Spaces:
Running
Running
add
Browse files
app.py
CHANGED
@@ -2,8 +2,256 @@ from fastapi import FastAPI
|
|
2 |
from fastapi.staticfiles import StaticFiles
|
3 |
from fastapi.responses import RedirectResponse
|
4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5 |
app = FastAPI()
|
6 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
# Servir les fichiers statiques (HTML, CSS, JS)
|
8 |
app.mount("/static", StaticFiles(directory="static", html=True), name="static")
|
9 |
|
|
|
2 |
from fastapi.staticfiles import StaticFiles
|
3 |
from fastapi.responses import RedirectResponse
|
4 |
|
5 |
+
from fastapi import FastAPI, File, UploadFile, Form
|
6 |
+
from fastapi.responses import JSONResponse, RedirectResponse
|
7 |
+
from fastapi.staticfiles import StaticFiles
|
8 |
+
from fastapi.middleware.cors import CORSMiddleware
|
9 |
+
from transformers import pipeline, M2M100ForConditionalGeneration, M2M100Tokenizer, MarianMTModel, MarianTokenizer
|
10 |
+
import shutil
|
11 |
+
#
|
12 |
+
import os
|
13 |
+
import logging
|
14 |
+
from PyPDF2 import PdfReader
|
15 |
+
import docx
|
16 |
+
from PIL import Image
|
17 |
+
import openpyxl # 📌 Pour lire les fichiers Excel (.xlsx)
|
18 |
+
from pptx import Presentation
|
19 |
+
import fitz # PyMuPDF
|
20 |
+
import io
|
21 |
+
from docx import Document
|
22 |
+
import matplotlib.pyplot as plt
|
23 |
+
import seaborn as sns
|
24 |
+
import torch
|
25 |
+
import re
|
26 |
+
import pandas as pd
|
27 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
28 |
+
from fastapi.responses import FileResponse
|
29 |
+
import os
|
30 |
+
from fastapi.middleware.cors import CORSMiddleware
|
31 |
+
import matplotlib
|
32 |
+
matplotlib.use('Agg')
|
33 |
+
|
34 |
+
import re
|
35 |
+
import torch
|
36 |
+
import pandas as pd
|
37 |
+
import matplotlib.pyplot as plt
|
38 |
+
import seaborn as sns
|
39 |
+
from transformers import AutoTokenizer, AutoModelForCausalLM
|
40 |
+
from fastapi import FastAPI, File, UploadFile, Form
|
41 |
+
from fastapi.responses import FileResponse
|
42 |
+
import os
|
43 |
+
from fastapi.middleware.cors import CORSMiddleware
|
44 |
+
from fastapi import FastAPI, File, UploadFile, Form
|
45 |
+
from fastapi.responses import JSONResponse, RedirectResponse
|
46 |
+
from fastapi.staticfiles import StaticFiles
|
47 |
+
from transformers import pipeline, M2M100ForConditionalGeneration, M2M100Tokenizer
|
48 |
+
import shutil
|
49 |
+
import os
|
50 |
+
import logging
|
51 |
+
from fastapi.middleware.cors import CORSMiddleware
|
52 |
+
from PyPDF2 import PdfReader
|
53 |
+
import docx
|
54 |
+
from PIL import Image # Pour ouvrir les images avant analyse
|
55 |
+
from transformers import MarianMTModel, MarianTokenizer
|
56 |
+
import os
|
57 |
+
import fitz
|
58 |
+
from transformers import M2M100ForConditionalGeneration, M2M100Tokenizer
|
59 |
+
|
60 |
+
import logging
|
61 |
+
import openpyxl
|
62 |
+
|
63 |
+
|
64 |
+
# Configuration du logging
|
65 |
+
logging.basicConfig(level=logging.INFO)
|
66 |
+
|
67 |
+
|
68 |
app = FastAPI()
|
69 |
|
70 |
+
# Configuration CORS
|
71 |
+
app.add_middleware(
|
72 |
+
CORSMiddleware,
|
73 |
+
allow_origins=["*"],
|
74 |
+
allow_credentials=True,
|
75 |
+
allow_methods=["*"],
|
76 |
+
allow_headers=["*"],
|
77 |
+
)
|
78 |
+
|
79 |
+
UPLOAD_DIR = "uploads"
|
80 |
+
os.makedirs(UPLOAD_DIR, exist_ok=True)
|
81 |
+
|
82 |
+
|
83 |
+
|
84 |
+
|
85 |
+
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
|
86 |
+
model_name = "facebook/m2m100_418M"
|
87 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
88 |
+
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
|
89 |
+
|
90 |
+
|
91 |
+
# Fonction pour extraire le texte
|
92 |
+
def extract_text_from_pdf(file):
|
93 |
+
doc = fitz.open(stream=file.file.read(), filetype="pdf")
|
94 |
+
return "\n".join([page.get_text() for page in doc]).strip()
|
95 |
+
|
96 |
+
def extract_text_from_docx(file):
|
97 |
+
doc = Document(io.BytesIO(file.file.read()))
|
98 |
+
return "\n".join([para.text for para in doc.paragraphs]).strip()
|
99 |
+
|
100 |
+
def extract_text_from_pptx(file):
|
101 |
+
prs = Presentation(io.BytesIO(file.file.read()))
|
102 |
+
return "\n".join([shape.text for slide in prs.slides for shape in slide.shapes if hasattr(shape, "text")]).strip()
|
103 |
+
|
104 |
+
def extract_text_from_excel(file):
|
105 |
+
wb = openpyxl.load_workbook(io.BytesIO(file.file.read()), data_only=True)
|
106 |
+
text = [str(cell) for sheet in wb.worksheets for row in sheet.iter_rows(values_only=True) for cell in row if cell]
|
107 |
+
return "\n".join(text).strip()
|
108 |
+
|
109 |
+
@app.post("/translate/")
|
110 |
+
async def translate_document(file: UploadFile = File(...), target_lang: str = Form(...)):
|
111 |
+
"""API pour traduire un document."""
