datavow 0.2.2

A solemn vow on your data. From YAML to verdict.

DataVow

A solemn vow on your data. From YAML to verdict.

Data contract enforcement for modern data teams. Define contracts in YAML. Validate with DuckDB. Block in CI. Report for stakeholders.

---

Why DataVow?

89% of data teams report pain points with data modeling and ownership. Data contracts are the answer — but the tooling is fragmented:

• dbt tests: SQL-only, no formal contract, no pre-ingestion validation
• Great Expectations: verbose Python, steep learning curve
• Soda: good YAML checks, but no CI-native workflow or stakeholder reporting
• ODCS v3.1: promising standard, but no complete implementation

DataVow fills the gap: one tool from contract definition to validation, CI blocking, and human-readable reports. Built on ODCS v3.1 and powered by DuckDB.

Install

pip install datavow

Quick start

# 1. Scaffold a project
datavow init my-project

# 2. Write a contract (or edit the example)
cat contracts/orders.yaml

# 3. Validate data against the contract
datavow validate contracts/orders.yaml data/orders.csv

# 4. Generate a stakeholder report
datavow report contracts/orders.yaml data/orders.csv

# 5. Run in CI — exit 1 on critical failures
datavow ci contracts/ data/

Commands

datavow init [project-name]

Scaffold a new project with a datavow.yaml config and a contracts/ directory.

datavow define <contract.yaml>

Validate a contract's YAML syntax and display its structure — fields, rules, SLA — without needing data.

✓ Contract orders is valid

Name              orders
Version           1.0.0
Domain            sales
Owner             [e-mail]

Schema: 5 fields (5 required, 1 PII)
  • order_id integer (required, unique)
  • customer_email string (required, pii)
  • total_amount decimal (required)
  • status string (required)
  • created_at timestamp (required)

Quality rules: 3
  • no_negative_totals CRITICAL (sql)
  • email_not_null CRITICAL (not_null)
  • daily_volume WARNING (row_count)

SLA: freshness=24h, completeness=99.5%

datavow validate <contract.yaml> <source>

Run schema, quality, and freshness checks against a data source (CSV, Parquet, JSON).

datavow validate contracts/orders.yaml data/orders.csv --verbose
datavow validate contracts/orders.yaml data/orders.csv --ci        # exit 1 on CRITICAL
datavow validate contracts/orders.yaml data/orders.csv -o json     # JSON output
datavow validate contracts/orders.yaml data/orders.csv -o summary  # one-liner

datavow report <contract.yaml> <source>

Generate a self-contained HTML or Markdown report. Share it with stakeholders, attach to deliveries, or publish.

datavow report contracts/orders.yaml data/orders.csv                    # HTML (default)
datavow report contracts/orders.yaml data/orders.csv -f md              # Markdown
datavow report contracts/orders.yaml data/orders.csv -o my-report.html  # custom path

datavow ci <contracts_dir> <sources_dir>

Batch-validate all contracts against matching data sources. Matches by name convention: contracts/orders.yaml → sources/orders.csv.

datavow ci contracts/ data/                    # exit 1 on CRITICAL
datavow ci contracts/ data/ --fail-on warning  # stricter: fail on WARNING too

Contract format

DataVow contracts are a superset of ODCS v3.1 — compatible but extended with severity, SLA, and PII flags.

apiVersion: datavow/v1
kind: DataContract
metadata:
  name: orders
  version: 1.0.0
  owner: [e-mail]
  domain: sales
  description: "Customer orders from the e-commerce platform"
  tags: [pii, financial, critical]

schema:
  type: table
  fields:
    - name: order_id
      type: integer
      required: true
      unique: true
    - name: customer_email
      type: string
      required: true
      pii: true
      pattern: "^[a-zA-Z0-9_.+-]+@[a-zA-Z0-9-]+\\.[a-zA-Z0-9-.]+$"
    - name: total_amount
      type: decimal
      required: true
      min: 0
    - name: status
      type: string
      required: true
      allowed_values: [pending, confirmed, shipped, delivered, cancelled]
    - name: created_at
      type: timestamp
      required: true

quality:
  rules:
    - name: no_negative_totals
      type: sql
      query: "SELECT COUNT(*) FROM {table} WHERE total_amount < 0"
      threshold: 0
      severity: CRITICAL
    - name: email_not_null
      type: not_null
      field: customer_email
      severity: CRITICAL
    - name: daily_volume
      type: row_count
      min: 1000
      max: 100000
      severity: WARNING

sla:
  freshness: 24h
  completeness: "99.5%"

Supported field types

string, integer, float, decimal, boolean, date, timestamp

Supported quality rule types

Type	Description	Required fields
sql	Custom SQL query returning a count	query, threshold
not_null	Field has no nulls	field
unique	Field values are unique	field
row_count	Row count within bounds	min, max
range	Field values within bounds	field, min_value, max_value
accepted_values	Field values in allowed set	field, values
regex	Field values match pattern	field, pattern

Severity levels

Each rule has a severity: CRITICAL, WARNING, or INFO.

Vow Score

Score = 100 - (20 × CRITICAL + 5 × WARNING + 1 × INFO)

95-100  ✅ Vow Kept       — fully compliant
80-94   ⚠️ Vow Strained   — action needed
50-79   🔧 Vow Broken     — blocking issues
0-49    ❌ Vow Shattered   — critical violations

Data Mesh ready

Contracts are organized by domain. Each contract has a metadata.domain field. Structure your repo naturally:

contracts/
├── sales/
│   ├── orders.yaml
│   └── invoices.yaml
├── logistics/
│   └── shipments.yaml
└── finance/
    └── transactions.yaml

Tech stack

Component	Technology
Language	Python 3.12+
CLI	Typer + Rich
Contract parsing	Pydantic v2
Data validation	DuckDB
Reporting	Jinja2
Data formats	CSV, Parquet, JSON (via DuckDB)

Development

git clone https://github.com/ludovicschmetz-stack/datavow.git
cd datavow
python -m venv .venv && source .venv/bin/activate
pip install -e '.[dev]'
pytest tests/ -v

Roadmap

• Phase 1 — CLI MVP (done)
• Phase 2 — Integrations: dbt post-hook, Airflow operator, GitHub Action, PostgreSQL/MySQL via DuckDB, Slack/Teams notifications
• Phase 3 — SaaS: web dashboard, contract catalogue, role-based access, API

License

Apache 2.0 — free and open source forever. The CLI stays free. Monetization comes from the SaaS (Phase 3).

Author

Built by Ludovic Schmetz — Senior Data Engineer/Architect, Luxembourg. Also the author of Olympus.