Ultra-simple version: direct CSV loading, no database

Return plain dict instead of Pydantic models to avoid validation

Bypass Pydantic validation to avoid datetime parse errors

Load data directly from CSV using pandas

Use mock data to show UI effect

Fix trade_date validation with field_validator

Use raw SQL to bypass SQLAlchemy datetime conversion

Fix datetime handling in trade.py to handle both datetime objects and strings

Fix: Use trade_hf router which has datetime handling logic

Initialize database at startup instead of build time to avoid Docker cache issues

Fix datetime handling: DROP TABLE before CREATE, handle both datetime objects and strings

Add request data cleaning to handle empty date strings

Add validator to handle empty string dates in request parameters

Change trade_date to String type to avoid SQLAlchemy datetime conversion errors

Add debug logging for trade_date

Handle empty trade_date in router before Pydantic validation

Use model_validator to handle empty trade_date

Add Pydantic validator to handle empty trade_date strings

Add detailed trade_date handling and error logging

Fix trade_date indentation

Handle empty trade_date fields

Fix indentation and add exception handling

Add database debug info to health check

Fix PaginatedResponse parameters

Fix missing source_record_id in API response

Add missing __init__.py files

Deploy to Hugging Face Spaces - SQLite version with 10k sample data

Enhances platform with robust monitoring and notifications

feat: 完成多国家海关数据源接入与核心功能全量迭代

feat(customs): 完成美墨两国海关数据连接器开发与文档适配

feat: 新增智利海关数据接入链路，完善系统框架与文档

feat: 新增贸易查询筛选条件并完善巴西贸易数据解析

feat: 接入巴西真实海关数据源并优化回补逻辑与HTTP客户端

Initial commit for Hugging Face Space