Update app.py
Browse files
app.py
CHANGED
@@ -6,12 +6,27 @@ import pandas as pd
|
|
6 |
import os
|
7 |
import nltk
|
8 |
import asyncio
|
9 |
-
nltk.download('
|
|
|
10 |
from nltk.tokenize import sent_tokenize
|
11 |
|
12 |
HF_TOKEN = os.getenv("HF_TOKEN")
|
13 |
|
14 |
-
#
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
15 |
async def call_zephyr_api(prompt, mode, hf_token=HF_TOKEN):
|
16 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=hf_token)
|
17 |
try:
|
@@ -21,27 +36,15 @@ async def call_zephyr_api(prompt, mode, hf_token=HF_TOKEN):
|
|
21 |
elif mode == "Équilibré":
|
22 |
max_new_tokens = 100
|
23 |
temperature = 0.5
|
24 |
-
else:
|
25 |
max_new_tokens = 150
|
26 |
temperature = 0.7
|
27 |
response = await asyncio.to_thread(client.text_generation, prompt, max_new_tokens=max_new_tokens, temperature=temperature)
|
28 |
return response
|
29 |
except Exception as e:
|
30 |
-
raise gr.Error(f"❌ Erreur
|
31 |
-
|
32 |
-
# Chargement du modèle de sentiment pour analyser les réponses
|
33 |
-
classifier = pipeline("sentiment-analysis", model="mrm8488/distilroberta-finetuned-financial-news-sentiment-analysis")
|
34 |
-
|
35 |
-
# Modèles de traduction (optionnels, désactivés pour optimisation)
|
36 |
-
translator_to_en = pipeline("translation", model="Helsinki-NLP/opus-mt-mul-en")
|
37 |
-
# Traduction en français désactivée pour l'instant
|
38 |
-
# translator_to_fr = pipeline("translation", model="Helsinki-NLP/opus-mt-en-fr")
|
39 |
|
40 |
-
#
|
41 |
-
def safe_translate_to_fr(text, max_length=512):
|
42 |
-
return "Traduction désactivée pour l'instant pour améliorer la vitesse."
|
43 |
-
|
44 |
-
# Fonction pour suggérer le meilleur modèle
|
45 |
def suggest_model(text):
|
46 |
word_count = len(text.split())
|
47 |
if word_count < 50:
|
@@ -51,42 +54,33 @@ def suggest_model(text):
|
|
51 |
else:
|
52 |
return "Précis"
|
53 |
|
54 |
-
#
|
55 |
def create_sentiment_gauge(sentiment, score):
|
56 |
score_percentage = score * 100
|
57 |
-
|
58 |
-
|
59 |
-
elif sentiment.lower() == "positive":
|
60 |
color = "#2E8B57"
|
61 |
elif sentiment.lower() == "negative":
|
62 |
color = "#DC143C"
|
63 |
-
else:
|
64 |
-
color = "#A9A9A9"
|
65 |
|
66 |
html = f"""
|
67 |
<div style='width: 100%; max-width: 300px; margin: 10px 0;'>
|
68 |
-
<div style='background-color: #D3D3D3; border-radius: 5px; height: 20px; position: relative;
|
69 |
-
<div style='background-color: {color}; width: {score_percentage}%; height: 100%; border-radius: 5px;
|
70 |
-
|
71 |
-
<span style='position: absolute; top: 0; left: 50%; transform: translateX(-50%); color: #0A1D37; font-size: 12px; line-height: 20px; font-weight: 600;'>
|
72 |
-
{score_percentage:.1f}%
|
73 |
-
</span>
|
74 |
-
</div>
|
75 |
-
<div style='text-align: center; font-size: 14px; margin-top: 5px; color: #E0E0E0;'>
