fabrikate 0.4.1

Generate contextually coherent mock data — people, companies, products, and transactions that actually make sense together. 11 locales, 1,000 names per locale, 25+ fields per person, realistic distributions, zero dependencies.

fabrikate 🏭

Contextually coherent mock data generation for Python.

Generate fake people, companies, products, and transactions where everything makes sense together. A person in Tokyo gets a Japanese name, Japanese phone number, Japanese bank, My Number, a Toyota, and pays via PayPay. A German company has a GmbH suffix, a VAT number, and pays salaries in EUR.

Zero dependencies. Pure Python. 11 locales. Up to 2 million unique names per locale (1,000 first names × 1,000 last names, per gender). 25+ fields per person.

pip install fabrikate

Quick Start

from fabrikate import World

# Create a world set in India
world = World(locale="hi_IN", seed=42, industry="tech")

# Generate a person — everything is Indian and coherent
person = world.person()
print(person.full_name)        # "Pooja Reddy"
print(person.phone)            # "[phone]"
print(person.phone_carrier)    # "Airtel"
print(person.national_id)      # "9600 1338 9083" (Aadhaar format)
print(person.salary)           # 1445867.41 (INR)
print(person.credit_card_type) # "RuPay"
print(person.vehicle.brand)    # "Maruti Suzuki"
print(person.medical.blood_type) # "O+"
print(person.education.university) # "IIT Delhi"

# Generate a company with employees
company = world.company()
employees = company.hire(10)

# Generate linked transactions
tx = world.transaction(sender=person, receiver=company)
print(tx.payment_method)  # "UPI"
print(tx.tax_amount)      # 18% GST applied

# Export everything as JSON
print(world.to_json())

Why fabrikate?

Faker generates random data — a person might have an American name, a Japanese phone number, and a German address. fabrikate keeps everything coherent within a cultural and business context.

Feature	Faker	fabrikate
Random names	✓	✓
Locale-aware formatting	✓	✓
Coherent person (name + phone + bank + address)	✗	✓
Locale-specific payment methods (UPI, Pix, Zelle)	✗	✓
Locale-weighted blood types, car brands, carriers	✗	✓
National ID formats (Aadhaar, SSN, NI Number)	✗	✓
Company → employee context inheritance	✗	✓
Industry-appropriate job titles and salaries	✗	✓
Realistic distributions (log-normal, bell curve)	✗	✓
Linked transactions with tax and invoices	✗	✓
Full relational dataset export	✗	✓
Zero dependencies	✗	✓

Supported Locales

Code	Country	Currency	Phone	National ID	Top Payment
en_US	United States	USD ($)	+1	SSN	Credit Card
ja_JP	Japan	JPY (¥)	+81	My Number	Credit Card
de_DE	Germany	EUR (€)	+49	Personalausweis	Bank Transfer
pt_BR	Brazil	BRL (R$)	+55	CPF	Pix
hi_IN	India	INR (₹)	+91	Aadhaar	UPI
en_GB	United Kingdom	GBP (£)	+44	NI Number	Debit Card
fr_FR	France	EUR (€)	+33	INSEE	Credit Card
ko_KR	South Korea	KRW (₩)	+82	RRN	Credit Card
es_ES	Spain	EUR (€)	+34	DNI	Credit Card
ar_SA	Saudi Arabia	SAR (﷼)	+966	National ID	Mada Card
en_NG	Nigeria	NGN (₦)	+234	NIN	Bank Transfer

Each locale has 1,000 male first names, 1,000 female first names, and 1,000 last names — yielding up to 2 million unique full name combinations (1,000 × 1,000 per gender). Plus 50 cities, locale-specific banks, email domains, company suffixes, street formats, and more.

