scoring/quality.py

import os
import openai
import requests
from bs4 import BeautifulSoup

openai.api_key = os.getenv("OPENAI_API_KEY")

def quality_score(entity: str) -> int:
    """
    If 'entity' is a URL:
      • Fetch the first 2000 characters of its visible text.
      • Ask GPT-4: "Rate this snippet 0–100 on combined Expertise, Authority, Trustworthiness."
      • Return the integer score.
    Else, return default 60.
    """
    if not entity.startswith("http"):
        return 60

    try:
        resp = requests.get(entity, timeout=5)
        text_snippet = BeautifulSoup(resp.text, "html.parser").get_text()[:2000]

        prompt = (
            "Below is the text snippet of a webpage. Rate it 0–100 on combined "
            "Expertise, Authority, and Trustworthiness (E-E-A-T), and return just the integer score.\n\n"
            f"{text_snippet}"
        )
        api_resp = openai.ChatCompletion.create(
            model="gpt-4",
            messages=[{"role": "user", "content": prompt}]
        )
        content = api_resp.choices[0].message.content.strip()
        digits = "".join(filter(str.isdigit, content))
        if digits:
            score = int(digits)
            return max(0, min(score, 100))
    except Exception:
        pass

    return 60

def quality_recommendation(entity: str, score: int) -> str:
    if score < 50:
        return (
            "Improve E-E-A-T: add author bylines, cite reputable sources, "
            "and maintain factual consistency."
        )
    return "Content quality is strong; keep updating and citing reputable references."