Nadine Elleuch 1 tahun lalu
induk
melakukan
f67956f54b
50 mengubah file dengan 4361 tambahan dan 413 penghapusan
  1. 0 413
      app.py
  2. 0 0
      src/__init__.py
  3. TEMPAT SAMPAH
      src/__pycache__/__init__.cpython-312.pyc
  4. TEMPAT SAMPAH
      src/__pycache__/app.cpython-312.pyc
  5. 0 0
      src/agent_ai/__init__.py
  6. TEMPAT SAMPAH
      src/agent_ai/__pycache__/__init__.cpython-312.pyc
  7. TEMPAT SAMPAH
      src/agent_ai/__pycache__/auth.cpython-310.pyc
  8. TEMPAT SAMPAH
      src/agent_ai/__pycache__/auth.cpython-312.pyc
  9. TEMPAT SAMPAH
      src/agent_ai/__pycache__/chunking.cpython-310.pyc
  10. TEMPAT SAMPAH
      src/agent_ai/__pycache__/chunking.cpython-312.pyc
  11. TEMPAT SAMPAH
      src/agent_ai/__pycache__/embedding_index.cpython-312.pyc
  12. TEMPAT SAMPAH
      src/agent_ai/__pycache__/qa_chain.cpython-312.pyc
  13. TEMPAT SAMPAH
      src/agent_ai/__pycache__/scraping.cpython-310.pyc
  14. TEMPAT SAMPAH
      src/agent_ai/__pycache__/scraping.cpython-312.pyc
  15. TEMPAT SAMPAH
      src/agent_ai/__pycache__/test_auth.cpython-312-pytest-8.4.1.pyc
  16. TEMPAT SAMPAH
      src/agent_ai/__pycache__/test_chunking.cpython-312-pytest-8.4.1.pyc
  17. TEMPAT SAMPAH
      src/agent_ai/__pycache__/test_embedding_index.cpython-312-pytest-8.4.1.pyc
  18. TEMPAT SAMPAH
      src/agent_ai/__pycache__/test_qa_chain.cpython-312-pytest-8.4.1.pyc
  19. TEMPAT SAMPAH
      src/agent_ai/__pycache__/test_scraping.cpython-312-pytest-8.4.1.pyc
  20. 43 0
      src/agent_ai/auth.py
  21. 86 0
      src/agent_ai/chunking.py
  22. 192 0
      src/agent_ai/scraping.py
  23. 0 0
      src/api/__init__.py
  24. 490 0
      src/app.py
  25. 10 0
      src/data/All_chunks.json
  26. TEMPAT SAMPAH
      src/data/faiss_index.index
  27. 1 0
      src/data/page_hashes.json
  28. 3 0
      src/data/raw_pages.json
  29. 438 0
      src/data/urls.json
  30. 25 0
      src/models/VLM.py
  31. 0 0
      src/models/__init__.py
  32. TEMPAT SAMPAH
      src/models/__pycache__/VLM.cpython-312.pyc
  33. TEMPAT SAMPAH
      src/models/__pycache__/__init__.cpython-312.pyc
  34. 540 0
      src/templates/index.html
  35. TEMPAT SAMPAH
      src/uploads/cos.png
  36. TEMPAT SAMPAH
      src/uploads/cross.png
  37. TEMPAT SAMPAH
      src/uploads/methodologie.png
  38. 24 0
      src/uploads/ocr_output.html
  39. 2123 0
      src/uploads/ocr_output.json
  40. 25 0
      src/uploads/ocr_output_words.html
  41. TEMPAT SAMPAH
      src/uploads/processus_.png
  42. TEMPAT SAMPAH
      src/uploads/reg_chunk.png
  43. 0 0
      src/utils/__init__.py
  44. TEMPAT SAMPAH
      src/utils/__pycache__/__init__.cpython-312.pyc
  45. TEMPAT SAMPAH
      src/utils/__pycache__/ocr.cpython-312.pyc
  46. TEMPAT SAMPAH
      src/utils/__pycache__/processing.cpython-312.pyc
  47. TEMPAT SAMPAH
      src/utils/__pycache__/security.cpython-312.pyc
  48. 162 0
      src/utils/ocr.py
  49. 115 0
      src/utils/processing.py
  50. 84 0
      src/utils/security.py

+ 0 - 413
app.py

@@ -1,413 +0,0 @@
-from flask import Flask, request, jsonify, send_from_directory
-from apscheduler.schedulers.background import BackgroundScheduler
-from sentence_transformers import CrossEncoder, SentenceTransformer
-from langchain_ollama import OllamaLLM
-from processing import full_update_pipeline
-from agent_ai.auth import login_to_wiki
-import faiss 
-import json
-import numpy as np
-import textwrap
-from sklearn.metrics.pairwise import cosine_similarity
-from tabulate import tabulate
-import os
-import threading
-from serpapi import GoogleSearch
-from werkzeug.utils import secure_filename
-import PyPDF2
-from PIL import Image
-from transformers import AutoTokenizer
-from ocr import run_ocr, generate_html , generate_word_level_html  
-import tiktoken
-from VLM import ask_blip
-import getpass
-
-
-app = Flask(__name__)
-
-llm_variants = [
-    {
-        "name": "config_1",
-        "model": "llama3",
-        "temperature": 0.1,
-        "top_p": 0.95,
-        "max_tokens": 1000
-    },
-    {
-        "name": "config_2",
-        "model": "llama3",
-        "temperature": 0.7,
-        "top_p": 0.85,
-        "max_tokens": 1000
-    },
-    {
-        "name": "config_3",
-        "model": "llama3",
-        "temperature": 0.3,
-        "top_p": 0.9,
-        "max_tokens": 1000
-    }
-]
-
-state = {
-    "index": None,
-    "all_chunks": [],
-    "chunk_urls": [],
-    "all_embeddings": None,
-    "embedding_model": SentenceTransformer('all-MiniLM-L6-v2'),
-    "cross_encoder": CrossEncoder("cross-encoder/ms-marco-MiniLM-L-6-v2"),
-    "is_updating": False,
-    "llm": llm_variants,
-    "chat_history": []
-   
-}
-
-def run_full_update():
-    print("🔄 Lancement du pipeline via le scheduler...")
-    full_update_pipeline()
-
-tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
-
-def chunk_text(text, chunk_size=400, overlap=100):
-    tokens = tokenizer.encode(text)
-    chunks = []
-    for i in range(0, len(tokens), chunk_size - overlap):
-        chunk = tokens[i:i + chunk_size]
-        chunks.append(tokenizer.decode(chunk))
-    return chunks
-
-
-UPLOAD_FOLDER = "uploads"
-os.makedirs(UPLOAD_FOLDER, exist_ok=True)  # Crée le dossier s'il n'existe pas
-app.config["UPLOAD_FOLDER"] = UPLOAD_FOLDER
-
-ALLOWED_EXTENSIONS = {'txt', 'pdf', 'png', 'jpg', 'jpeg'}
-
-def allowed_file(filename):
-    return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS
-
-@app.route("/upload_file", methods=["POST"])
-def upload_file():
-    if 'file' not in request.files:
-        return jsonify({"error": "Aucun fichier détecté."}), 400
-
-    file = request.files['file']
-    if file.filename == '':
-        return jsonify({"error": "Nom de fichier vide."}), 400
-
-    if file and allowed_file(file.filename):
-        filename = secure_filename(file.filename)
-        filepath = os.path.join(app.config['UPLOAD_FOLDER'], filename)
-        file.save(filepath)
-
-        question = request.form.get("question", "")
-        if not question:
-            return jsonify({"error": "Veuillez poser une question liée au fichier."}), 400
-
-        try:
-            # ✅ 1. Exécuter l'OCR
-            document, ocr_output = run_ocr(filepath)
-
-            # ✅ 2. Sauvegarder le JSON OCR
-            json_path = os.path.join(app.config['UPLOAD_FOLDER'], "ocr_output.json")
-            with open(json_path, "w", encoding="utf-8") as fjson:
-                json.dump(ocr_output, fjson, ensure_ascii=False, indent=2)
-
-            # ✅ 3. Générer le HTML avec bounding boxes par mot
-            html_path = os.path.join(app.config['UPLOAD_FOLDER'], "ocr_output_words.html")
-            generate_word_level_html(ocr_output, output_html_path=html_path)
-
-            # ✅ 4. Extraire le texte brut pour le traitement RAG
-            text = ""
-            for page in ocr_output['pages']:
-                for block in page['blocks']:
-                    for line in block['lines']:
-                        for word in line['words']:
-                            text += word['value'] + " "
-                        text += "\n"
-                    text += "\n"
-                text += "\n"
-
-        except Exception as e:
-            return jsonify({"error": f"OCR échoué : {str(e)}"}), 500
-
-        if text.strip():
-            chunks = chunk_text(text)
-            embeddings = state["embedding_model"].encode(chunks)
-            question_embedding = state["embedding_model"].encode([question])[0]
-            similarities = cosine_similarity([question_embedding], embeddings)[0]
-            top_indices = np.argsort(similarities)[::-1][:5]
-            top_chunks = [chunks[i] for i in top_indices]
-            top_context = "\n\n".join(top_chunks)
-
-            prompt = (
-                "Réponds en français à la question suivante en utilisant uniquement le contenu fourni. "
-                "Si la réponse est absente, dis-le clairement.\n\n"
-                f"{top_context}\n\nQuestion : {question}"
-            )
-
-            answer = state["llm"](prompt).strip()
-            return jsonify({
-                "answer": answer,
-                "html_file": f"uploads/ocr_output_words.html"
-            })
-
-        else:
-            try:
-                answer = ask_blip(filepath, question)
-                return jsonify({"answer": answer})
-            except Exception as e:
-                return jsonify({"error": f"LLaVA échoué : {str(e)}"}), 500
-
-    return jsonify({"error": "Format de fichier non supporté."}), 400
-
-@app.route("/delete_file", methods=["POST"])
-def delete_file():
-    data = request.get_json()
-    filename = data.get("filename")
-
-    if not filename:
-        return jsonify({"error": "Nom de fichier manquant."}), 400
-
-    filepath = os.path.join(app.config["UPLOAD_FOLDER"], secure_filename(filename))
-
-    if os.path.exists(filepath):
-        os.remove(filepath)
-        return jsonify({"message": "Fichier supprimé avec succès."})
-    else:
-        return jsonify({"error": "Fichier introuvable."}), 404
-
-
-@app.route('/')
-def index():
-    return send_from_directory(os.getcwd(), 'index.html')
-
-@app.route("/login", methods=["POST"])
-def login():
-    data = request.json
-    username = data.get("username")
-    password = data.get("password")
-
-    if not username or not password:
-        return jsonify({"error": "username and password required"}), 400
-
-    if not login_to_wiki(username, password):
-        return jsonify({"error": "Authentication failed"}), 401
-
-    return jsonify({"message": "Login successful"})
-
-@app.route("/chat_history", methods=["GET"])
-def chat_history():
-    return jsonify(state["chat_history"])
-
-@app.route("/qa_chain", methods=["POST"])
-def qa_chain_route():
-    if state.get("llm") is None:
-        return jsonify({"error": "Le modèle LLM n'est pas initialisé."}), 500
-
-    if state["index"] is None or not state["all_chunks"]:
-        if state["is_updating"]:
-            return jsonify({"error": "Le système est encore en train d’indexer les documents. Réessayez dans quelques instants."}), 503
-        else:
-            return jsonify({"error": "L’index est vide. Veuillez attendre la mise à jour ou réinitialiser le système."}), 500
-
-    data = request.get_json()
-    question = data.get("question", "")
-
-    if not question:
-        return jsonify({"error": "Question vide."}), 400
-
-    try:
-        query_embedding = state["embedding_model"].encode([question])[0]
-
-        if state["all_embeddings"] is None:
-            state["all_embeddings"] = state["embedding_model"].encode(state["all_chunks"])
-
-        similarities = cosine_similarity([query_embedding], state["all_embeddings"])[0]
-        top_indices = np.argsort(similarities)[::-1][:20]
-
-        rerank_data = [(question, state["all_chunks"][i]) for i in top_indices]
-        rerank_scores = state["cross_encoder"].predict(rerank_data)
-        reranked = sorted(
-            zip([state["all_chunks"][i] for i in top_indices], top_indices, rerank_scores),
-            key=lambda x: x[2], reverse=True
-        )
-
-        enc = tiktoken.encoding_for_model("gpt-3.5-turbo")
-        cache_data = []
-
-        for rank, (chunk, idx, score) in enumerate(reranked[:20], start=1):
-            url = state["chunk_urls"][idx] if idx < len(state["chunk_urls"]) else "URL inconnue"
-            token_count = len(enc.encode(chunk))
-
-            cache_data.append({
-                "rank": rank,
-                "url": url,
-                "chunk": chunk,
-                "tokens": token_count
-            })
-
-        with open("cache_context.json", "w", encoding="utf-8") as f:
-            json.dump(cache_data, f, indent=2, ensure_ascii=False)
-
-        table = []
-        top_chunks = []
-        for rank, (chunk, idx, score) in enumerate(reranked, start=1):
-            url = state["chunk_urls"][idx] if idx < len(state["chunk_urls"]) else "URL inconnue"
-            wrapped_chunk = "\n".join(textwrap.wrap(chunk.strip(), width=80))[:1000] + "..."
-            wrapped_url = "\n".join(textwrap.wrap(url, width=80))
-            table.append([rank, round(score, 4), wrapped_chunk, wrapped_url])
-            top_chunks.append(chunk)
-
-        print("\n================== RÉSULTATS CLASSÉS PAR PERTINENCE ==================")
-        print(tabulate(table, headers=["Rank", "Score", "Chunk", "URL"], tablefmt="fancy_grid"))
-        print("=======================================================================\n")
-
-        max_chars = 15000
-        top_chunks = []
-        current_length = 0
-        for chunk, _, _ in reranked:
-            if current_length + len(chunk) > max_chars:
-                break
-            top_chunks.append(chunk)
-            current_length += len(chunk)
-
-        top_context = "\n\n".join(top_chunks)
-
-        llm_variants = [
-            {
-                "name": "config_1",
-                "model": "llama3",
-                "temperature": 0.1,
-                "top_p": 0.95,
-                "max_tokens": 600
-            },
-            {
-                "name": "config_2",
-                "model": "llama3",
-                "temperature": 0.7,
-                "top_p": 0.85,
-                "max_tokens": 600
-            },
-            {
-                "name": "config_3",
-                "model": "llama3",
-                "temperature": 0.3,
-                "top_p": 0.9,
-                "max_tokens": 600
-            }
-        ]
-
-        all_answers = []
-
-        for config in llm_variants:
-            llm = OllamaLLM(
-                model=config["model"],
-                temperature=config["temperature"],
-                top_p=config["top_p"],
-                max_tokens=config["max_tokens"]
-            )
-
-            prompt = f"""
-Tu es un assistant intelligent. Réponds uniquement avec les informations données dans le contexte suivant. 
-Ta réponse doit être factuelle, concise et précise. Si la réponse n’est pas présente dans le contexte, dis :
-"Information non disponible."
-
-Ne reformule pas la question. N’invente rien.
-
-Contexte :
-{top_context}
-
-Question : {question}
-Réponse :
-"""
-
-            response = llm(prompt).strip()
-            all_answers.append({
-                "config": config["name"],
-                "response": response
-            })
-
-        print("\n====================== PROMPTS & RÉPONSES ======================")
-        for ans in all_answers:
-            print(f"\nConfig: {ans['config']}\nRéponse : {ans['response']}")
-        print("==============================================================\n")
-
-        return jsonify({"answers": all_answers})
-
-    except Exception as e:
-        print(f"[Erreur Wiki] : {e}")
-       
-
-    #  ÉTAPE 2 : Fallback vers recherche web (SerpAPI) 
-    try:
-        params = {
-            "engine": "google",
-            "q": question,
-            "api_key": "1b0f09bd610259b457b08c5fad06ded73a9e108e6bcc932e02ac88b2f3840171",
-            "gl": "fr",
-            "hl": "fr"
-        }
-        search = GoogleSearch(params)
-        results = search.get_dict()
-
-        answer = "Je n'ai pas trouvé de réponse précise."
-        if "answer_box" in results and "answer" in results["answer_box"]:
-            answer = results["answer_box"]["answer"]
-        elif "organic_results" in results and results["organic_results"]:
-            answer = results["organic_results"][0].get("snippet", answer)
-
-        #  Résumer ou filtrer la réponse via le LLM 
-        if state.get("llm"):
-            prompt = (
-    "Réponds uniquement à la question de manière directe et concise.\n"
-    "Sans reformuler la question, sans phrase d’introduction, sans politesse et sans contenu promotionnel. "
-    "Utilise uniquement les faits du texte ci-dessous. "
-    "Si l’information n’est pas présente, réponds : 'Information non disponible.'\n\n"
-    f"Texte : {answer}\n\n"
-    f"Question : {question}"
-)
-
-            try:
-                answer = state["llm"](prompt).strip()
-            except Exception as llm_error:
-                print("[Erreur LLM sur reformulation] :", llm_error)
-
-        # Historique
-        state["chat_history"].append({
-            "question": question,
-            "answer": answer,
-            "sources": ["Google Search"]
-        })
-
-        return jsonify({
-            "answer": answer,
-            "sources": ["Google Search"]
-        })
-
-    except Exception as e:
-        return jsonify({"error": f"Erreur SerpAPI : {str(e)}"}), 500
-
-
-
-# Scheduler
-scheduler = BackgroundScheduler()
-scheduler.add_job(run_full_update, 'interval', minutes=150)
-print("🔁 Scheduler activé pour mises à jour toutes les 30 min")
-
-def prompt_for_login():
-    print("🔐 Please log in to the Wiki before starting the app.")
-    while True:
-        username = input("Username: ")
-        password = getpass.getpass("Password: ")
-        if login_to_wiki(username, password):
-            print("Login successful, starting the app...")
-            break
-        else:
-            print("Login failed, please try again.\n")
-if __name__ == "__main__":
-    
-    prompt_for_login()
-    full_update_pipeline()
-    scheduler.start()
-    app.run(debug=True, use_reloader=False)

