Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -6,182 +6,112 @@ import torch
|
|
6 |
import pandas as pd
|
7 |
import time
|
8 |
import marqo
|
|
|
9 |
from scipy.sparse import csr_matrix
|
10 |
from transformers import AutoModel, AutoProcessor
|
11 |
from sklearn.feature_extraction.text import TfidfVectorizer
|
12 |
|
13 |
-
# Vérifier si CUDA est disponible
|
14 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
15 |
print(f"🔹 Utilisation du périphérique : {device}")
|
16 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
17 |
# Définition des fichiers JSON
|
18 |
PRODUCTS_FILE = "products.json"
|
19 |
QA_FILE = "qa_sequences_output.json"
|
20 |
|
21 |
-
#
|
22 |
-
|
23 |
-
|
|
|
|
|
24 |
|
25 |
-
for attempt in range(MAX_RETRIES):
|
26 |
try:
|
27 |
-
|
28 |
-
|
29 |
-
|
30 |
-
|
31 |
-
|
32 |
-
|
33 |
-
print(f"❌ Erreur de chargement : {e}")
|
34 |
-
if attempt < MAX_RETRIES - 1:
|
35 |
-
print("🔁 Nouvelle tentative dans 5 secondes...")
|
36 |
-
time.sleep(5)
|
37 |
-
else:
|
38 |
-
print("⛔ Échec final du chargement du modèle.")
|
39 |
-
model, processor = None, None
|
40 |
-
|
41 |
-
# Fonction pour charger les fichiers JSON
|
42 |
-
def load_data():
|
43 |
-
products_data, qa_data = [], []
|
44 |
-
|
45 |
-
if os.path.exists(PRODUCTS_FILE):
|
46 |
-
with open(PRODUCTS_FILE, "r", encoding="utf-8") as f:
|
47 |
-
products_data = json.load(f).get("products", [])
|
48 |
-
else:
|
49 |
-
print(f"⛔ Fichier introuvable : {PRODUCTS_FILE}")
|
50 |
-
|
51 |
-
if os.path.exists(QA_FILE):
|
52 |
-
with open(QA_FILE, "r", encoding="utf-8") as f:
|
53 |
-
qa_data = json.load(f)
|
54 |
-
else:
|
55 |
-
print(f"⛔ Fichier introuvable : {QA_FILE}")
|
56 |
-
|
57 |
-
return products_data, qa_data
|
58 |
-
|
59 |
-
products_data, qa_data = load_data()
|
60 |
-
|
61 |
-
# Associer les questions-réponses aux produits
|
62 |
-
def associate_qa_with_products(products, qa_data):
|
63 |
-
for product in products:
|
64 |
-
product["qa_info"] = []
|
65 |
-
product_name = product.get("title", "").lower()
|
66 |
-
product_desc = product.get("description", "").lower()
|
67 |
-
|
68 |
-
for qa in qa_data:
|
69 |
-
question = qa.get("question", "").lower()
|
70 |
-
if product_name in question or product_desc in question:
|
71 |
-
product["qa_info"].append(qa)
|
72 |
-
|
73 |
-
return products
|
74 |
-
|
75 |
-
products_data = associate_qa_with_products(products_data, qa_data)
|
76 |
-
|
77 |
-
# Connexion au serveur Marqo
|
78 |
-
mq = marqo.Client(url="http://localhost:8882") # Port par défaut de Marqo
|
79 |
-
INDEX_NAME = "ecommerce_products"
|
80 |
|
81 |
-
|
82 |
-
|
|
|
|
|
|
|
|
|
83 |
mq.delete_index(INDEX_NAME)
|
84 |
-
except:
|
85 |
-
pass
|
86 |
|
87 |
-
|
88 |
-
mq.create_index(INDEX_NAME)
|
89 |
print("✅ Index Marqo créé avec succès !")
|
90 |
|
91 |
# Ajouter les produits à Marqo
|
92 |
-
documents = [
|
93 |
-
|
94 |
-
|
95 |
-
"
|
96 |
-
"
|
97 |
-
"
|
98 |
-
"
|
99 |
-
"
|
100 |
-
"category": product["category"],
|
101 |
-
"qa_info": product.get("qa_info", []),
|
102 |
-
"_model": "open_clip/ViT-B-32/laion2B-b79K", # Spécifier le modèle ici
|
103 |
}
|
104 |
-
|
|
|
105 |
|
106 |
mq.index(INDEX_NAME).add_documents(documents, tensor_fields=["title", "description"])
|
107 |
print("✅ Produits indexés dans Marqo avec succès !")
