#!/usr/bin/env python3
"""Walidacja: dla proby URL-i z klasy SAFE_HTTPS porownuje odpowiedz http i https
(dlugosc, typ, finalny URL), zeby wykryc falszywe 200 (strona bledu HTML)."""
import json
import concurrent.futures as cf
import ssl
import random
import urllib.request
import urllib.error

UA = {"User-Agent": "curl/8.7.1"}
d = json.load(open("/tmp/http-urlaudit.json"))
safe = [i["expanded"] for i in d["SAFE_HTTPS"]]
random.seed(11)
sample = random.sample(safe, 40)


def info(url):
    ctx = ssl.create_default_context()
    req = urllib.request.Request(url, headers=dict(UA, Range="bytes=0-1"))
    try:
        with urllib.request.urlopen(req, timeout=25, context=ctx) as r:
            return {
                "status": r.status,
                "final": r.geturl(),
                "ctype": r.headers.get("Content-Type", ""),
                "clen": r.headers.get("Content-Length", ""),
                "crange": r.headers.get("Content-Range", ""),
            }
    except urllib.error.HTTPError as e:
        return {"status": e.code, "final": e.geturl(), "ctype": e.headers.get("Content-Type", ""),
                "clen": e.headers.get("Content-Length", ""), "crange": ""}
    except Exception as e:
        return {"status": "ERR", "final": "", "ctype": type(e).__name__, "clen": "", "crange": ""}


def check(u):
    h = info(u)
    s = info("https://" + u[len("http://"):])
    return u, h, s


print(f"probka: {len(sample)} URL-i\n")
suspect = 0
with cf.ThreadPoolExecutor(max_workers=10) as pool:
    for u, h, s in pool.map(check, sample):
        flag = "  "
        if h["status"] == 200 and s["status"] == 200:
            if h["clen"] != s["clen"] or h["ctype"] != s["ctype"]:
                flag = "!!"
                suspect += 1
        print(f"{flag} http={h['status']:<4} https={s['status']:<4} "
              f"len {h['clen'][:10]} vs {s['clen'][:10]} | {h['ctype'][:28]} vs {s['ctype'][:28]} | {u[:70]}")

print()
print(f"podejrzane (rozny content http vs https): {suspect}/{len(sample)}")