llm_ticket3/agents/RAG/agent_rag_indexer.py

import os
import pandas as pd
import logging
from typing import Optional
from ..base_agent import BaseAgent

logger = logging.getLogger("AgentRagIndexer")

class AgentRagIndexer(BaseAgent):
    def __init__(self, ragflow, llm):
        super().__init__("AgentRagIndexer", llm)
        self.ragflow = ragflow

    def executer(self, chemin_csv: str, ticket_id: Optional[str] = None) -> int:
        if not os.path.exists(chemin_csv):
            logger.error(f"Fichier CSV introuvable : {chemin_csv}")
            return 0

        df = pd.read_csv(chemin_csv)
        lignes_indexees = 0

        for index, row in df.iterrows():
            question = str(row.get("question", "")).strip()
            reponse = str(row.get("reponse", "")).strip()

            if not question or not reponse:
                continue

            contenu = f"Question : {question}\nRéponse : {reponse}"
            metadata = {
                "ticket_id": ticket_id or "UNKNOWN",
                "ligne": index + 1
            }

            try:
                self.ragflow.indexer(contenu, metadata)
                lignes_indexees += 1
            except Exception as e:
                logger.warning(f"Erreur lors de l'indexation ligne {index}: {e}")

        logger.info(f"{lignes_indexees} lignes indexées depuis {chemin_csv}")
        return lignes_indexees