-essential \
    ffmpeg \
    libgl1 \
    chromium \
    && apt-get clean

# Python package manager (use uv for acceleration)
RUN curl -LsSf https://astral.sh/uv/install.sh | sh

⚠️ Notes

1. Python Version Lock

• ✅ Use Python 3.11 (recommended)
• ❌ Do not use Python 3.14 (compatibility issues)
• ⚠️ Python 3.12 untested

2. Dependency Installation Issues

Some packages need to be installed after container startup:

# cryptography (MySQL authentication)
docker exec bettafish pip install cryptography --quiet

# If encountering permission issues
docker exec -u root bettafish pip install cryptography

Permanent solution: Add cryptography to requirements.txt and rebuild image.

2. Container Configuration

📍 Port Mapping

docker run -d \
  --name bettafish \
  -p 15888:5000   # Flask main app
  -p 8501:8501    # InsightEngine (Streamlit)
  -p 8502:8502    # MediaEngine (Streamlit)
  -p 8503:8503    # QueryEngine (Streamlit)
  bettafish:latest

⚠️ Port Conflict Issues

Issue: Default port 5000 may be occupied by proxy (e.g., Clash)

Symptoms:

HTTP 403 Forbidden
localhost:5000 cannot be accessed

Solutions:

# Method 1: Use non-standard port
-p 15888:5000

# Method 2: Set no_proxy
export no_proxy="localhost,127.0.0.1"
open -a "Google Chrome" "http://localhost:15888"

3. Network Configuration

📍 Docker Network

BettaFish uses a separate Docker network to connect containers:

# Create network
docker network create bettafish-network

# MySQL container
docker run -d \
  --name bettafish-mysql \
  --network bettafish-network \
  -p 3307:3306 \
  -e MYSQL_ROOT_PASSWORD=root \
  -e MYSQL_DATABASE=bettafish_db \
  mysql:8.0

# Application container
docker run -d \
  --name bettafish \
  --network bettafish-network \
  -p 15888:5000 \
  bettafish:latest

⚠️ Database Connection Configuration

.env file configuration:

# ❌ Wrong (localhost doesn't work inside container)
DB_HOST=localhost

# ✅ Correct (use container name)
DB_HOST=bettafish-mysql
DB_PORT=3306

4. Environment Variable Management

📋 .env File Sync

Issue: Docker image contains old .env file

Symptoms:

• After updating host .env file, container configuration doesn’t take effect
• docker restart doesn’t reload .env

Solutions:

# Method 1: Manual copy (recommended)
docker cp .env bettafish:/app/.env
docker restart bettafish

# Method 2: Use --env-file (when creating container)
docker run -d --env-file .env bettafish:latest

# Method 3: Rebuild image
docker build -t bettafish:latest .

⚠️ Configuration Update Workflow

After modifying .env or code:

# 1. Stop and remove old container
docker stop bettafish && docker rm bettafish

# 2. Rebuild image (if code changed)
docker build -t bettafish:latest .

# 3. Start new container
docker run -d \
  --name bettafish \
  --network bettafish-network \
  -p 15888:5000 \
  -p 8501-8503:8501-8503 \
  --env-file .env \
  bettafish:latest

# 4. Install additional dependencies
docker exec bettafish pip install cryptography --quiet

# 5. Restart to apply
docker restart bettafish

5. Data Persistence

📍 Important Directories

/app/logs/              # Runtime logs
/app/final_reports/     # Generated reports
/app/forum.log          # Forum discussion logs
/app/templates/         # Custom templates

💡 Mount Volumes (Optional)

docker run -d \
  -v /host/path/logs:/app/logs \
  -v /host/path/reports:/app/final_reports \
  bettafish:latest

🔧 Common Issues & Solutions

1. ModuleNotFoundError: No module named ‘config’

Cause: Missing config.py file in container

Solution:

# Create config.py
docker exec bettafish bash -c 'cat > /app/config.py << EOF
import os
from dotenv import load_dotenv

load_dotenv()

# Database configuration
DB_HOST = os.getenv("DB_HOST", "localhost")
DB_PORT = int(os.getenv("DB_PORT", "3306"))
DB_USER = os.getenv("DB_USER", "root")
DB_PASSWORD = os.getenv("DB_PASSWORD", "")
DB_NAME = os.getenv("DB_NAME", "bettafish_db")

# LLM configuration (add all required config items)
# ...
EOF'

# Restart container
docker restart bettafish

Permanent solution: Create config.py in project root directory, rebuild image.

