fakedata_referenced_table.py

from faker import Faker
import random
import csv

fake = Faker()

# Number of users and orders to generate
num_users = 100
num_orders = 20000

# Step 1: Generate Data for the users Table (Parent Table)
users = []
for user_id in range(1, num_users + 1):
    user = {
        'user_id': str(fake.random_int(min=1, max=99999999)).zfill(8),
        'first_name': fake.first_name(),
        'last_name': fake.last_name(),
        'email': fake.email()
    }
    users.append(user)

# Save users data to a CSV file (Optional)
with open('users.csv', mode='w', newline='') as file:
    writer = csv.writer(file)
    writer.writerow(['user_id', 'first_name', 'last_name', 'email'])
    for user in users:
        writer.writerow([user['user_id'], user['first_name'], user['last_name'], user['email']])

# Step 2: Generate Data for the orders Table (Child Table)
orders = []
for order_id in range(1, num_orders + 1):
    order = {
        'order_id': order_id,
        'user_id': random.choice(users)['user_id'],  # Pick a random user_id from the users list
        'product_name': fake.word(),
        'order_date': fake.date_this_year().strftime('%Y-%m-%d')
    }
    orders.append(order)

# Save orders data to a CSV file (Optional)
with open('orders.csv', mode='w', newline='') as file:
    writer = csv.writer(file)
    writer.writerow(['order_id', 'user_id', 'product_name', 'order_date'])
    for order in orders:
        writer.writerow([order['order_id'], order['user_id'], order['product_name'], order['order_date']])

print("Data for 'users' and 'orders' tables generated successfully.")