|
76 |
-
Sentiment: {sentiment}
|
77 |
</div>
|
|
|
78 |
</div>
|
79 |
"""
|
80 |
return html
|
81 |
|
82 |
-
#
|
83 |
async def full_analysis(text, mode, detail_mode, count, history):
|
84 |
if not text:
|
85 |
-
yield "Entrez une phrase.", "", "", 0, history,
|
86 |
return
|
87 |
|
88 |
-
|
89 |
-
yield "Analyse en cours... (Étape 1 : Détection de la langue)", "", "", count, history, None, ""
|
90 |
|
91 |
try:
|
92 |
lang = detect(text)
|
@@ -98,17 +92,15 @@ async def full_analysis(text, mode, detail_mode, count, history):
|
|
98 |
else:
|
99 |
text_en = text
|
100 |
|
101 |
-
yield "Analyse en cours... (Étape 2 : Analyse du sentiment)", "", "", count, history,
|
102 |
|
103 |
-
# Analyse du sentiment avec RoBERTa sur le texte d'entrée
|
104 |
result = await asyncio.to_thread(classifier, text_en)
|
105 |
result = result[0]
|
106 |
sentiment_output = f"Sentiment prédictif : {result['label']} (Score: {result['score']:.2f})"
|
107 |
sentiment_gauge = create_sentiment_gauge(result['label'], result['score'])
|
108 |
|
109 |
-
yield "Analyse en cours... (Étape 3 : Explication
|
110 |
|
111 |
-
# Appel à Zephyr pour expliquer l'impact basé sur le sentiment
|
112 |
explanation_prompt = f"""<|system|>
|
113 |
You are a professional financial analyst AI with expertise in economic forecasting.
|
114 |
</s>
|
@@ -117,15 +109,14 @@ Given the following question about a potential economic event: "{text}"
|
|
117 |
|
118 |
The predicted sentiment for this event is: {result['label'].lower()}.
|
119 |
|
120 |
-
Assume the event happens
|
121 |
</s>
|
122 |
<|assistant|>"""
|
123 |
explanation_en = await call_zephyr_api(explanation_prompt, mode)
|
124 |
|
125 |
-
yield "Analyse en cours... (Étape 4 : Traduction)", "", "", count, history,
|
126 |
|
127 |
-
|
128 |
-
explanation_fr = safe_translate_to_fr(explanation_en, max_length=512)
|
129 |
|
130 |
count += 1
|
131 |
history.append({
|
@@ -136,9 +127,9 @@ Assume the event happens (e.g., if the question is "Will the Federal Reserve rai
|
|
136 |
"Explication_FR": explanation_fr
|
137 |
})
|
138 |
|
139 |
-
yield sentiment_output, explanation_en, explanation_fr, count, history, sentiment_gauge, ""
|
140 |
|
141 |
-
#
|
142 |
def download_history(history):
|
143 |
if not history:
|
144 |
return None
|
@@ -147,222 +138,62 @@ def download_history(history):
|
|
147 |
df.to_csv(file_path, index=False)
|
148 |
return file_path
|
149 |
|
150 |
-
#
|
151 |
def launch_app():
|
152 |
custom_css = """
|
153 |
-
|
154 |
-
@import url('https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css');
|
155 |
-
|
156 |
body {
|
157 |
-
background
|
158 |
-
background-size: cover;
|
159 |
-
background-position: center;
|
160 |
-
background-attachment: fixed;
|
161 |
-
background-color: #0A1D37;
|
162 |
-
