riskengine / src /bmc /fetcher.py
datavorous's picture
intial commit
28779c1 verified
import json
import requests
from bs4 import BeautifulSoup
from .config_loader import load_app_config
from .models import ProfileSnapshot
APP_CONFIG = load_app_config()
REQUESTS_CONFIG = APP_CONFIG["requests"]
HEADERS = {"User-Agent": REQUESTS_CONFIG["user_agent"]}
TIMEOUT = REQUESTS_CONFIG["timeout_seconds"]
class ProfileNotFound(Exception):
pass
def fetch_profile(url: str) -> ProfileSnapshot:
resp = requests.get(url, headers=HEADERS, timeout=TIMEOUT)
if resp.status_code == 404:
raise ProfileNotFound("HTTP 404")
soup = BeautifulSoup(resp.text, "html.parser")
app = soup.find("div", id="app")
if not app or not app.has_attr("data-page"):
raise RuntimeError("data-page JSON not found")
data = json.loads(app["data-page"])
if data.get("component") == "Error":
status = data.get("props", {}).get("status")
if status == 404:
raise ProfileNotFound("Profile not found (app-level 404)")
creator = data["props"]["creator_data"]["data"]
return ProfileSnapshot(
bmc_url=url,
name=creator.get("name"),
slug=creator.get("slug"),
bio=creator.get("description"),
external_links=creator.get("links", []),
has_nsfw=creator.get("has_nsfw"),
payment_method=creator.get("payment_method"),
)