2. MySQL Authentication Failure

Error Message:

Authentication plugin 'caching_sha2_password' cannot be loaded

Cause: PyMySQL requires cryptography package for MySQL 8.0

Solution:

docker exec bettafish pip install cryptography --quiet
docker restart bettafish

3. Bocha API 403 Error

Error Message:

{"message": "You do not have enough money or package quota", "code": "403"}

Troubleshooting Steps:

1. Check API Key:

   docker exec bettafish grep "BOCHA" /app/.env
   

2. Confirm Account Balance: Login to Bocha dashboard

3. Confirm API Endpoint: Check if using correct endpoint (web-search vs ai-search)

4. Test API:

   curl -X POST "https://api.bochaai.com/v1/web-search" \
     -H "Authorization: Bearer YOUR_KEY" \
     -H "Content-Type: application/json" \
     -d '{"query":"test"}'
   

4. LLM API High Load (500 Error)

Error Message:

{"error": {"message": "Current model load is high, please retry later or switch to another model"}}

Solutions:

Method 1: Wait and Retry (system will auto-retry)

Method 2: Switch to Backup Model

# Edit .env file
# Replace unstable model with backup
QUERY_ENGINE_MODEL_NAME=deepseek-chat  # Change from deepseek-reasoner to deepseek-chat

# Sync to container
docker cp .env bettafish:/app/.env
docker restart bettafish

5. Report Format Issues

Issue: Generated reports are plain text, no visualization

Cause:

1. System visualization is built-in, not template-dependent
2. Templates only define report outline structure

Explanation:

• ✅ System has 6 built-in templates
• ✅ ReportEngine auto-selects appropriate template
• ✅ Visualization effects (charts, data cards) are code-generated
• ⚠️ “Upload template” feature is for customizing report structure, not adding visualization

Confirm System Normal:

• Check logs for [HTMLGenerationNode] Starting HTML report generation...
• Check /app/final_reports/ directory for report files

🚀 Quick Start Guide ⭐️ Core Content

💡 Important: This is the fastest way to deploy BettaFish, bookmark this script!

Use Case: Initial deployment completed, need to quickly start system

Complete Startup Script

📦 One-Click Startup Script - Copy and Use

Save the following as start-bettafish.sh, give execute permission and run directly to start the entire system.

#!/bin/bash

echo "======================================"
echo "BettaFish Micro-Sentiment System - Quick Start"
echo "======================================"

# 1. Start MySQL
echo "1️⃣  Starting MySQL container..."
docker start bettafish-mysql
sleep 3

# 2. Start application
echo "2️⃣  Starting BettaFish container..."
docker start bettafish
sleep 10

# 3. Sync config file
echo "3️⃣  Syncing config file..."
docker cp .env bettafish:/app/.env 2>/dev/null || echo "   .env synced"

# 4. Install dependencies
echo "4️⃣  Installing required dependencies..."
docker exec bettafish pip install cryptography --quiet 2>/dev/null

# 5. Restart application
echo "5️⃣  Restarting application to load config..."
docker restart bettafish >/dev/null 2>&1
sleep 30

# 6. Check status
echo "6️⃣  Checking service status..."
docker ps --filter "name=bettafish" --format "   ✓ {{.Names}}: {{.Status}}"

echo ""
echo "======================================"
echo "✅ Startup Complete!"
echo "======================================"
echo ""
echo "🌐 Access URLs:"
echo "   Main App: http://localhost:15888"
echo "   InsightEngine: http://localhost:8501"
echo "   MediaEngine: http://localhost:8502"
echo "   QueryEngine: http://localhost:8503"
echo ""
echo "💡 Opening browser:"
export no_proxy="localhost,127.0.0.1"
export NO_PROXY="localhost,127.0.0.1"
open -a "Google Chrome" "http://localhost:15888"

Usage:

chmod +x start-bettafish.sh
./start-bettafish.sh

📚 Complete Configuration File Example

.env File Template

# ====================== Database Configuration ======================
DB_HOST=bettafish-mysql
DB_PORT=3306
DB_USER=bettafish_user
DB_PASSWORD=BettaFish2024!
DB_NAME=bettafish_db
DB_CHARSET=utf8mb4
DB_DIALECT=mysql

# ======================= LLM Configuration =======================
# Insight Agent - Database Mining
INSIGHT_ENGINE_API_KEY=sk-YOUR_API_KEY
INSIGHT_ENGINE_BASE_URL=https://vibecodingapi.ai/v1
INSIGHT_ENGINE_MODEL_NAME=claude-3-5-sonnet-20241022

# Media Agent - Multimodal Analysis
MEDIA_ENGINE_API_KEY=sk-YOUR_API_KEY
MEDIA_ENGINE_BASE_URL=https://vibecodingapi.ai/v1
MEDIA_ENGINE_MODEL_NAME=gemini-pro

# Query Agent - Web Search
QUERY_ENGINE_API_KEY=sk-YOUR_API_KEY
QUERY_ENGINE_BASE_URL=https://vibecodingapi.ai/v1
QUERY_ENGINE_MODEL_NAME=deepseek-chat