|
108 |
|
109 |
-
# Prétraitement du texte
|
110 |
-
def preprocess(text: str) -> str:
|
111 |
-
text = text.lower()
|
112 |
-
text = re.sub(r'\s+', ' ', text)
|
113 |
-
text = re.sub(r'[^\w\s]', '', text)
|
114 |
-
return text.strip()
|
115 |
-
|
116 |
-
# TF-IDF Vectorizer
|
117 |
-
vectorizer = TfidfVectorizer(stop_words="english")
|
118 |
-
tfidf_matrix = vectorizer.fit_transform([prod["title"] + " " + prod["description"] for prod in products_data])
|
119 |
-
|
120 |
-
# Recherche hybride avec Marqo + TF-IDF
|
121 |
-
def search_products(query, category, min_price, max_price, weight_tfidf=0.5, weight_marqo=0.5):
|
122 |
-
query = preprocess(query)
|
123 |
-
|
124 |
-
# Recherche Marqo (top 50 résultats)
|
125 |
-
marqo_results = mq.index(INDEX_NAME).search(query, searchable_attributes=["title", "description"], limit=50)
|
126 |
-
|
127 |
-
# Récupérer les résultats Marqo
|
128 |
-
marqo_products = []
|
129 |
-
marqo_scores = []
|
130 |
-
for hit in marqo_results["hits"]:
|
131 |
-
marqo_products.append(hit)
|
132 |
-
marqo_scores.append(hit["_score"])
|
133 |
-
|
134 |
-
# Normaliser les scores Marqo
|
135 |
-
if len(marqo_scores) > 0:
|
136 |
-
marqo_scores = (pd.Series(marqo_scores) - min(marqo_scores)) / (max(marqo_scores) - min(marqo_scores) + 1e-6)
|
137 |
-
else:
|
138 |
-
marqo_scores = [0] * len(marqo_products)
|
139 |
-
|
140 |
-
# TF-IDF Similarité
|
141 |
-
query_vector_sparse = csr_matrix(vectorizer.transform([query]))
|
142 |
-
tfidf_scores = (tfidf_matrix * query_vector_sparse.T).toarray().flatten()
|
143 |
-
|
144 |
-
# Normaliser les scores TF-IDF
|
145 |
-
if len(tfidf_scores) > 0:
|
146 |
-
tfidf_scores_norm = (tfidf_scores - min(tfidf_scores)) / (max(tfidf_scores) - min(tfidf_scores) + 1e-6)
|
147 |
-
else:
|
148 |
-
tfidf_scores_norm = [0] * len(marqo_products)
|
149 |
-
|
150 |
-
# Fusionner les scores TF-IDF et Marqo
|
151 |
-
final_scores = weight_tfidf * tfidf_scores_norm[:len(marqo_products)] + weight_marqo * marqo_scores
|
152 |
-
|
153 |
-
# Ajouter le score final aux produits
|
154 |
-
for i, product in enumerate(marqo_products):
|
155 |
-
product["score"] = final_scores[i]
|
156 |
-
|
157 |
-
# Convertir en DataFrame
|
158 |
-
results_df = pd.DataFrame(marqo_products)
|
159 |
-
|
160 |
-
# Filtrer les résultats par prix et disponibilité
|
161 |
-
results_df = results_df[
|
162 |
-
(results_df["price"] >= min_price) &
|
163 |
-
(results_df["price"] <= max_price) &
|
164 |
-
(results_df["availability"] == "in stock")
|
165 |
-
]
|
166 |
-
|
167 |
-
if category and category != "Toutes":
|
168 |
-
results_df = results_df[results_df["category"].str.contains(category, case=False, na=False)]
|
169 |
-
|
170 |
-
return results_df.sort_values(by="score", ascending=False).head(20)
|
171 |
-
|
172 |
# Interface Gradio
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
173 |
app = gr.Interface(
|
174 |
fn=search_products,
|
175 |
-
inputs=[
|
176 |
-
|
177 |
-
gr.Textbox(label="Catégorie"),
|
178 |
-
gr.Number(label="Prix min"),
|
179 |
-
gr.Number(label="Prix max")
|
180 |
-
],
|
181 |
-
outputs=[
|
182 |
-
gr.Dataframe(headers=["ID", "Titre", "Description", "Prix", "Disponibilité", "Questions/Réponses"],
|
183 |
-
datatype=["str", "str", "str", "number", "str", "json"])
|
184 |
-
]
|
185 |
)
|
186 |
|
187 |
app.launch()
|
|
|
6 |
import pandas as pd
|
7 |
import time
|
8 |
import marqo
|
9 |
+
import requests
|
10 |
from scipy.sparse import csr_matrix
|
11 |
from transformers import AutoModel, AutoProcessor
|
12 |
from sklearn.feature_extraction.text import TfidfVectorizer
|
13 |
|
14 |
+
# Vérifier si CUDA est disponible
|
15 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
16 |
print(f"🔹 Utilisation du périphérique : {device}")
|
17 |
|
18 |
+
# Lancer Marqo si nécessaire
|
19 |
+
os.system("docker run -d -p 8882:8882 marqoai/marqo")
|
20 |
+
|
21 |
+
# Vérifier que Marqo est bien lancé
|
22 |
+
def wait_for_marqo(timeout=30):
|
23 |
+
start_time = time.time()
|
24 |
+
while time.time() - start_time < timeout:
|
25 |
+
try:
|
26 |
+
response = requests.get("http://localhost:8882")
|
27 |
+
if response.status_code == 200:
|
28 |
+
print("✅ Marqo est prêt !")
