Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,201 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from flask import Flask, render_template, request, jsonify, send_file
|
2 |
+
import threading
|
3 |
+
import time
|
4 |
+
import os
|
5 |
+
from datetime import datetime
|
6 |
+
from google import genai
|
7 |
+
import uuid
|
8 |
+
|
9 |
+
app = Flask(__name__)
|
10 |
+
|
11 |
+
# Configuration
|
12 |
+
GOOGLE_API_KEY = "AIzaSyCVCvvAJbf1oLJV81fNhl56-N2s2f0ZeBc" # Remplacez par votre clé API
|
13 |
+
MODEL_ID = "gemini-2.5-flash-preview-05-20" # Ou le modèle que vous utilisez
|
14 |
+
UPLOAD_FOLDER = 'uploads'
|
15 |
+
RESULTS_FOLDER = 'results'
|
16 |
+
|
17 |
+
# Créer les dossiers s'ils n'existent pas
|
18 |
+
os.makedirs(UPLOAD_FOLDER, exist_ok=True)
|
19 |
+
os.makedirs(RESULTS_FOLDER, exist_ok=True)
|
20 |
+
|
21 |
+
# Stockage des tâches en cours
|
22 |
+
tasks = {}
|
23 |
+
|
24 |
+
class TaskManager:
|
25 |
+
def __init__(self):
|
26 |
+
self.tasks = {}
|
27 |
+
|
28 |
+
def create_task(self, task_id):
|
29 |
+
self.tasks[task_id] = {
|
30 |
+
'status': 'running',
|
31 |
+
'progress': 0,
|
32 |
+
'total': 470,
|
33 |
+
'results_file': f'results_{task_id}.txt',
|
34 |
+
'start_time': datetime.now(),
|
35 |
+
'errors': []
|
36 |
+
}
|
37 |
+
|
38 |
+
def update_progress(self, task_id, progress):
|
39 |
+
if task_id in self.tasks:
|
40 |
+
self.tasks[task_id]['progress'] = progress
|
41 |
+
|
42 |
+
def add_error(self, task_id, error):
|
43 |
+
if task_id in self.tasks:
|
44 |
+
self.tasks[task_id]['errors'].append(error)
|
45 |
+
|
46 |
+
def complete_task(self, task_id):
|
47 |
+
if task_id in self.tasks:
|
48 |
+
self.tasks[task_id]['status'] = 'completed'
|
49 |
+
|
50 |
+
def get_task(self, task_id):
|
51 |
+
return self.tasks.get(task_id)
|
52 |
+
|
53 |
+
task_manager = TaskManager()
|
54 |
+
|
55 |
+
def generate_synthetic_data(file_path, task_id):
|
56 |
+
"""Fonction qui exécute les 470 requêtes en arrière-plan"""
|
57 |
+
try:
|
58 |
+
# Initialiser le client Google AI
|
59 |
+
client = genai.Client(api_key=GOOGLE_API_KEY)
|
60 |
+
|
61 |
+
# Uploader le fichier
|
62 |
+
file_ref = client.files.upload(file=file_path)
|
63 |
+
|
64 |
+
# Prompt pour chaque requête
|
65 |
+
prompt = ("J'aimerais générer des nouvelles données synthétiques à partir de ça. "
|
66 |
+
"Une fang, une français. Je veux 400 phrases longues.\n\n"
|
67 |
+
"Selon le format : fang :.... / Français :...")
|
68 |
+
|
69 |
+
# Fichier de résultats
|
70 |
+
results_file = os.path.join(RESULTS_FOLDER, f'results_{task_id}.txt')
|
71 |
+
|
72 |
+
with open(results_file, 'w', encoding='utf-8') as f:
|
73 |
+
f.write(f"Génération de données synthétiques - Démarré le {datetime.now()}\n")
|
74 |
+
f.write("="*60 + "\n\n")
|
75 |
+
|
76 |
+
for i in range(470):
|
77 |
+
try:
|
78 |
+
# Faire la requête
|
79 |
+
response = client.models.generate_content(
|
80 |
+
model=MODEL_ID,
|
81 |
+
contents=[file_ref, prompt]
|
82 |
+
)
|
83 |
+
|
84 |
+
# Écrire la réponse dans le fichier
|
85 |
+
f.write(f"--- Requête {i+1}/470 ---\n")
|
86 |
+
f.write(response.text)
|
87 |
+
f.write("\n" + "="*60 + "\n\n")
|
88 |
+
f.flush() # Forcer l'écriture
|
89 |
+
|
90 |
+
# Mettre à jour le progrès
|
91 |
+
task_manager.update_progress(task_id, i + 1)
|
92 |
+
|
93 |
+
print(f"Requête {i+1}/470 complétée")
|
94 |
+
|
95 |
+
# Pause pour éviter de surcharger l'API
|
96 |
+
time.sleep(1)
|
97 |
+
|
98 |
+
except Exception as e:
|
99 |
+
error_msg = f"Erreur requête {i+1}: {str(e)}"
|
100 |
+
task_manager.add_error(task_id, error_msg)