background-blend-mode: overlay;
|
163 |
-
color: #E0E0E0;
|
164 |
font-family: 'Inter', sans-serif;
|
165 |
-
|
166 |
padding: 20px;
|
167 |
}
|
168 |
-
|
169 |
.gr-box {
|
170 |
-
background: #
|
171 |
-
border-radius: 12px !important;
|
172 |
border: 1px solid #FFD700 !important;
|
173 |
-
|
174 |
-
padding:
|
175 |
-
|
176 |
-
transition: transform 0.2s ease, box-shadow 0.3s ease !important;
|
177 |
-
}
|
178 |
-
|
179 |
-
.gr-box:hover {
|
180 |
-
transform: translateY(-3px) !important;
|
181 |
-
box-shadow: 0 8px 20px rgba(255, 215, 0, 0.3) !important;
|
182 |
-
}
|
183 |
-
|
184 |
-
.gr-textbox, .gr-dropdown {
|
185 |
-
background: #2E4A43 !important;
|
186 |
-
border: 1px solid #FFD700 !important;
|
187 |
-
border-radius: 8px !important;
|
188 |
-
color: #E0E0E0 !important;
|
189 |
-
font-size: 16px !important;
|
190 |
-
padding: 12px !important;
|
191 |
-
transition: border-color 0.3s ease, box-shadow 0.3s ease !important;
|
192 |
-
}
|
193 |
-
|
194 |
-
.gr-textbox:focus, .gr-dropdown:focus {
|
195 |
-
border-color: #FFD700 !important;
|
196 |
-
box-shadow: 0 0 10px rgba(255, 215, 0, 0.4) !important;
|
197 |
}
|
198 |
-
|
199 |
.gr-button {
|
200 |
-
background: linear-gradient(90deg, #FFD700
|
201 |
-
color: #0A1D37
|
202 |
-
|
203 |
-
border
|
204 |
-
|
205 |
-
|
206 |
-
|
207 |
-
transition: transform 0.1s ease, box-shadow 0.3s ease !important;
|
208 |
-
box-shadow: 0 3px 10px rgba(255, 215, 0, 0.3) !important;
|
209 |
}
|
210 |
-
|
211 |
.gr-button:hover {
|
212 |
-
transform: translateY(-2px)
|
213 |
-
box-shadow: 0 6px
|
214 |
-
}
|
215 |
-
|
216 |
-
h1, h2, h3 {
|
217 |
-
color: #FFD700 !important;
|
218 |
-
font-weight: 700 !important;
|
219 |
-
text-shadow: 0 2px 4px rgba(0, 0, 0, 0.3) !important;
|
220 |
-
animation: fadeIn 1s ease-in-out;
|
221 |
-
}
|
222 |
-
|
223 |
-
@keyframes fadeIn {
|
224 |
-
from { opacity: 0; transform: translateY(-10px); }
|
225 |
-
to { opacity: 1; transform: translateY(0); }
|
226 |
-
}
|
227 |
-
|
228 |
-
.gr-row {
|
229 |
-
margin: 20px 0 !important;
|
230 |
-
}
|
231 |
-
|
232 |
-
.gr-column {
|
233 |
-
padding: 15px !important;
|
234 |
-
}
|
235 |
-
|
236 |
-
label {
|
237 |
-
color: #FFD700 !important;
|
238 |
-
font-weight: 600 !important;
|
239 |
-
font-size: 16px !important;
|
240 |
-
margin-bottom: 8px !important;
|
241 |
-
display: flex !important;
|
242 |
-
align-items: center !important;
|
243 |
-
}
|
244 |
-
|
245 |
-
label::before {
|
246 |
-
font-family: "Font Awesome 6 Free";
|
247 |
-
font-weight: 900;
|
248 |
-
margin-right: 8px;
|
249 |
-
}
|
250 |
-
|
251 |
-
.gr-textbox label::before {
|
252 |
-
content: '\\f201';
|
253 |
-
}
|
254 |
-
|
255 |
-
.gr-html label::before {
|
256 |
-
content: '\\f080';
|
257 |
-
}
|
258 |
-
|
259 |
-
.gr-file label::before {
|
260 |
-
content: '\\f019';
|
