romanbredehoft-zama's picture
Adding deployment files and updating app
c119738
raw
history blame
15.9 kB
"""A local gradio app that filters images using FHE."""
import os
import shutil
import subprocess
import time
import gradio as gr
import numpy
import requests
from itertools import chain
from settings import (
REPO_DIR,
SERVER_URL,
FHE_KEYS,
CLIENT_FILES,
SERVER_FILES,
DEPLOYMENT_PATH,
INITIAL_INPUT_SHAPE,
INPUT_INDEXES,
START_POSITIONS,
)
from development.client_server_interface import MultiInputsFHEModelClient
subprocess.Popen(["uvicorn", "server:app"], cwd=REPO_DIR)
time.sleep(3)
def shorten_bytes_object(bytes_object, limit=500):
"""Shorten the input bytes object to a given length.
Encrypted data is too large for displaying it in the browser using Gradio. This function
provides a shorten representation of it.
Args:
bytes_object (bytes): The input to shorten
limit (int): The length to consider. Default to 500.
Returns:
str: Hexadecimal string shorten representation of the input byte object.
"""
# Define a shift for better display
shift = 100
return bytes_object[shift : limit + shift].hex()
def get_client(client_id, client_type):
"""Get the client API.
Args:
client_id (int): The client ID to consider.
client_type (str): The type of user to consider (either 'user', 'bank' or 'third_party').
Returns:
FHEModelClient: The client API.
"""
key_dir = FHE_KEYS / f"{client_type}_{client_id}"
return MultiInputsFHEModelClient(DEPLOYMENT_PATH, key_dir=key_dir)
def get_client_file_path(name, client_id, client_type):
"""Get the correct temporary file path for the client.
Args:
name (str): The desired file name (either 'evaluation_key' or 'encrypted_inputs').
client_id (int): The client ID to consider.
client_type (str): The type of user to consider (either 'user', 'bank' or 'third_party').
Returns:
pathlib.Path: The file path.
"""
return CLIENT_FILES / f"{name}_{client_type}_{client_id}"
def clean_temporary_files(n_keys=20):
"""Clean keys and encrypted images.
A maximum of n_keys keys and associated temporary files are allowed to be stored. Once this
limit is reached, the oldest files are deleted.
Args:
n_keys (int): The maximum number of keys and associated files to be stored. Default to 20.
"""
# Get the oldest key files in the key directory
key_dirs = sorted(FHE_KEYS.iterdir(), key=os.path.getmtime)
# If more than n_keys keys are found, remove the oldest
user_ids = []
if len(key_dirs) > n_keys:
n_keys_to_delete = len(key_dirs) - n_keys
for key_dir in key_dirs[:n_keys_to_delete]:
user_ids.append(key_dir.name)
shutil.rmtree(key_dir)
# Get all the encrypted objects in the temporary folder
client_files = CLIENT_FILES.iterdir()
server_files = SERVER_FILES.iterdir()
# Delete all files related to the ids whose keys were deleted
for file in chain(client_files, server_files):
for user_id in user_ids:
if user_id in file.name:
file.unlink()
def keygen(client_id, client_type):
"""Generate the private key associated to a filter.
Args:
client_id (int): The client ID to consider.
client_type (str): The type of client to consider (either 'user', 'bank' or 'third_party').
"""
# Clean temporary files
clean_temporary_files()
# Retrieve the client instance
client = get_client(client_id, client_type)
# Generate a private key
client.generate_private_and_evaluation_keys(force=True)
# Retrieve the serialized evaluation key. In this case, as circuits are fully leveled, this
# evaluation key is empty. However, for software reasons, it is still needed for proper FHE
# execution
evaluation_key = client.get_serialized_evaluation_keys()
# Save evaluation_key as bytes in a file as it is too large to pass through regular Gradio
# buttons (see https://github.com/gradio-app/gradio/issues/1877)
evaluation_key_path = get_client_file_path("evaluation_key", client_id, client_type)
with evaluation_key_path.open("wb") as evaluation_key_file:
evaluation_key_file.write(evaluation_key)
def send_input(client_id, client_type):
"""Send the encrypted input image as well as the evaluation key to the server.
Args:
client_id (int): The client ID to consider.
client_type (str): The type of client to consider (either 'user', 'bank' or 'third_party').
"""
# Get the paths to the evaluation key and encrypted inputs
evaluation_key_path = get_client_file_path("evaluation_key", client_id, client_type)
encrypted_input_path = get_client_file_path("encrypted_inputs", client_id, client_type)
# Define the data and files to post
data = {
"client_id": client_id,
"client_type": client_type,
}
files = [
("files", open(encrypted_input_path, "rb")),
("files", open(evaluation_key_path, "rb")),
]
# Send the encrypted input image and evaluation key to the server
url = SERVER_URL + "send_input"
with requests.post(
url=url,
data=data,
files=files,
) as response:
return response.ok
def keygen_encrypt_send(inputs, client_type):
"""Encrypt the given inputs for a specific client.
