Update app.py

app.py

@@ -1,77 +1,60 @@
+import gradio as gr
 from transformers import pipeline
 from langdetect import detect
-import gradio as gr
 import pandas as pd
+import textstat
 
-# Chargement modèle de sentiment
+# Chargement du modèle de sentiment
 classifier = pipeline(
     "sentiment-analysis",
     model="mrm8488/distilroberta-finetuned-financial-news-sentiment-analysis"
 )
 
 # Modèles de traduction
-translator_to_en = pipeline(
-    "translation",
-    model="Helsinki-NLP/opus-mt-mul-en"
-)
-translator_to_fr = pipeline(
-    "translation",
-    model="Helsinki-NLP/opus-mt-en-fr"
-)
-
-# Modèles disponibles
-MODEL_OPTIONS = {
-    "Flan-T5 Small (rapide)" : "google/flan-t5-small",
-    "Flan-T5 Base (équilibré)" : "google/flan-t5-base",
-    "Flan-T5 Large (précis)" : "google/flan-t5-large",
-}
+translator_to_en = pipeline("translation", model="Helsinki-NLP/opus-mt-mul-en")
+translator_to_fr = pipeline("translation", model="Helsinki-NLP/opus-mt-en-fr")
 
-# Fonction pour charger un modèle d'explication
-def load_explainer(model_choice):
-    model_name = MODEL_OPTIONS.get(model_choice, "google/flan-t5-small")
-    return pipeline("text2text-generation", model=model_name)
+# Modèle explicatif CPU-friendly
+explainer = pipeline("text2text-generation", model="facebook/blenderbot-1B-distill")
 
 # Fonction pour suggérer le meilleur modèle
 def suggest_model(text):
     word_count = len(text.split())
     if word_count < 50:
-        suggestion = "Flan-T5 Small (rapide)"
+        return "Rapide"
     elif word_count <= 200:
-        suggestion = "Flan-T5 Base (équilibré)"
+        return "Équilibré"
     else:
-        suggestion = "Flan-T5 Large (précis)"
-    return suggestion
+        return "Précis"
 
-# Fonction complète d'analyse + explication
-def full_analysis(text, model_choice, count, history):
+# Fonction d'analyse
+def full_analysis(text, mode, detail_mode, count, history):
     if not text:
-        return "Entrez une phrase.", "", "", count, history, None
+        return "Entrez une phrase.", "", "", 0, history, None
 
-    # Charger modèle sélectionné
-    explainer = load_explainer(model_choice)
-
-    # Détection de langue
     try:
         lang = detect(text)
     except:
         lang = "unknown"
 
-    # Traduction si besoin
     if lang != "en":
         text = translator_to_en(text, max_length=512)[0]['translation_text']
 
-    # Analyse du sentiment
     result = classifier(text)[0]
     sentiment_output = f"Sentiment : {result['label']} (Score: {result['score']:.2f})"
 
-    # Prompt amélioré pour meilleure explication
-    prompt = f"""Given the following financial news: \"{text}\", explain in natural, clear and detailed language why the sentiment is {result['label'].lower()}. Focus on the main financial aspects and keep it concise."""
-    explanation_en = explainer(prompt, max_length=150)[0]['generated_text']
+    prompt = f"""
+    You are a financial analyst AI.
+    Based on the following financial news: \"{text}\",
+    explain clearly why the sentiment is {result['label'].lower()}.
+    {"Write a concise paragraph." if detail_mode == "Normal" else "Write a detailed explanation over multiple paragraphs."}
+    """
 
-    # Traduction française
+    explanation_en = explainer(prompt, max_length=300 if detail_mode == "Expert" else 150)[0]['generated_text']
     explanation_fr = translator_to_fr(explanation_en, max_length=512)[0]['translation_text']
 
-    # Mise à jour historique
+    clarity_score = textstat.flesch_reading_ease(explanation_en)
+
     count += 1
     history.append({
         "Texte": text,
@@ -79,12 +62,12 @@ def full_analysis(text, model_choice, count, history):
         "Score": f"{result['score']:.2f}",
         "Explication_EN": explanation_en,
         "Explication_FR": explanation_fr,
-        "Modèle utilisé": model_choice
+        "Clarté": f"{clarity_score:.1f}"
     })
 