+ 0 - 0
src/__init__.py


TEMPAT SAMPAH
src/__pycache__/__init__.cpython-312.pyc


TEMPAT SAMPAH
src/__pycache__/app.cpython-312.pyc


+ 0 - 0
src/agent_ai/__init__.py


TEMPAT SAMPAH
src/agent_ai/__pycache__/__init__.cpython-312.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/auth.cpython-310.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/auth.cpython-312.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/chunking.cpython-310.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/chunking.cpython-312.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/embedding_index.cpython-312.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/qa_chain.cpython-312.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/scraping.cpython-310.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/scraping.cpython-312.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/test_auth.cpython-312-pytest-8.4.1.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/test_chunking.cpython-312-pytest-8.4.1.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/test_embedding_index.cpython-312-pytest-8.4.1.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/test_qa_chain.cpython-312-pytest-8.4.1.pyc


TEMPAT SAMPAH
src/agent_ai/__pycache__/test_scraping.cpython-312-pytest-8.4.1.pyc


+ 43 - 0
src/agent_ai/auth.py

@@ -0,0 +1,43 @@
+"""
+auth.py 
+
+Gestion de l’authentification sur le wiki Intersel
+
+Ce module permet de se connecter au wiki en utilisant une session HTTP persistante et il récupère tout seul un code spécial (le token CSRF) nécessaire pour se connecter en toute sécurité
+"""
+
+import requests
+from bs4 import BeautifulSoup
+
+session = requests.Session() # permet de conserver les cookies entre les requêtes pour rester connecté
+
+def login_to_wiki(username, password):
+    login_url = "https://wiki.intersel.fr/doku.php?id=start&do=login"
+    resp = session.get(login_url)
+    soup = BeautifulSoup(resp.text, "html.parser")
+    sectok_input = soup.find("input", {"name": "sectok"})
+
+    if not sectok_input:
+        print("❌ Impossible de trouver le token CSRF.")
+        return False
+
+    login_data = {
+        "u": username,
+        "p": password,
+        "do": "login",
+        "sectok": sectok_input.get("value")
+    }
+
+    headers = {
+        "User-Agent": "Mozilla/5.0",
+        "Referer": login_url
+    }
+
+    login_resp = session.post(login_url, data=login_data, headers=headers)
+
+    if "logout" in login_resp.text.lower():
+        print("✅ Connexion réussie.")
+        return True
+    else:
+        print("❌ Échec de la connexion.")
+        return False

+ 86 - 0
src/agent_ai/chunking.py

@@ -0,0 +1,86 @@
+from sentence_transformers import SentenceTransformer
+from sklearn.metrics.pairwise import cosine_similarity
+import nltk
+import tiktoken
+import numpy as np
+from typing import List
+
+nltk.download('punkt')
+
+tokenizer = tiktoken.get_encoding("cl100k_base")
+model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2', trust_remote_code=True)
+
+MAX_TOKENS = 350
+CHUNK_OVERLAP = 50
+SIMILARITY_THRESHOLD = 0.75  # adjust as needed
+
+def tokenize_text(text):
+    return tokenizer.encode(text)
+
+def count_tokens(text):
+    return len(tokenize_text(text))
+
+def embed_sentences(sentences: List[str]):
+    return model.encode(sentences)
+
+def semantic_chunking(text: str, max_tokens=MAX_TOKENS, overlap=CHUNK_OVERLAP, with_metadata=True) -> List[dict]:
+    sentences = nltk.sent_tokenize(text)
+    embeddings = embed_sentences(sentences)
+    chunks = []
+    current_chunk = []
+    current_tokens = 0
+    chunk_id = 0
+
+    for i in range(len(sentences)):
+        sent = sentences[i]
+        sent_tokens = count_tokens(sent)
+        sent_embedding = embeddings[i]
+
+        if not current_chunk:
+            current_chunk.append(sent)
+            current_tokens += sent_tokens
+            continue
+
+        # Compute similarity to previous sentence in chunk
+        prev_embedding = embed_sentences([" ".join(current_chunk)])[-1]
+        similarity = cosine_similarity([sent_embedding], [prev_embedding])[0][0]
+
+        if similarity >= SIMILARITY_THRESHOLD and current_tokens + sent_tokens <= max_tokens:
+            current_chunk.append(sent)
+            current_tokens += sent_tokens
+        else:
+            chunk_text = " ".join(current_chunk)
+            if with_metadata:
+                chunks.append({
+                    "chunk_id": chunk_id,
+                    "text": chunk_text,
+                    "word_count": len(chunk_text.split()),
+                    "token_count": current_tokens
+                })
+            else:
+                chunks.append(chunk_text)
+            chunk_id += 1
+
+            # Create overlapping window
+            overlap_chunk = current_chunk[-1:] if overlap > 0 else []
+            current_chunk = overlap_chunk.copy()
+            current_tokens = sum(count_tokens(s) for s in current_chunk)
+
+            # Add the new sentence to start a new chunk
+            current_chunk.append(sent)
+            current_tokens += sent_tokens
+
+    # Final chunk
+    if current_chunk:
+        chunk_text = " ".join(current_chunk)
+        if with_metadata:
+            chunks.append({
+                "chunk_id": chunk_id,
+                "text": chunk_text,
+                "word_count": len(chunk_text.split()),
+                "token_count": current_tokens
+            })
+        else:
+            chunks.append(chunk_text)
+
+    return chunks

+ 192 - 0
src/agent_ai/scraping.py

@@ -0,0 +1,192 @@
+from bs4 import BeautifulSoup
+from src.agent_ai.auth import session
+from urllib.parse import urljoin, urlparse
+import time
+import hashlib
+import json
+import random
+from concurrent.futures import ThreadPoolExecutor, as_completed
+import os
+
+# 🧭 Config
+BASE_URL = "https://wiki.intersel.fr/"
+USER_AGENTS = [
+    "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/115.0.0.0 Safari/537.36"
+]
+MAX_PAGES = 500
+
+def random_headers():
+    return {
+        "User-Agent": random.choice(USER_AGENTS)
+    }
+
+def hash_text(text):
+    return hashlib.md5(text.encode("utf-8")).hexdigest()
+
+def fetch_single_page(url, old_hashes):
+    try:
+        headers = random_headers()
+        edit_url = url + "&do=edit" if "?" in url else url + "?do=edit"
+        response = session.get(edit_url, headers=headers, timeout=10)
+
+        if response.status_code != 200:
+            return url, None, None
+
+        response.encoding = 'utf-8'
+
+        if "text/html" not in response.headers.get("Content-Type", ""):
+            return url, None, None
+
+        soup = BeautifulSoup(response.text, "html.parser")
+        textarea = soup.find("textarea", {"name": "wikitext"})
+
+        if textarea:
+            raw_text = textarea.get_text(separator="\n").strip()
+            current_hash = hash_text(raw_text)
+
+            if old_hashes.get(url) != current_hash:
+                print(f"🆕 Modifiée : {url}")
+                return url, raw_text, current_hash
+            else:
+                print(f"✅ Inchangée : {url}")
+        else:
+            print(f"⚠️ Pas de <textarea> pour {url}")
+    except Exception as e:
+        print(f"❌ Erreur pour {url}: {e}")
+
+    time.sleep(random.uniform(0.5, 2))
+    return url, None, None
+
+def fetch_all_pages_content(urls, old_hashes):
+    url_texts = {}
+    new_hashes = {}
+
+    with ThreadPoolExecutor(max_workers=2) as executor:
+        futures = [executor.submit(fetch_single_page, url, old_hashes) for url in urls]
+
+        for future in as_completed(futures):
+            url, text, current_hash = future.result()
+            if text:
+                url_texts[url] = text
+                new_hashes[url] = current_hash
+            elif current_hash:
+                new_hashes[url] = current_hash
+
+    return url_texts, new_hashes
+
+def is_valid_wiki_url(url):
+    parsed = urlparse(url)
+    return (
+        parsed.netloc == "wiki.intersel.fr"
+        and "do=" not in parsed.query
+        and not url.endswith((".jpg", ".png", ".pdf", ".docx", ".xlsx"))
+        and "#" not in url
+    )
+
+def crawl_wiki(start_url, existing_urls):
+    visited = set(existing_urls)
+    to_visit = [start_url]
+
+    while to_visit and len(visited) < MAX_PAGES:
+        current_url = to_visit.pop(0)
+        if current_url in visited:
+            continue
+
+        print(f"🔎 Visiting: {current_url}")
+        visited.add(current_url)
+
+        try:
+            headers = random_headers()
+            response = session.get(current_url, headers=headers, timeout=10)
+            if response.status_code != 200:
+                continue
+
+            soup = BeautifulSoup(response.text, "html.parser")
+
+            for link in soup.find_all("a", href=True):
+                absolute_url = urljoin(current_url, link["href"])
+                if (
+                    is_valid_wiki_url(absolute_url)
+                    and absolute_url not in visited
+                    and absolute_url not in to_visit
+                ):
+                    to_visit.append(absolute_url)
+
+            time.sleep(random.uniform(1, 2))
+
+            if len(visited) >= MAX_PAGES:
+                break
+
+        except Exception as e:
+            print(f"❌ Error with {current_url}: {e}")
+
+    return list(visited)
+
+def scrape_pages():
+    start_url = BASE_URL
+    urls_path = os.path.join("data", "urls.json")
+    raw_pages_path = "data/raw_pages.json"
+    hashes_path = "data/page_hashes.json"
+
+    # Charger anciens hashs
+    try:
+        with open(hashes_path, "r", encoding="utf-8") as f:
+            old_hashes = json.load(f)
+    except FileNotFoundError:
+        old_hashes = {}
+
+    # Charger anciennes URLs
+    try:
+        with open(urls_path, "r", encoding="utf-8") as f:
+            urls = json.load(f)
+        print("✅ URLs chargées depuis urls.json")
+    except FileNotFoundError:
+        print("🚀 Crawling initial du wiki")
+        urls = []
+
+    # Nouveau crawl
+    crawled_urls = crawl_wiki(start_url, urls)
+    all_urls = sorted(set(urls + crawled_urls))[:MAX_PAGES]
+
+    print(f"🌐 Total URLs à vérifier : {len(all_urls)}")
+
+    # Scraping des pages
+    pages, new_hashes = fetch_all_pages_content(all_urls, old_hashes)
+
+    # Charger anciens raw_pages
+    try:
+        with open(raw_pages_path, "r", encoding="utf-8") as f:
+            old_raw_pages = {p["url"]: p["text"] for p in json.load(f)}
+    except FileNotFoundError:
+        old_raw_pages = {}
+
+    # Mettre à jour uniquement les pages modifiées
+    for url, text in pages.items():
+        old_raw_pages[url] = text
+
+    # Nettoyer (facultatif)
+    cleaned_raw_pages = {url: text for url, text in old_raw_pages.items() if url in all_urls}
+
+    # Sauvegarder le contenu
+    os.makedirs("data", exist_ok=True)
+
+    with open(raw_pages_path, "w", encoding="utf-8") as f:
+        json.dump(
+            [{"url": url, "text": text} for url, text in cleaned_raw_pages.items()],
+            f, ensure_ascii=False, indent=2
+        )
+
+    with open(hashes_path, "w", encoding="utf-8") as f:
+        json.dump(new_hashes, f, ensure_ascii=False, indent=2)
+
+    with open(urls_path, "w", encoding="utf-8") as f:
+        json.dump(all_urls, f, ensure_ascii=False, indent=2)
+
+    print("📄 Pages modifiées sauvegardées")
+    print("📄 Contenu brut mis à jour enregistré dans 'raw_pages.json'")
+    print("🔗 URLs enregistrées dans 'data/urls.json'")
+    print("🔐 Hashes des pages enregistrés dans 'page_hashes.json'")
+
+    return [{"url": url, "text": text} for url, text in cleaned_raw_pages.items()]
+
+

