def compare_files(file1, file2, msg_callback=None, cmd_output=None):
"""
Compare files by calculating and comparing SHA-256 checksums
:param file1: First file to compare
:param file2: Second file to compare
:param msg_callback: message callback object, can be used to collect additional data via .echo()
:param cmd_output: Additional data to be used for JSON output
:return: True if matches, False if doesn't match
"""
# Calculate the checksums
hasher = filehash.FileHash(DEFAULT_HASH_ALGORITHM)
try:
file1_hash = hasher.hash_file(filename=file1)
file2_hash = hasher.hash_file(filename=file2)
except FileNotFoundError as err:
if msg_callback:
msg_callback.echo(str(err))
return False
# Output additional information if needed
if msg_callback:
msg_callback.echo('File1 checksum: ' + file1_hash)
msg_callback.echo('File2 checksum: ' + file2_hash)
# Compare the checksums and return the result
if file1_hash == file2_hash:
return True
else:
if cmd_output is not None:
cmd_output.append('File1 checksum: ' + file1_hash)
cmd_output.append('File2 checksum: ' + file2_hash)
return False
def getFileHash(self, client_socket, command_list):
'''
Function to get file hash and send a response to the client
check decode_command for info of how this will be called.
Input:
client_socket - client socket object
command_list - list object
Returns:
None
'''
if len(command_list) == 3 or len(command_list) == 2:
arg = command_list[1].lower()
if arg.lower() == 'verify':
filename = command_list[2]
if os.path.isfile(self.file_storage_path + '/' + filename):
arg_file = open(self.file_storage_path + '/' + filename,
'rb')
file_stats = os.stat(self.file_storage_path + '/' +
filename)
size = file_stats.st_size
file_mtime = time.localtime(
os.path.getmtime(self.file_storage_path + '/' +
filename))
file_mtime = time.strftime('%Y-%m-%d %H:%M:%S', file_mtime)
# hasher = hashlib.md5(arg_file.read()).hexdigest()
hasher = filehash.FileHash('md5')
hasher = hasher.hash_file(self.file_storage_path + '/' +
filename)
return_value = 'hash: ' + hasher + ' size: ' + str(
size) + ' last modified: ' + file_mtime
print(return_value)
client_socket.send(return_value.encode('utf-8'))
return None
else:
return_value = '0'
client_socket.send(return_value.encode('utf-8'))
return None
elif arg.lower() == 'checkall':
return_value = ''
for file_name in os.listdir(self.file_storage_path):
hasher = filehash.FileHash('md5')
hasher = hasher.hash_file(self.file_storage_path + '/' +
file_name)
file_mtime = time.localtime(
os.path.getmtime(self.file_storage_path + '/' +
file_name))
file_mtime = time.strftime('%Y-%m-%d %H:%M:%S', file_mtime)
return_value = return_value + file_name + ' hash value: ' + hasher + ' last modified: ' + file_mtime + '\n'
client_socket.send(return_value.encode('utf-8'))
return None
return_value = "Command error. Usage: FileHash verify <filename>\n FileHash checkall"
client_socket.send(return_value.encode('utf-8'))
def sendFile(self, client_socket, command_list):
'''
Function to send files to client(client download file)
Input:
arg - TCP or UDP
client_socket - client socket object
filename - filename to the respective file
Returns:
None
'''
if len(command_list) == 3:
if command_list[1].lower() != 'udp' and command_list[1].lower(
) != 'tcp':
print('arg error')
string_to_send = 'Command error. Usage: FileDownload tcp/udp <filename>'
client_socket.send(string_to_send.encode('utf-8'))
return None
path = self.file_storage_path + '/' + command_list[2]
arg = command_list[1]
string_to_send = "Requested file not present in server"
if not os.path.isfile(path):
print(string_to_send)
client_socket.send(string_to_send.encode('utf-8'))
else:
filename = command_list[2]
file_stats = os.stat(path)
size = file_stats.st_size
file_mtime = time.localtime(os.path.getmtime(path))
file_mtime = time.strftime('%Y-%m-%d %H:%M:%S', file_mtime)
hasher = filehash.FileHash('md5')
hasher = hasher.hash_file(path)
print("Sending...")