Core Concepts

World

A World is your mock data universe. Everything generated within a world shares context and can reference each other.

world = World(
    locale="de_DE",     # Country/culture context
    seed=123,           # Random seed for reproducibility (optional)
    year=2025,          # Reference year for dates and ages
    industry="finance", # Default industry for companies/products
)

Parameters:

• locale (str, default "en_US"): Determines the cultural context — names, phone formats, currency, banks, addresses, ID formats, payment methods, car brands, blood type distributions, and everything else. Use codes like "hi_IN", "ja_JP", "de_DE".
• seed (int or None): When set, the same seed always produces identical data. Critical for reproducible test fixtures. World(seed=42) will generate the exact same people, companies, and transactions every time.
• year (int, default 2025): The reference year. Ages, dates of birth, graduation years, transaction dates, and vehicle model years are all calculated relative to this.
• industry (str or None): Sets the default industry context. Affects company names, job titles, product names, and salary ranges. Options: "tech", "finance", "health", "retail", "food", "manufacturing".

Person

A person has 25+ fields, all coherent within the locale:

person = world.person()

# --- Identity ---
person.id                # UUID
person.first_name        # Locale-appropriate first name
person.last_name         # Locale-appropriate last name
person.full_name         # Ordered correctly (family-first in Japan/Korea)
person.gender            # "male" or "female"
person.age               # Bell curve centered at 35
person.date_of_birth     # "1990-03-15" — consistent with age and year
person.nationality       # "India", "Deutschland", "日本", etc.
person.national_id       # Locale format: "9600 1338 9083" (Aadhaar)
person.marital_status    # Age-dependent: "single", "married", "divorced", "widowed"
person.children          # Locale-weighted: higher in Nigeria/Saudi, lower in Japan/Korea

# --- Contact ---
person.email             # Derived from name + locale domain (yahoo.co.jp, gmail.com)
person.phone             # Locale format: "[phone]"
person.phone_carrier     # "Jio", "Airtel", "Verizon", "Docomo", etc.
person.username          # Derived from name: "pooja.reddy128"
person.address           # Address object with street, city, postal_code, country
person.address.street    # Locale street format: "1675, Nehru Nagar"
person.address.city      # "Pune", "München", "東京"
person.address.postal_code  # Locale format: "559407", "SW1A 1AA"

# --- Financial ---
person.bank              # Locale bank: "HDFC Bank", "三菱UFJ銀行"
person.credit_card_type  # Locale-weighted: "RuPay" in India, "JCB" in Japan
person.salary            # Log-normal distribution, locale-scaled
person.salary_currency   # "INR", "EUR", "USD"

# --- Employment ---
person.job_title         # Industry-appropriate: "Software Engineer", "Portfolio Manager"
person.department        # "Engineering", "Finance", "Marketing"
person.company_id        # Links to company (set by company.hire())

# --- Education ---
person.education.university      # Locale university: "IIT Delhi", "MIT", "東京大学"
person.education.degree          # Weighted: "Bachelor's" (55%), "Master's" (25%), "PhD" (5%)
person.education.field_of_study  # "Computer Science", "Economics", etc.
person.education.graduation_year # Consistent with age and degree type

# --- Vehicle (70% chance) ---
person.vehicle                # None if no vehicle
person.vehicle.brand          # Locale-weighted: "Maruti Suzuki" in India, "Toyota" in Japan
person.vehicle.year           # Recent model year
person.vehicle.plate_number   # Locale format: "CH 16 IO 5255"

# --- Medical ---
person.medical.blood_type          # Real regional distributions (O+ 51% in Nigeria, A+ 40% in Japan)
person.medical.height_cm           # Bell curve by locale and gender
person.medical.weight_kg           # Bell curve by locale and gender
person.medical.insurance_provider  # "Star Health" (India), "NHS" (UK), "Blue Cross" (US)

Override any field:

# Force specific values
person = world.person(
    gender="female",
    age=30,
    job_title="CEO",
    salary=500000,
)