Args:
inputs (numpy.ndarray): The inputs to encrypt.
client_type (str): The type of client to consider (either 'user', 'bank' or 'third_party').
Returns:
"""
# Create an ID for the current client to consider
client_id = numpy.random.randint(0, 2**32)
keygen(client_id, client_type)
# Retrieve the client instance
client = get_client(client_id, client_type)
# TODO : pre-process the data first
# Quantize, encrypt and serialize the inputs
encrypted_inputs = client.quantize_encrypt_serialize_multi_inputs(
inputs,
input_index=INPUT_INDEXES[client_type],
initial_input_shape=INITIAL_INPUT_SHAPE,
start_position=START_POSITIONS[client_type],
)
# Save encrypted_inputs to bytes in a file, since too large to pass through regular Gradio
# buttons, https://github.com/gradio-app/gradio/issues/1877
encrypted_inputs_path = get_client_file_path("encrypted_inputs", client_id, client_type)
with encrypted_inputs_path.open("wb") as encrypted_inputs_file:
encrypted_inputs_file.write(encrypted_inputs)
# Create a truncated version of the encrypted image for display
encrypted_inputs_short = shorten_bytes_object(encrypted_inputs)
send_input(client_id, client_type)
# TODO: also return private key representation if possible
return encrypted_inputs_short
def run_fhe(client_id):
"""Run the model on the encrypted inputs previously sent using FHE.
Args:
client_id (int): The client ID to consider.
"""
# TODO : add a warning for users to send all client types' inputs
data = {
"client_id": client_id,
}
# Trigger the FHE execution on the encrypted inputs previously sent
url = SERVER_URL + "run_fhe"
with requests.post(
url=url,
data=data,
) as response:
if response.ok:
return response.json()
else:
raise gr.Error("Please wait for the inputs to be sent to the server.")
def get_output(client_id):
"""Retrieve the encrypted output.
Args:
client_id (int): The client ID to consider.
Returns:
output_encrypted_representation (numpy.ndarray): A representation of the encrypted output.
"""
data = {
"client_id": client_id,
}
# Retrieve the encrypted output image
url = SERVER_URL + "get_output"
with requests.post(
url=url,
data=data,
) as response:
if response.ok:
encrypted_output = response.content
# Save the encrypted output to bytes in a file as it is too large to pass through regular
# Gradio buttons (see https://github.com/gradio-app/gradio/issues/1877)
# TODO : check if output to user is relevant
encrypted_output_path = get_client_file_path("encrypted_output", client_id, "user")
with encrypted_output_path.open("wb") as encrypted_output_file:
encrypted_output_file.write(encrypted_output)
# TODO
# Decrypt the output using a different (wrong) key for display
# output_encrypted_representation = decrypt_output_with_wrong_key(encrypted_output, client_type)
# return output_encrypted_representation
return None
else:
raise gr.Error("Please wait for the FHE execution to be completed.")
def decrypt_output(client_id, client_type):
"""Decrypt the result.
Args:
client_id (int): The client ID to consider.
client_type (str): The type of client to consider (either 'user', 'bank' or 'third_party').
Returns:
output(numpy.ndarray): The decrypted output
"""
# Get the encrypted output path
encrypted_output_path = get_client_file_path("encrypted_output", client_id, client_type)
if not encrypted_output_path.is_file():
raise gr.Error("Please run the FHE execution first.")
# Load the encrypted output as bytes
with encrypted_output_path.open("rb") as encrypted_output_file:
encrypted_output_proba = encrypted_output_file.read()
# Retrieve the client API
client = get_client(client_id, client_type)
# Deserialize, decrypt and post-process the encrypted output
output_proba = client.deserialize_decrypt_post_process(encrypted_output_proba)
# Determine the predicted class
output = numpy.argmax(output_proba, axis=1)
return output
demo = gr.Blocks()
print("Starting the demo...")
with demo:
gr.Markdown(
"""
<h1 align="center">Credit Card Approval Prediction Using Fully Homomorphic Encryption</h1>
"""
)
gr.Markdown("## Client side")
gr.Markdown("### Step 1: Infos. ")
with gr.Row():
with gr.Column():
gr.Markdown("### User")
# TODO : change infos
choice_1 = gr.Dropdown(choices=["Yes, No"], label="Choose", interactive=True)
slide_1 = gr.Slider(2, 20, value=4, label="Count", info="Choose between 2 and 20")
with gr.Column():
gr.Markdown("### Bank ")
# TODO : change infos
checkbox_1 = gr.CheckboxGroup(["USA", "Japan", "Pakistan"], label="Countries", info="Where are they from?")