# Report Agent - Report Generation
REPORT_ENGINE_API_KEY=sk-YOUR_API_KEY
REPORT_ENGINE_BASE_URL=https://vibecodingapi.ai/v1
REPORT_ENGINE_MODEL_NAME=gemini-2.5-pro

# Forum Host - Forum Hosting
FORUM_HOST_API_KEY=sk-YOUR_API_KEY
FORUM_HOST_BASE_URL=https://vibecodingapi.ai/v1
FORUM_HOST_MODEL_NAME=claude-3-5-sonnet-20241022

# MindSpider - Crawler Analysis
MINDSPIDER_API_KEY=sk-YOUR_API_KEY
MINDSPIDER_BASE_URL=https://vibecodingapi.ai/v1
MINDSPIDER_MODEL_NAME=deepseek-chat

# Keyword Optimizer - Keyword Optimization
KEYWORD_OPTIMIZER_API_KEY=sk-YOUR_API_KEY
KEYWORD_OPTIMIZER_BASE_URL=https://vibecodingapi.ai/v1
KEYWORD_OPTIMIZER_MODEL_NAME=gpt-3.5-turbo

# ================== Web Tools Configuration ====================
# Tavily API (Web Search)
TAVILY_API_KEY=tvly-YOUR_KEY

# Bocha API (Multimodal Search)
BOCHA_BASE_URL=https://api.bochaai.com
BOCHA_WEB_SEARCH_API_KEY=sk-YOUR_BOCHA_KEY

🎯 Deployment Checklist

Pre-Deployment Checks

• Python 3.11 environment
• Docker and Docker Compose installed
• At least 4GB available memory
• Ports 15888, 8501-8503 not occupied

API Configuration Checks

• VibeCodingAPI API Key obtained
• Tavily API Key obtained
• Bocha API Key obtained (optional)
• All models tested and available

Container Configuration Checks

• Docker network created
• MySQL container running normally
• BettaFish container running normally
• Container network connectivity

Function Testing Checks

• Main app (15888) accessible
• 3 Streamlit Engines (8501-8503) accessible
• Database connection successful
• Test query generates report
• 5 Engine status indicators show green

📞 Get Help

Official Resources

• GitHub: https://github.com/666ghj/BettaFish
• Documentation: README.md
• Discussion: https://linux.do/t/topic/1009280
• Sample Report: final_reports/final_report__20250827_131630.html

View Logs

# Real-time logs
docker logs -f bettafish

# Last 100 lines
docker logs bettafish --tail 100

# Error logs
docker logs bettafish 2>&1 | grep -i error

✅ Summary

Signs of Successful BettaFish Deployment:

1. ✅ All containers running
2. ✅ Main app accessible (http://localhost:15888)
3. ✅ All 5 Engine indicators green
4. ✅ Test query successfully generates report
5. ✅ Report includes visualization content

Key Configuration Points:

• 🔑 API Keys Configured Correctly (VibeCodingAPI, Tavily, Bocha)
• 🎯 Models Selected Appropriately (choose best models based on tasks)
• 🐳 Docker Configuration Complete (network, ports, environment variables)
• 📁 Files Synced Timely (.env, config.py)

📝 Changelog

v1.0.1 (2025-11-07)

New Content:

• ✨ Added Gemini 2.5 Pro model configuration guide
• ✨ Added Gemini 2.5 Pro feature introduction and testing methods
• 📝 Updated ReportEngine recommended model to gemini-2.5-pro
• 📝 Updated model priority recommendation list
• 📝 Updated tested model list
• 📝 Updated .env configuration file template

Optimized Content:

• 🔧 Marked DeepSeek Reasoner as unstable, not recommended
• 🔧 Optimized report generation model selection recommendations

v1.0.0 (2025-11-06)

Initial Version:

• 📚 Complete BettaFish system deployment guide
• 🔑 Third-party API configuration detailed explanation (VibeCodingAPI, Tavily, Bocha)
• 🐳 Docker deployment notes
• 🔧 Common issues and solutions
• 🚀 Quick start script

Enjoy using it! 🎉

💬 Get Help & Services

Recommended API Service

This document is tested and verified on the following platforms, recommended for BettaFish Micro-Sentiment System deployment:

LionCC API (VibeCodingAPI.ai) - Fully compatible with all BettaFish models

• Claude 3.5 Sonnet / GPT-4o / Gemini 2.5 Pro full coverage
• Optimized configuration for BettaFish Micro-Sentiment System
• Pricing: ¥100 = $100 API credits (1:1, no exchange rate loss)
• 🎁 Community users get $20 testing credits (limited to first 1000 users)

CodeCodex.ai - Claude Code Carpool Service

• No IP blocking, direct connection in China, starting from ¥398/month
• Suitable for users who need stable Claude Code service

Technical Support

For deployment guidance or issues:

• WeChat: HSQBJ088888888 (Click to view QR code, note: BettaFish)

• Visit: codecodex.ai

  vibecodingapi.ai