# dkim_verifier_sim.py - Flawed DKIM verifier
import re, hashlib

# Simulated email with DKIM signature
raw_email = b"""From: [email protected]
To: [email protected]
Subject: Hello
DKIM-Signature: v=1; a=rsa-sha256; d=example.com; s=mail; h=from:to:subject;
        b=abc123; bh=def456;
X-Extra: injected

This is a test.
"""

def parse_headers(raw):
    headers = {}
    lines = raw.decode().split('\r\n')
    for line in lines:
        if ': ' in line:
            key, val = line.split(': ', 1)
            headers[key.lower()] = val
    return headers

def verify_dkim(raw):
    headers = parse_headers(raw)
    # Vulnerability: canonicalisation does not remove extra headers not in the 'h' list
    # According to RFC, only headers listed in 'h' are signed, but the verifier should exclude others.
    # Here we simulate that extra header 'x-extra' is mistakenly included in the hash computation
    # because the verifier canonicalizes all headers instead of just the listed ones.
    signed_headers = headers['dkim-signature'].split('h=')[1].split(';')[0].split(':')
    # Build header list for hash
    header_block = ""
    for h in signed_headers:
        header_block += f"{h}:{headers[h]}\r\n"
    # Flaw: include extra header X-Extra because it's present in the actual headers
    if 'x-extra' in headers:
        header_block += f"x-extra:{headers['x-extra']}\r\n"
    # Now compute hash and compare... For demo, we'll just print that verification succeeds incorrectly.
    print("Verification passed (incorrectly includes extra header)")

verify_dkim(raw_email)