+ 0 - 0
src/api/__init__.py


+ 490 - 0
src/app.py

@@ -0,0 +1,490 @@
+from dotenv import load_dotenv
+import os
+
+from flask import Flask, request, jsonify, render_template
+from apscheduler.schedulers.background import BackgroundScheduler
+from sentence_transformers import CrossEncoder, SentenceTransformer
+from langchain_ollama import OllamaLLM
+from .utils.processing import full_update_pipeline
+from .agent_ai.auth import login_to_wiki
+import json
+import numpy as np
+import textwrap
+from sklearn.metrics.pairwise import cosine_similarity
+from tabulate import tabulate
+import os
+from serpapi import GoogleSearch
+from werkzeug.utils import secure_filename
+import PyPDF2
+from PIL import Image
+from transformers import AutoTokenizer
+from .utils.ocr import run_ocr, generate_html , generate_word_level_html  
+import tiktoken
+from .models.VLM import ask_blip
+import getpass
+import faiss
+from langchain_huggingface import HuggingFaceEndpointEmbeddings
+import traceback
+from transformers import AutoTokenizer
+import getpass
+import requests
+from openai import OpenAI
+import traceback
+#from utils.security import advanced_security_check, log_suspicious_query, send_security_alert
+
+load_dotenv(override=True)
+
+
+app = Flask(__name__)
+
+
+def get_embedding_function():
+    return HuggingFaceEndpointEmbeddings(
+        model="sentence-transformers/all-MiniLM-L6-v2",  
+        task="feature-extraction"    )
+
+
+# Clé d'API Scaleway
+api_key = os.getenv("SCW_API_KEY")
+
+# Vérification de la clé
+if not api_key:
+    print("❌ La clé API est manquante.")
+else:
+    print("✅ Clé API chargée avec succès.")
+
+# Initialisation du client Scaleway
+client = OpenAI(
+    api_key=api_key,
+    base_url='https://api.scaleway.ai/v1'  # Base URL Scaleway
+)
+
+
+def call_scaleway_llama3(prompt, model="llama-3.3-70b-instruct", temperature=0.5, top_p=0.95, max_tokens=800):
+    try:
+        completion = client.chat.completions.create(
+            model=model,
+            messages=[{"role": "user", "content": prompt}],
+            temperature=temperature,
+            top_p=top_p,
+            max_tokens=max_tokens,
+            presence_penalty=0.3,
+            frequency_penalty=0.2
+        )
+        return completion.choices[0].message.content
+    except Exception as e:
+        print("🔴 Erreur lors de l'appel à l'API Scaleway :", e)
+        traceback.print_exc()  
+        return "Erreur : Impossible d'obtenir une réponse du modèle."
+
+
+'''
+def call_ollama_llama3(prompt, model="llama3", temperature=0.3, top_p=0.9, max_tokens=600):
+    try:
+        url = f"http://{ollama_host}/api/generate" if ollama_host else "http://localhost:11434/api/generate"
+        response = requests.post(
+            url,
+            json={
+                'model': model,
+                'prompt': prompt,
+                'temperature': temperature,
+                'top_p': top_p,
+                'max_tokens': max_tokens,
+                'stream': False
+            }
+        )
+        for line in response.iter_lines():
+            if line:
+                print(line.decode())
+        response.raise_for_status()
+        return response.json()['response']
+    except requests.exceptions.RequestException as e:
+        print("🔴 Error calling Ollama API:", e)
+        return "Erreur : Impossible d'obtenir une réponse du modèle."
+'''
+state = {
+    "index": None,
+    "all_chunks": [],
+    "chunk_urls": [],
+    "all_embeddings": None,
+    "embedding_model": get_embedding_function(),
+    "cross_encoder": CrossEncoder("cross-encoder/ms-marco-MiniLM-L-6-v2"),
+    "is_updating": False,
+    "llm": "llama-3.3-70b-instruct",
+    "chat_history": [],
+    "top_k": 20,
+    "rerank_k": 10
+}
+
+@app.route('/')
+def home():
+    return render_template('index.html')
+
+@app.route("/login", methods=["POST"])
+def login():
+    data = request.json
+    username = data.get("username")
+    password = data.get("password")
+
+    if not username or not password:
+        return jsonify({"error": "username and password required"}), 400
+
+    if not login_to_wiki(username, password):
+        return jsonify({"error": "Authentication failed"}), 401
+
+    return jsonify({"message": "Login successful"})
+
+@app.route("/chat_history", methods=["GET"])
+def chat_history():
+    return jsonify(state["chat_history"])
+
+def load_index_and_chunks():
+    base_path = os.path.dirname(os.path.abspath(__file__))
+    data_path = os.path.join(base_path, "data")
+    chunk_path = os.path.join(data_path, "All_chunks.json")
+    index_path = os.path.join(data_path, "faiss_index.index")
+
+    chunk_exists = os.path.exists(chunk_path)
+    index_exists = os.path.exists(index_path)
+
+    if chunk_exists:
+        print("✅ Fichier All_chunks.json trouvé.")
+    else:
+        print("❌ Fichier All_chunks.json manquant.")
+
+    if index_exists:
+        print("✅ Fichier faiss_index.index trouvé.")
+    else:
+        print("❌ Fichier faiss_index.index manquant.")
+
+    # Si les deux fichiers existent, charger directement
+    if chunk_exists and index_exists:
+        with open(chunk_path, "r", encoding="utf-8") as f:
+            state["all_chunks"] = json.load(f)
+        state["index"] = faiss.read_index(index_path)
+        print(f"✅ Chunks chargés : {len(state['all_chunks'])}")
+        return
+
+    # Sinon, lancer la mise à jour
+    print("⚠️ Index ou chunks manquants — lancement du scraping complet...")
+    run_full_update()  # Cette fonction doit générer les deux fichiers correctement
+
+    # Recharger après génération
+    if os.path.exists(chunk_path) and os.path.exists(index_path):
+        with open(chunk_path, "r", encoding="utf-8") as f:
+            state["all_chunks"] = json.load(f)
+        state["index"] = faiss.read_index(index_path)
+        print(f"✅ Index et chunks rechargés après mise à jour : {len(state['all_chunks'])} chunks.")
+    else:
+        print("❌ Les fichiers n'ont pas été générés correctement après mise à jour.")
+        state["index"] = None
+        state["all_chunks"] = []
+
+        
+def run_full_update():
+    print("🔄 Lancement du pipeline via le scheduler...")
+    full_update_pipeline()
+
+
+tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
+
+def chunk_text(text, chunk_size=400, overlap=100):
+    tokens = tokenizer.encode(text, add_special_tokens=False)  # Pas d’[CLS], [SEP]
+    chunks = []
+    step = chunk_size - overlap
+    for i in range(0, len(tokens), step):
+        chunk_tokens = tokens[i:i + chunk_size]
+        chunk_text = tokenizer.decode(chunk_tokens, skip_special_tokens=True)
+        chunks.append(chunk_text)
+        if i + chunk_size >= len(tokens):
+            break
+    return chunks
+
+
+
+UPLOAD_FOLDER = "uploads"
+os.makedirs(UPLOAD_FOLDER, exist_ok=True)  
+app.config["UPLOAD_FOLDER"] = UPLOAD_FOLDER
+
+ALLOWED_EXTENSIONS = {'txt', 'pdf', 'png', 'jpg', 'jpeg'}
+
+def allowed_file(filename):
+    return '.' in filename and filename.rsplit('.', 1)[1].lower() in ALLOWED_EXTENSIONS
+
+@app.route("/upload_file", methods=["POST"])
+def upload_file():
+    if 'file' not in request.files:
+        return jsonify({"error": "Aucun fichier détecté."}), 400
+
+    file = request.files['file']
+    if file.filename == '':
+        return jsonify({"error": "Nom de fichier vide."}), 400
+
+    if file and allowed_file(file.filename):
+        filename = secure_filename(file.filename)
+        filepath = os.path.join(app.config['UPLOAD_FOLDER'], filename)
+        file.save(filepath)
+
+        question = request.form.get("question", "")
+        if not question:
+            return jsonify({"error": "Veuillez poser une question liée au fichier."}), 400
+
+        try:
+            # ✅ 1. Exécuter l'OCR
+            document, ocr_output = run_ocr(filepath)
+
+            # ✅ 2. Sauvegarder le JSON OCR
+            json_path = os.path.join(app.config['UPLOAD_FOLDER'], "ocr_output.json")
+            with open(json_path, "w", encoding="utf-8") as fjson:
+                json.dump(ocr_output, fjson, ensure_ascii=False, indent=2)
+
+            # ✅ 3. Générer le HTML avec bounding boxes par mot
+            html_path = os.path.join(app.config['UPLOAD_FOLDER'], "ocr_output_words.html")
+            generate_word_level_html(ocr_output, output_html_path=html_path)
+
+            # ✅ 4. Extraire le texte brut pour le traitement RAG
+            text = ""
+            for page in ocr_output['pages']:
+                for block in page['blocks']:
+                    for line in block['lines']:
+                        for word in line['words']:
+                            text += word['value'] + " "
+                        text += "\n"
+                    text += "\n"
+                text += "\n"
+
+        except Exception as e:
+            return jsonify({"error": f"OCR échoué : {str(e)}"}), 500
+
+        if text.strip():
+            chunks = chunk_text(text)
+            embedding_model = state["embedding_model"]
+
+            # embeddings de tous les chunks (liste de vecteurs)
+            embeddings = embedding_model.embed_documents(chunks)
+
+            # embedding de la question
+            question_embedding = embedding_model.embed_query(question)
+
+            # calcul similarité cosine
+            similarities = cosine_similarity([question_embedding], embeddings)[0]
+
+            # récupérer top 5 indices les plus similaires
+            top_indices = np.argsort(similarities)[::-1][:5]
+            top_chunks = [chunks[i] for i in top_indices]
+
+            # contexte final concaténé
+            top_context = "\n\n".join(top_chunks)
+
+            prompt = (
+                "Réponds en français à la question suivante en utilisant uniquement le contenu fourni. "
+                "Si la réponse est absente, dis-le clairement.\n\n"
+                f"{top_context}\n\nQuestion : {question}"
+            )
+
+            answer = call_scaleway_llama3(prompt)
+
+            return jsonify({
+                "answer": answer,
+                "html_file": f"uploads/ocr_output_words.html"
+            })
+
+        else:
+            try:
+                answer = ask_blip(filepath, question)
+                return jsonify({"answer": answer})
+            except Exception as e:
+                return jsonify({"error": f"LLaVA échoué : {str(e)}"}), 500
+
+    return jsonify({"error": "Format de fichier non supporté."}), 400
+
+@app.route("/delete_file", methods=["POST"])
+def delete_file():
+    data = request.get_json()
+    filename = data.get("filename")
+
+    if not filename:
+        return jsonify({"error": "Nom de fichier manquant."}), 400
+
+    filepath = os.path.join(app.config["UPLOAD_FOLDER"], secure_filename(filename))
+
+    if os.path.exists(filepath):
+        os.remove(filepath)
+        return jsonify({"message": "Fichier supprimé avec succès."})
+    else:
+        return jsonify({"error": "Fichier introuvable."}), 404
+
+
+@app.route("/qa_chain", methods=["POST"])
+def qa_chain():
+    if not state.get("embedding_model") or not state.get("llm"):
+        return jsonify({"error": "Modèles non chargés."}), 500
+
+    data = request.get_json()
+    question = str(data.get("query", "")).strip()
+
+    if not question:
+        return jsonify({"error": "Question vide."}), 400
+
+    try:
+        chunk_texts = []
+        for i, chunk in enumerate(state["all_chunks"]):
+            if isinstance(chunk, dict):
+                text = chunk.get("text", "")
+                if text:
+                    chunk_texts.append(text)
+                else:
+                    return jsonify({"error": f"Chunk vide à l'index {i}."}), 500
+            elif isinstance(chunk, str):
+                chunk_texts.append(chunk)
+            else:
+                return jsonify({"error": f"Chunk invalide à l'index {i}."}), 500
+
+        embedding_model = state["embedding_model"]
+        question_embedding = embedding_model.embed_query(question)
+        chunk_embeddings = embedding_model.embed_documents(chunk_texts)
+
+        if question_embedding is None or not isinstance(chunk_embeddings, list) or len(chunk_embeddings) == 0:
+            return jsonify({"error": "Les embeddings sont vides ou mal formatés."}), 500
+
+        similarities = cosine_similarity([question_embedding], chunk_embeddings)[0]
+
+        try:
+            rerank_k_value = int(state.get("rerank_k", 20))
+        except (ValueError, TypeError):
+            rerank_k_value = 20
+
+        top_indices = similarities.argsort()[-rerank_k_value:][::-1]
+
+        # Création d'une liste de tuples (chunk, index, score)
+        reranked_chunks = [(chunk_texts[i], i, similarities[i]) for i in top_indices]
+
+        table = []
+        top_chunks = []
+        for rank, (chunk, idx, score) in enumerate(reranked_chunks):
+            url = state["chunk_urls"][idx] if idx < len(state["chunk_urls"]) else "URL inconnue"
+            wrapped_chunk = "\n".join(textwrap.wrap(chunk.strip(), width=80))[:500] + "..."
+            wrapped_url = "\n".join(textwrap.wrap(url, width=80))           
+            tokenizer = AutoTokenizer.from_pretrained("gpt2")  
+            num_tokens = len(tokenizer.encode(chunk))
+            table.append([rank, round(score, 4), num_tokens, wrapped_chunk, wrapped_url])
+            top_chunks.append(chunk)
+            top_context = "\n\n".join(top_chunks)
+
+        print("\n================== RÉSULTATS CLASSÉS PAR PERTINENCE ==================")
+        print(tabulate(table, headers=["Rank", "Score", "Tokens", "Chunk", "URL"], tablefmt="fancy_grid"))
+        print("=======================================================================\n")
+
+        
+        prompt = (
+        "Fournis une réponse factuelle, précise et compléte en français en te basant uniquement sur le contexte fourni. "
+        "Si tu n'es pas certain de la réponse ou si les données sont ambiguës, indique clairement que l'information n'est pas disponible ou nécessite une vérification. "
+        "Ne fais aucune supposition ou inférence qui pourrait être incorrecte, et précise les sources utilisées lorsque possible.\n\n"
+        f"{top_context}\n\n"
+        f"Question : {question}"
+    )
+
+        answer = call_scaleway_llama3(prompt)
+
+        print("\n====================== PROMPT UTILISÉ ======================")
+        print(f"CONTEXTE :\n{top_context}\nQUESTION :\n{question}\nRÉPONSE :\n{answer}")
+        print("============================================================\n")
+
+
+        # Detect no good answer from RAG
+        if not answer or "Information non disponible" in answer or len(answer.strip()) < 10:
+            raise ValueError("Pas de réponse fiable, fallback vers SerpAPI.")
+
+        state["chat_history"].append({"query": question, "answer": answer})
+        return jsonify({"answer": answer})
+
+    except Exception as rag_error:
+        print("Erreur RAG:", rag_error)
+        # Fallback to SerpAPI
+      
+        #  ÉTAPE 2 : Fallback vers recherche web (SerpAPI) 
+        try:
+            params = {
+                "engine": "google",
+                "q": question,
+                "api_key": os.getenv("SERPAPI_API_KEY"),
+            }
+            search = GoogleSearch(params)
+            results = search.get_dict()
+
+            answer = None
+            if "answer_box" in results and "answer" in results["answer_box"]:
+                answer = results["answer_box"]["answer"]
+            elif "organic_results" in results and results["organic_results"]:
+                answer = results["organic_results"][0].get("snippet")
+
+            if not answer:
+                answer = "Information non disponible."
+
+            # Summarize or filter with LLM if available
+            if state.get("llm"):
+                prompt = (
+                    "Réponds uniquement à la question de manière directe et concise.\n"
+                    "Sans reformuler la question, sans phrase d’introduction, sans politesse et sans contenu promotionnel. "
+                    "Utilise uniquement les faits du texte ci-dessous. "
+                    "Si l’information n’est pas présente, réponds : 'Information non disponible.'\n\n"
+                    f"Texte : {answer}\n\n"
+                    f"Question : {question}"
+                )
+                try:
+                    answer = call_scaleway_llama3(prompt).strip()
+                except Exception as llm_error:
+                    print("[Erreur LLM sur reformulation] :", llm_error)
+
+            state["chat_history"].append({
+                "question": question,
+                "answer": answer,
+                "sources": ["Google Search"]
+            })
+
+            return jsonify({
+                "answer": answer,
+                "sources": ["Google Search"]
+            })
+
+        except Exception as e:
+            return jsonify({"error": f"Erreur SerpAPI : {str(e)}"}), 500
+
+
+
+# Scheduler
+scheduler = BackgroundScheduler()
+scheduler.add_job(run_full_update, 'interval', hours=5)
+print("🔁 Scheduler activé pour mises à jour toutes les 5h")
+
+def prompt_for_login():
+    print("🔐 Please log in to the Wiki before starting the app.")
+    while True:
+        username = input("Username: ")
+        password = getpass.getpass("Password: ")
+        if login_to_wiki(username, password):
+            print("Login successful, starting the app...")
+            break
+        else:
+            print("Login failed, please try again.\n")
+
+basedir = os.path.abspath(os.path.dirname(__file__))
+load_dotenv(os.path.join(basedir, '.env'))
+def auto_login():
+    username = os.getenv("WIKI_USERNAME")
+    password = os.getenv("WIKI_PASSWORD")
+
+    if not username or not password:
+        raise ValueError("Missing WIKI_USERNAME or WIKI_PASSWORD in the .env file")
+
+    if login_to_wiki(username, password):
+        print("✅ Auto login successful.")
+    else:
+        raise Exception("❌ Auto login failed. Please check your credentials.")
+
+
+if __name__ == "__main__":
+    auto_login()
+    load_index_and_chunks()
+    scheduler.start()
+    app.run(debug=True)

