# import gradio as gr # from langchain_mistralai.chat_models import ChatMistralAI # from langchain.prompts import ChatPromptTemplate # import os # from pathlib import Path # from typing import List, Dict, Optional # import json # import faiss # import numpy as np # from langchain.schema import Document # from sentence_transformers import SentenceTransformer # import pickle # import re # os.environ.get('HUGGINGFACE_TOKEN') # class RAGLoader: # def __init__(self, # docs_folder: str = "./docs", # splits_folder: str = "./splits", # index_folder: str = "./index",): # # model_name: str = "intfloat/multilingual-e5-large") # """ # Initialise le RAG Loader # Args: # docs_folder: Dossier contenant les documents sources # splits_folder: Dossier où seront stockés les morceaux de texte # index_folder: Dossier où sera stocké l'index FAISS # model_name: Nom du modèle SentenceTransformer à utiliser # """ # self.docs_folder = Path(docs_folder) # self.splits_folder = Path(splits_folder) # self.index_folder = Path(index_folder) # # self.model_name = model_name # # Créer les dossiers s'ils n'existent pas # self.splits_folder.mkdir(parents=True, exist_ok=True) # self.index_folder.mkdir(parents=True, exist_ok=True) # # Chemins des fichiers # self.splits_path = self.splits_folder / "splits.json" # self.index_path = self.index_folder / "faiss.index" # self.documents_path = self.index_folder / "documents.pkl" # # Initialiser le modèle # # self.model = None # self.index = None # self.indexed_documents = None # def encode(self,payload): # token = os.environ.get('HUGGINGFACE_TOKEN') # API_URL = "https://api-inference.huggingface.co/models/intfloat/multilingual-e5-large" # headers = {"Authorization": "Bearer {token}"} # response = requests.post(API_URL, headers=headers, json=payload) # return response.json() # def load_and_split_texts(self) -> List[Document]: # """ # Charge les textes du dossier docs, les découpe en morceaux et les sauvegarde # dans un fichier JSON unique. # Returns: # Liste de Documents contenant les morceaux de texte et leurs métadonnées # """ # documents = [] # # Vérifier d'abord si les splits existent déjà # if self._splits_exist(): # print("Chargement des splits existants...") # return self._load_existing_splits() # print("Création de nouveaux splits...") # # Parcourir tous les fichiers du dossier docs # for file_path in self.docs_folder.glob("*.txt"): # with open(file_path, 'r', encoding='utf-8') as file: # text = file.read() # # Découper le texte en phrases # # chunks = [chunk.strip() for chunk in text.split('.') if chunk.strip()] # chunks = [s.strip() for s in re.split(r'(?<=[.!?])\s+', text) if s.strip()] # # Créer un Document pour chaque morceau # for i, chunk in enumerate(chunks): # doc = Document( # page_content=chunk, # metadata={ # 'source': file_path.name, # 'chunk_id': i, # 'total_chunks': len(chunks) # } # ) # documents.append(doc) # # Sauvegarder tous les splits dans un seul fichier JSON # self._save_splits(documents) # print(f"Nombre total de morceaux créés: {len(documents)}") # return documents # def _splits_exist(self) -> bool: # """Vérifie si le fichier de splits existe""" # return self.splits_path.exists() # def _save_splits(self, documents: List[Document]): # """Sauvegarde tous les documents découpés dans un seul fichier JSON""" # splits_data = { # 'splits': [ # { # 'text': doc.page_content, # 'metadata': doc.metadata # } # for doc in documents # ] # } # with open(self.splits_path, 'w', encoding='utf-8') as f: # json.dump(splits_data, f, ensure_ascii=False, indent=2) # def _load_existing_splits(self) -> List[Document]: # """Charge les splits depuis le fichier JSON unique""" # with open(self.splits_path, 'r', encoding='utf-8') as f: # splits_data = json.load(f) # documents = [ # Document( # page_content=split['text'], # metadata=split['metadata'] # ) # for split in splits_data['splits'] # ] # print(f"Nombre de splits chargés: {len(documents)}") # return documents # def load_index(self) -> bool: # """ # Charge l'index FAISS