convert.py

# This is a script for converting between the CSV traces provided
# by the authors of CS23 into the hdf5 format we use.

import h5py
import numpy as np
from binascii import unhexlify
import os
import csv

def convert(path, num_traces=50000):
    name = os.path.splitext(os.path.basename(path))[0]
    metadata = name.split("-")

    with open(path) as csv_file:
        with h5py.File(f"traces/{name}.hdf5", "w") as f:
            f.attrs["trace_type"] = "attack"
            f.attrs["side"] = "start"
            f.attrs["implementation"] = "circuit"

            f.attrs["tk1"] = np.array(bytearray(unhexlify(metadata[0])), dtype=np.uint8)
            f.attrs["tk2"] = np.array(bytearray(unhexlify(metadata[1])), dtype=np.uint8)
            f.attrs["key"] = np.array(bytearray(unhexlify(metadata[2])), dtype=np.uint8)
            plaintext_set = f.create_dataset("plaintext", (num_traces, 16), dtype=np.uint8)
            ciphertext_set = f.create_dataset(
                "ciphertext", (num_traces, 16), dtype=np.uint8
            )
            trace_set = f.create_dataset("trace", (num_traces, 5000), dtype=np.float64)

            for i, row in enumerate(csv.reader(csv_file, delimiter=";")):
                plaintext_set[i] = np.array(bytearray(unhexlify(row[0])))
                ciphertext_set[i] = np.array(bytearray(unhexlify(row[1])))
                trace_set[i] = np.array(np.fromstring(row[2], sep=","))