https://github.com/jaguarliuu/rookie_text2data
Dify插件 - 自然语言获取数据库数据
https://github.com/jaguarliuu/rookie_text2data
dify llm
Last synced: 4 months ago
JSON representation
Dify插件 - 自然语言获取数据库数据
- Host: GitHub
- URL: https://github.com/jaguarliuu/rookie_text2data
- Owner: jaguarliuu
- License: other
- Created: 2025-03-12T06:49:50.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-10-24T10:59:20.000Z (8 months ago)
- Last Synced: 2025-10-24T12:31:11.383Z (8 months ago)
- Topics: dify, llm
- Language: Python
- Homepage: https://marketplace.dify.ai/plugins/jaguarliuu/rookie_text2data
- Size: 41.7 MB
- Stars: 241
- Watchers: 7
- Forks: 58
- Open Issues: 26
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
## rookie_text2data
**Author:** jaguarliuu
**Version:** 1.0.1
**Type:** tool
### Description
A tool that converts natural language into secure and optimized SQL queries, supporting multiple mainstream databases including MySQL, PostgreSQL, Oracle, SQL Server, and Chinese domestic databases (GaussDB, KingbaseES, DM).
### 声明
We are truly grateful for the overwhelming interest in this experimental project. Your feedback is invaluable for improving this plugin. Join our WeChat group for discussions and collaboration opportunities!
> Contact:
> - Wechat： L1763077056
### ✨ Core Features
#### Multi-Database Support
- Native support for 7 database types with automatic syntax adaptation
- International databases: MySQL, PostgreSQL, Oracle, SQL Server
- Chinese domestic databases: GaussDB (Huawei), KingbaseES (KINGBASE), DM (Dameng)
- Automatic SQL syntax adaptation based on database type (e.g., LIMIT vs FETCH FIRST)
- Schema-aware query generation for databases that support schemas
#### Security Mechanisms
- Mandatory result set limits (default LIMIT 100, configurable up to 100,000)
- DML operation prohibition (SELECT statements only)
- Field whitelist validation (based on database metadata)
- Least privilege principle for query execution
### Supported Databases
- **MySQL** - Full support with optimized query generation
- **PostgreSQL** - Native support with schema awareness
- **Oracle** - Enterprise database with schema support
- **SQL Server** - Microsoft SQL Server support
- **GaussDB** - Huawei's enterprise-grade database (openGauss compatible)
- **KingbaseES** - Kingbase's PostgreSQL-compatible database
- **DM (Dameng)** - Dameng's Oracle-compatible database
### Supported LLMs
Compatible with all non-deep-thinking models
- ChatGLM-6B
- DeepSeek V3
- Qwen-max
...
### Quick Start
#### SQL Generation Component
1. Import the rookie_text2data plugin
2. Configure basic parameters:
| Parameter | Type | Required | Description | Multilingual Support |
|----------------|----------------|----------|----------------------------------------------------------------|--------------------------|
| db_type | select | Yes | Database type (MySQL/PostgreSQL/Oracle/SQL Server/GaussDB/KingbaseES/DM) | CN/EN/PT |
| host | string | Yes | Database host/IP address | CN/EN/PT |
| port | number | Yes | Database port (1-65535) | CN/EN/PT |
| db_name | string | Yes | Target database name | CN/EN/PT |
| table_names | string | No | Comma-separated table names (empty for all tables) | CN (format hints) |
| schema_name | string | No | Schema name (PostgreSQL default: public, Oracle/DM: uppercase username) | CN/EN/PT |
| username | string | Yes | Database username | CN/EN/PT |
| password | secret-input | Yes | Database password | CN/EN/PT |
| model | model-selector | Yes | LLM model configuration | CN/EN/PT |
| query | string | Yes | Natural language query statement | CN/EN/PT |
| limit | number | No | Query result limit (1-100000, default 100) | CN/EN/PT |
| result_format | select | No | Result format (JSON/TEXT, default JSON) | CN/EN/PT |
| custom_prompt | string | No | Custom prompt for fine-tuning query generation | CN/EN/PT |
| with_comment | boolean | No | Include database comments in schema metadata | CN/EN/PT |
3. Select Model - We recommend using the Qwen-max model. Other models can be tested but deep-thinking models are unsupported.
4. Generate SQL queries using natural language
#### SQL Execution Component
1. Import the rookie_execute_sql plugin
2. Configure basic parameters:
| Parameter | Type | Required | Description | Multilingual Support |
|----------------|--------------|----------|----------------------------------------------------------------|--------------------------|
| db_type | select | Yes | Database type (MySQL/PostgreSQL/Oracle/SQL Server/GaussDB/KingbaseES/DM) | CN/EN/PT |
| host | string | Yes | Database host/IP address | CN/EN/PT |
| port | number | Yes | Database port (1-65535) | CN/EN/PT |
| db_name | string | Yes | Target database name | CN/EN/PT |
| username | string | Yes | Database username | CN/EN/PT |
| password | secret-input | Yes | Database password | CN/EN/PT |
| schema | string | No | Schema name (PostgreSQL default: public, Oracle/DM: uppercase username) | CN/EN/PT |
| sql | string | Yes | SQL query to execute | CN/EN/PT |
| result_format | select | No | Result format (JSON/TEXT/CSV, default JSON) | CN/EN/PT |
3. Click "Execute" to run the SQL statement
### Database-Specific Notes
#### Chinese Domestic Databases
- **GaussDB**: Compatible with PostgreSQL protocol, uses SCRAM-SHA-256 authentication
- **KingbaseES**: PostgreSQL-compatible, supports schema-based organization
- **DM (Dameng)**: Oracle-compatible, uses Oracle-style schema management
#### Schema Configuration
- **PostgreSQL/GaussDB/KingbaseES**: Default schema is `public`, can be customized
- **Oracle/DM**: Default schema is the uppercase version of your username (e.g., user `admin` → schema `ADMIN`)
- **MySQL/SQL Server**: Schema parameter not required
### License
This project is licensed under the Apache License 2.0 - see the [LICENSE](LICENSE) file for details.