Add MariaDB integration with native MariaDB VECTOR support

[CroWzblooD]

Add MariaDB Integration with Native VECTOR Support

This PR adds MariaDB database and vector store integration to MindSQL with native VECTOR(384) data type support.

What's Added

MariaDB Database Connector (mindsql/databases/mariadb.py)

• Uses official mariadb Python connector
• Works with MariaDB 11.7+ (includes VECTOR support)
• Handles schema discovery and DDL extraction
• Connection management with proper error handling

MariaDB Vector Store (mindsql/vectorstores/mariadb_vector.py)

• Native VECTOR(384) data type support - no JSON serialization needed
• FULLTEXT indexing for hybrid search (combines vector similarity with keyword search)
• Stores data and vectors in the same database
• ACID compliant with transaction support

What's Modified

Constants (mindsql/_utils/constants.py)

• Added MariaDB-specific query constants for database introspection
• SHOW DATABASES, table info schema, and CREATE TABLE queries

Helper Functions (mindsql/_helper/helper.py)

• Improved SQL extraction to handle different LLM response formats
• Better parsing of SQLQuery labels and code blocks

Core Pipeline (mindsql/core/mindsql_core.py)

• Enhanced DDL retrieval with fallback logic
• Tries vector search first, falls back to full schema if needed

Module Exports (mindsql/databases/__init__.py and mindsql/vectorstores/__init__.py)

• Added MariaDB and MariaDBVectorStore to public API

Architecture

Usage Example

from mindsql.core import MindSQLCore
from mindsql.databases import MariaDB
from mindsql.vectorstores import MariaDBVectorStore
from mindsql.llms import GoogleGenAi

# Setup
database = MariaDB()
vectorstore = MariaDBVectorStore(config={
    'host': 'localhost',
    'port': 3306,
    'user': 'root',
    'password': 'password',
    'database': 'mydb'
})
llm = GoogleGenAi(config={'api_key': 'your-key'})

mindsql = MindSQLCore(database=database, vectorstore=vectorstore, llm=llm)

# Connect and query
connection = database.create_connection('mariadb://user:pass@localhost:3306/mydb')
tables = database.get_table_names(connection, 'mydb')['table_name'].tolist()

# Ask questions in natural language
sql = mindsql.create_database_query(
    question="Show me customers who ordered this month",
    connection=connection,
    tables=tables
)
results = database.execute_sql(connection, sql)