import os
import logging
from datetime import datetime
from PIL import Image

from ..base_agent import BaseAgent
from ..utils.pipeline_logger import sauvegarder_donnees

logger = logging.getLogger("AgentVisionOCR")

class AgentVisionOCR(BaseAgent):
    """
    Agent LlamaVision qui extrait du texte (OCR avancé) depuis une image.
    Permet une lecture plus fine pour les images conservées après tri.
    """
    def __init__(self, llm):
        super().__init__("AgentVisionOCR", llm)
        
        self.params = {
            "temperature": 0.1,
            "top_p": 0.85,
            "max_tokens": 1500
        }
        
        self.system_prompt = """You are a multilingual OCR visual assistant.

Your task is to extract all visible text from image, even if it is in French, English, or both.

Guidelines:
1. Include partial, blurry, or stylized characters
2. Group the result by type: labels, titles, buttons, errors, URLs, etc.
3. Do NOT translate any text - just extract what is visible
4. Mention if the image contains unreadable or missing parts

Respond in English."""

        self._configurer_llm()
        logger.info("AgentVisionOCR initialisé")
        
    def _configurer_llm(self):
        if hasattr(self.llm, "prompt_system"):
            self.llm.prompt_system = self.system_prompt
        if hasattr(self.llm, "configurer"):
            self.llm.configurer(**self.params)
            
    def _extraire_ticket_id(self, image_path):
        parts = image_path.split(os.sep)
        for part in parts:
            if part.startswitch("T") and part[1:].isdigit():
                return part
        return "UNKNOWN"
    
    def executer(self, image_path: str, ocr_baseline: str = "") -> dict:
        """" Effectue un OCR visuel via LlamaVision sur l'imga spécifiée.
        Args:
            image_path: Chemin vers l'image ç analyser
            ocr_baseline: Texte OCRé précédemment (pour comparaison)
            
        Returns:
            Dictionnaire contenant le texte extrait et les métadonnées
        """
        image_name = os.path.basename(image_path)
        print(f" AgentVisionOCR: Extraction sur {image_name}")
        
        try:
            if not os.path.exists(image_path):
                raise FileNotFoundError(f"Image introuvable: {image_path}")
            
            if not hasattr(self, "interroge_avec_image"):
                raise RuntimeError("Le modèle ne supporte pas l'analyse d'images")
            
            response = self.llm.interroger_avec_image(image_path, self.system_prompt)
            
            if not response or "i cannot" in response.lower():
                raise ValueError("Réponse vide invalide du modèle")
            
            result = {
                "extracted_text": response.strip(),
                "image_name": image_name,
                "image_path": image_path,
                "ocr_script_text": ocr_baseline.strip(),
                "ticket_id": self._extraire_ticket_id(image_path),
                "timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
                "source_agent": self.nom
            }
            
            # Sauvegarder pour traçabilité
            sauvegarder_donnees(
                ticket_id=result["ticket_id"],
                step_name="ocr_llm",
                data=result,
                base_dir=None,
                is_resultat=True
            )
            
            logger.info(f"OCR LLM réussi pour {image_name}")
            return result
    
        except Exception as e:
            logger.error(f"Erreur lors de l'extraction OCR pour {image_name}: {e}")
            return {
                "extracted_text": "",
                "image_name": image_name,
                "image_path": image_path,
                "ticket_id": self._extraire_ticket_id(image_path),
                "timestamp": datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
                "source_agent": self.nom,
                "error": str(e)
            }