Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Updating the swarm sample, to include actual product data. #7

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

Updating the swarm sample, to include actual product data. #7

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
253 changes: 139 additions & 114 deletions src/app/azure_cosmos_db.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,73 +1,99 @@
import json
import os
import sys
import uuid
from typing import List, Optional

import azure_open_ai
from azure.cosmos import ContainerProxy, CosmosClient, PartitionKey, exceptions

# Initialize CosmosDB Client

from azure.cosmos import CosmosClient, PartitionKey, exceptions
COSMOS_DB_URL = os.getenv("COSMOS_DB_URL")
COSMOS_DB_KEY = os.getenv("COSMOS_DB_KEY")
AZURE_OPENAI_API_KEY = os.getenv("AZURE_OPENAI_API_KEY")
DATABASE_NAME = "ProductAssistant"
PRODUCTS_CONTAINER = "Products"
USERS_CONTAINER = "Users"
PURCHASE_HISTORY_CONTAINER = "PurchaseHistory"

# Initialize the Cosmos client
# reference environment variables for the values of these variables
endpoint = os.environ['AZURE_COSMOSDB_ENDPOINT']
key = os.environ['AZURE_COSMOSDB_KEY']
client = CosmosClient(endpoint, key)
client = CosmosClient(COSMOS_DB_URL, COSMOS_DB_KEY)
client.create_database_if_not_exists(DATABASE_NAME)
database = client.get_database_client(DATABASE_NAME)

# Database and container names
database_name = "MultiAgentDemoDB"
users_container_name = "Users"
purchase_history_container_name = "PurchaseHistory"
products_container_name = "Products"

# Create database and containers if they don't exist
def create_database():
# Create Containers with Vector and Full-Text Indexing Policies
def create_containers():
try:
database = client.create_database_if_not_exists(id=database_name)
users_container = database.create_container_if_not_exists(
id=users_container_name,
id=USERS_CONTAINER,
partition_key=PartitionKey(path="/user_id"),
offer_throughput=400
offer_throughput=400,
)

print(
f"Container {USERS_CONTAINER} created."
)

purchase_history_container = database.create_container_if_not_exists(
id=purchase_history_container_name,
id=PURCHASE_HISTORY_CONTAINER,
partition_key=PartitionKey(path="/user_id"),
offer_throughput=400
)

print(
f"Container {PURCHASE_HISTORY_CONTAINER} created."
)

vector_embedding_policy = {
"vectorEmbeddings": [
{
"path": "/product_description_vector",
"path": "/embedding",
"dataType": "float32",
"dimensions": 1536,
"distanceFunction": "cosine",
"dimensions": 1536
},
}
]
}
diskann_indexing_policy = {
"includedPaths": [
{"path": "/*"}
],
"excludedPaths": [
{"path": "/\"_etag\"/?"}
],
"vectorIndexes": [

full_text_policy = {
Copy link
Contributor

@TheovanKraay TheovanKraay Feb 24, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You have added full_text_policy but I can't see full text search being used anywhere? If not being used, I would say better to remove it.

"defaultLanguage": "en-US",
"fullTextPaths": [
{
"path": "/product_description_vector",
"type": "diskANN",
"path": "/product_description",
"language": "en-US",
}
]
}

indexing_policy = {
"indexingMode": "consistent",
"includedPaths": [{"path": "/*"}],
"excludedPaths": [{"path": '/"_etag"/?'}],
"vectorIndexes": [{"path": "/embedding", "type": "diskANN"}],
"fullTextIndexes": [{"path": "/product_description"}],
}

products_container = database.create_container_if_not_exists(
id=products_container_name,
partition_key=PartitionKey(path="/product_id"),
id=PRODUCTS_CONTAINER,
partition_key=PartitionKey(path="/category"),
offer_throughput=400,
vector_embedding_policy=vector_embedding_policy,
indexing_policy=diskann_indexing_policy
full_text_policy=full_text_policy,
indexing_policy=indexing_policy,
)

print(
f"Container {PRODUCTS_CONTAINER} created with vector and full-text search indexing."
)
except exceptions.CosmosHttpResponseError as e:
print(f"Database creation failed: {e}")
print(f"Container creation failed: {e}")


def add_user(user_id, first_name, last_name, email, phone):
database = client.get_database_client(database_name)
container = database.get_container_client(users_container_name)
container = database.get_container_client(USERS_CONTAINER)
user = {
"id": str(user_id),
"id": str(uuid.uuid4()),
"user_id": user_id,
"first_name": first_name,
"last_name": last_name,
Expand All @@ -79,29 +105,32 @@ def add_user(user_id, first_name, last_name, email, phone):
except exceptions.CosmosResourceExistsError:
print(f"User with user_id {user_id} already exists.")

def add_purchase(user_id, date_of_purchase, item_id, amount):
database = client.get_database_client(database_name)
container = database.get_container_client(purchase_history_container_name)

def add_purchase(user_id, date_of_purchase, item_id, amount, product_name, category):
container = database.get_container_client(PURCHASE_HISTORY_CONTAINER)
purchase = {
"id": f"{user_id}_{item_id}_{date_of_purchase}",
"id": str(uuid.uuid4()),
"user_id": user_id,
"date_of_purchase": date_of_purchase,
"item_id": item_id,
"product_id": item_id,
"product_name": product_name,
"category": category,
"amount": amount
}
try:
container.create_item(body=purchase)
except exceptions.CosmosResourceExistsError:
print(f"Purchase already exists for user_id {user_id} on {date_of_purchase} for item_id {item_id}.")

