SchemaX Python SDK & CLI

Declarative schema management and migration for Databricks Unity Catalog. Version control your catalog structure, generate SQL migrations, and deploy consistently across environments — from a single developer workflow through CI/CD.

What is schema management?

Schema management is defining and maintaining the structure of your data catalog — catalogs, schemas, tables, views, columns, constraints, and their metadata. In Unity Catalog, this is the layer that determines what exists and who can access it.

Schema migration is applying changes to that structure over time in a controlled way: add a column, create a table, change grants — and track those changes so they can be reviewed, versioned, and deployed per environment (dev → test → prod).

Without it, schema changes are ad hoc, hard to audit, and risky when promoting to production. With it, changes are declarative, stored in Git, and applied consistently via CI/CD or manual deployment.

Works alongside Spark and Lakeflow Declarative Pipelines

If your tables are created by Spark jobs or DLT pipelines, SchemaX complements that workflow. Use governance-only mode to version and deploy comments, tags, grants, row filters, and column masks on existing objects — without touching CREATE TABLE statements. Your pipelines handle the data; SchemaX handles the governance layer.

Features

Full Unity Catalog object support

• Catalogs and schemas — create, rename, update managed locations, comments, tags, grants
• Tables — managed and external (Delta, Iceberg), partitioning, liquid clustering, column mapping
• Views — definitions, dependency tracking with automatic SQL extraction via sqlglot
• Materialized views — definitions, refresh schedules, dependency ordering
• Volumes — managed and external with storage locations
• Functions — SQL and Python UDFs, table functions, parameters with types and defaults
• Columns — add, rename, drop, reorder, change types, nullability, comments, tags
• Constraints — primary keys, foreign keys, check constraints (with NOT ENFORCED / RELY)

Data governance

• Grants — GRANT and REVOKE on all securable types
• Tags — governance tags on catalogs, schemas, tables, views, and columns
• Row filters — row-level security policies
• Column masks — column-level data masking
• Table and view properties — TBLPROPERTIES configuration
• Governance-only mode — deploy only governance DDL, skip CREATE for pipeline-managed objects

Deployment and CI/CD

• Multi-environment — logical catalog names mapped to physical names per environment
• Apply with tracking — deploy via Databricks Statement Execution API with database-backed audit trail
• Rollback — partial (revert a failed deployment) and complete (to a snapshot), with safety classification
• Dry run — preview SQL without executing
• Auto-rollback — automatically revert on partial failure
• Deployment scope — governance-only mode, existing-object awareness
• CI/CD templates — GitHub Actions, GitLab CI, Azure DevOps
• Databricks Asset Bundles — generate DAB resource YAML with schemax bundle

Version control

• Snapshots — semantic versioned state captures
• Changelogs — every change tracked as a typed operation
• State diffing — compute minimal operations between any two snapshots
• Dependency-ordered SQL — correct creation ordering for views and materialized views
• Stale detection — detect and rebase snapshots after Git rebases

Installation

pip install schemaxpy

Quick start

1. Initialize a project

schemax init

Creates a .schemax/ directory with project configuration.

2. Import from a live workspace

schemax import --profile my-databricks --warehouse-id abc123

Brings your existing Unity Catalog hierarchy into a SchemaX project.

3. Make changes and generate SQL

Use the VS Code extension to design visually, or modify the state directly. Then:

# Preview SQL for an environment
schemax sql --target dev

# Save to file
schemax sql --target prod --output migration.sql

4. Deploy

# Dry run
schemax apply --target dev --profile my-databricks --warehouse-id abc123 --dry-run

# Apply
schemax apply --target dev --profile my-databricks --warehouse-id abc123

# With auto-rollback on failure
schemax apply --target prod --profile my-databricks --warehouse-id abc123 --auto-rollback

5. Rollback if needed

# Partial — revert a failed deployment
schemax rollback --partial --deployment <id> --target dev --profile my-databricks --warehouse-id abc123

# Complete — rollback to a snapshot
schemax rollback --to-snapshot v0.2.0 --target dev --profile my-databricks --warehouse-id abc123

CLI reference

Run schemax <command> --help for detailed options.

CI/CD integration

GitHub Actions

name: Deploy Schema
on:
  push:
    branches: [main]

jobs:
  deploy:
    runs-on: ubuntu-latest
    environment: production
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-python@v5
        with:
          python-version: '3.11'

      - run: pip install schemaxpy
      - run: schemax validate

      - name: Apply to production
        env:
          DATABRICKS_HOST: ${{ secrets.DATABRICKS_HOST }}
          DATABRICKS_TOKEN: ${{ secrets.DATABRICKS_TOKEN }}
        run: |
          schemax apply \
            --target prod \
            --profile default \
            --warehouse-id ${{ secrets.WAREHOUSE_ID }} \
            --no-interaction \
            --auto-rollback

Python API

from pathlib import Path
from schemax.core.storage import load_current_state, read_project, get_environment_config
from schemax.providers.base.operations import Operation

workspace = Path.cwd()
state, changelog, provider = load_current_state(workspace)

# Generate SQL with environment-specific catalog names
project = read_project(workspace)
env_config = get_environment_config(project, "prod")

catalog_mapping = {}
for catalog in state.get("catalogs", []):
    logical = str(catalog.get("name"))
    physical = env_config.get("catalogMappings", {}).get(logical, logical)
    catalog_mapping[logical] = physical

generator = provider.get_sql_generator(state)
generator.catalog_name_mapping = catalog_mapping

operations = [Operation(**op) for op in changelog["ops"]]
print(generator.generate_sql(operations))

Multi-provider roadmap

SchemaX is built on a provider architecture. Unity Catalog is fully supported today (v0.2.x). Lakebase (PostgreSQL) support is in active development for v0.3.x.

Requirements

• Python 3.11+
• A SchemaX project (.schemax/ directory)
• For deployment: Databricks workspace with SQL Warehouse access

Links

• Documentation — Setup, quickstart, and reference
• VS Code Extension — Visual schema designer
• GitHub Repository — Source code and issues
• PyPI

Apache License 2.0 — see LICENSE for details.