mirror of
https://github.com/creyD/entro.py.git
synced 2026-04-16 05:20:29 +02:00
Prepared for packaging
This commit is contained in:
87
entro.py
87
entro.py
@@ -1,87 +0,0 @@
|
|||||||
'''
|
|
||||||
calc_entro.py calculates the entropy of a given string or file
|
|
||||||
|
|
||||||
This uses the negative sum of the log (to the base of 2) of the probability
|
|
||||||
times the probability of a char to occur in a certain string as the entropy.
|
|
||||||
'''
|
|
||||||
|
|
||||||
import math
|
|
||||||
import argparse
|
|
||||||
|
|
||||||
|
|
||||||
# Calculates the entropy of a given string (as described in the docstring)
|
|
||||||
def calculateEntropy(input_string):
|
|
||||||
alphabet, alphabet_size, entropy = {}, 0, 0
|
|
||||||
|
|
||||||
for char in input_string:
|
|
||||||
if char in alphabet:
|
|
||||||
alphabet[char] += 1
|
|
||||||
else:
|
|
||||||
alphabet[char] = 1
|
|
||||||
alphabet_size += 1
|
|
||||||
|
|
||||||
for char in alphabet:
|
|
||||||
alphabet[char] = alphabet[char] / alphabet_size
|
|
||||||
entropy -= alphabet[char] * math.log(alphabet[char], 2)
|
|
||||||
|
|
||||||
return entropy, alphabet
|
|
||||||
|
|
||||||
|
|
||||||
# Outputs a given entropy including the original text and the alphabet with probabilities
|
|
||||||
def printEntropy(original_string, entropy_value, alphabet_dict, simple_bool):
|
|
||||||
print('---')
|
|
||||||
if simple_bool == False:
|
|
||||||
print('Content: ' + original_string)
|
|
||||||
print('Probabilities: ' + str(alphabet_dict))
|
|
||||||
print('Entropy: ' + str(entropy_value) + ' bits')
|
|
||||||
print('---')
|
|
||||||
|
|
||||||
|
|
||||||
# Reads a file by a given path
|
|
||||||
def readEntropyFile(path_string):
|
|
||||||
f = open(path_string, 'r')
|
|
||||||
content = f.read().replace('\n', ' ')
|
|
||||||
f.close()
|
|
||||||
return content.strip()
|
|
||||||
|
|
||||||
|
|
||||||
# List of the arguments one can use to influence the behavior of the program
|
|
||||||
parser = argparse.ArgumentParser(description='Calculate the information entropy of some strings.')
|
|
||||||
|
|
||||||
# INPUT ARGUMENTS
|
|
||||||
parser.add_argument('strings', nargs='*', default='', type=str, help='Strings to calculate the entropy of.')
|
|
||||||
parser.add_argument('--files', nargs='*', type=str, default='', help='Provide file path(s) to calculate the entropy of.')
|
|
||||||
|
|
||||||
# OUTPUT OPTIONS
|
|
||||||
parser.add_argument('--simple', nargs='?', type=bool, default=False, help='Determines the explicitness of the output. (True = only entropy shown)')
|
|
||||||
|
|
||||||
# CONVERT OPTIONS
|
|
||||||
parser.add_argument('--lower', nargs='?', type=bool, default=False, help='Converts given strings or textfiles to lowercase before calculating.')
|
|
||||||
parser.add_argument('--upper', nargs='?', type=bool, default=False, help='Converts given strings or textfiles to uppercase before calculating.')
|
|
||||||
parser.add_argument('--squash', nargs='?', type=bool, default=False, help='Removes all whitespaces before calculating.')
