Files
atc-agents/infra/airflow/scripts/generate_mongodb_events_data.py

91 lines
2.4 KiB
Python

#!/usr/bin/env python3
"""
Script to generate fake event data for MongoDB
Generates approximately 1GB of data
"""
import pymongo
import random
from datetime import datetime, timedelta
import uuid
import sys
# Database connection details
DB_HOST = "10.0.21.51"
DB_PORT = "27017"
DB_NAME = "supplychain"
COLLECTION_NAME = "events"
# Data generation settings
import os
TARGET_DOCUMENTS = int(os.getenv("GEN_ROWS", "5000")) # light, configurable via GEN_ROWS
BATCH_SIZE = min(5000, max(500, TARGET_DOCUMENTS))
# Sample data
EVENT_TYPES = ["INSERT", "UPDATE", "DELETE", "CREATE", "MODIFY"]
REGIONS = ["EU", "APAC", "LATAM", "NA", "EMEA"]
SOURCES = ["ERP", "WMS", "CRM", "SCM", "TMS"]
def generate_fake_event():
"""Generate a single fake event"""
event_id = str(uuid.uuid4())
event_type = random.choice(EVENT_TYPES)
region = random.choice(REGIONS)
source = random.choice(SOURCES)
# Random timestamp within the last year
days_ago = random.randint(0, 365)
ts = datetime.now() - timedelta(days=days_ago, hours=random.randint(0, 23),
minutes=random.randint(0, 59))
amount = random.uniform(100.0, 50000.0)
# Generate a long payload field (like the existing data)
payload = "X" * 500
return {
"event_id": event_id,
"type": event_type,
"region": region,
"source": source,
"amount": amount,
"ts": ts,
"payload": payload
}
def main():
print(f"Connecting to MongoDB at {DB_HOST}:{DB_PORT}...")
client = pymongo.MongoClient(f"mongodb://{DB_HOST}:{DB_PORT}/")
db = client[DB_NAME]
collection = db[COLLECTION_NAME]
print(f"Generating {TARGET_DOCUMENTS} events...")
print(f"Batch size: {BATCH_SIZE}")
total_generated = 0
batch = []
for i in range(TARGET_DOCUMENTS):
batch.append(generate_fake_event())
if len(batch) >= BATCH_SIZE:
collection.insert_many(batch)
total_generated += len(batch)
batch = []
if total_generated % 100000 == 0:
print(f"Generated {total_generated} documents...")
# Insert remaining documents
if batch:
collection.insert_many(batch)
total_generated += len(batch)
client.close()
print(f"Completed! Generated {total_generated} events.")
if __name__ == "__main__":
main()