core/sources/fema.py

import os
from urllib.parse import quote_plus, urljoin

import httpx
from bs4 import BeautifulSoup

_PARAM_CACHE: dict[str, tuple[str, str, str]] = {}


def _discover_params(base: str) -> tuple[str, str, str]:
    if base in _PARAM_CACHE:
        return _PARAM_CACHE[base]

    cas_param = os.getenv("FEMA_CAS_PARAM", "field_cas_tid_1")
    name_param = os.getenv("FEMA_NAME_PARAM", "field_chemical_synonym_tid")
    action = os.getenv("FEMA_FORM_ACTION", base)

    try:
        r = httpx.get(base, timeout=15, headers={"User-Agent": "Mozilla/5.0"})
        if r.status_code < 400:
            soup = BeautifulSoup(r.text, "lxml")
            form = soup.find("form")
            if form and form.get("action"):
                action = urljoin(base, form.get("action"))
            inputs = soup.find_all("input")
            for inp in inputs:
                name = (inp.get("name") or "").strip()
                if not name:
                    continue
                placeholder = (inp.get("placeholder") or "").lower()
                lower_name = name.lower()
                if "cas" in placeholder or lower_name == "cas" or "cas" in lower_name:
                    cas_param = name
                if (
                    "synonym" in placeholder
                    or "chemical" in placeholder
                    or "synonym" in lower_name
                    or "chemical" in lower_name
                ):
                    name_param = name
    except Exception:
        pass

    _PARAM_CACHE[base] = (cas_param, name_param, action)
    return cas_param, name_param, action


def fema_link(cas_or_query: str, name_query: str | None = None) -> dict:
    """Build the FEMA / Fragrance Materials Safety Resource search URL.

    Production uses Elsevier's Fragrance Materials Safety Resource with CAS query params.
    """

    q = (cas_or_query or "").strip()
    name_q = (name_query or "").strip()
    if not q and not name_q:
        return {"ok": False, "error": "Empty query"}

    # NOTE: domain spelling matters; the older '...materialssafety...' variant often 404s.
    base = os.getenv("FEMA_BASE_URL", "https://fragrancematerialsafetyresource.elsevier.com/")
    cas_param, name_param, action = _discover_params(base)

    cas_value = quote_plus(q) if q else ""
    name_value = quote_plus(name_q or q)

    cas_url = f"{action}?{cas_param}={cas_value}&{name_param}=" if cas_value else ""
    name_url = f"{action}?{cas_param}=&{name_param}={name_value}" if name_value else ""
    combo_url = (
        f"{action}?{cas_param}={cas_value}&{name_param}={name_value}"
        if cas_value and name_value
        else ""
    )
    # Generic search fallback (some deployments ignore filter params)
    search_term = name_q or q
    search_url = f"{base}search/node?keys={quote_plus(search_term)}" if search_term else ""
    search_api_url = (
        f"{base}search/node?search_api_fulltext={quote_plus(search_term)}" if search_term else ""
    )
    return {
        "ok": True,
        "cas_url": cas_url,
        "name_url": name_url,
        "combo_url": combo_url,
        "alt_url": search_url,
        "search_api_url": search_api_url,
    }