|
|
||||||
args = parser.parse_args()
|
|
||||||
|
|
||||||
# Prepares the queue of different strings
|
|
||||||
queue = []
|
|
||||||
|
|
||||||
# Add all the provided strings to the list
|
|
||||||
for string in args.strings:
|
|
||||||
queue.append(string)
|
|
||||||
|
|
||||||
# Add all the provided files to the list
|
|
||||||
for file in args.files:
|
|
||||||
string = readEntropyFile(file)
|
|
||||||
queue.append(string)
|
|
||||||
|
|
||||||
# Interates over the collected strings and prints the entropies
|
|
||||||
for string in queue:
|
|
||||||
if args.lower != False:
|
|
||||||
string = string.lower()
|
|
||||||
elif args.upper != False:
|
|
||||||
string = string.upper()
|
|
||||||
|
|
||||||
if args.squash != False:
|
|
||||||
string = string.replace(" ", "")
|
|
||||||
|
|
||||||
a, b = calculateEntropy(string)
|
|
||||||
printEntropy(string, a, b, args.simple)
|
|
||||||
1
entro.py/__init__.py
Normal file
1
entro.py/__init__.py
Normal file
@@ -0,0 +1 @@
|
|||||||
|
name = "entro.py"
|
||||||
38
entro.py/entro.py
Normal file
38
entro.py/entro.py
Normal file
@@ -0,0 +1,38 @@
|
|||||||
|
import math
|
||||||
|
|
||||||
|
# Calculates the entropy of a given string
|
||||||
|
# Returns the entropy and an alphabet with the calculated probabilities
|
||||||
|
def calculateEntropy(input_string):
|
||||||
|
alphabet, alphabet_size, entropy = {}, 0, 0
|
||||||
|
|
||||||
|
for char in input_string:
|
||||||
|
if char in alphabet:
|
||||||
|
alphabet[char] += 1
|
||||||
|
else:
|
||||||
|
alphabet[char] = 1
|
||||||
|
alphabet_size += 1
|
||||||
|
|
||||||
|
for char in alphabet:
|
||||||
|
alphabet[char] = alphabet[char] / alphabet_size
|
||||||
|
entropy -= alphabet[char] * math.log(alphabet[char], 2)
|
||||||
|
|
||||||
|
return entropy, alphabet
|
||||||
|
|
||||||
|
|
||||||
|
# Calculates the entropy of a given string
|
||||||
|
# Returns only the entropy in bits as this is the minimal function
|
||||||
|
def calculateEntropyMin(input_string):
|
||||||
|
alphabet, alphabet_size, entropy = {}, 0, 0
|
||||||
|
|
||||||
|
for char in input_string:
|
||||||
|
if char in alphabet:
|
||||||
|
alphabet[char] += 1
|
||||||
|
else:
|
||||||
|
alphabet[char] = 1
|
||||||
|
alphabet_size += 1
|
||||||
|
|
||||||
|
for char in alphabet:
|
||||||
|
i = alphabet[char] / alphabet_size
|
||||||
|
entropy -= i * math.log(i, 2)
|
||||||
|
|
||||||
|
return entropy
|
||||||
21
setup.py
Normal file
21
setup.py
Normal file
@@ -0,0 +1,21 @@
|
|||||||
|
import setuptools
|
||||||
|
|
||||||
|
with open("README.md", "r") as fh:
|
||||||
|
long_description = fh.read()
|
||||||
|
|
||||||
|
setuptools.setup(
|
||||||
|
name="entro.py-creyD",
|
||||||
|
version="1.0",
|
||||||
|
author="Conrad Großer",
|
||||||
|
author_email="grosserconrad@gmail.com",
|
||||||
|
description="Small Information Entropy Calculator",
|
||||||
|
long_description=long_description,
|
||||||
|
long_description_content_type="text/markdown",
|
||||||
|
url="https://github.com/creyD/entro.py",
|
||||||
|
packages=setuptools.find_packages(),
|
||||||
|
classifiers=[
|
||||||
|
"Programming Language :: Python :: 3",
|
||||||
|
"License :: OSI Approved :: MIT License",
|
||||||
|
"Operating System :: OS Independent",
|
||||||
|
],
|
||||||
|
)
|
||||||
Reference in New Issue
Block a user