File diff ditekan karena terlalu besar
+ 10 - 0
src/data/All_chunks.json


TEMPAT SAMPAH
src/data/faiss_index.index


+ 1 - 0
src/data/page_hashes.json

@@ -0,0 +1 @@
+{}

File diff ditekan karena terlalu besar
+ 3 - 0
src/data/raw_pages.json


+ 438 - 0
src/data/urls.json

@@ -0,0 +1,438 @@
+[
+  "https://wiki.intersel.fr/",
+  "https://wiki.intersel.fr/doku.php?id=dossiers_thematiques:https_et_ssl",
+  "https://wiki.intersel.fr/doku.php?id=dossiers_thematiques:poste_pc:reparation_pc",
+  "https://wiki.intersel.fr/doku.php?id=dossiers_thematiques:produit:e-marketing:dossier_luxe",
+  "https://wiki.intersel.fr/doku.php?id=dossiers_thematiques:reseaux_sociaux",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:accountable",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:acl",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:adresse_ip",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:adu",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:agent",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:ant",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:archiving",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:a:az",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b:back_office",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b:backup",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b:bert",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b:bien",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b:brm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:b:but",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cab",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cd",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:ceb",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:change",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:change_management",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:chapeaux_de_bono",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:charm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:ci",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:ci_cd",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:client_leger",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:client_lourd",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cloud",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cluster",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cmdb",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cmis",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cmmi",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cmp",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cms",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cobit",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:confidential_information",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:consulted",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:conteneur",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:conteneurisation",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:copywriter",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:copywriting",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:corrective_action",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:crm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cross-site_scripting_xss",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:cso",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:c:css",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:db",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:demand_management",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:developpeur_web",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:dfs",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:dkim",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:dmarc",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:dns",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:docker",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:document",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:d:downtime",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e:edi",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e:elevation_des_privileges",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e:emotion",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e:etl",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e:evolution_cms",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:e:extranet",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:f",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:f:fai",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:f:fbl",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:f:fqdn",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:f:front_office",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:f:frontend",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:g",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:g:gti",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:g:gtm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:g:gtr",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:h",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:h:hebergement",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:h:hostname",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:h:hr",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:h:html",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:h:hw",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:ihm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:information_system",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:information_system_security",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:informed",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:injection_sql",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:ip",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:is",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:isp",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:it",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:itil",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:i:itscm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:j",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:j:json",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:k",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:k:kpi",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:k:kubelet",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:k:kubernetes",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:l",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:l:intelligence_artificielle",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:l:lamp",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:l:lcid",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:l:lessor",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:m",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:m:machine_virtuelle",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:m:maintenance",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:m:map",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:m:modx",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:m:moteur_de_conteneurs",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:n",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:n:nat",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:n:nlp",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:n:no_code",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:n:node",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:n:nsd",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:o",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:o:objectif",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:o:okr",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:o:ola",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:o:one_page",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:o:orchestration_de_conteneurs",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:php",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:pl",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:plan",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:plan_de_taggage",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:pm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:pmo",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:pod",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:pra",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:preventive_action",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:procedure",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:process",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:produit",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:proposition_de_valeur",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:p:proxy",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:q",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:q:qms",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:q:quality_plan",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:raci",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:rca",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:rce",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:resources",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:responsible",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:rest",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:r:rfc",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:scd",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:scm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:sdp",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:secret_information",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:sem",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:serveur",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:service",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:service_transition",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:si",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:siem",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:sim",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:sip",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:skms",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:sla",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:slm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:slp",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:smart",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:soa",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:soar",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:spm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:ssl",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:storage",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:s:sw",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:start",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:t",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:t:tbd",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:t:tma",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:t:tmc",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:t:tms",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:t:top_secret_information",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:tags",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u:uat",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u:ueba",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u:uml",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u:uptime",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u:utm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:u:ux",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:v",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:v:vm",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:v:vpn",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:w",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:w:waf",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:w:web_services",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:x",
+  "https://wiki.intersel.fr/doku.php?id=glossaire:x:xss",
+  "https://wiki.intersel.fr/doku.php?id=intersel:aides:aide_commande_mysql:generation_historisation_sql",
+  "https://wiki.intersel.fr/doku.php?id=intersel:aides:aide_integration_fckeditor",
+  "https://wiki.intersel.fr/doku.php?id=intersel:aides:aide_roundcube",
+  "https://wiki.intersel.fr/doku.php?id=intersel:aides:sites_internet_utiles",
+  "https://wiki.intersel.fr/doku.php?id=intersel:aides:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:aides:tutorials:git:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:divers:exemples_courriers:firewall",
+  "https://wiki.intersel.fr/doku.php?id=intersel:divers:exemples_courriers:modifications_dns",
+  "https://wiki.intersel.fr/doku.php?id=intersel:divers:exemples_courriers:voeux",
+  "https://wiki.intersel.fr/doku.php?id=intersel:frontend_des_applications_de_suivi_et_de_gestion_des_projet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:entretiens:stagiaires",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:configurations_serveur_mail",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:cycle_de_developpement",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:finitions_hebergement",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:git:gogs:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:html5:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:internet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:l_espace_projet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:l_identification_documentaire_des_projets",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:la_conception_logicielle",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:la_conception_technique_d_une_newsletter",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:la_correction_d_une_anomalie",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:le_referencement_d_un_site_web",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:le_suivi_des_anomalies_sur_les_projets",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:le_versionnement_des_projets",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:les_finitions_d_un_site_web",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:les_logiciels",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:les_logiciels:phpmyadmin",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:complement_information",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:compterendu-reunion",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:conceptionlogicielle",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:configuration_emails",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:creation_email",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:creation_hebergement",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:demande_documents_startssl",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:document_standard",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:envoi_proposition",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:hebergement",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:homepage_projet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:pagemodele:specificationlogiciel",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:specifications_logicielles",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:guide_de_developpement_intersel:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:guides:liste_des_guides_d_aide",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:accessibilite:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:cdc:cahier_des_charges_v_aa",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:cdc:cdc_pour_la_realisation_d_une_application_logicielle",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:cgv_simplifiees",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:cgv_site_web",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:description_infogerance_standard",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:gestion_demandes_client:gestion_de_la_relation_client",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:la_gestion_des_demandes_client",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:accord_confidentialite",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:accord_confidentialite_fournisseur",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:accord_confidentialite_gb",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:bureaucracy_forms:annee_clients",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:bureaucracy_forms:projet_clients",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:bureaucracy_forms:prospects_project",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:google_comment",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:lettre_de_mise_en_demeure",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:lettre_de_relance_avec_ar",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:lettre_de_relance_simple",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:lettre_desistement_cheque",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:modele_footer_mail",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:modele_proposition",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:modele_proposition_english",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:modele_proposition_hebergement",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:phishing_attaques",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:reserves_proposition",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:modeles_des_documents:support",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:ordre_mission",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:procedures:gestion_developpements:modx:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:procedures:gestion_livraisons_projet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:procedures:gestion_sauvegarde_projet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:qualite:exigences_fonctionnelles",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:entete_fichier_fonction",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:javadoc",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_codage_css",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_codage_html",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_codage_javascript",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_codage_modx",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_codage_php",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_nomination_modx",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:regles_de_codage:regles_nommage_fichier",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:reunionsdelancement:identite_graphique",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:le_referentiel_qualite:workflow_livraison_git",
+  "https://wiki.intersel.fr/doku.php?id=intersel:marketing:modx:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:marketing:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:php",
+  "https://wiki.intersel.fr/doku.php?id=intersel:presentation",
+  "https://wiki.intersel.fr/doku.php?id=intersel:presentation_equipe",
+  "https://wiki.intersel.fr/doku.php?id=intersel:projets_internes:psi:sbl:aa:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:projets_internes:sas:conception_ab",
+  "https://wiki.intersel.fr/doku.php?id=intersel:projets_internes:site_intersel",
+  "https://wiki.intersel.fr/doku.php?id=intersel:projets_internes:slc_suivi_livrables_client:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:resume_annuaire",
+  "https://wiki.intersel.fr/doku.php?id=intersel:resume_clients",
+  "https://wiki.intersel.fr/doku.php?id=intersel:resume_viadeo",
+  "https://wiki.intersel.fr/doku.php?id=intersel:reunions_internes",
+  "https://wiki.intersel.fr/doku.php?id=intersel:rh:formulaires",
+  "https://wiki.intersel.fr/doku.php?id=intersel:rh:guide_pour_les_nouveaux_arrivants",
+  "https://wiki.intersel.fr/doku.php?id=intersel:securite:securite_site_internet",
+  "https://wiki.intersel.fr/doku.php?id=intersel:start",
+  "https://wiki.intersel.fr/doku.php?id=intersel:technique",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:animation_lettres:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:automation:preparation:lwstarter:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:automation:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:cold_calling:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:conception:import-de-ressources-csv",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:conception:import-export-de-ressources",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:conception_mx4:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:documentations:admin_manual:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:documentations:dev_manual:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:documentations:manuel-utilisateur-tres-simple:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:documentations:user_manual:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:documentations:user_manual:video:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:documentations:user_manual_expert:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:imx_demo:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:marketing:accroches:annuaire:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:marketing:accroches:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:nouveau_menu_panel",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:plan-type-contenu-imx",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projects:template_slider_avec_swiper1",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:amelioration_vitesse_generation_lw:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:blapy:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:commerce:specs",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:evol_babel_202501",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:imx_otp:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:integration_metronic:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:panier:specs",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:panier:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:solefy:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:supergen:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projets:webtimepilot:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:projetsimx",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:start",
+  "https://wiki.intersel.fr/doku.php?id=produit:developpement:livinweb:vente:cahier_des_charges:start",
+  "https://wiki.intersel.fr/doku.php?id=projets:samsic:anticipia:start",
+  "https://wiki.intersel.fr/doku.php?id=projets:samsic:start",
+  "https://wiki.intersel.fr/doku.php?id=securite:virus:virus_et_nettoyage_pc",
+  "https://wiki.intersel.fr/doku.php?id=start",
+  "https://wiki.intersel.fr/doku.php?id=technique:codes_source_utiles",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:codes_source_utiles:shell:scripts_migration_versions",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_limesurvey:manuel_utilisateur_limesurvey",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_limesurvey:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:affectation_header_content_type",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:apimodx",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:arborescence",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:auditor",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:displaymessagealert",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:ditto",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:ditto_probleme_cache",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:docmanager",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:documentpublisher",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:editarea",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:evolution_cms",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:fichiers_style_css",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:fileupload",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:formations:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:generation_de_fichiers_incorrecte",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:gestion_de_documents",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:insertion_editeur",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:integration_modx_dokuwiki",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:linkedin",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:loginlogout",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:bug_ie",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:droits_acces",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:preparemovieplayer.snippet",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:snippet:wayfinder",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:snippets_a_installer",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:templates_de_base",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:upgrade:modx_1_0_5_vers_1_0_12",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_evolution:upgrade:modx_1_x_vers_1_4.x",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_probleme_cache",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_revolution",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_revolution:architecture_modx_revo",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:modx_revolution:intersel:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:molecule",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:moteur_de_recherche",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:page_navigation",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:phx",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:picture_editor",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:probleme_appel_snippet",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:probleme_editeur_riche",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:probleme_eventlog",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:probleme_filtrage_modx",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:probleme_htaccess",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:probleme_utf8_ansi",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:quickedit",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:sitemap",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:snippet_probleme_cache",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:template_reference:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:tinymce",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_modx:weblogin",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_phplist:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_prestashop:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:developpement:developpement_wordpress:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:aides_hebergement",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:archinet:archinet-mutu1",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:archinet:archinet-mutu2",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:archinet:archinet-mutu3",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:archinet:archinet-mutu4",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:archinet:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:architecture_intersel_2014:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:backups:scripts:intersel_dump",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ikoula:mutu-iko-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ikoula:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:jobsc-ovh-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:mutu-ovh-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:mutu-ovh-2.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:nas-ovh-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:oc-ovh-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:rl-ovh-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:ovh:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:scaleway:mutu-scw-1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:scaleway:mutu-scw-2.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:scaleway:mutu-scw-3.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:scaleway:mutu-scw-o1.intersel.net",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:scaleway:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:hebergement:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:aide_commandes_linux",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:apt-get_ou_aptitude",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:architecture_ha_lb",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:fichiers_linux",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:fichiers_linux:htaccess",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:nagios",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:serveur_dedie_linux",
+  "https://wiki.intersel.fr/doku.php?id=technique:linux:start:aides_hebergement:copie_vm_mutu:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:matomo:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:mysql:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:postgresql:start",
+  "https://wiki.intersel.fr/doku.php?id=technique:start"
+]

+ 25 - 0
src/models/VLM.py

@@ -0,0 +1,25 @@
+
+from transformers import BlipProcessor, BlipForQuestionAnswering
+import torch
+from PIL import Image
+
+device = "cuda" if torch.cuda.is_available() else "cpu"
+
+# Chargement du modèle BLIP pré-entraîné (VQA)
+processor = BlipProcessor.from_pretrained("Salesforce/blip-vqa-base", use_fast=True)
+model = BlipForQuestionAnswering.from_pretrained("Salesforce/blip-vqa-base").to(device)
+model.eval()
+
+def ask_blip(image_path, question):
+    image = Image.open(image_path).convert("RGB")
+
+    # Préparer les inputs (image + question)
+    inputs = processor(image, question, return_tensors="pt").to(device)
+
+    # Générer la réponse
+    with torch.no_grad():
+        outputs = model.generate(**inputs)
+        answer = processor.tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+    return answer
+