|
29 |
+
return True
|
30 |
+
except requests.exceptions.ConnectionError:
|
31 |
+
print("⏳ En attente du démarrage de Marqo...")
|
32 |
+
time.sleep(3)
|
33 |
+
print("⛔ Marqo ne répond pas après 30 secondes. Vérifiez le démarrage.")
|
34 |
+
return False
|
35 |
+
|
36 |
+
if not wait_for_marqo():
|
37 |
+
exit(1)
|
38 |
+
|
39 |
+
# Connexion à Marqo avec gestion des erreurs
|
40 |
+
MAX_RETRIES = 5
|
41 |
+
for attempt in range(MAX_RETRIES):
|
42 |
+
try:
|
43 |
+
mq = marqo.Client(url="http://localhost:8882")
|
44 |
+
print("✅ Connexion à Marqo réussie !")
|
45 |
+
break
|
46 |
+
except marqo.errors.BackendCommunicationError:
|
47 |
+
print(f"⚠️ Erreur de connexion à Marqo (tentative {attempt + 1}/{MAX_RETRIES})")
|
48 |
+
time.sleep(5)
|
49 |
+
else:
|
50 |
+
print("⛔ Impossible de se connecter à Marqo après plusieurs tentatives.")
|
51 |
+
exit(1)
|
52 |
+
|
53 |
# Définition des fichiers JSON
|
54 |
PRODUCTS_FILE = "products.json"
|
55 |
QA_FILE = "qa_sequences_output.json"
|
56 |
|
57 |
+
# Fonction pour charger les fichiers JSON de manière sécurisée
|
58 |
+
def safe_load_json(file_path):
|
59 |
+
if not os.path.exists(file_path):
|
60 |
+
print(f"⛔ Fichier introuvable : {file_path}")
|
61 |
+
return []
|
62 |
|
|
|
63 |
try:
|
64 |
+
with open(file_path, "r", encoding="utf-8") as f:
|
65 |
+
data = json.load(f)
|
66 |
+
return data.get("products", []) if "products" in data else data
|
67 |
+
except json.JSONDecodeError:
|
68 |
+
print(f"⚠️ Erreur de décodage JSON dans {file_path}")
|
69 |
+
return []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
70 |
|
71 |
+
products_data = safe_load_json(PRODUCTS_FILE)
|
72 |
+
qa_data = safe_load_json(QA_FILE)
|
73 |
+
|
74 |
+
# Création de l'index Marqo avec la bonne configuration
|
75 |
+
INDEX_NAME = "ecommerce_products"
|
76 |
+
if INDEX_NAME in [index["index_name"] for index in mq.get_indexes()["results"]]:
|
77 |
mq.delete_index(INDEX_NAME)
|
|
|
|
|
78 |
|
79 |
+
mq.create_index(INDEX_NAME, model="open_clip/ViT-B-32/laion2B-b79K", normalize_embeddings=True)
|
|
|
80 |
print("✅ Index Marqo créé avec succès !")
|
81 |
|
82 |
# Ajouter les produits à Marqo
|
83 |
+
documents = [
|
84 |
+
{
|
85 |
+
"id": product.get("id", ""),
|
86 |
+
"title": product.get("title", ""),
|
87 |
+
"description": product.get("description", ""),
|
88 |
+
"price": product.get("price", 0),
|
89 |
+
"availability": product.get("availability", ""),
|
90 |
+
"category": product.get("category", ""),
|
|
|
|
|
|
|
91 |
}
|
92 |
+
for product in products_data
|
93 |
+
]
|
94 |
|
95 |
mq.index(INDEX_NAME).add_documents(documents, tensor_fields=["title", "description"])
|
96 |
print("✅ Produits indexés dans Marqo avec succès !")
|
97 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
98 |
# Interface Gradio
|
99 |
+
def search_products(query, category, min_price, max_price):
|
100 |
+
query = query.strip()
|
101 |
+
if not query:
|
102 |
+
return "❌ Veuillez entrer un terme de recherche valide."
|
103 |
+
|
104 |
+
min_price = float(min_price) if isinstance(min_price, (int, float)) else 0
|
105 |
+
max_price = float(max_price) if isinstance(max_price, (int, float)) else float("inf")
|
106 |
+
|
107 |
+
marqo_results = mq.index(INDEX_NAME).search(query, searchable_attributes=["title", "description"], limit=20)
|
108 |
+
results_df = pd.DataFrame(marqo_results["hits"])
|
109 |
+
return results_df
|
110 |
+
|
111 |
app = gr.Interface(
|
112 |
fn=search_products,
|
113 |
+
inputs=[gr.Textbox(label="Rechercher un produit"), gr.Textbox(label="Catégorie"), gr.Number(label="Prix min"), gr.Number(label="Prix max")],
|
114 |
+
outputs=gr.Dataframe(),
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
115 |
)
|
116 |
|
117 |
app.launch()
|