261 |
-
}
|
262 |
-
|
263 |
-
.economic-question-section {
|
264 |
-
background: rgba(26, 60, 52, 0.9) !important;
|
265 |
-
border-radius: 12px;
|
266 |
-
padding: 25px;
|
267 |
-
margin: 20px 0;
|
268 |
-
box-shadow: 0 6px 16px rgba(0, 0, 0, 0.4);
|
269 |
-
}
|
270 |
-
|
271 |
-
.economic-question-section .gr-textbox {
|
272 |
-
background: rgba(46, 74, 67, 0.8) !important;
|
273 |
-
border: 2px solid #FFD700 !important;
|
274 |
-
box-shadow: 0 3px 10px rgba(255, 215, 0, 0.3) !important;
|
275 |
-
font-size: 18px !important;
|
276 |
-
padding: 15px !important;
|
277 |
-
}
|
278 |
-
|
279 |
-
.economic-question-section .gr-textbox:focus {
|
280 |
-
border-color: #FFD700 !important;
|
281 |
-
box-shadow: 0 0 12px rgba(255, 215, 0, 0.5) !important;
|
282 |
-
}
|
283 |
-
|
284 |
-
.prompt-box {
|
285 |
-
background: rgba(46, 74, 67, 0.6) !important;
|
286 |
-
border: 1px solid #FFD700 !important;
|
287 |
-
border-radius: 8px !important;
|
288 |
-
color: #E0E0E0 !important;
|
289 |
-
font-size: 16px !important;
|
290 |
-
padding: 12px !important;
|
291 |
-
margin-top: 10px !important;
|
292 |
-
}
|
293 |
-
|
294 |
-
.prompt-box label::before {
|
295 |
-
content: '\\f075';
|
296 |
-
}
|
297 |
-
|
298 |
-
.options-section {
|
299 |
-
display: flex;
|
300 |
-
flex-direction: column;
|
301 |
-
gap: 15px;
|
302 |
-
margin-top: 15px;
|
303 |
-
}
|
304 |
-
|
305 |
-
.options-section .gr-dropdown {
|
306 |
-
width: 200px !important;
|
307 |
-
}
|
308 |
-
|
309 |
-
.options-section .gr-dropdown label::before {
|
310 |
-
content: '\\f0c9';
|
311 |
-
}
|
312 |
-
|
313 |
-
.progress-message {
|
314 |
-
color: #FFD700 !important;
|
315 |
-
font-style: italic;
|
316 |
-
margin-bottom: 10px;
|
317 |
}
|
318 |
"""
|
319 |
|
320 |
with gr.Blocks(theme=gr.themes.Base(), css=custom_css) as iface:
|
321 |
-
gr.Markdown("# 📈 Analyse Financière Premium
|
322 |
-
gr.Markdown("Posez une question
|
323 |
|
324 |
count = gr.State(0)
|
325 |
history = gr.State([])
|
326 |
|
327 |
-
with gr.Row(
|
328 |
with gr.Column(scale=2):
|
329 |
-
|
330 |
-
|
331 |
-
|
332 |
-
|
333 |
-
)
|
334 |
-
prompt_display = gr.Textbox(
|
335 |
-
value="Une hausse des taux causerait-elle une récession ?",
|
336 |
-
label="Exemple de Prompt",
|
337 |
-
interactive=False,
|
338 |
-
elem_classes=["prompt-box"]
|
339 |
-
)
|
340 |
-
with gr.Column(scale=1, elem_classes=["options-section"]):
|
341 |
-
mode_selector = gr.Dropdown(
|
342 |
-
choices=["Rapide", "Équilibré", "Précis"],
|
343 |
-
value="Équilibré",
|
344 |
-
label="Mode (longueur et style de réponse)"
|
345 |
-
)
|
346 |
-
detail_mode_selector = gr.Dropdown(
|
347 |
-
choices=["Normal", "Expert"],
|
348 |
-
value="Normal",
|
349 |
-
label="Niveau de détail (simplicité ou technicité)"
|
350 |
-
)
|
351 |
|
352 |
-
|
353 |
-
|
354 |
-
