from __future__ import annotations

import bleach
from bs4 import BeautifulSoup

ALLOWED_TAGS = [
    "p",
    "br",
    "ul",
    "ol",
    "li",
    "strong",
    "b",
    "em",
    "i",
    "h2",
    "h3",
    "h4",
    "blockquote",
    "code",
    "pre",
    "a",
]
ALLOWED_ATTRIBUTES = {"a": ["href", "title", "rel"]}


def sanitize_job_html(value: str) -> str:
    cleaned = bleach.clean(
        value or "",
        tags=ALLOWED_TAGS,
        attributes=ALLOWED_ATTRIBUTES,
        protocols=["http", "https", "mailto"],
        strip=True,
        strip_comments=True,
    )
    soup = BeautifulSoup(cleaned, "lxml")
    for anchor in soup.find_all("a"):
        anchor["rel"] = "noopener noreferrer nofollow"
    body = soup.body
    return "".join(str(child) for child in body.children) if body else str(soup)