et les documents associés s'ils existent # Returns: # bool: True si l'index a été chargé, False sinon # """ # if not self._index_exists(): # print("Aucun index trouvé.") # return False # print("Chargement de l'index existant...") # try: # # Charger l'index FAISS # self.index = faiss.read_index(str(self.index_path)) # # Charger les documents associés # with open(self.documents_path, 'rb') as f: # self.indexed_documents = pickle.load(f) # print(f"Index chargé avec {self.index.ntotal} vecteurs") # return True # except Exception as e: # print(f"Erreur lors du chargement de l'index: {e}") # return False # def create_index(self, documents: Optional[List[Document]] = None) -> bool: # """ # Crée un nouvel index FAISS à partir des documents. # Si aucun document n'est fourni, charge les documents depuis le fichier JSON. # Args: # documents: Liste optionnelle de Documents à indexer # Returns: # bool: True si l'index a été créé avec succès, False sinon # """ # try: # # # Initialiser le modèle si nécessaire # # if self.model is None: # # print("Chargement du modèle...") # # self.model = SentenceTransformer(self.model_name) # # Charger les documents si non fournis # if documents is None: # documents = self.load_and_split_texts() # if not documents: # print("Aucun document à indexer.") # return False # print("Création des embeddings...") # texts = [doc.page_content for doc in documents] # embeddings = self.encode(texts) # # Initialiser l'index FAISS # dimension = embeddings.shape[1] # self.index = faiss.IndexFlatL2(dimension) # # Ajouter les vecteurs à l'index # self.index.add(np.array(embeddings).astype('float32')) # # Sauvegarder l'index # print("Sauvegarde de l'index...") # faiss.write_index(self.index, str(self.index_path)) # # Sauvegarder les documents associés # self.indexed_documents = documents # with open(self.documents_path, 'wb') as f: # pickle.dump(documents, f) # print(f"Index créé avec succès : {self.index.ntotal} vecteurs") # return True # except Exception as e: # print(f"Erreur lors de la création de l'index: {e}") # return False # def _index_exists(self) -> bool: # """Vérifie si l'index et les documents associés existent""" # return self.index_path.exists() and self.documents_path.exists() # def get_retriever(self, k: int = 10): # """ # Crée un retriever pour l'utilisation avec LangChain # Args: # k: Nombre de documents similaires à retourner # Returns: # Callable: Fonction de recherche compatible avec LangChain # """ # if self.index is None: # if not self.load_index(): # if not self.create_index(): # raise ValueError("Impossible de charger ou créer l'index") # # if self.model is None: # # self.model = SentenceTransformer(self.model_name) # def retriever_function(query: str) -> List[Document]: # # Créer l'embedding de la requête # query_embedding = self.encode([query])[0] # # Rechercher les documents similaires # distances, indices = self.index.search( # np.array([query_embedding]).astype('float32'), # k # ) # # Retourner les documents trouvés # results = [] # for idx in indices[0]: # if idx != -1: # FAISS retourne -1 pour les résultats invalides # results.append(self.indexed_documents[idx]) # return results # return retriever_function # # Initialize the RAG system # llm = ChatMistralAI(model="mistral-large-latest", mistral_api_key="QK0ZZpSxQbCEVgOLtI6FARQVmBYc6WGP") # rag_loader = RAGLoader() # retriever = rag_loader.get_retriever(k=10) # prompt_template = ChatPromptTemplate.from_messages([ # ("system", """أنت مساعد مفيد يجيب على الأسئلة باللغة العربية باستخدام المعلومات المقدمة. # استخدم المعلومات التالية للإجابة على السؤال: # {context} # إذا لم تكن المعلومات كافية للإجابة على السؤال بشكل كامل، قم بتوضيح ذلك. # أجب بشكل موجز ودقيق."""), # ("human", "{question}") # ]) # def process_question(question: str) -> tuple[str, str]: # """ # Process a question and return both the answer and the relevant context # """ # relevant_docs = retriever(question) # context = "\n".join([doc.page_content for doc in relevant_docs]) # prompt = prompt_template.format_messages( # context=context, # question=question # ) # response = llm(prompt) # return response.content, context # def gradio_interface(question: str) -> tuple[str, str]: # """ # Gradio interface function that returns both answer and context as a tuple # """ # return process_question(question) # # Custom CSS for right-aligned text in textboxes # custom_css = """ # .rtl-text { # text-align: right !important; # direction: rtl !important; # } # .rtl-text textarea { # text-align: right !important; # direction: rtl !important; # } # """ # # Define the Gradio interface # with gr.Blocks(css=custom_css) as iface: # with gr.Column(): # input_text = gr.Textbox( # label="السؤال", # placeholder="اكتب سؤالك هنا...", # lines=2, # elem_classes="rtl-text" # ) # answer_box = gr.Textbox( # label="الإجابة", # lines=4, # elem_classes="rtl-text" # ) # context_box = gr.Textbox( # label="السياق المستخدم", # lines=8, # elem_classes="rtl-text" # ) # submit_btn = gr.Button("إرسال") # submit_btn.click( # fn=gradio_interface, # inputs=input_text, # outputs=[answer_box, context_box] # ) # # Launch the interface # if __name__ == "__main__": # iface.launch(share=True) import gradio as gr from langchain_mistralai.chat_models import ChatMistralAI from langchain.prompts import ChatPromptTemplate import os from pathlib import Path import json import faiss import numpy as np from langchain.schema import Document import pickle import re import requests from functools import lru_cache import torch from sentence_transformers import SentenceTransformer import threading from queue import Queue import concurrent.futures class OptimizedRAGLoader: def __init__(self, docs_folder: str = "./docs", splits_folder: str = "./splits", index_folder: str = "./index"): self.docs_folder = Path(docs_folder) self.splits_folder = Path(splits_folder) self.index_folder = Path(index_folder) # Create folders if they don't exist for folder in [self.splits_folder, self.index_folder]: folder.mkdir(parents=True, exist_ok=True) # File paths self.splits_path = self.splits_folder / "splits.json" self.index_path = self.index_folder / "faiss.index" self.documents_path = self.index_folder / "documents.pkl" # Initialize components self.index = None self.indexed_documents = None # Initialize encoder model self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu") self.encoder = SentenceTransformer("intfloat/multilingual-e5-large") self.encoder.to(self.device) # Initialize thread pool self.executor = concurrent.futures.ThreadPoolExecutor(max_workers=4) # Initialize response cache self.response_cache = {} @lru_cache(maxsize=1000) def encode(self, text: str): """Cached encoding function""" with torch.no_grad(): embeddings = self.encoder.encode( text, convert_to_numpy=True, normalize_embeddings=True ) return embeddings def batch_encode(self, texts: list): """Batch encoding for multiple texts""" with torch.no_grad(): embeddings = self.encoder.encode( texts, batch_size=32, convert_to_numpy=True, normalize_embeddings=True, show_progress_bar=False ) return embeddings def load_and_split_texts(self): if self._splits_exist(): return self._load_existing_splits() documents = [] futures = [] for file_path in self.docs_folder.glob("*.txt"): future = self.executor.submit(self._process_file, file_path) futures.append(future) for future in concurrent.futures.as_completed(futures): documents.extend(future.result()) self._save_splits(documents) return documents def _process_file(self, file_path): with open(file_path, 'r', encoding='utf-8') as file: text = file.read() chunks = [s.strip() for s in re.split(r'(?<=[.!?])