reset_graph_btn = gr.Button("Réinitialiser")
|
355 |
-
download_btn = gr.Button("Télécharger CSV")
|
356 |
|
357 |
with gr.Row():
|
358 |
-
|
359 |
-
|
360 |
-
|
361 |
-
|
362 |
-
|
363 |
-
|
364 |
-
explanation_output_en = gr.Textbox(label="Explication en Anglais")
|
365 |
-
explanation_output_fr = gr.Textbox(label="Explication en Français")
|
366 |
|
367 |
download_file = gr.File(label="Fichier CSV")
|
368 |
|
@@ -371,7 +202,7 @@ def launch_app():
|
|
371 |
analyze_btn.click(
|
372 |
full_analysis,
|
373 |
inputs=[input_text, mode_selector, detail_mode_selector, count, history],
|
374 |
-
outputs=[sentiment_output, explanation_output_en, explanation_output_fr, count, history, sentiment_gauge, progress_message]
|
375 |
)
|
376 |
|
377 |
download_btn.click(
|
@@ -383,4 +214,4 @@ def launch_app():
|
|
383 |
iface.launch(share=True)
|
384 |
|
385 |
if __name__ == "__main__":
|
386 |
-
launch_app()
|
|
|
6 |
import os
|
7 |
import nltk
|
8 |
import asyncio
|
9 |
+
nltk.download('punkt') # CORRECT : 'punkt' !
|
10 |
+
|
11 |
from nltk.tokenize import sent_tokenize
|
12 |
|
13 |
HF_TOKEN = os.getenv("HF_TOKEN")
|
14 |
|
15 |
+
# Chargement modèles
|
16 |
+
translator_to_en = pipeline("translation", model="Helsinki-NLP/opus-mt-mul-en")
|
17 |
+
translator_to_fr = pipeline("translation", model="Helsinki-NLP/opus-mt-en-fr")
|
18 |
+
classifier = pipeline("sentiment-analysis", model="mrm8488/distilroberta-finetuned-financial-news-sentiment-analysis")
|
19 |
+
|
20 |
+
# Fonction traduction segmentée
|
21 |
+
def safe_translate_to_fr(text, max_length=512):
|
22 |
+
sentences = sent_tokenize(text)
|
23 |
+
translated_sentences = []
|
24 |
+
for sentence in sentences:
|
25 |
+
translated = translator_to_fr(sentence, max_length=max_length)[0]['translation_text']
|
26 |
+
translated_sentences.append(translated)
|
27 |
+
return " ".join(translated_sentences)
|
28 |
+
|
29 |
+
# Appel API Zephyr
|
30 |
async def call_zephyr_api(prompt, mode, hf_token=HF_TOKEN):
|
31 |
client = InferenceClient("HuggingFaceH4/zephyr-7b-beta", token=hf_token)
|
32 |
try:
|
|
|
36 |
elif mode == "Équilibré":
|
37 |
max_new_tokens = 100
|
38 |
temperature = 0.5
|
39 |
+
else:
|
40 |
max_new_tokens = 150
|
41 |
temperature = 0.7
|
42 |
response = await asyncio.to_thread(client.text_generation, prompt, max_new_tokens=max_new_tokens, temperature=temperature)
|
43 |
return response
|
44 |
except Exception as e:
|
45 |
+
raise gr.Error(f"❌ Erreur API Hugging Face : {str(e)}")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
|
47 |
+
# Suggestion mode
|
|
|
|
|
|
|
|
|
48 |
def suggest_model(text):
|
49 |
word_count = len(text.split())
|
50 |
if word_count < 50:
|
|
|
54 |
else:
|
55 |
return "Précis"
|
56 |
|
57 |
+
# Jauge sentiment
|
58 |
def create_sentiment_gauge(sentiment, score):
|
59 |
