agents/requirements_extractor.py

import os, json, re
from .utils_llm import call_llm

PROMPT = """
You are a senior business analyst.
Read the document text and output structured JSON for a PowerApps-style app.

Rules:
1️⃣ Output only JSON with these exact top-level keys:
   entities, workflows, roles, actions, ui_notes
2️⃣ Do not add any text, comments, or markdown.
3️⃣ If information is missing, return empty lists.
Document:
{text}
"""

GEMINI_KEY_EXTRACTOR = os.getenv("GEMINI_API_KEY_EXTRACTOR")

async def extract_requirements(text: str) -> dict:
    raw = await call_llm(
        PROMPT.format(text=text[:15000]),
        system="Return strictly valid JSON with top-level keys entities, workflows, roles, actions, ui_notes",
        api_key=GEMINI_KEY_EXTRACTOR,
        temperature=0.0,
    )

    # --- clean raw output ---
    raw = (raw or "").strip()
    # remove anything before first '{' and after last '}'
    start = raw.find("{")
    end = raw.rfind("}") + 1
    json_candidate = raw[start:end] if start != -1 and end != -1 else "{}"

    try:
        data = json.loads(json_candidate)
    except Exception:
        # final fallback: safe empty schema
        data = {}

    # guarantee required keys
    for k in ["entities", "workflows", "roles", "actions", "ui_notes"]:
        if k not in data or not isinstance(data.get(k), list):
            data[k] = []

    # optional debug
    if os.getenv("DEBUG_EXTRACTOR") == "1":
        print("--- RAW EXTRACTOR OUTPUT ---")
        print(raw[:400])

    return data