|
112 |
+
try:
|
113 |
+
logging.info(f"📥 Fichier reçu : {file.filename}")
|
114 |
+
logging.info(f"🌍 Langue cible reçue : {target_lang}")
|
115 |
+
|
116 |
+
if model is None or tokenizer is None:
|
117 |
+
return JSONResponse(status_code=500, content={"error": "Modèle de traduction non chargé"})
|
118 |
+
|
119 |
+
# Extraction du texte
|
120 |
+
if file.filename.endswith(".pdf"):
|
121 |
+
text = extract_text_from_pdf(file)
|
122 |
+
elif file.filename.endswith(".docx"):
|
123 |
+
text = extract_text_from_docx(file)
|
124 |
+
elif file.filename.endswith(".pptx"):
|
125 |
+
text = extract_text_from_pptx(file)
|
126 |
+
elif file.filename.endswith(".xlsx"):
|
127 |
+
text = extract_text_from_excel(file)
|
128 |
+
else:
|
129 |
+
return JSONResponse(status_code=400, content={"error": "Format non supporté"})
|
130 |
+
|
131 |
+
logging.info(f"📜 Texte extrait : {text[:50]}...")
|
132 |
+
|
133 |
+
if not text:
|
134 |
+
return JSONResponse(status_code=400, content={"error": "Aucun texte trouvé dans le document"})
|
135 |
+
|
136 |
+
# Vérifier si la langue cible est supportée
|
137 |
+
target_lang_id = tokenizer.get_lang_id(target_lang)
|
138 |
+
|
139 |
+
if target_lang_id is None:
|
140 |
+
return JSONResponse(
|
141 |
+
status_code=400,
|
142 |
+
content={"error": f"Langue cible '{target_lang}' non supportée. Langues disponibles : {list(tokenizer.lang_code_to_id.keys())}"}
|
143 |
+
)
|
144 |
+
|
145 |
+
# Traduction
|
146 |
+
tokenizer.src_lang = "fr"
|
147 |
+
encoded_text = tokenizer(text, return_tensors="pt", padding=True, truncation=True)
|
148 |
+
|
149 |
+
logging.info(f"🔍 ID de la langue cible : {target_lang_id}")
|
150 |
+
|
151 |
+
generated_tokens = model.generate(**encoded_text, forced_bos_token_id=target_lang_id)
|
152 |
+
|
153 |
+
translated_text = tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0]
|
154 |
+
|
155 |
+
logging.info(f"✅ Traduction réussie : {translated_text[:50]}...")