file_info_text = (
"Filename: %s, Timestamp: %s, MD5hash: %s, Filesize(Bytes): %s"
% (filename, str(file_mtime), str(hasher), str(size)))
print(file_info_text)
# string_to_send = "Requested file has been found. Filesize: " + str(size)
string_to_send = "Requested file has been found.\n" + file_info_text
print(string_to_send)
client_socket.send(string_to_send.encode('utf-8'))
time.sleep(0.02)
if arg == 'tcp' or arg == 'TCP':
file = open(path, 'rb')
reading = file.read(config.BUFFER_SIZE)
# print (reading)
while (reading):
client_socket.send(reading)
reading = file.read(config.BUFFER_SIZE)
# print (reading)
file.close()
elif arg == 'udp' or arg == 'UDP':
file = open(path, 'rb')
client_udp_socket = socket.socket(socket.AF_INET,
socket.SOCK_DGRAM)
client_ip = client_socket.getpeername()[0]
port_number = client_socket.getpeername()
if client_ip == '':
client_ip = '127.0.0.1'
reading = file.read(config.BUFFER_SIZE)
while (reading):
if (client_udp_socket.sendto(
reading, (client_ip, self.udp_port))):
reading = file.read(config.BUFFER_SIZE)
time.sleep(0.02)
client_udp_socket.close()
file.close()
else:
string_to_send = 'Command error. Usage: FileDownload tcp/udp <filename>'
client_socket.send(string_to_send.encode('utf-8'))
return None
def verify_checksum(filename,
algorithm,
msg_callback=None,
cmd_output=None,
checksum_value=None,
checksumfile=None):
"""
Calculates a filename hash and compares against the provided checksum or checksums file
:param filename: Filename used to calculate the hash
:param algorithm: Algorithm to use for hashing
:param msg_callback: message callback object, can be used to collect additional data via .echo()
:param cmd_output: Additional data to be used for JSON output
:param checksum_value: Checksum value
:param checksumfile: Filename of the file containing checksums, follows the format from shasum
:return: True if matches, False if doesn't match
"""
# Check algorithm for valid values
if algorithm not in filehash.SUPPORTED_ALGORITHMS:
raise ValueError('Unsupported algorithm value')
# Make sure either checksum or checksumfile arguments are set
if checksum_value is None and checksumfile is None:
raise ValueError(
'Either checksum_value or checksumfile arguments must be set')
# Calculate the hash
try:
calculated_hash = filehash.FileHash(algorithm).hash_file(
filename=filename)
except FileNotFoundError as err:
if msg_callback:
msg_callback.echo(str(err))
return False
# Output additional information if needed
if msg_callback:
msg_callback.echo('Algorithm: ' + algorithm)
msg_callback.echo('File hash: ' + calculated_hash)
# Compare the checksums and return the result
if checksum_value:
if calculated_hash == checksum_value.lower().strip():
return True
else:
if cmd_output is not None:
cmd_output.append('Algorithm: ' + algorithm)
cmd_output.append('File checksum: ' + calculated_hash)
cmd_output.append('Checksum to check against: ' +
checksum_value)
return False
else:
try:
checksums_content = str(Path(checksumfile).read_bytes())
except (FileNotFoundError, TypeError) as err:
if msg_callback:
msg_callback.echo(str(err))
return False
# Process verification results
if calculated_hash in checksums_content:
return True
else:
if cmd_output is not None:
cmd_output.append('Algorithm: ' + algorithm)
cmd_output.append('File checksum: ' + calculated_hash)
cmd_output.append('No match found in checksum file')
return False
import itertools
import multiprocessing
import matplotlib
import matplotlib.pyplot as plt
import filehash
import subprocess
cpu_amount = multiprocessing.cpu_count()
import numpy as np
from PIL import Image, ImageFile
hasher = filehash.FileHash('sha512')
# import decimal
# prec = 50
# decimal.getcontext().prec = prec
# from decimal import Decimal as Dec
ROOT_PATH = os.path.dirname(os.path.abspath(__file__)) + "/"
if __name__ == "__main__":
# n = 4
# for page_num in range(100*n, 100*(n+1)):
# 1237
# n = 1000
n1 = 4200
os.chdir(this_file.parents[1])
cache_file = this_file.parent / "cache.json"
if cache_file.exists():
with open(cache_file, 'r') as f:
cache = json.load(f)
else:
cache = dict()
root = Path(os.getcwd())
svgs = [str(x) for x in root.glob("**/svg_source/*.svg")]
changed = list()
sha512hasher = filehash.FileHash()
for file in svgs:
_hash = sha512hasher.hash_file(file)
if file in cache.keys():
if cache[file] != _hash:
cache[file] = _hash
changed.append(file)
else:
cache[file] = _hash
changed.append(file)
for change in changed:
_path = Path(change).parents[1]
command = [
import multiprocessing
import matplotlib
import matplotlib.pyplot as plt
import hashlib
import filehash
import subprocess
cpu_amount = multiprocessing.cpu_count()
import numpy as np
from PIL import Image, ImageFile
hasher512 = filehash.FileHash('sha512')
hasher256 = filehash.FileHash('sha256')
# import decimal
# prec = 50
# decimal.getcontext().prec = prec
# from decimal import Decimal as Dec
ROOT_PATH = os.path.dirname(os.path.abspath(__file__))+"/"
def get_image_size_from_path(file_path):
ImPar=ImageFile.Parser()
with open(file_path, "rb") as f:
ImPar=ImageFile.Parser()
chunk = f.read(2048)
def Cache(self, command_list):
'''
Cache command
Input :
command_list - list object - contains command separated at spaces
Return :
None
'''
if len(command_list) == 1:
print(
'Command error. Usage :\nCache verify <filename>\nCache show')
return None
if command_list[1].lower() == 'show':
files = os.scandir(self.cache_directory_path)
# print (files)
string_to_display = ''
for entry in files:
stats_entry = entry.stat()
size = stats_entry.st_size
string = entry.name + " | " + str(size)
string_to_display = string_to_display + string + '\n'
if string_to_display == '':
string_to_display = 'No files to display'
print(string_to_display)
elif command_list[1].lower() == 'verify':
download_flag = 1
file_name = command_list[2]
file_name = file_name.replace(' ', '\\ ')
path = self.cache_directory_path + '/' + command_list[2]
if os.path.exists(path):
hasher = filehash.FileHash('md5')
hasher = hasher.hash_file(path)
command = 'FileHash verify ' + file_name
hash_value, size = self.decode_command(command)
if hash_value == 0 and size == 0:
download_flag = 0
return 0
elif hasher == hash_value:
print('File exists.')
download_flag = 0
else:
print('File changed on server downloading again.')
download_flag = 1
if download_flag:
command = 'FileHash verify ' + file_name
hash_value, size = self.decode_command(command)
if hash_value == 0 and size == 0:
return 0
download_flag = helper_functions.clear_cache(
self.cache_directory_path, size, self.cache_size)
if download_flag:
command = 'FileDownload tcp ' + file_name
print('Downloading file\n')
self.client_socket.send(command.encode('utf-8'))
command_list = helper_functions.string_split(command)
self.FileDownload(command_list, cache=1)
else:
print(
'Command error. Usage :\nCache verify <filename>\nCache show')
return None