score_percentage = score * 100
|
60 |
+
color = "#A9A9A9"
|
61 |
+
if sentiment.lower() == "positive":
|
|
|
62 |
color = "#2E8B57"
|
63 |
elif sentiment.lower() == "negative":
|
64 |
color = "#DC143C"
|
|
|
|
|
65 |
|
66 |
html = f"""
|
67 |
<div style='width: 100%; max-width: 300px; margin: 10px 0;'>
|
68 |
+
<div style='background-color: #D3D3D3; border-radius: 5px; height: 20px; position: relative;'>
|
69 |
+
<div style='background-color: {color}; width: {score_percentage}%; height: 100%; border-radius: 5px;'></div>
|
70 |
+
<span style='position: absolute; top: 0; left: 50%; transform: translateX(-50%); font-weight: bold;'>{score_percentage:.1f}%</span>
|
|
|
|
|
|
|
|
|
|
|
|
|
71 |
</div>
|
72 |
+
<div style='text-align: center; margin-top: 5px;'>Sentiment : {sentiment}</div>
|
73 |
</div>
|
74 |
"""
|
75 |
return html
|
76 |
|
77 |
+
# Analyse principale
|
78 |
async def full_analysis(text, mode, detail_mode, count, history):
|
79 |
if not text:
|
80 |
+
yield "Entrez une phrase.", "", "", "", 0, history, "", "Aucune analyse effectuée."
|
81 |
return
|
82 |
|
83 |
+
yield "Analyse en cours... (Étape 1 : Détection de la langue)", "", "", "", count, history, "", "Détection de la langue"
|
|
|
84 |
|
85 |
try:
|
86 |
lang = detect(text)
|
|
|
92 |
else:
|
93 |
text_en = text
|
94 |
|
95 |
+
yield "Analyse en cours... (Étape 2 : Analyse du sentiment)", "", "", "", count, history, "", "Analyse du sentiment"
|
96 |
|
|
|
97 |
result = await asyncio.to_thread(classifier, text_en)
|
98 |
result = result[0]
|
99 |
sentiment_output = f"Sentiment prédictif : {result['label']} (Score: {result['score']:.2f})"
|
100 |
sentiment_gauge = create_sentiment_gauge(result['label'], result['score'])
|
101 |
|
102 |
+
yield "Analyse en cours... (Étape 3 : Explication IA)", "", "", "", count, history, "", "Génération de l'explication"
|
103 |
|
|
|
104 |
explanation_prompt = f"""<|system|>
|
105 |
You are a professional financial analyst AI with expertise in economic forecasting.
|
106 |
</s>
|
|
|
109 |
|
110 |
The predicted sentiment for this event is: {result['label'].lower()}.
|
111 |
|
112 |
+
Assume the event happens. Explain why this event would likely have a {result['label'].lower()} economic impact.
|
113 |
</s>
|
114 |
<|assistant|>"""
|
115 |
explanation_en = await call_zephyr_api(explanation_prompt, mode)
|
116 |
|
117 |
+
yield "Analyse en cours... (Étape 4 : Traduction en français)", "", "", "", count, history, "", "Traduction en français"
|
118 |
|
119 |
+
explanation_fr = safe_translate_to_fr(explanation_en)
|
|
|
120 |
|
121 |
count += 1
|
122 |
history.append({
|
|
|
127 |
"Explication_FR": explanation_fr
|
128 |
})
|
129 |
|
130 |
+
yield sentiment_output, text, explanation_en, explanation_fr, count, history, sentiment_gauge, "✅ Analyse terminée."