|
156 |
+
return {"translated_text": translated_text}
|
157 |
+
|
158 |
+
except Exception as e:
|
159 |
+
logging.error(f"❌ Erreur lors de la traduction : {e}")
|
160 |
+
return JSONResponse(status_code=500, content={"error": "Échec de la traduction"})
|
161 |
+
|
162 |
+
|
163 |
+
|
164 |
+
|
165 |
+
# Charger le modèle pour la génération de code
|
166 |
+
codegen_model_name = "Salesforce/codegen-350M-mono"
|
167 |
+
device = "cuda" if torch.cuda.is_available() else "cpu"
|
168 |
+
|
169 |
+
codegen_tokenizer = AutoTokenizer.from_pretrained(codegen_model_name)
|
170 |
+
codegen_model = AutoModelForCausalLM.from_pretrained(codegen_model_name).to(device)
|
171 |
+
|
172 |
+
VALID_PLOTS = {"histplot", "scatterplot", "barplot", "lineplot", "boxplot"}
|
173 |
+
|
174 |
+
@app.post("/generate_viz/")
|
175 |
+
async def generate_viz(file: UploadFile = File(...), query: str = Form(...)):
|
176 |
+
try:
|
177 |
+
if query not in VALID_PLOTS:
|
178 |
+
return {"error": f"Type de graphique invalide. Choisissez parmi : {', '.join(VALID_PLOTS)}"}
|
179 |
+
|
180 |
+
df = pd.read_excel(file.file)
|
181 |
+
|
182 |
+
numeric_cols = df.select_dtypes(include=["number"]).columns
|
183 |
+
if len(numeric_cols) < 2:
|
184 |
+
return {"error": "Le fichier doit contenir au moins deux colonnes numériques."}
|
185 |
+
|
186 |
+
x_col, y_col = numeric_cols[:2]
|
187 |
+
|
188 |
+
# Contraintes spécifiques pour éviter l'erreur avec histplot
|
189 |
+
if query == "histplot":
|
190 |
+
prompt_y = ""
|
191 |
+
else:
|
192 |
+
prompt_y = f', y="{y_col}"'
|
193 |
+
|
194 |
+
# Générer l'invite pour le modèle
|
195 |
+
prompt = f"""
|
196 |
+
### Génère uniquement du code Python fonctionnel pour tracer un {query} avec Matplotlib et Seaborn ###
|
197 |
+
# Contraintes :
|
198 |
+
# - Utilise 'df' sans recréer de nouvelles données
|
199 |
+
# - Axe X : '{x_col}'
|
200 |
+
# - Enregistre le graphique sous 'plot.png'
|
201 |
+
# - Ne génère que du code Python valide, sans texte explicatif
|
202 |
+
# Contraintes spécifiques pour sns.histplot :
|
203 |
+
# - N'inclut pas "y=" car histplot ne supporte qu'un axe
|
204 |
+
import matplotlib.pyplot as plt
|
205 |
+
import seaborn as sns
|
206 |
+
plt.figure(figsize=(8,6))
|
207 |
+
sns.{query}(data=df, x="{x_col}"{prompt_y})
|
208 |
+
plt.savefig("plot.png")
|
209 |
+
plt.close()
|
210 |
+
"""
|
211 |
+
|
212 |
+
# Génération du code
|
213 |
+
inputs = codegen_tokenizer(prompt, return_tensors="pt").to(device)
|
214 |
+
outputs = codegen_model.generate(**inputs, max_new_tokens=120, pad_token_id=codegen_tokenizer.eos_token_id)
|
215 |
+
generated_code = codegen_tokenizer.decode(outputs[0], skip_special_tokens=True).strip()
|
216 |
+
# Nettoyage du code
|
217 |
+
generated_code = re.sub(r"(import matplotlib.pyplot as plt\nimport seaborn as sns\n)+", "import matplotlib.pyplot as plt\nimport seaborn as sns\n", generated_code)
|
218 |
+
if generated_code.strip().endswith("sns."):
|
219 |
+
generated_code = generated_code.rsplit("\n", 1)[0] # Supprime la dernière ligne incomplète
|
220 |
+
|
221 |
+
print("🔹 Code généré par l'IA :\n", generated_code)
|
222 |
+
|
223 |
+
# Vérification syntaxique avant exécution
|
224 |
+
try:
|
225 |
+
compile(generated_code, "<string>", "exec")
|
226 |
+
except SyntaxError as e:
|
227 |
+
return {"error": f"Erreur de syntaxe détectée : {e}\nCode généré :\n{generated_code}"}
|
228 |
+
|
229 |
+
# Vérification des données
|
230 |
+
print(df.head()) # Affiche les premières lignes du dataframe
|
231 |
+
print(df.dtypes) # Vérifie les types de colonnes
|
232 |
+
print(f"Colonne '{x_col}' - Valeurs uniques:", df[x_col].unique())
|
233 |
+
|
234 |
+
if df.empty or x_col not in df.columns or df[x_col].isnull().all():
|
235 |
+
return {"error": f"La colonne '{x_col}' est absente ou ne contient pas de données valides."}
|
236 |
+
|
237 |
+
# Exécution du code généré
|
238 |
+
exec_env = {"df": df, "plt": plt, "sns": sns, "pd": pd}
|
239 |
+
exec(generated_code, exec_env)
|
240 |
+
|
241 |
+
# Vérification de l'image générée
|
242 |
+
img_path = "plot.png"
|
243 |
+
if not os.path.exists(img_path):
|
244 |
+
return {"error": "Le fichier plot.png n'a pas été généré."}
|
245 |
+
if os.path.getsize(img_path) == 0:
|
246 |
+
return {"error": "Le fichier plot.png est vide."}
|
247 |
+
|
248 |
+
plt.close()
|
249 |
+
return FileResponse(img_path, media_type="image/png")
|
250 |
+
|
251 |
+
except Exception as e:
|
252 |
+
return {"error": f"Erreur lors de la génération du graphique : {str(e)}"}
|
253 |
+
|
254 |
+
|
255 |
# Servir les fichiers statiques (HTML, CSS, JS)
|
256 |
app.mount("/static", StaticFiles(directory="static", html=True), name="static")
|
257 |
|