+ 0 - 0
src/models/__init__.py


TEMPAT SAMPAH
src/models/__pycache__/VLM.cpython-312.pyc


TEMPAT SAMPAH
src/models/__pycache__/__init__.cpython-312.pyc


+ 540 - 0
src/templates/index.html

@@ -0,0 +1,540 @@
+<!DOCTYPE html>
+<html lang="fr">
+<head>
+  <meta charset="UTF-8" />
+  <meta name="viewport" content="width=device-width, initial-scale=1" />
+  <title>Chatbot Documentaire</title>
+  <link rel="stylesheet" href="https://fonts.googleapis.com/css2?family=Inter:wght@400;600&display=swap" />
+  <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css">
+
+  <style>
+    body {
+      margin: 0;
+      font-family: 'Inter', sans-serif;
+      background-color: #f4f6f8;
+      height: 100vh;
+      width: 100vw;
+      overflow: hidden;
+      display: flex;
+      flex-direction: column;
+    }
+    .chat-container {
+      flex: 1;
+      display: flex;
+      flex-direction: column;
+      height: 100vh;
+      width: 100vw;
+    }
+    .chat-header {
+      height: 64px;
+      background: #10a37f;
+      color: white;
+      padding: 1rem;
+      font-size: 1.25rem;
+      font-weight: bold;
+      text-align: center;
+      flex-shrink: 0;
+    }
+    .chat-body {
+      display: flex;
+      flex: 1;
+      height: calc(100vh - 64px);
+      overflow: hidden;
+    }
+    .chat-sidebar {
+      width: 20%;
+      background-color: #e9f5f2;
+      border-right: 1px solid #ccc;
+      padding: 1rem;
+      overflow-y: auto;
+    }
+    .chat-title {
+      font-weight: bold;
+      margin-bottom: 1rem;
+    }
+    .chat-main {
+      flex: 1;
+      display: flex;
+      flex-direction: column;
+      background-color: #f7f9fa;
+    }
+    .chat-history {
+      flex: 1;
+      padding: 1rem;
+      overflow-y: auto;
+      display: flex;
+      flex-direction: column;
+      gap: 1rem;
+    }
+    .chat-input {
+      padding: 1rem;
+      background: white;
+      border-top: 1px solid #ddd;
+      flex-shrink: 0;
+    }
+    form {
+      display: flex;
+      flex-direction: column;
+      gap: 1rem;
+    }
+    input[type="text"] {
+      padding: 0.75rem;
+      border-radius: 8px;
+      border: 1px solid #ccc;
+      font-size: 1rem;
+      width: 100%;
+      box-sizing: border-box;
+    }
+    button {
+      background-color: #10a37f;
+      color: white;
+      padding: 0.75rem;
+      border: none;
+      border-radius: 8px;
+      cursor: pointer;
+      font-weight: bold;
+      transition: background-color 0.3s;
+    }
+    button:hover {
+      background-color: #0d8f6a;
+    }
+    .upload-label {
+      display: inline-block;
+      background-color: #e0f7f1;
+      color: #10a37f;
+      padding: 0.5rem 1rem;
+      border-radius: 8px;
+      cursor: pointer;
+      font-weight: bold;
+      font-size: 0.95rem;
+      width: fit-content;
+    }
+    input[type="file"] {
+      display: none;
+    }
+    .file-preview {
+      background-color: #f0f0f0;
+      padding: 0.5rem;
+      border-radius: 6px;
+      font-size: 0.9rem;
+      color: #555;
+      max-width: 100%;
+      word-break: break-word;
+    }
+    .message {
+      display: flex;
+      max-width: 70%;
+      line-height: 1.4;
+      position: relative;
+      gap: 0.5rem;
+    }
+    .message.user {
+      align-self: flex-end;
+      flex-direction: row-reverse;
+    }
+    .message.bot {
+      align-self: flex-start;
+      flex-direction: row;
+    }
+    .message .avatar {
+      font-size: 1.5rem;
+      user-select: none;
+      flex-shrink: 0;
+      align-self: flex-end;
+      margin: 0 0.5rem;
+    }
+    .message .bubble {
+      padding: 0.8rem 1rem;
+      border-radius: 16px;
+      box-shadow: 0 1px 3px rgba(0,0,0,0.1);
+      white-space: pre-wrap;
+      word-wrap: break-word;
+      font-size: 1rem;
+      line-height: 1.4;
+      position: relative;
+    }
+    .message.user .bubble {
+      background-color: #d1e7dd;
+      color: #0f5132;
+      border-bottom-right-radius: 0;
+    }
+    .message.user .bubble::after {
+      content: "";
+      position: absolute;
+      bottom: 0;
+      right: -8px;
+      width: 0;
+      height: 0;
+      border-top: 12px solid #d1e7dd;
+      border-left: 8px solid transparent;
+    }
+    .message.bot .bubble {
+      background-color: #e2e3e5;
+      color: #41464b;
+      border-bottom-left-radius: 0;
+    }
+    .message.bot .bubble::after {
+      content: "";
+      position: absolute;
+      bottom: 0;
+      left: -8px;
+      width: 0;
+      height: 0;
+      border-top: 12px solid #e2e3e5;
+      border-right: 8px solid transparent;
+    }
+
+
+
+    body.dark {
+      background-color: #1e1e1e;
+      color: #f0f0f0;
+    }
+
+    body.dark .chat-header {
+      background: #333;
+    }
+
+    body.dark .chat-sidebar {
+      background-color: #2d2d2d;
+      border-color: #444;
+    }
+
+    body.dark .chat-main {
+      background-color: #1e1e1e;
+    }
+
+    body.dark .chat-history {
+      background-color: #1e1e1e;
+    }
+
+    body.dark .chat-input {
+      background-color: #2c2c2c;
+      border-color: #444;
+    }
+
+    body.dark input[type="text"] {
+      background-color: #3a3a3a;
+      color: #f0f0f0;
+      border-color: #555;
+    }
+
+    body.dark button {
+      background-color: #10a37f;
+      color: white;
+    }
+
+    body.dark .upload-label {
+      background-color: #37474f;
+      color: #10a37f;
+    }
+
+    body.dark .file-preview {
+      background-color: #333;
+      color: #ccc;
+    }
+
+    body.dark .message.user .bubble {
+      background-color: #264d3d;
+      color: #c8e6c9;
+    }
+
+    body.dark .message.user .bubble::after {
+      border-top-color: #264d3d;
+    }
+
+    body.dark .message.bot .bubble {
+      background-color: #444;
+      color: #eee;
+    }
+
+    body.dark .message.bot .bubble::after {
+      border-top-color: #444;
+    }
+
+/* Header layout */
+.header {
+  display: flex;
+  justify-content: space-between;
+  align-items: center;
+  padding: 12px 24px;
+  background-color: var(--header-bg, #f5f5f5);
+  border-bottom: 1px solid #ccc;
+  color: var(--text-color, #333);
+}
+
+.header h1 {
+  font-size: 1.5em;
+  margin: 0;
+}
+
+/* Right section layout */
+.header-right {
+  display: flex;
+  gap: 16px;
+}
+
+/* Shared button/icon style */
+.icon-button {
+  background: none;
+  border: none;
+  font-size: 1.4em;
+  color: var(--text-color, #333);
+  cursor: pointer;
+  transition: color 0.3s ease;
+  text-decoration: none;
+}
+
+.icon-button:hover {
+  color: #007bff;
+}
+
+/* Dark mode support */
+@media (prefers-color-scheme: dark) {
+  .header {
+    --header-bg: #1e1e1e;
+    --text-color: #f0f0f0;
+  }
+
+  .icon-button {
+    color: var(--text-color);
+  }
+
+  .icon-button:hover {
+    color: #66b0ff;
+  }
+}
+
+.header-buttons {
+  position: absolute;
+  right: 1rem;
+  top: 1rem;
+  display: flex;
+  align-items: center;
+  gap: 10px;
+   z-index: 1000;
+}
+
+.theme-toggle,
+.signup-btn {
+  background-color: rgba(255, 255, 255, 0.2);
+  color: white;
+  border: none;
+  border-radius: 6px;
+  width: 40px;
+  height: 40px;
+  font-size: 20px;
+  line-height: 1;
+  display: flex;
+  align-items: center;
+  justify-content: center;
+  cursor: pointer;
+}
+
+.theme-toggle:hover,
+.signup-btn:hover {
+  background-color: rgba(255, 255, 255, 0.35);
+}
+.chat-sidebar {
+  width: 200px;
+  background-color: #202020;
+  color: white;
+  padding: 1rem;
+}
+
+.chat-title {
+  font-weight: bold;
+  margin-bottom: 1rem;
+}
+
+.chat-item {
+  padding: 0.5rem;
+  cursor: pointer;
+  border-radius: 4px;
+}
+
+.chat-item:hover {
+  background-color: #333;
+}
+
+  </style>
+</head>
+
+<body>
+  <div class="chat-container">
+    <div class="chat-header">📎 Chatbot Documentaire</div>
+<div class="header-buttons">
+  <button class="theme-toggle" id="themeToggle" title="Changer de thème">🌙</button>
+  <button class="signup-btn" id="signupBtn" title="Créer un compte">📝</button>
+</div>
+
+    <div class="chat-body">
+
+      <div class="chat-sidebar" id="chatSidebar">
+        <div class="chat-title">💬 Vos Chats</div>
+<div id="chatList"></div>
+  <button onclick="addNewChat()">➕ Nouveau Chat</button>
+
+        <div class="chat-item">🗂️ Chat du 1 août</div>
+        <div class="chat-item">🗂️ Chat du 2 août</div>
+      </div>
+
+      <div class="chat-main">
+        <div class="chat-history" id="chatHistory"></div>
+
+        <div class="chat-input">
+          <form id="questionForm" enctype="multipart/form-data">
+            <label class="upload-label" for="file">📁 Ajouter un fichier (PDF, TXT, Image)</label>
+            <input type="file" id="file" name="file" accept=".pdf,.txt,image/*" />
+            <div class="file-preview" id="filePreview" style="display: none;"></div>
+
+            <input type="text" id="question" name="question" placeholder="Posez votre question…" autocomplete="off" required />
+            <button type="submit">Envoyer</button>
+          </form>
+        </div>
+      </div>
+    </div>
+
+    <div id="status" style="color: gray; margin-top: 0.5rem;"></div>
+  </div>
+
+  <script>
+    const chatHistoryDiv = document.getElementById("chatHistory");
+    const questionForm = document.getElementById("questionForm");
+    const questionInput = document.getElementById("question");
+    const status = document.getElementById("status");
+    const fileInput = document.getElementById("file");
+    const filePreview = document.getElementById("filePreview");
+
+    // Affiche nom fichier sélectionné
+    fileInput.addEventListener('change', () => {
+      if (fileInput.files.length > 0) {
+        filePreview.style.display = 'block';
+        filePreview.textContent = `Fichier sélectionné : ${fileInput.files[0].name}`;
+      } else {
+        filePreview.style.display = 'none';
+        filePreview.textContent = '';
+      }
+    });
+
+    // Fonction ajout message
+    function appendMessage(sender, text) {
+      const message = document.createElement("div");
+      message.className = `message ${sender}`;
+
+      const avatar = document.createElement("div");
+      avatar.className = "avatar";
+      avatar.textContent = sender === "user" ? "👤" : "🤖";
+
+      const bubble = document.createElement("div");
+      bubble.className = "bubble";
+      bubble.textContent = text;
+
+      message.appendChild(avatar);
+      message.appendChild(bubble);
+      chatHistoryDiv.appendChild(message);
+
+      chatHistoryDiv.scrollTop = chatHistoryDiv.scrollHeight;
+    }
+
+    // Charger historique si tu as cette route backend
+    async function loadChatHistory() {
+      try {
+        const response = await fetch("/chat_history");
+        if (response.ok) {
+          const chatHistory = await response.json();
+          chatHistoryDiv.innerHTML = "";
+          chatHistory.forEach(({ question, answer }) => {
+            appendMessage("user", question);
+            appendMessage("bot", answer);
+          });
+        }
+      } catch (error) {
+        console.warn("Erreur chargement historique :", error);
+      }
+    }
+    loadChatHistory();
+
+    // Gestion submit formulaire unique
+    questionForm.addEventListener("submit", async (e) => {
+      e.preventDefault();
+
+      const question = questionInput.value.trim();
+      if (!question) return;
+
+      appendMessage("user", question);
+      status.textContent = "En attente de la réponse...";
+
+      // Si fichier sélectionné, on envoie en multipart/form-data vers /upload_file
+      if (fileInput.files.length > 0) {
+        const formData = new FormData();
+        formData.append("file", fileInput.files[0]);
+        formData.append("question", question);
+
+        try {
+          const res = await fetch("/upload_file", {
+            method: "POST",
+            body: formData,
+          });
+          if (res.ok) {
+            const data = await res.json();
+            appendMessage("bot", data.answer || "Réponse reçue.");
+          } else {
+            appendMessage("bot", `Erreur serveur: ${res.status}`);
+          }
+        } catch (error) {
+          appendMessage("bot", "Erreur réseau ou serveur.");
+          console.error(error);
+        }
+      } else {
+        // Pas de fichier, on envoie JSON vers /qa_chain
+        try {
+          const res = await fetch("/qa_chain", {
+            method: "POST",
+            headers: { "Content-Type": "application/json" },
+            body: JSON.stringify({ query: question }),
+          });
+          if (res.ok) {
+            const data = await res.json();
+            appendMessage("bot", data.answer || "Réponse reçue.");
+          } else {
+            appendMessage("bot", `Erreur serveur: ${res.status}`);
+          }
+        } catch (error) {
+          appendMessage("bot", "Erreur réseau ou serveur.");
+          console.error(error);
+        }
+      }
+
+      status.textContent = "";
+      questionInput.value = "";
+      fileInput.value = "";
+      filePreview.style.display = 'none';
+    });
+
+        const themeToggle = document.getElementById("themeToggle");
+
+    function setTheme(mode) {
+      document.body.classList.toggle("dark", mode === "dark");
+      themeToggle.textContent = mode === "dark" ? "☀️" : "🌙";
+      localStorage.setItem("theme", mode);
+    }
+
+    themeToggle.addEventListener("click", () => {
+      const currentMode = document.body.classList.contains("dark") ? "dark" : "light";
+      const newMode = currentMode === "dark" ? "light" : "dark";
+      setTheme(newMode);
+    });
+
+    // Appliquer le thème sauvegardé
+    const savedTheme = localStorage.getItem("theme") || "light";
+    setTheme(savedTheme);
+
+  function addNewChat() {
+    const today = new Date().toLocaleDateString('fr-FR', { day: 'numeric', month: 'long' });
+    chatData.push({ label: `Chat du ${today}`, icon: "🗂️" });
+    renderChatList();
+  }
+  </script>
+</body>
+</html>

TEMPAT SAMPAH
src/uploads/cos.png


TEMPAT SAMPAH
src/uploads/cross.png


TEMPAT SAMPAH
src/uploads/methodologie.png


+ 24 - 0
src/uploads/ocr_output.html

@@ -0,0 +1,24 @@
+
+    <html><head><meta charset="utf-8"><title>Résultat OCR Organisé</title>
+    <style>
+      body { font-family: Arial, sans-serif; background: #f0f0f0; }
+      .page {
+        position: relative;
+        width: 800px; height: 1120px; /* ratio approx A4 */
+        margin: 20px auto;
+        background: white;
+        box-shadow: 0 0 8px rgba(0,0,0,0.2);
+      }
+      .block {
+        position: absolute;
+        border: 1px solid rgba(255,0,0,0.5);
+        background: rgba(255,255,255,0.7);
+        padding: 4px;
+        font-size: 14px;
+        line-height: 1.2em;
+        overflow-wrap: break-word;
+      }
+    </style>
+    </head><body>
+    <h1>Résultat OCR Organisé</h1>
+    <div class="page"><h2>Page 1</h2><div class="block" style="left:25.78125px; top:53.81754322766571px; width:749.21875px; height:981.0417417146973px;">Collecte des<br>Preparation des<br>Exploration des<br>données<br>données<br>données<br>(Collecter les<br>(nettoyage jeu de<br>(Analyse<br>donneés pertinentes<br>données, créer des<br>exploratoire et<br>comme la méteo,<br>variables<br>visualisation des<br>date du vol...)<br>supplementaires...)<br>donnees)<br>Evaluation des<br>Visualisation des<br>Modelisation<br>performances<br>resultats<br>(Prediction)<br>modele<br>(Presenter les<br>(Choisir un modele<br>(voir la performance<br>predictions dans<br>et l'entrainer sur des<br>du modele sur des<br>dashboard)<br>données histiriques)<br>données non vues)<br>Deploiement<br>(déploie le modèle<br>sur un serveur pour<br>qu'il fonctionne<br>automatiquement - )<br></div></div></body></html>