|
131 |
|
132 |
+
# Historique CSV
|
133 |
def download_history(history):
|
134 |
if not history:
|
135 |
return None
|
|
|
138 |
df.to_csv(file_path, index=False)
|
139 |
return file_path
|
140 |
|
141 |
+
# Lancement Gradio
|
142 |
def launch_app():
|
143 |
custom_css = """
|
144 |
+
/* Ici tout ton CSS personnalisé pour style de fond, couleurs, boutons */
|
|
|
|
|
145 |
body {
|
146 |
+
background: linear-gradient(135deg, #0A1D37 0%, #1A3C34 100%);
|
|
|
|
|
|
|
|
|
|
|
|
|
147 |
font-family: 'Inter', sans-serif;
|
148 |
+
color: #E0E0E0;
|
149 |
padding: 20px;
|
150 |
}
|
|
|
151 |
.gr-box {
|
152 |
+
background: #2A4A43 !important;
|
|
|
153 |
border: 1px solid #FFD700 !important;
|
154 |
+
border-radius: 12px !important;
|
155 |
+
padding: 20px !important;
|
156 |
+
box-shadow: 0px 4px 12px rgba(255, 215, 0, 0.4);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
157 |
}
|
|
|
158 |
.gr-button {
|
159 |
+
background: linear-gradient(90deg, #FFD700, #D4AF37);
|
160 |
+
color: #0A1D37;
|
161 |
+
font-weight: bold;
|
162 |
+
border: none;
|
163 |
+
border-radius: 8px;
|
164 |
+
padding: 12px 24px;
|
165 |
+
transition: transform 0.2s;
|
|
|
|
|
166 |
}
|
|
|
167 |
.gr-button:hover {
|
168 |
+
transform: translateY(-2px);
|
169 |
+
box-shadow: 0 6px 12px rgba(255, 215, 0, 0.5);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
170 |
}
|
171 |
"""
|
172 |
|
173 |
with gr.Blocks(theme=gr.themes.Base(), css=custom_css) as iface:
|
174 |
+
gr.Markdown("# 📈 Analyse Financière Premium avec IA")
|
175 |
+
gr.Markdown("**Posez une question économique.** L'IA analyse et explique l'impact.")
|
176 |
|
177 |
count = gr.State(0)
|
178 |
history = gr.State([])
|
179 |
|
180 |
+
with gr.Row():
|
181 |
with gr.Column(scale=2):
|
182 |
+
input_text = gr.Textbox(lines=4, label="Votre question économique")
|
183 |
+
with gr.Column(scale=1):
|
184 |
+
mode_selector = gr.Dropdown(choices=["Rapide", "Équilibré", "Précis"], value="Équilibré", label="Mode de réponse")
|
185 |
+
detail_mode_selector = gr.Dropdown(choices=["Normal", "Expert"], value="Normal", label="Niveau de détail")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
186 |
|
187 |
+
analyze_btn = gr.Button("Analyser")
|
188 |
+
download_btn = gr.Button("Télécharger l'historique")
|
|
|
|
|
189 |
|
190 |
with gr.Row():
|
191 |
+
sentiment_output = gr.Textbox(label="Sentiment prédictif")
|
192 |
+
displayed_prompt = gr.Textbox(label="Texte utilisé (anglais)", interactive=False)
|
193 |
+
explanation_output_en = gr.Textbox(label="Explication en anglais")
|
194 |
+
explanation_output_fr = gr.Textbox(label="Explication en français")
|
195 |
+
sentiment_gauge = gr.HTML()
|
196 |
+
progress_message = gr.Textbox(label="Progression", interactive=False)
|
|
|
|
|
197 |
|
198 |
download_file = gr.File(label="Fichier CSV")
|
199 |
|
|
|
202 |
analyze_btn.click(
|
203 |
full_analysis,
|
204 |
inputs=[input_text, mode_selector, detail_mode_selector, count, history],
|
205 |
+
outputs=[sentiment_output, displayed_prompt, explanation_output_en, explanation_output_fr, count, history, sentiment_gauge, progress_message]
|
206 |
)
|
207 |
|
208 |
download_btn.click(
|
|
|
214 |
iface.launch(share=True)
|
215 |
|
216 |
if __name__ == "__main__":
|
217 |
+
launch_app()
|