Files
AI-Trader/docker-compose.yml
Bill fb9583b374 feat: transform to REST API service with SQLite persistence (v0.3.0)
Major architecture transformation from batch-only to API service with
database persistence for Windmill integration.

## REST API Implementation
- POST /simulate/trigger - Start simulation jobs
- GET /simulate/status/{job_id} - Monitor job progress
- GET /results - Query results with filters (job_id, date, model)
- GET /health - Service health checks

## Database Layer
- SQLite persistence with 6 tables (jobs, job_details, positions,
  holdings, reasoning_logs, tool_usage)
- Foreign key constraints with cascade deletes
- Replaces JSONL file storage

## Backend Components
- JobManager: Job lifecycle management with concurrency control
- RuntimeConfigManager: Thread-safe isolated runtime configs
- ModelDayExecutor: Single model-day execution engine
- SimulationWorker: Date-sequential, model-parallel orchestration

## Testing
- 102 unit and integration tests (85% coverage)
- Database: 98% coverage
- Job manager: 98% coverage
- API endpoints: 81% coverage
- Pydantic models: 100% coverage
- TDD approach throughout

## Docker Deployment
- Dual-mode: API server (persistent) + batch (one-time)
- Health checks with 30s interval
- Volume persistence for database and logs
- Separate entrypoints for each mode

## Validation Tools
- scripts/validate_docker_build.sh - Build validation
- scripts/test_api_endpoints.sh - Complete API testing
- scripts/test_batch_mode.sh - Batch mode validation
- DOCKER_API.md - Deployment guide
- TESTING_GUIDE.md - Testing procedures

## Configuration
- API_PORT environment variable (default: 8080)
- Backwards compatible with existing configs
- FastAPI, uvicorn, pydantic>=2.0 dependencies

Co-Authored-By: AI Assistant <noreply@example.com>
2025-10-31 11:47:10 -04:00

94 lines
2.9 KiB
YAML

services:
# Batch mode: Run one-time simulations with config file
ai-trader-batch:
image: ghcr.io/xe138/ai-trader:latest
# Uncomment to build locally instead of pulling:
# build: .
container_name: ai-trader-batch
volumes:
- ${VOLUME_PATH:-.}/data:/app/data
- ${VOLUME_PATH:-.}/logs:/app/logs
- ${VOLUME_PATH:-.}/configs:/app/configs
environment:
# AI Model API Configuration
- OPENAI_API_BASE=${OPENAI_API_BASE}
- OPENAI_API_KEY=${OPENAI_API_KEY}
# Data Source Configuration
- ALPHAADVANTAGE_API_KEY=${ALPHAADVANTAGE_API_KEY}
- JINA_API_KEY=${JINA_API_KEY}
# System Configuration
- RUNTIME_ENV_PATH=/app/data/runtime_env.json
# MCP Service Ports (fixed internally)
- MATH_HTTP_PORT=8000
- SEARCH_HTTP_PORT=8001
- TRADE_HTTP_PORT=8002
- GETPRICE_HTTP_PORT=8003
# Agent Configuration
- AGENT_MAX_STEP=${AGENT_MAX_STEP:-30}
ports:
# Format: "HOST:CONTAINER" - container ports are fixed, host ports configurable via .env
- "${MATH_HTTP_PORT:-8000}:8000"
- "${SEARCH_HTTP_PORT:-8001}:8001"
- "${TRADE_HTTP_PORT:-8002}:8002"
- "${GETPRICE_HTTP_PORT:-8003}:8003"
- "${WEB_HTTP_PORT:-8888}:8888"
restart: "no" # Batch jobs should not auto-restart
profiles:
- batch # Only start with: docker-compose --profile batch up
# API mode: REST API server for Windmill integration
ai-trader-api:
image: ghcr.io/xe138/ai-trader:latest
# Uncomment to build locally instead of pulling:
# build: .
container_name: ai-trader-api
entrypoint: ["./entrypoint-api.sh"]
volumes:
- ${VOLUME_PATH:-.}/data:/app/data
- ${VOLUME_PATH:-.}/logs:/app/logs
- ${VOLUME_PATH:-.}/configs:/app/configs
environment:
# AI Model API Configuration
- OPENAI_API_BASE=${OPENAI_API_BASE}
- OPENAI_API_KEY=${OPENAI_API_KEY}
# Data Source Configuration
- ALPHAADVANTAGE_API_KEY=${ALPHAADVANTAGE_API_KEY}
- JINA_API_KEY=${JINA_API_KEY}
# System Configuration
- RUNTIME_ENV_PATH=/app/data/runtime_env.json
# MCP Service Ports (fixed internally)
- MATH_HTTP_PORT=8000
- SEARCH_HTTP_PORT=8001
- TRADE_HTTP_PORT=8002
- GETPRICE_HTTP_PORT=8003
# API Configuration
- API_PORT=${API_PORT:-8080}
# Agent Configuration
- AGENT_MAX_STEP=${AGENT_MAX_STEP:-30}
ports:
# MCP service ports (internal)
- "${MATH_HTTP_PORT:-8000}:8000"
- "${SEARCH_HTTP_PORT:-8001}:8001"
- "${TRADE_HTTP_PORT:-8002}:8002"
- "${GETPRICE_HTTP_PORT:-8003}:8003"
# API server port (primary interface)
- "${API_PORT:-8080}:8080"
# Web dashboard
- "${WEB_HTTP_PORT:-8888}:8888"
restart: unless-stopped # Keep API server running
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
interval: 30s
timeout: 10s
retries: 3
start_period: 40s