+ 2123 - 0
src/uploads/ocr_output.json

@@ -0,0 +1,2123 @@
+{
+  "pages": [
+    {
+      "page_idx": 0,
+      "dimensions": [
+        694,
+        795
+      ],
+      "orientation": {
+        "value": null,
+        "confidence": null
+      },
+      "language": {
+        "value": null,
+        "confidence": null
+      },
+      "blocks": [
+        {
+          "geometry": [
+            [
+              0.0322265625,
+              0.04805137788184438
+            ],
+            [
+              0.96875,
+              0.9239815044128242
+            ]
+          ],
+          "objectness_score": 0.816634056688864,
+          "lines": [
+            {
+              "geometry": [
+                [
+                  0.0810546875,
+                  0.04805137788184438
+                ],
+                [
+                  0.2568359375,
+                  0.07601849558717577
+                ]
+              ],
+              "objectness_score": 0.8275238573551178,
+              "words": [
+                {
+                  "value": "Collecte",
+                  "confidence": 0.9877678751945496,
+                  "geometry": [
+                    [
+                      0.0810546875,
+                      0.04917006259005763
+                    ],
+                    [
+                      0.1953125,
+                      0.07489981087896252
+                    ]
+                  ],
+                  "objectness_score": 0.805518388748169,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.9382218718528748,
+                  "geometry": [
+                    [
+                      0.205078125,
+                      0.04805137788184438
+                    ],
+                    [
+                      0.2568359375,
+                      0.07601849558717577
+                    ]
+                  ],
+                  "objectness_score": 0.8495293259620667,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.3896484375,
+                  0.04805137788184438
+                ],
+                [
+                  0.61328125,
+                  0.07937454971181557
+                ]
+              ],
+              "objectness_score": 0.8656781613826752,
+              "words": [
+                {
+                  "value": "Preparation",
+                  "confidence": 0.6832982301712036,
+                  "geometry": [
+                    [
+                      0.3896484375,
+                      0.050288747298270875
+                    ],
+                    [
+                      0.55078125,
+                      0.07937454971181557
+                    ]
+                  ],
+                  "objectness_score": 0.9028745293617249,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.985428512096405,
+                  "geometry": [
+                    [
+                      0.5625,
+                      0.04805137788184438
+                    ],
+                    [
+                      0.61328125,
+                      0.07713718029538907
+                    ]
+                  ],
+                  "objectness_score": 0.8284817934036255,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.72265625,
+                  0.04805137788184438
+                ],
+                [
+                  0.9443359375,
+                  0.07937454971181557
+                ]
+              ],
+              "objectness_score": 0.8440051674842834,
+              "words": [
+                {
+                  "value": "Exploration",
+                  "confidence": 0.9612176418304443,
+                  "geometry": [
+                    [
+                      0.72265625,
+                      0.050288747298270875
+                    ],
+                    [
+                      0.8828125,
+                      0.07937454971181557
+                    ]
+                  ],
+                  "objectness_score": 0.8654814958572388,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.98373943567276,
+                  "geometry": [
+                    [
+                      0.8935546875,
+                      0.04805137788184438
+                    ],
+                    [
+                      0.9443359375,
+                      0.07713718029538907
+                    ]
+                  ],
+                  "objectness_score": 0.8225288391113281,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.107421875,
+                  0.0883240273775216
+                ],
+                [
+                  0.228515625,
+                  0.1140537756664265
+                ]
+              ],
+              "objectness_score": 0.8256243467330933,
+              "words": [
+                {
+                  "value": "données",
+                  "confidence": 0.8609145283699036,
+                  "geometry": [
+                    [
+                      0.107421875,
+                      0.0883240273775216
+                    ],
+                    [
+                      0.228515625,
+                      0.1140537756664265
+                    ]
+                  ],
+                  "objectness_score": 0.8256243467330933,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.439453125,
+                  0.0883240273775216
+                ],
+                [
+                  0.560546875,
+                  0.1140537756664265
+                ]
+              ],
+              "objectness_score": 0.8341006636619568,
+              "words": [
+                {
+                  "value": "données",
+                  "confidence": 0.8609145283699036,
+                  "geometry": [
+                    [
+                      0.439453125,
+                      0.0883240273775216
+                    ],
+                    [
+                      0.560546875,
+                      0.1140537756664265
+                    ]
+                  ],
+                  "objectness_score": 0.8341006636619568,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.767578125,
+                  0.0883240273775216
+                ],
+                [
+                  0.8876953125,
+                  0.1140537756664265
+                ]
+              ],
+              "objectness_score": 0.824738621711731,
+              "words": [
+                {
+                  "value": "données",
+                  "confidence": 0.9612035751342773,
+                  "geometry": [
+                    [
+                      0.767578125,
+                      0.0883240273775216
+                    ],
+                    [
+                      0.8876953125,
+                      0.1140537756664265
+                    ]
+                  ],
+                  "objectness_score": 0.824738621711731,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.0771484375,
+                  0.12524062274855907
+                ],
+                [
+                  0.2607421875,
+                  0.15880116399495675
+                ]
+              ],
+              "objectness_score": 0.8089526891708374,
+              "words": [
+                {
+                  "value": "(Collecter",
+                  "confidence": 0.9168035984039307,
+                  "geometry": [
+                    [
+                      0.0771484375,
+                      0.12524062274855907
+                    ],
+                    [
+                      0.21484375,
+                      0.15880116399495675
+                    ]
+                  ],
+                  "objectness_score": 0.743909478187561,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "les",
+                  "confidence": 0.9869842529296875,
+                  "geometry": [
+                    [
+                      0.2216796875,
+                      0.12524062274855907
+                    ],
+                    [
+                      0.2607421875,
+                      0.155445109870317
+                    ]
+                  ],
+                  "objectness_score": 0.8739959001541138,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.3857421875,
+                  0.12524062274855907
+                ],
+                [
+                  0.6171875,
+                  0.1610385334113833
+                ]
+              ],
+              "objectness_score": 0.8517009615898132,
+              "words": [
+                {
+                  "value": "(nettoyage",
+                  "confidence": 0.7188050746917725,
+                  "geometry": [
+                    [
+                      0.3857421875,
+                      0.12635930745677232
+                    ],
+                    [
+                      0.5244140625,
+                      0.15991984870317005
+                    ]
+                  ],
+                  "objectness_score": 0.8291909694671631,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "jeu",
+                  "confidence": 0.9746283888816833,
+                  "geometry": [
+                    [
+                      0.5322265625,
+                      0.12524062274855907
+                    ],
+                    [
+                      0.5751953125,
+                      0.1610385334113833
+                    ]
+                  ],
+                  "objectness_score": 0.8322051167488098,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "de",
+                  "confidence": 0.9980244636535645,
+                  "geometry": [
+                    [
+                      0.58203125,
+                      0.12524062274855907
+                    ],
+                    [
+                      0.6171875,
+                      0.155445109870317
+                    ]
+                  ],
+                  "objectness_score": 0.8937067985534668,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.7763671875,
+                  0.12859667687319887
+                ],
+                [
+                  0.888671875,
+                  0.15880116399495675
+                ]
+              ],
+              "objectness_score": 0.8885737657546997,
+              "words": [
+                {
+                  "value": "(Analyse",
+                  "confidence": 0.9893469214439392,
+                  "geometry": [
+                    [
+                      0.7763671875,
+                      0.12859667687319887
+                    ],
+                    [
+                      0.888671875,
+                      0.15880116399495675
+                    ]
+                  ],
+                  "objectness_score": 0.8885737657546997,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.0322265625,
+                  0.16439458753602304
+                ],
+                [
+                  0.3017578125,
+                  0.19683644407420747
+                ]
+              ],
+              "objectness_score": 0.8225630521774292,
+              "words": [
+                {
+                  "value": "donneés",
+                  "confidence": 0.9632758498191833,
+                  "geometry": [
+                    [
+                      0.0322265625,
+                      0.16439458753602304
+                    ],
+                    [
+                      0.146484375,
+                      0.19348038994956773
+                    ]
+                  ],
+                  "objectness_score": 0.8078020215034485,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "pertinentes",
+                  "confidence": 0.6743541955947876,
+                  "geometry": [
+                    [
+                      0.158203125,
+                      0.1666319569524496
+                    ],
+                    [
+                      0.3017578125,
+                      0.19683644407420747
+                    ]
+                  ],
+                  "objectness_score": 0.8373240828514099,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.3720703125,
+                  0.1632759028278098
+                ],
+                [
+                  0.6298828125,
+                  0.19795512878242078
+                ]
+              ],
+              "objectness_score": 0.8254188895225525,
+              "words": [
+                {
+                  "value": "données,",
+                  "confidence": 0.9874939322471619,
+                  "geometry": [
+                    [
+                      0.3720703125,
+                      0.1632759028278098
+                    ],
+                    [
+                      0.4931640625,
+                      0.19795512878242078
+                    ]
+                  ],
+                  "objectness_score": 0.7603554725646973,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "créer",
+                  "confidence": 0.9589602947235107,
+                  "geometry": [
+                    [
+                      0.50390625,
+                      0.1655132722442363
+                    ],
+                    [
+                      0.572265625,
+                      0.19236170524135449
+                    ]
+                  ],
+                  "objectness_score": 0.8975253105163574,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.9953073859214783,
+                  "geometry": [
+                    [
+                      0.5810546875,
+                      0.16439458753602304
+                    ],
+                    [
+                      0.6298828125,
+                      0.19348038994956773
+                    ]
+                  ],
+                  "objectness_score": 0.8183758854866028,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.73828125,
+                  0.16439458753602304
+                ],
+                [
+                  0.9296875,
+                  0.19795512878242078
+                ]
+              ],
+              "objectness_score": 0.820769876241684,
+              "words": [
+                {
+                  "value": "exploratoire",
+                  "confidence": 0.9378995299339294,
+                  "geometry": [
+                    [
+                      0.73828125,
+                      0.16439458753602304
+                    ],
+                    [
+                      0.8935546875,
+                      0.19795512878242078
+                    ]
+                  ],
+                  "objectness_score": 0.7567607760429382,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "et",
+                  "confidence": 0.8651161193847656,
+                  "geometry": [
+                    [
+                      0.90234375,
+                      0.1666319569524496
+                    ],
+                    [
+                      0.9296875,
+                      0.19348038994956773
+                    ]
+                  ],
+                  "objectness_score": 0.8847789764404297,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.052734375,
+                  0.20131118290706052
+                ],
+                [
+                  0.2822265625,
+                  0.2326343547370317
+                ]
+              ],
+              "objectness_score": 0.8012694120407104,
+              "words": [
+                {
+                  "value": "comme",
+                  "confidence": 0.9968090653419495,
+                  "geometry": [
+                    [
+                      0.052734375,
+                      0.2069046064481268
+                    ],
+                    [
+                      0.1513671875,
+                      0.23151567002881845
+                    ]
+                  ],
+                  "objectness_score": 0.7506269216537476,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "la",
+                  "confidence": 0.9957412481307983,
+                  "geometry": [
+                    [
+                      0.1591796875,
+                      0.20131118290706052
+                    ],
+                    [
+                      0.1845703125,
+                      0.23151567002881845
+                    ]
+                  ],
+                  "objectness_score": 0.8192916512489319,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "méteo,",
+                  "confidence": 0.9663773775100708,
+                  "geometry": [
+                    [
+                      0.1923828125,
+                      0.20242986761527376
+                    ],
+                    [
+                      0.2822265625,
+                      0.2326343547370317
+                    ]
+                  ],
+                  "objectness_score": 0.8338896632194519,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.4404296875,
+                  0.203548552323487
+                ],
+                [
+                  0.55859375,
+                  0.2292783006123919
+                ]
+              ],
+              "objectness_score": 0.8311287760734558,
+              "words": [
+                {
+                  "value": "variables",
+                  "confidence": 0.5077484250068665,
+                  "geometry": [
+                    [
+                      0.4404296875,
+                      0.203548552323487
+                    ],
+                    [
+                      0.55859375,
+                      0.2292783006123919
+                    ]
+                  ],
+                  "objectness_score": 0.8311287760734558,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.7216796875,
+                  0.20242986761527376
+                ],
+                [
+                  0.9423828125,
+                  0.2303969853206052
+                ]
+              ],
+              "objectness_score": 0.8476908206939697,
+              "words": [
+                {
+                  "value": "visualisation",
+                  "confidence": 0.9528688788414001,
+                  "geometry": [
+                    [
+                      0.7216796875,
+                      0.20242986761527376
+                    ],
+                    [
+                      0.8837890625,
+                      0.2303969853206052
+                    ]
+                  ],
+                  "objectness_score": 0.7872260808944702,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.9943244457244873,
+                  "geometry": [
+                    [
+                      0.89453125,
+                      0.20242986761527376
+                    ],
+                    [
+                      0.9423828125,
+                      0.2303969853206052
+                    ]
+                  ],
+                  "objectness_score": 0.9081555604934692,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.0771484375,
+                  0.24046514769452448
+                ],
+                [
+                  0.2587890625,
+                  0.27402568894092216
+                ]
+              ],
+              "objectness_score": 0.8521583477656046,
+              "words": [
+                {
+                  "value": "date",
+                  "confidence": 0.9927149415016174,
+                  "geometry": [
+                    [
+                      0.0771484375,
+                      0.24158383240273773
+                    ],
+                    [
+                      0.134765625,
+                      0.2695509501080692
+                    ]
+                  ],
+                  "objectness_score": 0.8823536038398743,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "du",
+                  "confidence": 0.993779718875885,
+                  "geometry": [
+                    [
+                      0.142578125,
+                      0.24046514769452448
+                    ],
+                    [
+                      0.177734375,
+                      0.2706696348162824
+                    ]
+                  ],
+                  "objectness_score": 0.8689723014831543,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "vol...)",
+                  "confidence": 0.817709743976593,
+                  "geometry": [
+                    [
+                      0.185546875,
+                      0.24158383240273773
+                    ],
+                    [
+                      0.2587890625,
+                      0.27402568894092216
+                    ]
+                  ],
+                  "objectness_score": 0.8051491379737854,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.373046875,
+                  0.24270251711095103
+                ],
+                [
+                  0.625,
+                  0.27514437364913547
+                ]
+              ],
+              "objectness_score": 0.7972156405448914,
+              "words": [
+                {
+                  "value": "supplementaires...)",
+                  "confidence": 0.7050409913063049,
+                  "geometry": [
+                    [
+                      0.373046875,
+                      0.24270251711095103
+                    ],
+                    [
+                      0.625,
+                      0.27514437364913547
+                    ]
+                  ],
+                  "objectness_score": 0.7972156405448914,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.771484375,
+                  0.24270251711095103
+                ],
+                [
+                  0.892578125,
+                  0.27290700423270897
+                ]
+              ],
+              "objectness_score": 0.8575900197029114,
+              "words": [
+                {
+                  "value": "donnees)",
+                  "confidence": 0.7150509357452393,
+                  "geometry": [
+                    [
+                      0.771484375,
+                      0.24270251711095103
+                    ],
+                    [
+                      0.892578125,
+                      0.27290700423270897
+                    ]
+                  ],
+                  "objectness_score": 0.8575900197029114,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.396484375,
+                  0.3590457267651297
+                ],
+                [
+                  0.60546875,
+                  0.38813152917867433
+                ]
+              ],
+              "objectness_score": 0.838508129119873,
+              "words": [
+                {
+                  "value": "Evaluation",
+                  "confidence": 0.9654068946838379,
+                  "geometry": [
+                    [
+                      0.396484375,
+                      0.3612830961815562
+                    ],
+                    [
+                      0.5439453125,
+                      0.38589415976224783
+                    ]
+                  ],
+                  "objectness_score": 0.8400536775588989,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.9658872485160828,
+                  "geometry": [
+                    [
+                      0.5546875,
+                      0.3590457267651297
+                    ],
+                    [
+                      0.60546875,
+                      0.38813152917867433
+                    ]
+                  ],
+                  "objectness_score": 0.8369625806808472,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.0458984375,
+                  0.37806336680475505
+                ],
+                [
+                  0.291015625,
+                  0.40603048451008644
+                ]
+              ],
+              "objectness_score": 0.8438914716243744,
+              "words": [
+                {
+                  "value": "Visualisation",
+                  "confidence": 0.8689184188842773,
+                  "geometry": [
+                    [
+                      0.0458984375,
+                      0.3791820515129683
+                    ],
+                    [
+                      0.228515625,
+                      0.40379311509365995
+                    ]
+                  ],
+                  "objectness_score": 0.8300719857215881,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.998823344707489,
+                  "geometry": [
+                    [
+                      0.23828125,
+                      0.37806336680475505
+                    ],
+                    [
+                      0.291015625,
+                      0.40603048451008644
+                    ]
+                  ],
+                  "objectness_score": 0.8577109575271606,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.7421875,
+                  0.3791820515129683
+                ],
+                [
+                  0.9208984375,
+                  0.40379311509365995
+                ]
+              ],
+              "objectness_score": 0.8251617550849915,
+              "words": [
+                {
+                  "value": "Modelisation",
+                  "confidence": 0.962478756904602,
+                  "geometry": [
+                    [
+                      0.7421875,
+                      0.3791820515129683
+                    ],
+                    [
+                      0.9208984375,
+                      0.40379311509365995
+                    ]
+                  ],
+                  "objectness_score": 0.8251617550849915,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.40234375,
+                  0.39819969155259366
+                ],
+                [
