Fixturize

Data Subsetting Without the Invoice

Subset and mask PostgreSQL production data in one step. Go from a 500 GB production database to a 2 GB masked test database with referential integrity intact and PII gone.

Get Fixturize GitHub Repository

How It Works

# Inspect your schema topology
fixturize inspect --connection "$DB" --root users

# Extract one org and everything connected to it
fixturize extract --connection "$DB" --root "organizations WHERE id = 42"

# Mask PII before it leaves the building
fixturize extract --connection "$DB" --root "organizations WHERE id = 42" --mask "users.email='u' || id || '@test.com'"

# Scan for PII you might have missed
fixturize analyze --connection "$DB" --root organizations

# Load into your test database
fixturize apply --connection "$TEST_DB" fixtures/organizations.json

# Wipe and reload for a fresh start
fixturize apply --connection "$TEST_DB" --force fixtures/organizations.json

Features

Database Subsetting

Follows foreign keys in both directions, children down and parents up, to extract self-contained data subgraphs. Go from terabytes to gigabytes without breaking a single reference.

Built-in Data Masking

Mask PII with SQL expressions during extraction. Emails, names, and credit cards are anonymized before data leaves production. No separate masking step, no extra tools.

Automatic PII Discovery

The analyze command scans column names and types for 30+ PII categories. Catches what you missed, with ready-to-use masking expressions and confidence scoring.

Referential Integrity

Topologically sorted inserts respect FK order. Circular dependencies handled automatically via deferred constraints. No dangling references, ever.

Developer-First Workflow

Store connection strings, mask policies, and extraction defaults in .fixturize.yaml. Fits into CI/CD pipelines and local dev loops alike.

Flexible Output

Export as JSON for inspection or SQL for direct loading. Cap child tables with --limit, control traversal depth with --depth.

Installation

go install github.com/boringSQL/fixturize/cmd/fixturize@latest

Why Fixturize?

Your staging database is probably a full copy of production. It's slow to refresh, expensive to store, and if anyone forgot to mask the PII, a compliance incident waiting to happen.

Fixturize takes a different approach: extract only the data you need, mask it on the way out, and load it anywhere. One tool, one pipeline, one command.

Subset, don't copy. Pick a root record and fixturize walks the foreign key graph in both directions to collect a self-contained slice. Terabytes become gigabytes.
Mask inline. PII is anonymized during extraction, so sensitive data never lands on a developer laptop or CI runner in the clear.
Stay in sync. Re-run the same extraction when production schema evolves. Your test data stays fresh and realistic without manual upkeep.

Use Cases

Production-realistic dev environments

Every developer gets a small, masked slice of production instead of a stale seed file from 2019. Bugs that only surface on real data shapes show up locally, before they reach staging.

Test data for CI/CD pipelines

Generate a fixture once, check it into your repo, and fixturize apply it in every pipeline run. Consistent, referentially intact, and small enough that CI stays fast.

Need to share data with a contractor or move it to a non-production environment? Masking happens at extraction time, so PII never leaves the production perimeter unprotected.