with gr.Column():
gr.Markdown("### Third Party ")
# TODO : change infos
radio_1 = gr.Radio(["park", "zoo", "road"], label="Location", info="Where did they go?")
gr.Markdown("### Step 2: Keygen, encrypt using FHE and send the inputs to the server.")
with gr.Row():
with gr.Column():
gr.Markdown("### User")
encrypt_button_user = gr.Button("Encrypt the inputs and send to server.")
keys_user = gr.Textbox(
label="Keys representation:", max_lines=2, interactive=False
)
encrypted_input_user = gr.Textbox(
label="Encrypted input representation:", max_lines=2, interactive=False
)
user_id = gr.Textbox(label="", max_lines=2, interactive=False, visible=False)
with gr.Column():
gr.Markdown("### Bank ")
encrypt_button_bank = gr.Button("Encrypt the inputs and send to server.")
keys_bank = gr.Textbox(
label="Keys representation:", max_lines=2, interactive=False
)
encrypted_input_bank = gr.Textbox(
label="Encrypted input representation:", max_lines=2, interactive=False
)
bank_id = gr.Textbox(label="", max_lines=2, interactive=False, visible=False)
with gr.Column():
gr.Markdown("### Third Party ")
encrypt_button_third_party = gr.Button("Encrypt the inputs and send to server.")
keys_3 = gr.Textbox(
label="Keys representation:", max_lines=2, interactive=False
)
encrypted_input__third_party = gr.Textbox(
label="Encrypted input representation:", max_lines=2, interactive=False
)
third_party_id = gr.Textbox(label="", max_lines=2, interactive=False, visible=False)
gr.Markdown("## Server side")
gr.Markdown(
"The encrypted values are received by the server. The server can then compute the prediction "
"directly over them. Once the computation is finished, the server returns "
"the encrypted result to the client."
)
gr.Markdown("### Step 6: Run FHE execution.")
execute_fhe_button = gr.Button("Run FHE execution.")
fhe_execution_time = gr.Textbox(
label="Total FHE execution time (in seconds):", max_lines=1, interactive=False
)
gr.Markdown("## Client side")
gr.Markdown(
"The encrypted output is sent back to the client, who can finally decrypt it with the "
"private key."
)
gr.Markdown("### Step 7: Receive the encrypted output from the server.")
gr.Markdown(
"The output displayed here is the encrypted result sent by the server, which has been "
"decrypted using a different private key. This is only used to visually represent an "
"encrypted output."
)
get_output_button = gr.Button("Receive the encrypted output from the server.")
encrypted_output_representation = gr.Textbox(
label="Encrypted output representation: ", max_lines=1, interactive=False
)
gr.Markdown("### Step 8: Decrypt the output.")
decrypt_button = gr.Button("Decrypt the output")
prediction_output = gr.Textbox(
label="Credit card approval decision: ", max_lines=1, interactive=False
)
# Button to encrypt inputs on the client side
# encrypt_button_user.click(
# encrypt,
# inputs=[user_id, input_image, filter_name],
# outputs=[original_image, encrypted_input],
# )
# # Button to encrypt inputs on the client side
# encrypt_button_bank.click(
# encrypt,
# inputs=[user_id, input_image, filter_name],
# outputs=[original_image, encrypted_input],
# )
# # Button to encrypt inputs on the client side
# encrypt_button_third_party.click(
# encrypt,
# inputs=[user_id, input_image, filter_name],
# outputs=[original_image, encrypted_input],
# )
# # Button to send the encodings to the server using post method
# send_input_button.click(
# send_input, inputs=[user_id, filter_name], outputs=[send_input_checkbox]
# )
# # Button to send the encodings to the server using post method
# execute_fhe_button.click(run_fhe, inputs=[user_id, filter_name], outputs=[fhe_execution_time])
# # Button to send the encodings to the server using post method
# get_output_button.click(
# get_output,
# inputs=[user_id, filter_name],
# outputs=[encrypted_output_representation]
# )
# # Button to decrypt the output on the client side
# decrypt_button.click(
# decrypt_output,
# inputs=[user_id, filter_name],
# outputs=[output_image, keygen_checkbox, send_input_checkbox],
# )
gr.Markdown(
"The app was built with [Concrete-ML](https://github.com/zama-ai/concrete-ml), a "
"Privacy-Preserving Machine Learning (PPML) open-source set of tools by [Zama](https://zama.ai/). "
"Try it yourself and don't forget to star on Github &#11088;."
)
demo.launch(share=False)