+                  0.5986328125,
+                  0.43176023279899134
+                ]
+              ],
+              "objectness_score": 0.6945668458938599,
+              "words": [
+                {
+                  "value": "performances",
+                  "confidence": 0.9785539507865906,
+                  "geometry": [
+                    [
+                      0.40234375,
+                      0.39819969155259366
+                    ],
+                    [
+                      0.5986328125,
+                      0.43176023279899134
+                    ]
+                  ],
+                  "objectness_score": 0.6945668458938599,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.107421875,
+                  0.41833601630043227
+                ],
+                [
+                  0.2294921875,
+                  0.44406576458933716
+                ]
+              ],
+              "objectness_score": 0.7634888291358948,
+              "words": [
+                {
+                  "value": "resultats",
+                  "confidence": 0.9981811046600342,
+                  "geometry": [
+                    [
+                      0.107421875,
+                      0.41833601630043227
+                    ],
+                    [
+                      0.2294921875,
+                      0.44406576458933716
+                    ]
+                  ],
+                  "objectness_score": 0.7634888291358948,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.75,
+                  0.417217331592219
+                ],
+                [
+                  0.9130859375,
+                  0.4507778728386167
+                ]
+              ],
+              "objectness_score": 0.7848072648048401,
+              "words": [
+                {
+                  "value": "(Prediction)",
+                  "confidence": 0.6085659265518188,
+                  "geometry": [
+                    [
+                      0.75,
+                      0.417217331592219
+                    ],
+                    [
+                      0.9130859375,
+                      0.4507778728386167
+                    ]
+                  ],
+                  "objectness_score": 0.7848072648048401,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.44921875,
+                  0.4384723410482709
+                ],
+                [
+                  0.5517578125,
+                  0.46420208933717577
+                ]
+              ],
+              "objectness_score": 0.8305306434631348,
+              "words": [
+                {
+                  "value": "modele",
+                  "confidence": 0.9977617263793945,
+                  "geometry": [
+                    [
+                      0.44921875,
+                      0.4384723410482709
+                    ],
+                    [
+                      0.5517578125,
+                      0.46420208933717577
+                    ]
+                  ],
+                  "objectness_score": 0.8305306434631348,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.0771484375,
+                  0.45525261167146974
+                ],
+                [
+                  0.2626953125,
+                  0.48657578350144093
+                ]
+              ],
+              "objectness_score": 0.8596967160701752,
+              "words": [
+                {
+                  "value": "(Presenter",
+                  "confidence": 0.6310722231864929,
+                  "geometry": [
+                    [
+                      0.0771484375,
+                      0.456371296379683
+                    ],
+                    [
+                      0.2138671875,
+                      0.48657578350144093
+                    ]
+                  ],
+                  "objectness_score": 0.8357106447219849,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "les",
+                  "confidence": 0.9968156218528748,
+                  "geometry": [
+                    [
+                      0.22265625,
+                      0.45525261167146974
+                    ],
+                    [
+                      0.2626953125,
+                      0.48433841408501443
+                    ]
+                  ],
+                  "objectness_score": 0.8836827874183655,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.70703125,
+                  0.4541339269632565
+                ],
+                [
+                  0.9580078125,
+                  0.4888131529178674
+                ]
+              ],
+              "objectness_score": 0.8198439876238505,
+              "words": [
+                {
+                  "value": "(Choisir",
+                  "confidence": 0.9063538908958435,
+                  "geometry": [
+                    [
+                      0.70703125,
+                      0.4541339269632565
+                    ],
+                    [
+                      0.8095703125,
+                      0.4888131529178674
+                    ]
+                  ],
+                  "objectness_score": 0.8057355284690857,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "un",
+                  "confidence": 0.9998005032539368,
+                  "geometry": [
+                    [
+                      0.8173828125,
+                      0.46084603521253603
+                    ],
+                    [
+                      0.8525390625,
+                      0.4854570987932277
+                    ]
+                  ],
+                  "objectness_score": 0.7646810412406921,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "modele",
+                  "confidence": 0.9942703247070312,
+                  "geometry": [
+                    [
+                      0.861328125,
+                      0.45748998108789624
+                    ],
+                    [
+                      0.9580078125,
+                      0.4821010446685879
+                    ]
+                  ],
+                  "objectness_score": 0.8891153931617737,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.3662109375,
+                  0.4742702517110951
+                ],
+                [
+                  0.6318359375,
+                  0.5111868470821326
+                ]
+              ],
+              "objectness_score": 0.8241175611813863,
+              "words": [
+                {
+                  "value": "(voir",
+                  "confidence": 0.9973810315132141,
+                  "geometry": [
+                    [
+                      0.3662109375,
+                      0.4742702517110951
+                    ],
+                    [
+                      0.4267578125,
+                      0.5111868470821326
+                    ]
+                  ],
+                  "objectness_score": 0.8463975191116333,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "la",
+                  "confidence": 0.9981163740158081,
+                  "geometry": [
+                    [
+                      0.43359375,
+                      0.4742702517110951
+                    ],
+                    [
+                      0.4580078125,
+                      0.5055934235410663
+                    ]
+                  ],
+                  "objectness_score": 0.8432861566543579,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "performance",
+                  "confidence": 0.9783651232719421,
+                  "geometry": [
+                    [
+                      0.466796875,
+                      0.47762630583573484
+                    ],
+                    [
+                      0.6318359375,
+                      0.5078307929574928
+                    ]
+                  ],
+                  "objectness_score": 0.7826690077781677,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.060546875,
+                  0.49552526116714696
+                ],
+                [
+                  0.27734375,
+                  0.525729748288905
+                ]
+              ],
+              "objectness_score": 0.8490830659866333,
+              "words": [
+                {
+                  "value": "predictions",
+                  "confidence": 0.5426515936851501,
+                  "geometry": [
+                    [
+                      0.060546875,
+                      0.49552526116714696
+                    ],
+                    [
+                      0.2021484375,
+                      0.525729748288905
+                    ]
+                  ],
+                  "objectness_score": 0.8191495537757874,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "dans",
+                  "confidence": 0.9975013136863708,
+                  "geometry": [
+                    [
+                      0.2138671875,
+                      0.49552526116714696
+                    ],
+                    [
+                      0.27734375,
+                      0.5223736941642652
+                    ]
+                  ],
+                  "objectness_score": 0.8790165781974792,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.693359375,
+                  0.4944065764589337
+                ],
+                [
+                  0.96875,
+                  0.525729748288905
+                ]
+              ],
+              "objectness_score": 0.8130820393562317,
+              "words": [
+                {
+                  "value": "et",
+                  "confidence": 0.8902291655540466,
+                  "geometry": [
+                    [
+                      0.693359375,
+                      0.4944065764589337
+                    ],
+                    [
+                      0.724609375,
+                      0.525729748288905
+                    ]
+                  ],
+                  "objectness_score": 0.7699534296989441,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "l'entrainer",
+                  "confidence": 0.91228187084198,
+                  "geometry": [
+                    [
+                      0.7314453125,
+                      0.49552526116714696
+                    ],
+                    [
+                      0.861328125,
+                      0.5212550094560519
+                    ]
+                  ],
+                  "objectness_score": 0.8133661150932312,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "sur",
+                  "confidence": 0.9974398612976074,
+                  "geometry": [
+                    [
+                      0.8701171875,
+                      0.5
+                    ],
+                    [
+                      0.9130859375,
+                      0.5223736941642652
+                    ]
+                  ],
+                  "objectness_score": 0.8441946506500244,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.9702199697494507,
+                  "geometry": [
+                    [
+                      0.919921875,
+                      0.4944065764589337
+                    ],
+                    [
+                      0.96875,
+                      0.5234923788724783
+                    ]
+                  ],
+                  "objectness_score": 0.824813961982727,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.375,
+                  0.5123055317903458
+                ],
+                [
+                  0.625,
+                  0.543628703620317
+                ]
+              ],
+              "objectness_score": 0.8126386255025864,
+              "words": [
+                {
+                  "value": "du",
+                  "confidence": 0.9909602403640747,
+                  "geometry": [
+                    [
+                      0.375,
+                      0.5123055317903458
+                    ],
+                    [
+                      0.41015625,
+                      0.543628703620317
+                    ]
+                  ],
+                  "objectness_score": 0.8536072373390198,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "modele",
+                  "confidence": 0.8897013664245605,
+                  "geometry": [
+                    [
+                      0.4189453125,
+                      0.5134242164985591
+                    ],
+                    [
+                      0.5166015625,
+                      0.5425100189121037
+                    ]
+                  ],
+                  "objectness_score": 0.8270806074142456,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "sur",
+                  "confidence": 0.9979320764541626,
+                  "geometry": [
+                    [
+                      0.5244140625,
+                      0.5190176400396254
+                    ],
+                    [
+                      0.5693359375,
+                      0.543628703620317
+                    ]
+                  ],
+                  "objectness_score": 0.7769684195518494,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "des",
+                  "confidence": 0.8242294788360596,
+                  "geometry": [
+                    [
+                      0.576171875,
+                      0.5134242164985591
+                    ],
+                    [
+                      0.625,
+                      0.5425100189121037
+                    ]
+                  ],
+                  "objectness_score": 0.7928982377052307,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.0927734375,
+                  0.5346792259546109
+                ],
+                [
+                  0.2412109375,
+                  0.5648837130763689
+                ]
+              ],
+              "objectness_score": 0.8164286017417908,
+              "words": [
+                {
+                  "value": "dashboard)",
+                  "confidence": 0.9945673942565918,
+                  "geometry": [
+                    [
+                      0.0927734375,
+                      0.5346792259546109
+                    ],
+                    [
+                      0.2412109375,
+                      0.5648837130763689
+                    ]
+                  ],
+                  "objectness_score": 0.8164286017417908,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.697265625,
+                  0.5335605412463977
+                ],
+                [
+                  0.96484375,
+                  0.5671210824927954
+                ]
+              ],
+              "objectness_score": 0.7987184822559357,
+              "words": [
+                {
+                  "value": "données",
+                  "confidence": 0.9885821342468262,
+                  "geometry": [
+                    [
+                      0.697265625,
+                      0.5335605412463977
+                    ],
+                    [
+                      0.8125,
+                      0.5626463436599424
+                    ]
+                  ],
+                  "objectness_score": 0.8079077005386353,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "histiriques)",
+                  "confidence": 0.950248658657074,
+                  "geometry": [
+                    [
+                      0.822265625,
+                      0.5335605412463977
+                    ],
+                    [
+                      0.96484375,
+                      0.5671210824927954
+                    ]
+                  ],
+                  "objectness_score": 0.7895292639732361,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.373046875,
+                  0.5536968659942363
+                ],
+                [
+                  0.626953125,
+                  0.5850200378242075
+                ]
+              ],
+              "objectness_score": 0.8351868987083435,
+              "words": [
+                {
+                  "value": "données",
+                  "confidence": 0.8836828470230103,
+                  "geometry": [
+                    [
+                      0.373046875,
+                      0.5536968659942363
+                    ],
+                    [
+                      0.4873046875,
+                      0.5794266142831412
+                    ]
+                  ],
+                  "objectness_score": 0.8597532510757446,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "non",
+                  "confidence": 0.9986007809638977,
+                  "geometry": [
+                    [
+                      0.4970703125,
+                      0.5581716048270894
+                    ],
+                    [
+                      0.5458984375,
+                      0.5805452989913544
+                    ]
+                  ],
+                  "objectness_score": 0.8361157774925232,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "vues)",
+                  "confidence": 0.9901139736175537,
+                  "geometry": [
+                    [
+                      0.5537109375,
+                      0.5536968659942363
+                    ],
+                    [
+                      0.626953125,
+                      0.5850200378242075
+                    ]
+                  ],
+                  "objectness_score": 0.8096916675567627,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.4130859375,
+                  0.713668779268732
+                ],
+                [
+                  0.5888671875,
+                  0.7427545816822767
+                ]
+              ],
+              "objectness_score": 0.850524365901947,
+              "words": [
+                {
+                  "value": "Deploiement",
+                  "confidence": 0.6806442737579346,
+                  "geometry": [
+                    [
+                      0.4130859375,
+                      0.713668779268732
+                    ],
+                    [
+                      0.5888671875,
+                      0.7427545816822767
+                    ]
+                  ],
+                  "objectness_score": 0.850524365901947,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.3740234375,
+                  0.7729590688040346
+                ],
+                [
+                  0.6259765625,
+                  0.8076382947586456
+                ]
+              ],
+              "objectness_score": 0.8079845110575358,
+              "words": [
+                {
+                  "value": "(déploie",
+                  "confidence": 0.9897415637969971,
+                  "geometry": [
+                    [
+                      0.3740234375,
+                      0.7729590688040346
+                    ],
+                    [
+                      0.4873046875,
+                      0.8076382947586456
+                    ]
+                  ],
+                  "objectness_score": 0.7893381118774414,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "le",
+                  "confidence": 0.9997684359550476,
+                  "geometry": [
+                    [
+                      0.49609375,
+                      0.7729590688040346
+                    ],
+                    [
+                      0.521484375,
+                      0.8031635559257926
+                    ]
+                  ],
+                  "objectness_score": 0.7943187355995178,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "modèle",
+                  "confidence": 0.9219999313354492,
+                  "geometry": [
+                    [
+                      0.5302734375,
+                      0.775196438220461
+                    ],
+                    [
+                      0.6259765625,
+                      0.800926186509366
+                    ]
+                  ],
+                  "objectness_score": 0.8402966856956482,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.369140625,
+                  0.8177064571325648
+                ],
+                [
+                  0.6318359375,
+                  0.8467922595461095
+                ]
+              ],
+              "objectness_score": 0.8303013145923615,
+              "words": [
+                {
+                  "value": "sur",
+                  "confidence": 0.9938849806785583,
+                  "geometry": [
+                    [
+                      0.369140625,
+                      0.8188251418407781
+                    ],
+                    [
+                      0.412109375,
+                      0.8411988360050433
+                    ]
+                  ],
+                  "objectness_score": 0.8680939674377441,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "un",
+                  "confidence": 0.9983118772506714,
+                  "geometry": [
+                    [
+                      0.4189453125,
+                      0.8188251418407781
+                    ],
+                    [
+                      0.4541015625,
+                      0.8411988360050433
+                    ]
+                  ],
+                  "objectness_score": 0.8339855670928955,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "serveur",
+                  "confidence": 0.9163511395454407,
+                  "geometry": [
+                    [
+                      0.4609375,
+                      0.8177064571325648
+                    ],
+                    [
+                      0.5625,
+                      0.8411988360050433
+                    ]
+                  ],
+                  "objectness_score": 0.7940940856933594,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "pour",
+                  "confidence": 0.9337925910949707,
+                  "geometry": [
+                    [
+                      0.5712890625,
+                      0.8188251418407781
+                    ],
+                    [
+                      0.6318359375,
+                      0.8467922595461095
+                    ]
+                  ],
+                  "objectness_score": 0.8250316381454468,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.3994140625,
+                  0.8512669983789625
+                ],
+                [
+                  0.599609375,
+                  0.8870649090417868
+                ]
+              ],
+              "objectness_score": 0.7906262278556824,
+              "words": [
+                {
+                  "value": "qu'il",
+                  "confidence": 0.5704904198646545,
+                  "geometry": [
+                    [
+                      0.3994140625,
+                      0.8512669983789625
+                    ],
+                    [
+                      0.4521484375,
+                      0.8870649090417868
+                    ]
+                  ],
+                  "objectness_score": 0.7620018124580383,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "fonctionne",
+                  "confidence": 0.9132682085037231,
+                  "geometry": [
+                    [
+                      0.462890625,
+                      0.8523856830871758
+                    ],
+                    [
+                      0.599609375,
+                      0.8781154313760806
+                    ]
+                  ],
+                  "objectness_score": 0.8192506432533264,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            },
+            {
+              "geometry": [
+                [
+                  0.375,
+                  0.8904209631664265
+                ],
+                [
+                  0.625,
+                  0.9239815044128242
+                ]
+              ],
+              "objectness_score": 0.5923550923665365,
+              "words": [
+                {
+                  "value": "automatiquement",
+                  "confidence": 0.7246941924095154,
+                  "geometry": [
+                    [
+                      0.375,
+                      0.8904209631664265
+                    ],
+                    [
+                      0.611328125,
+                      0.9239815044128242
+                    ]
+                  ],
+                  "objectness_score": 0.7002537250518799,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": "-",
+                  "confidence": 0.4151031970977783,
+                  "geometry": [
+                    [
+                      0.6083984375,
+                      0.9027264949567724
+                    ],
+                    [
+                      0.6162109375,
+                      0.913913342038905
+                    ]
+                  ],
+                  "objectness_score": 0.5287280082702637,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                },
+                {
+                  "value": ")",
+                  "confidence": 0.4633558988571167,
+                  "geometry": [
+                    [
+                      0.615234375,
+                      0.892658332582853
+                    ],
+                    [
+                      0.625,
+                      0.9072012337896254
+                    ]
+                  ],
+                  "objectness_score": 0.5480835437774658,
+                  "crop_orientation": {
+                    "value": 0,
+                    "confidence": null
+                  }
+                }
+              ]
+            }
+          ],
+          "artefacts": []
+        }
+      ]
+    }
+  ]
+}