def add_product(product_id, product_name, product_description, price):
database = client.get_database_client(database_name)
container = database.get_container_client(products_container_name)

def add_product(product_id, product_name, category, product_description, price):
container = database.get_container_client(PRODUCTS_CONTAINER)
product_description_vector = azure_open_ai.generate_embedding(product_description)
product = {
"id": str(product_id),
"id": str(uuid.uuid4()),
"product_id": product_id,
"product_name": product_name,
"category": category,
"product_description": product_description,
"product_description_vector": product_description_vector,
"price": price
Expand All @@ -111,72 +140,68 @@ def add_product(product_id, product_name, product_description, price):
except exceptions.CosmosResourceExistsError:
print(f"Product with product_id {product_id} already exists.")

def preview_table(container_name):
database = client.get_database_client(database_name)
container = database.get_container_client(container_name)
items = container.query_items(
query="SELECT * FROM c",
enable_cross_partition_query=True

def process_and_insert_data(
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this could do with being parallelized and increasing default RU to 10000 for the container, or it takes quite a long time.

filename: str,
container: ContainerProxy,
vector_field: Optional[str] = None,
full_text_fields: Optional[List[str]] = None,
):
if not os.path.exists(filename):
print(f"File {filename} not found.")
return

with open(filename, "r") as f:
data = json.load(f)

if len(data) > 300:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think this logic means that some records are skipped if number of json docs in the file is greater than 300? What is this for?

data = data[226:]

for entry in data:
if full_text_fields is not None:
for field in full_text_fields:
if field in entry and isinstance(entry[field], list):
entry[field] = [", ".join(map(str, entry[field]))]

# Generate vector embedding
if vector_field and vector_field in entry and isinstance(entry[vector_field], str):
entry["embedding"] = azure_open_ai.generate_embedding(entry[vector_field])

# Insert into CosmosDB
entry["id"] = str(uuid.uuid4())
size = sys.getsizeof(json.dumps(entry))
if size > 2 * 1024 * 1024: # 2MB in bytes
print(f"Document {entry['id']} is too large: {size} bytes")
container.upsert_item(entry)

print(f"Inserted data from {filename} into {container.id}.")


def main():
# Create Containers
create_containers()

products_container = database.get_container_client(PRODUCTS_CONTAINER)
users_container = database.get_container_client(USERS_CONTAINER)
purchase_history_container = database.get_container_client(PURCHASE_HISTORY_CONTAINER)

# Insert data into CosmosDB with embedding and indexing
file_prefix = "/Users/aayushkataria/git/multi-agent-swarm/src/data/"
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

hardcoded

process_and_insert_data(
file_prefix + "final_products.json",
products_container,
"product_description",
["product_description"],
)
for item in items:
if (container_name == products_container_name):
# redact the product description vector
item.pop("product_description_vector", None)
print(item)

# Initialize and load database
def initialize_database():
create_database()

# Add some initial users
initial_users = [
(1, "Alice", "Smith", "alice@test.com", "123-456-7890"),
(2, "Bob", "Johnson", "bob@test.com", "234-567-8901"),
(3, "Sarah", "Brown", "sarah@test.com", "555-567-8901"),
# Add more initial users here
]

for user in initial_users:
add_user(*user)

# Add some initial purchases
initial_purchases = [
(1, "2024-01-01", 101, 99.99),
(2, "2023-12-25", 100, 39.99),
(3, "2023-11-14", 307, 49.99),
]

for purchase in initial_purchases:
add_purchase(*purchase)

initial_products = [
(7, "Hat", "A hat is a stylish and functional accessory designed to shield the "
"head from the elements while adding a touch of personality to any outfit. "
"Crafted from materials such as wool, cotton, straw, or synthetic blends, hats come "
"in a variety of shapes and designs, from wide-brimmed sun hats to snug beanies and classic fedoras. "
"They offer versatile use, providing protection from sun, rain, or cold while serving as a "
"fashionable statement piece. Whether for outdoor adventures, formal occasions, "
"or casual outings, a hat combines practicality and style, making it a "
"timeless wardrobe essential", 19.99),
(8, "Wool socks", "Wool socks are premium, cozy footwear accessories designed "
"to provide exceptional warmth, comfort, and moisture-wicking properties. "
"Made from natural wool fibers, they are ideal for keeping feet insulated in "
"cold weather while remaining breathable in warmer conditions. These socks are soft, "
"durable, and naturally odor-resistant, making them perfect for everyday wear, "
"outdoor adventures, or lounging at home. With their ability to regulate "
"temperature and cushion feet, wool socks offer unparalleled comfort, "
"making them an essential addition to any wardrobe, whether for hiking, working, "
"or simply relaxing.",29.99),
(9, "Shoes","Shoes are versatile footwear designed to protect and comfort "
"the feet while enabling effortless movement and style. They "
"come in a wide range of designs, materials, and functions, catering "
"to various activities, from formal occasions to rugged outdoor adventures. "
"Crafted from durable materials such as leather, canvas, or synthetic blends, "
"shoes provide support, cushioning, and stability through features like rubber soles, "
"padded insoles, and secure fastenings. Available in diverse styles such as sneakers, boots, "
"sandals, and dress shoes, they blend functionality with aesthetic appeal, making them a staple "
"for every wardrobe",39.99),
]

for product in initial_products:
add_product(*product)
process_and_insert_data(file_prefix + "users.json", users_container)
process_and_insert_data(
file_prefix + "purchase_history.json",
purchase_history_container)

print(
"Data successfully inserted into CosmosDB with embeddings, vector search, and full-text search indexing!"
)


if __name__ == "__main__":
main()
Loading