DataBridge Discovery Engine

Automated SQL parsing, CASE statement extraction, and hierarchy generation for data warehouse modeling.

Features

• SQL Parsing: Multi-dialect SQL parsing using sqlglot (Snowflake, PostgreSQL, T-SQL, MySQL, BigQuery)
• CASE Extraction: Automatic extraction of CASE WHEN statements with hierarchy detection
• Semantic Graph: Graph-based semantic modeling with NetworkX
• Entity Detection: Detects 12 standard entity types (account, cost_center, department, etc.)
• Librarian Integration: Direct export to Librarian hierarchy project format

Installation

# Basic installation
pip install databridge-discovery

# With embeddings support
pip install databridge-discovery[embeddings]

# With MCP tools
pip install databridge-discovery[mcp]

# Full installation
pip install databridge-discovery[all]

Quick Start

from databridge_discovery import SQLParser, CaseExtractor, DiscoverySession

# Parse SQL
parser = SQLParser(dialect="snowflake")
ast = parser.parse(sql_query)

# Extract CASE statements
extractor = CaseExtractor()
cases = extractor.extract(ast)

# Start discovery session
session = DiscoverySession()
session.add_sql_source(sql_query)
session.analyze()

# Get proposed hierarchies
hierarchies = session.get_proposed_hierarchies()

MCP Tools

The library provides 50 MCP tools across 7 phases:

Phase 1: SQL Parser & Session (6 tools)

• parse_sql - Parse SQL and return AST
• extract_case_statements - Extract CASE WHEN logic
• analyze_sql_complexity - Query complexity metrics
• start_discovery_session - Initialize discovery session
• get_discovery_session - Get session state
• export_discovery_evidence - Export evidence

Phase 2: Semantic Graph (8 tools)

• build_semantic_graph - Build from schema
• add_graph_relationship - Add edge
• find_join_paths - Find join candidates
• And more...

Phase 3-7: See full documentation

Entity Types

The discovery engine detects 12 standard entity types:

License

MIT