File diff ditekan karena terlalu besar
+ 25 - 0
src/uploads/ocr_output_words.html


TEMPAT SAMPAH
src/uploads/processus_.png


TEMPAT SAMPAH
src/uploads/reg_chunk.png


+ 0 - 0
src/utils/__init__.py


TEMPAT SAMPAH
src/utils/__pycache__/__init__.cpython-312.pyc


TEMPAT SAMPAH
src/utils/__pycache__/ocr.cpython-312.pyc


TEMPAT SAMPAH
src/utils/__pycache__/processing.cpython-312.pyc


TEMPAT SAMPAH
src/utils/__pycache__/security.cpython-312.pyc


+ 162 - 0
src/utils/ocr.py

@@ -0,0 +1,162 @@
+import os
+import json
+from doctr.io import DocumentFile
+from doctr.models import ocr_predictor
+
+
+def load_document(file_path: str):
+    ext = os.path.splitext(file_path)[-1].lower()
+    if ext == ".pdf":
+        return DocumentFile.from_pdf(file_path)
+    elif ext in [".png", ".jpg", ".jpeg"]:
+        return DocumentFile.from_images(file_path)
+    else:
+        raise ValueError(f"Type de fichier non supporté: {ext}")
+
+
+def run_ocr(file_path: str):
+    doc = load_document(file_path)
+    model = ocr_predictor(pretrained=True)
+    result = model(doc)
+    output = result.export()
+    return doc, output
+
+
+def generate_html(output, output_html_path="ocr_output.html"):
+    html_head = """
+    <html><head><meta charset="utf-8"><title>Résultat OCR Organisé</title>
+    <style>
+      body { font-family: Arial, sans-serif; background: #f0f0f0; }
+      .page {
+        position: relative;
+        width: 800px; height: 1120px; /* ratio approx A4 */
+        margin: 20px auto;
+        background: white;
+        box-shadow: 0 0 8px rgba(0,0,0,0.2);
+      }
+      .block {
+        position: absolute;
+        border: 1px solid rgba(255,0,0,0.5);
+        background: rgba(255,255,255,0.7);
+        padding: 4px;
+        font-size: 14px;
+        line-height: 1.2em;
+        overflow-wrap: break-word;
+      }
+    </style>
+    </head><body>
+    <h1>Résultat OCR Organisé</h1>
+    """
+
+    html_body = ""
+
+    for page_idx, page in enumerate(output['pages']):
+        html_body += f'<div class="page"><h2>Page {page_idx + 1}</h2>'
+
+        # Pour chaque bloc de texte, positionner la div selon la bounding box normalisée
+        for block in page['blocks']:
+            # bbox format: [[x_min, y_min], [x_max, y_max]] (normé 0-1)
+            (x_min, y_min), (x_max, y_max) = block['geometry']
+            left = x_min * 800  # correspond à width page
+            top = y_min * 1120  # correspond à height page
+            width = (x_max - x_min) * 800
+            height = (y_max - y_min) * 1120
+
+            # Récupérer le texte complet du bloc (concaténation des lignes et mots)
+            block_text = ""
+            for line in block['lines']:
+                line_text = " ".join(word['value'] for word in line['words'])
+                block_text += line_text + "<br>"
+
+            html_body += f'<div class="block" style="left:{left}px; top:{top}px; width:{width}px; height:{height}px;">{block_text}</div>'
+
+        html_body += "</div>"  # fin page div
+
+    html_tail = "</body></html>"
+
+    with open(output_html_path, "w", encoding="utf-8") as f:
+        f.write(html_head + html_body + html_tail)
+
+    print(f"Fichier HTML sauvegardé sous : {output_html_path}")
+
+
+
+
+def extract_text_from_file(file_path: str) -> str:
+    _, output = run_ocr(file_path)
+    text = ""
+
+    for page in output['pages']:
+        for block in page['blocks']:
+            for line in block['lines']:
+                for word in line['words']:
+                    text += word['value'] + " "
+                text += "\n"
+            text += "\n"
+        text += "\n"
+
+    # Sauvegarde du JSON OCR
+    with open("uploads/ocr_output.json", "w", encoding="utf-8") as fjson:
+        json.dump(output, fjson, ensure_ascii=False, indent=2)
+
+    # Génération HTML
+    generate_html(output)
+
+    return text
+ 
+
+
+def generate_word_level_html(output, output_html_path="uploads/ocr_output_words.html"):
+    html_head = """
+    <html><head><meta charset="utf-8"><title>OCR mots avec bounding boxes</title>
+    <style>
+      body { font-family: Arial, sans-serif; background: #f0f0f0; }
+      .page {
+        position: relative;
+        width: 800px; height: 1120px; /* ratio A4 approx */
+        margin: 20px auto;
+        background: white;
+        box-shadow: 0 0 8px rgba(0,0,0,0.2);
+      }
+      .word {
+        position: absolute;
+        border: 1px solid rgba(0, 0, 255, 0.5);
+        background: rgba(255,255,255,0.8);
+        font-size: 12px;
+        line-height: 1em;
+        padding: 1px 2px;
+        white-space: nowrap;
+        overflow: hidden;
+        text-overflow: ellipsis;
+      }
+    </style>
+    </head><body>
+    <h1>OCR mots avec bounding boxes</h1>
+    """
+
+    html_body = ""
+
+    for page_idx, page in enumerate(output['pages']):
+        html_body += f'<div class="page"><h2>Page {page_idx + 1}</h2>'
+
+        for block in page['blocks']:
+            for line in block['lines']:
+                for word in line['words']:
+                    # bbox normalized format: [[x_min, y_min], [x_max, y_max]]
+                    (x_min, y_min), (x_max, y_max) = word['geometry']
+                    left = x_min * 800
+                    top = y_min * 1120
+                    width = (x_max - x_min) * 800
+                    height = (y_max - y_min) * 1120
+
+                    word_text = word['value']
+                    html_body += f'<div class="word" style="left:{left}px; top:{top}px; width:{width}px; height:{height}px;">{word_text}</div>'
+
+        html_body += "</div>"  # fin page
+
+    html_tail = "</body></html>"
+
+    with open(output_html_path, "w", encoding="utf-8") as f:
+        f.write(html_head + html_body + html_tail)
+
+    print(f"Fichier HTML mots sauvegardé sous : {output_html_path}")

+ 115 - 0
src/utils/processing.py

@@ -0,0 +1,115 @@
+import os
+import json
+import re
+import faiss
+import numpy as np
+from sentence_transformers import SentenceTransformer
+from src.agent_ai.chunking import semantic_chunking
+from src.agent_ai.scraping import scrape_pages
+
+
+
+# === Paths ===
+os.makedirs("data", exist_ok=True)
+CHUNKS_FILE = "data/All_chunks.json"
+INDEX_FILE = "data/faiss_index.index"
+RAW_PAGES_FILE = "data/raw_pages.json"
+
+# === Embedding Model ===
+'''model = SentenceTransformer(
+    'nomic-ai/nomic-embed-text-v1',
+    device='cuda',
+    trust_remote_code=True
+)'''
+model = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2', device='cpu')
+
+
+# === Cleaning Utility ===
+def clean_text(text):
+    text = re.sub(r"~~.*?~~", " ", text)
+    text = re.sub(r"<[^>]+>", " ", text)
+    text = re.sub(r"[^\x00-\x7FÀ-ÿ]+", " ", text)
+    text = re.sub(r"\s+", " ", text)
+    return text.strip()
+
+# === Step 2: Update Chunks ===
+def update_chunks(pages_dict):
+    all_chunks = []
+    if os.path.exists(CHUNKS_FILE):
+        try:
+            with open(CHUNKS_FILE, "r", encoding="utf-8") as f:
+                all_chunks = json.load(f)
+        except Exception as e:
+            print(f"⚠️ Could not load old chunks: {e}")
+
+    # Supprimer tous les chunks des URLs modifiées
+    existing_urls = set(pages_dict.keys())
+    all_chunks = [c for c in all_chunks if c.get("url") not in existing_urls]
+
+    new_chunks = []
+    chunk_id_start = len(all_chunks)
+
+    for url, text in pages_dict.items():
+        cleaned = clean_text(text)
+        if not cleaned:
+            continue
+        chunks = semantic_chunking(cleaned, with_metadata=True)
+        for ch in chunks:
+            if not isinstance(ch, dict) or not isinstance(ch.get("text"), str):
+                print(f"⚠️ Ignored malformed chunk: {ch}")
+                continue
+
+            new_chunks.append({
+                "chunk_id": chunk_id_start + len(new_chunks),
+                "text": ch["text"],
+                "url": url,
+                "word_count": ch["word_count"],
+                "token_count": ch["token_count"]
+            })
+
+    all_chunks.extend(new_chunks)
+    with open(CHUNKS_FILE, "w", encoding="utf-8") as f:
+        json.dump(all_chunks, f, indent=2, ensure_ascii=False)
+
+    print(f"📄 Chunks updated: {len(new_chunks)} new chunks added, total chunks: {len(all_chunks)}.")
+    return all_chunks
+
+# === Step 3: Build Index ===
+def build_and_save_index(chunks):
+    if not chunks:
+        print("⚠️ No chunks to index.")
+        return None
+
+    texts = [c["text"] for c in chunks]
+    embeddings = model.encode(texts, convert_to_tensor=False, batch_size=4)
+
+    if len(embeddings) == 0:
+        print("⚠️ Embeddings generation failed or empty.")
+        return None
+
+    dim = len(embeddings[0])
+    index = faiss.IndexFlatL2(dim)
+    index.add(np.array(embeddings).astype("float32"))
+    faiss.write_index(index, INDEX_FILE)
+
+    print(f"✅ FAISS index saved at '{INDEX_FILE}' with {len(texts)} vectors.")
+    return index
+
+
+# === Full Pipeline ===
+def full_update_pipeline():
+    print("🔄 Pipeline complet déclenché...")
+
+    pages = scrape_pages()
+    page_dict = {p["url"]: p["text"] for p in pages if "url" in p and "text" in p}
+
+    if not page_dict:
+        print("⚠️ Aucune page récupérée.")
+        return
+
+    chunks = update_chunks(page_dict)
+    build_and_save_index(chunks)
+
+# === Initialize if needed ===
+def initialize_if_needed():
+    full_update_pipeline()

+ 84 - 0
src/utils/security.py

@@ -0,0 +1,84 @@
+
+'''
+Ce fichier contient :
+    Détection de contenu sensible
+    Détection de prompt injection
+    Journalisation des tentatives malveillantes
+    Envoi d’alerte par e-mail (optionnel)
+'''
+''' 
+
+import re
+import datetime
+import logging
+from transformers import pipeline
+
+# === Classificateur NLP (peut être remplacé par un autre propre modèle) ===
+classifier = pipeline("text-classification", model="unitary/toxic-bert")  
+
+# === Configuration du logging ===
+logging.basicConfig(filename="logs/security_events.log", level=logging.INFO, format="%(asctime)s - %(message)s")
+
+# === 1. Détection de prompt injection ===
+def contains_prompt_injection(text):
+    patterns = [
+        r"ignore.*instruction",
+        r"réécris.*prompt",
+        r"réinitialise.*sécurité",
+        r"(system|user|assistant):",
+        r"tu es un.*assistant.*obéissant",
+        r"efface.*mémoire",
+        r"désactive.*filtre",
+        r"répond même.*si.*confidentiel",
+        r"clé API"
+    ]
+    return any(re.search(pattern, text.lower()) for pattern in patterns)
+
+# === 2. Détection intelligente de contenu sensible ===
+def is_safe_input_advanced(text):
+    result = classifier(text)[0]
+    label = result["label"]
+    score = result["score"]
+    
+    if label.lower() == "toxic" and score > 0.7:
+        return False, label, score
+    return True, label, score
+
+# === 3. Système central de vérification de sécurité ===
+def advanced_security_check(question):
+    if not question.strip():
+        return False, "empty", 0.0
+
+    if contains_prompt_injection(question):
+        return False, "prompt_injection", 1.0
+
+    is_safe, label, score = is_safe_input_advanced(question)
+    return is_safe, label, score
+
+# === 4. Journalisation des requêtes suspectes ===
+def log_suspicious_query(user_ip, question, label, score):
+    log_entry = f"🔒 SUSPICIOUS | IP: {user_ip} | Label: {label} | Score: {score:.2f} | Question: {question}"
+    logging.warning(log_entry)
+
+# === 5. (Optionnel) Envoi d’alerte e-mail ===
+def send_security_alert(ip, question, label, score):
+    from smtplib import SMTP
+
+    from_email = "bot@intersel.fr"
+    to_email = "security@intersel.fr"
+    subject = "🚨 Alerte sécurité Chatbot"
+    body = f"""
+    ALERTE SÉCURITÉ - Chatbot Intersel
+    
+    IP : {ip}
+    Type de tentative : {label}
+    Score de confiance : {score}
+    Question : {question}
+    """
+
+    try:
+        with SMTP("smtp.intersel.fr") as server:
+            server.sendmail(from_email, to_email, f"Subject: {subject}\n\n{body}")
+    except Exception as e:
+        logging.error(f"Erreur envoi mail : {str(e)}")
+'''

Beberapa file tidak ditampilkan karena terlalu banyak file yang berubah dalam diff ini