\s+', text) if s.strip()] return [ Document( page_content=chunk, metadata={ 'source': file_path.name, 'chunk_id': i, 'total_chunks': len(chunks) } ) for i, chunk in enumerate(chunks) ] def load_index(self) -> bool: """ Charge l'index FAISS et les documents associés s'ils existent Returns: bool: True si l'index a été chargé, False sinon """ if not self._index_exists(): print("Aucun index trouvé.") return False print("Chargement de l'index existant...") try: # Charger l'index FAISS self.index = faiss.read_index(str(self.index_path)) # Charger les documents associés with open(self.documents_path, 'rb') as f: self.indexed_documents = pickle.load(f) print(f"Index chargé avec {self.index.ntotal} vecteurs") return True except Exception as e: print(f"Erreur lors du chargement de l'index: {e}") return False def create_index(self, documents=None): if documents is None: documents = self.load_and_split_texts() if not documents: return False texts = [doc.page_content for doc in documents] embeddings = self.batch_encode(texts) dimension = embeddings.shape[1] self.index = faiss.IndexFlatL2(dimension) if torch.cuda.is_available(): # Use GPU for FAISS if available res = faiss.StandardGpuResources() self.index = faiss.index_cpu_to_gpu(res, 0, self.index) self.index.add(np.array(embeddings).astype('float32')) self.indexed_documents = documents # Save index and documents cpu_index = faiss.index_gpu_to_cpu(self.index) if torch.cuda.is_available() else self.index faiss.write_index(cpu_index, str(self.index_path)) with open(self.documents_path, 'wb') as f: pickle.dump(documents, f) return True def _index_exists(self) -> bool: """Vérifie si l'index et les documents associés existent""" return self.index_path.exists() and self.documents_path.exists() def get_retriever(self, k: int = 10): if self.index is None: if not self.load_index(): if not self.create_index(): raise ValueError("Unable to load or create index") def retriever_function(query: str) -> list: # Check cache first cache_key = f"{query}_{k}" if cache_key in self.response_cache: return self.response_cache[cache_key] query_embedding = self.encode(query) distances, indices = self.index.search( np.array([query_embedding]).astype('float32'), k ) results = [ self.indexed_documents[idx] for idx in indices[0] if idx != -1 ] # Cache the results self.response_cache[cache_key] = results return results return retriever_function # Initialize components llm = ChatMistralAI( model="mistral-large-latest", mistral_api_key="QK0ZZpSxQbCEVgOLtI6FARQVmBYc6WGP", temperature=0.1 # Lower temperature for faster responses ) rag_loader = OptimizedRAGLoader() retriever = rag_loader.get_retriever(k=5) # Reduced k for faster retrieval # Cache for processed questions question_cache = {} prompt_template = ChatPromptTemplate.from_messages([ ("system", """أنت مساعد مفيد يجيب على الأسئلة باللغة العربية باستخدام المعلومات المقدمة. استخدم المعلومات التالية للإجابة على السؤال: {context} إذا لم تكن المعلومات كافية للإجابة على السؤال بشكل كامل، قم بتوضيح ذلك. أجب بشكل موجز ودقيق."""), ("human", "{question}") ]) def process_question(question: str) -> tuple[str, str]: # Check cache first if question in question_cache: return question_cache[question] relevant_docs = retriever(question) context = "\n".join([doc.page_content for doc in relevant_docs]) prompt = prompt_template.format_messages( context=context, question=question ) response = llm(prompt) result = (response.content, context) # Cache the result question_cache[question] = result return result # Custom CSS for right-aligned text in textboxes custom_css = """ .rtl-text { text-align: right !important; direction: rtl !important; } .rtl-text textarea { text-align: right !important; direction: rtl !important; } """ # Gradio interface with queue with gr.Blocks(css=custom_css) as iface: with gr.Column(): input_text = gr.Textbox( label="السؤال", placeholder="اكتب سؤالك هنا...", lines=2, elem_classes="rtl-text" ) with gr.Row(): answer_box = gr.Textbox( label="الإجابة", lines=4, elem_classes="rtl-text" ) context_box = gr.Textbox( label="السياق المستخدم", lines=8, elem_classes="rtl-text" ) submit_btn = gr.Button("إرسال") submit_btn.click( fn=process_question, inputs=input_text, outputs=[answer_box, context_box], api_name="predict" ) if __name__ == "__main__": iface.launch( share=True, server_name="0.0.0.0", server_port=7860, max_threads=3, # Controls concurrency show_error=True )