mirror of
https://git.sr.ht/~tsileo/microblog.pub
synced 2024-12-22 05:04:27 +00:00
39 lines
974 B
Python
39 lines
974 B
Python
from urllib.parse import urlparse
|
|
|
|
from bs4 import BeautifulSoup # type: ignore
|
|
from loguru import logger
|
|
|
|
from app.config import PRIVACY_REPLACE
|
|
|
|
|
|
def replace_content(content: str) -> str:
|
|
if not PRIVACY_REPLACE:
|
|
return content
|
|
|
|
soup = BeautifulSoup(content, "html5lib")
|
|
links = list(soup.find_all("a", href=True))
|
|
if not links:
|
|
return content
|
|
|
|
for link in links:
|
|
link.attrs["href"] = replace_url(link.attrs["href"])
|
|
|
|
return soup.find("body").decode_contents()
|
|
|
|
|
|
def replace_url(u: str) -> str:
|
|
if not PRIVACY_REPLACE:
|
|
return u
|
|
|
|
try:
|
|
parsed_href = urlparse(u)
|
|
if not parsed_href.hostname:
|
|
raise ValueError("Missing hostname")
|
|
except Exception:
|
|
logger.warning(f"Failed to parse url={u}")
|
|
return u
|
|
|
|
if new_netloc := PRIVACY_REPLACE.get(parsed_href.hostname.removeprefix("www.")):
|
|
return parsed_href._replace(netloc=new_netloc).geturl()
|
|
|
|
return u
|