Company

company = world.company()

company.id               # UUID
company.name             # Industry + locale appropriate: "CloudPrime GmbH"
company.industry         # "tech", "finance", "health", etc.
company.founded_year     # Random year between 1970 and now
company.employee_count   # 5, 12, 25, 50, 100, 250, 500, or 1000
company.currency         # Locale currency
company.bank             # Locale bank
company.city             # Locale city
company.country          # Locale country name
company.tax_id           # Locale format: "DE960013389" (German VAT), "##-#######" (US EIN)
company.annual_revenue   # Scales with employee count, log-normal
company.website          # Derived from company name
company.departments      # List of departments, scales with size
company.employees        # List of hired Person objects

Hiring employees:

company = world.company(industry="tech")
team = company.hire(10)  # 10 employees linked to this company

for person in team:
    print(person.company_id)  # == company.id
    print(person.job_title)   # Tech job titles
    print(person.address.country)  # Same locale as company

Product

product = world.product(company=company)

product.id          # UUID
product.name        # Industry-appropriate: "Cloud Platform", "Vita Formula"
product.sku         # "CLO-4832"
product.price       # Industry-appropriate range
product.currency    # Locale currency
product.category    # Matches industry
product.company_id  # Links to company

Transaction

tx = world.transaction(sender=person, receiver=company)

tx.id               # UUID
tx.type             # "purchase", "refund", "subscription", "transfer", "payment", "invoice"
tx.amount           # Log-normal: many small, few large
tx.tax_amount       # Locale tax rate applied (18% GST in India, 19% in Germany, 8% in US)
tx.total_amount     # amount + tax
tx.currency         # Locale currency
tx.status           # Weighted: 70% completed, 15% pending, 10% failed, 5% cancelled
tx.date             # Random date within the context year
tx.payment_method   # Locale-weighted: "UPI" (India), "Pix" (Brazil), "EC Card" (Germany)
tx.invoice_number   # "INV-2025-74042"
tx.sender_id        # Links to sender
tx.receiver_id      # Links to receiver
tx.product_id       # Links to product (optional)
tx.description      # Human-readable summary

Context and Child Contexts

The Context is the engine behind coherence. You usually don't create one directly — World handles it. But you can create child contexts for scenarios like multinational companies:

# Japanese HQ
world = World(locale="ja_JP", seed=42, industry="tech")
hq = world.company()

# American branch — child context overrides locale but inherits industry
us_ctx = world.ctx.child(locale="en_US")
branch_employees = [Person.generate(us_ctx, company_id=hq.id) for _ in range(5)]
# These employees have American names/phones but work for the Japanese company

Full Dataset Export

Everything you generate is tracked and exportable:

world = World(locale="hi_IN", seed=42)
company = world.company(industry="tech")
company.hire(50)
world.products(10, company=company)
world.transactions(200)

# As a Python dict
data = world.dataset()
# {
#   "metadata": {"locale": "hi_IN", "seed": 42, "year": 2025, "industry": None},
#   "people": [...],      # 50 people
#   "companies": [...],   # 1 company
#   "products": [...],    # 10 products
#   "transactions": [...] # 200 transactions
# }

# As a JSON string
json_str = world.to_json(indent=2)

# Write to file
with open("test_fixtures.json", "w") as f:
    f.write(world.to_json())

Overrides

Every generator accepts **overrides to force specific values:

# Override person fields
person = world.person(
    first_name="Raj",
    age=28,
    gender="male",
    salary=1200000,
    job_title="CTO",
    marital_status="single",
)

# Override company fields
company = world.company(
    name="Acme Corp",
    industry="tech",
    employee_count=500,
    founded_year=2010,
)

# Override transaction fields
tx = world.transaction(
    type="purchase",
    amount=9999.99,
    payment_method="UPI",
    status="completed",
)

Realistic Distributions

fabrikate doesn't use uniform random — it uses real-world statistical distributions:

• Names: Common names appear more frequently (skewed distribution)
• Ages: Bell curve centered at 35 for working populations
• Salaries: Log-normal (many average, few very high) scaled per locale
• Transaction amounts: Log-normal (many small purchases, few large ones)
• Blood types: Real regional frequencies (O+ is 51% in Nigeria, A+ is 40% in Japan)
• Children: Locale-weighted (avg 4-5 in Nigeria/Saudi, 1-2 in Japan/Korea)
• Marital status: Age-dependent (younger = more single, older = more married/widowed)
• Car brands: Market-share weighted (Maruti 35% in India, Toyota 30% in Japan)
• Payment methods: Usage-share weighted (UPI 45% in India, Pix 40% in Brazil)
• Credit cards: Regional preference (RuPay 35% in India, JCB 35% in Japan)

Locale-Specific Data

Each locale includes culturally accurate data for:

Data Point	Example (India)	Example (Japan)	Example (Germany)
Phone format	+91 #####-#####	+81 ##-####-####	+49 ### #######
National ID	Aadhaar (12 digits)	My Number (12 digits)	Personalausweis
Tax ID	GSTIN	Corporate Number	USt-IdNr (DE#########)
Top bank	HDFC Bank	三菱UFJ銀行	Deutsche Bank
Top carrier	Jio	NTT Docomo	Telekom
Top car brand	Maruti Suzuki (35%)	Toyota (30%)	Volkswagen (22%)
Payment method	UPI (45%)	Credit Card (30%)	Bank Transfer (30%)
Credit card	RuPay (35%)	JCB (35%)	Girocard (20%)
Blood type O+	36.5%	30%	35%
Avg children	2-3	1-2	1-2
Salary range	₹3L – ₹35L	¥3M – ¥15M	€30K – €140K
Universities	IIT Bombay, BITS	東京大学, 早稲田	TU München, RWTH
Insurance	Star Health, LIC	National Health	TK, AOK, Barmer
Company suffix	Pvt. Ltd., LLP	株式会社	GmbH, AG
Street format	MG Road, Nehru Nagar	渋谷区本町3丁目	Goethestraße 42

Supported Industries

Six industries with appropriate job titles, company names, product names, and price ranges:

Industry	Job Titles	Company Keywords	Product Price Range
tech	Software Engineer, CTO, Data Scientist	Cloud, Digital, AI, Quantum	$29.99 – $999.99
finance	Financial Analyst, Portfolio Manager	Capital, Trust, Equity	$9.99 – $499.99
health	Doctor, Pharmacist, Surgeon	Health, Vita, Bio, Care	$12.99 – $149.99
retail	Store Manager, Buyer, E-commerce Manager	Market, Store, Fresh	$4.99 – $299.99
food	Head Chef, Food Scientist, Sommelier	Fresh, Harvest, Kitchen	$3.99 – $79.99
manufacturing	Production Manager, Quality Engineer	Steel, Forge, Precision	$99.99 – $9,999.99

Use Cases

Test fixtures: Generate deterministic, realistic test data for your app:

world = World(locale="en_US", seed=42)
users = world.people(100)
# Same 100 users every time, with coherent data

Database seeding: Populate a dev database with realistic data:

world = World(locale="hi_IN", seed=1, industry="tech")
for _ in range(5):
    company = world.company()
    company.hire(20)
world.transactions(500)
data = world.dataset()
# Insert into your DB

Demo environments: Create convincing demo data for presentations:

world = World(locale="de_DE", industry="finance")
company = world.company(name="FinanzPrime GmbH", employee_count=250)
company.hire(50)

Load testing: Generate large volumes of realistic data:

world = World(locale="en_US", seed=1)
people = world.people(10000)     # 10K people — tiny fraction of 2M unique name combinations
txns = world.transactions(100000) # 100K transactions

API mocking: Return realistic responses in mock APIs:

world = World(locale="ja_JP", seed=1)
person = world.person()
return {"user": person.to_dict()}

Requirements

• Python 3.9+
• Zero external dependencies

License

MIT — Sai Aditya Datta Tipirneni

Links

• PyPI: https://pypi.org/project/fabrikate/
• GitHub: https://github.com/aditya36900/fabrikate
• Issues: https://github.com/aditya36900/fabrikate/issues