agents.py

# agents.py
import os
import json
import random

# Try to import zhipuai client
ZHIPU_AVAILABLE = False
client = None
ZHIPU_KEY = os.getenv("zhipuai_api_key")
if ZHIPU_KEY:
    try:
        from zhipuai import ZhipuAI
        client = ZhipuAI(api_key=ZHIPU_KEY)
        ZHIPU_AVAILABLE = True
    except Exception:
        ZHIPU_AVAILABLE = False

class AnalyzerAgent:
    """Simple objective analyzer for MCQ answers."""
    def analyze(self, student_answers, correct_answers):
        """
        student_answers: list of strings like 'A', 'B', 'C' or actual choice text
        correct_answers: list of strings (prefer letter 'A'..'D' or exact text)
        Returns (score, analysis_list)
        """
        score = 0
        analysis = []
        for i, (s, c) in enumerate(zip(student_answers, correct_answers), start=1):
            s_norm = (s or "").strip()
            c_norm = (c or "").strip()
            ok = False
            if c_norm == "":
                ok = False
            elif len(c_norm) == 1 and c_norm.upper() in ["A","B","C","D"]:
                ok = (s_norm.upper() == c_norm.upper())
            else:
                ok = (s_norm.lower() == c_norm.lower())
            if ok:
                score += 1
                analysis.append(f"Q{i}: ✅")
            else:
                analysis.append(f"Q{i}: ❌ (Your: {s_norm} | Key: {c_norm})")
        return score, analysis

class CoachAgent:
    """Provides AI feedback using ZhipuAI if available."""
    def __init__(self):
        self.client = client

    def coach(self, context_text):
        """
        context_text: a small JSON/string summarizing student's performance.
        Returns an advice string.
        """
        if not ZHIPU_AVAILABLE or self.client is None:
            # Fallback advice
            return "AI coach not configured. Set environment variable 'zhipuai_api_key' to enable AI analysis.\nGeneral advice: Review the wrong questions, focus on weak topics, and practice time management."
        try:
            prompt = (
                "You are an experienced SPM coach. Given the student's summary, provide concise actionable advice (3-5 bullets).\n\n"
                f"Student summary:\n{context_text}\n\nAdvice:"
            )
            resp = self.client.chat.completions.create(
                model="glm-4",
                messages=[{"role":"user","content":prompt}],
                temperature=0.2,
            )
            out = resp.choices[0].message["content"].strip()
            return out
        except Exception as e:
            return f"AI coach error: {str(e)}"

class PredictiveAgent:
    """(Optional) Use LLM to forecast likely topics/questions."""
    def __init__(self):
        self.client = client

    def predict(self, subject, year_range="2018-2024"):
        if not ZHIPU_AVAILABLE or self.client is None:
            return "Predicted question (LLM not configured)."
        prompt = (
            f"Analyze SPM {subject} past papers between {year_range}. "
            "Identify recurring topics and predict 5 high-probability short objective questions."
        )
        try:
            resp = self.client.chat.completions.create(
                model="glm-4",
                messages=[{"role":"user","content":prompt}],
                temperature=0.3,
            )
            return resp.choices[0].message["content"].strip()
        except Exception as e:
            return f"Prediction error: {str(e)}"