pypitwist.py

#!/usr/bin/env python

import json
import crayons
import requests

PYPI_URL = 'https://pypi.org/pypi/{package_name}/json'

homoglyphs = {
    'a': ['ɑ'],
    'c': ['ᴄ', 'ⅽ'],
    'd': ['ⅾ'],
    'f': ['ſ'],
    'g': ['ɡ'],
    'i': ['ⅰ'],
    'l': ['I', 'Ⅰ', 'ⅼ'],
    'o': ['ᴏ'],
    'u': ['ᴜ'],
    'v': ['ᴠ', 'ⅴ'],
    'w': ['ᴡ'],
    'x': ['ⅹ'],
    'y': ['ʏ'],
    'z': ['ᴢ']
}

def get_top_packages(n=4000):
    with open('top-pypi-packages-365-days.json', 'r') as f:
        data = json.loads(f.read())
    return [p['project'] for p in data['rows']][:n]

def get_homoglyphs(package):
    hgs = []
    for letter in package:
        if letter in homoglyphs:
            for new_letter in homoglyphs[letter]:
                hgs.append(package.replace(letter, new_letter))
    return hgs

def main():
    packages = get_top_packages(n=100)
    for package in packages:
        hgs = get_homoglyphs(package)
        for homoglyph in hgs:
            url = PYPI_URL.format(package_name=homoglyph)
            response = requests.get(url)
            if response.status_code == 200:
                print(crayons.green(f'[+] {homoglyph} found!'))
                package_data = response.json()
                with open(f'{package}-{homoglyph}-data.json', 'w') as f:
                    f.write(json.dumps(package_data, indent=2))
            else:
                print(crayons.red(f'[-] {homoglyph} NOT found.'))

if __name__ == '__main__':
    main()