|
101 |
+
f.write(f"ERREUR - Requête {i+1}: {str(e)}\n")
|
102 |
+
f.write("="*60 + "\n\n")
|
103 |
+
print(error_msg)
|
104 |
+
|
105 |
+
task_manager.complete_task(task_id)
|
106 |
+
print(f"Tâche {task_id} terminée avec succès")
|
107 |
+
|
108 |
+
except Exception as e:
|
109 |
+
error_msg = f"Erreur générale: {str(e)}"
|
110 |
+
task_manager.add_error(task_id, error_msg)
|
111 |
+
print(error_msg)
|
112 |
+
|
113 |
+
@app.route('/')
|
114 |
+
def index():
|
115 |
+
return render_template('index.html')
|
116 |
+
|
117 |
+
@app.route('/upload', methods=['POST'])
|
118 |
+
def upload_file():
|
119 |
+
if 'file' not in request.files:
|
120 |
+
return jsonify({'error': 'Aucun fichier sélectionné'}), 400
|
121 |
+
|
122 |
+
file = request.files['file']
|
123 |
+
if file.filename == '':
|
124 |
+
return jsonify({'error': 'Aucun fichier sélectionné'}), 400
|
125 |
+
|
126 |
+
if file:
|
127 |
+
# Générer un ID unique pour cette tâche
|
128 |
+
task_id = str(uuid.uuid4())
|
129 |
+
|
130 |
+
# Sauvegarder le fichier
|
131 |
+
filename = f"input_{task_id}.txt"
|
132 |
+
file_path = os.path.join(UPLOAD_FOLDER, filename)
|
133 |
+
file.save(file_path)
|
134 |
+
|
135 |
+
# Créer la tâche
|
136 |
+
task_manager.create_task(task_id)
|
137 |
+
|
138 |
+
# Démarrer le traitement en arrière-plan
|
139 |
+
thread = threading.Thread(
|
140 |
+
target=generate_synthetic_data,
|
141 |
+
args=(file_path, task_id)
|
142 |
+
)
|
143 |
+
thread.daemon = True
|
144 |
+
thread.start()
|
145 |
+
|
146 |
+
return jsonify({
|
147 |
+
'task_id': task_id,
|
148 |
+
'message': 'Traitement démarré en arrière-plan'
|
149 |
+
})
|
150 |
+
|
151 |
+
@app.route('/status/<task_id>')
|
152 |
+
def get_status(task_id):
|
153 |
+
task = task_manager.get_task(task_id)
|
154 |
+
if not task:
|
155 |
+
return jsonify({'error': 'Tâche non trouvée'}), 404
|
156 |
+
|
157 |
+
return jsonify({
|
158 |
+
'status': task['status'],
|
159 |
+
'progress': task['progress'],
|
160 |
+
'total': task['total'],
|
161 |
+
'percentage': round((task['progress'] / task['total']) * 100, 2),
|
162 |
+
'errors_count': len(task['errors']),
|
163 |
+
'start_time': task['start_time'].strftime('%Y-%m-%d %H:%M:%S')
|
164 |
+
})
|
165 |
+
|
166 |
+
@app.route('/download/<task_id>')
|
167 |
+
def download_results(task_id):
|
168 |
+
task = task_manager.get_task(task_id)
|
169 |
+
if not task:
|
170 |
+
return jsonify({'error': 'Tâche non trouvée'}), 404
|
171 |
+
|
172 |
+
results_file = os.path.join(RESULTS_FOLDER, f'results_{task_id}.txt')
|
173 |
+
|
174 |
+
if not os.path.exists(results_file):
|
175 |
+
return jsonify({'error': 'Fichier de résultats non trouvé'}), 404
|
176 |
+
|
177 |
+
return send_file(
|
178 |
+
results_file,
|
179 |
+
as_attachment=True,
|
180 |
+
download_name=f'donnees_synthetiques_{task_id}.txt'
|
181 |
+
)
|
182 |
+
|
183 |
+
@app.route('/tasks')
|
184 |
+
def list_tasks():
|
185 |
+
"""Liste toutes les tâches"""
|
186 |
+
task_list = []
|
187 |
+
for task_id, task_info in task_manager.tasks.items():
|
188 |
+
task_list.append({
|
189 |
+
'id': task_id,
|
190 |
+
'status': task_info['status'],
|
191 |
+
'progress': task_info['progress'],
|
192 |
+
'total': task_info['total'],
|
193 |
+
'percentage': round((task_info['progress'] / task_info['total']) * 100, 2),
|
194 |
+
'start_time': task_info['start_time'].strftime('%Y-%m-%d %H:%M:%S'),
|
195 |
+
'errors_count': len(task_info['errors'])
|
196 |
+
})
|
197 |
+
|
198 |
+
return jsonify(task_list)
|
199 |
+
|
200 |
+
if __name__ == '__main__':
|
201 |
+
app.run(debug=True, threaded=True)
|