-    return sentiment_output, explanation_en, explanation_fr, count, history, None
+    return sentiment_output, explanation_en, explanation_fr, clarity_score, count, history, None
 
-# Fonction pour générer historique CSV
+# Fonction pour télécharger historique CSV
 def download_history(history):
     if not history:
         return None
@@ -93,76 +76,61 @@ def download_history(history):
     df.to_csv(file_path, index=False)
     return file_path
 
-# Interface
-with gr.Blocks() as iface:
-    gr.Markdown("# 📈 Analyse de Sentiment Financier Multilingue + Raisonnement 🌍")
-    gr.Markdown("Entrez une actualité financière. L'analyse détecte automatiquement le sentiment et génère une explication **bilingue**.\nChoisissez ou laissez l'IA vous suggérer le meilleur modèle selon la longueur de votre texte.")
+# Interface Gradio
+def launch_app():
+    with gr.Blocks(theme=gr.themes.Base(), css="body {background-color: #0D1117; color: white;} .gr-button {background-color: #161B22; border: 1px solid #30363D;}") as iface:
+
+        gr.Markdown("# 📈 Analyse Financière Premium + Explication IA", elem_id="title")
+        gr.Markdown("Entrez une actualité financière. L'IA analyse et explique en anglais/français. Choisissez votre mode d'explication.")
+
+        count = gr.State(0)
+        history = gr.State([])
 
-    count = gr.State(0)
-    history = gr.State([])
+        with gr.Row():
+            input_text = gr.Textbox(lines=4, placeholder="Entrez une actualité ici...", label="Texte à analyser")
 
-    with gr.Row():
-        input_text = gr.Textbox(lines=4, placeholder="Entrez votre actualité ici...")
+        with gr.Row():
+            mode_selector = gr.Dropdown(
+                choices=["Rapide", "Équilibré", "Précis"],
+                value="Équilibré",
+                label="Mode recommandé selon la taille"
+            )
+            detail_mode_selector = gr.Dropdown(
+                choices=["Normal", "Expert"],
+                value="Normal",
+                label="Niveau de détail"
+            )
 
-    with gr.Row():
-        model_selector = gr.Dropdown(
-            choices=list(MODEL_OPTIONS.keys()),
-            label="Choisissez le modèle d'explication",
-            value="Flan-T5 Base (équilibré)"
+        analyze_btn = gr.Button("Analyser")
+        download_btn = gr.Button("Télécharger CSV")
+
+        with gr.Row():
+            sentiment_output = gr.Textbox(label="Résultat du Sentiment")
+
+        with gr.Row():
+            with gr.Column():
+                explanation_output_en = gr.Textbox(label="Explication en Anglais")
+            with gr.Column():
+                explanation_output_fr = gr.Textbox(label="Explication en Français")
+
+        clarity_score_output = gr.Textbox(label="Score de Clarté (Flesch Reading Ease)")
+        download_file = gr.File(label="Fichier CSV")
+
+        input_text.change(lambda t: gr.update(value=suggest_model(t)), inputs=[input_text], outputs=[mode_selector])
+
+        analyze_btn.click(
+            full_analysis,
+            inputs=[input_text, mode_selector, detail_mode_selector, count, history],
+            outputs=[sentiment_output, explanation_output_en, explanation_output_fr, clarity_score_output, count, history, download_file]
         )
-        model_suggestion = gr.Markdown("")
-
-    analyze_btn = gr.Button("Analyser")
-    download_btn = gr.Button("Télécharger l'historique CSV")
-
-    loading_text = gr.Markdown("⏳ Analyse en cours...", visible=False)
-
-    with gr.Row():
-        sentiment_output = gr.Textbox(label="Résultat du sentiment")
-
-    with gr.Row():
-        with gr.Column():
-            explanation_output_en = gr.Textbox(label="Explication en Anglais")
-        with gr.Column():
-            explanation_output_fr = gr.Textbox(label="Explication en Français")
-
-    with gr.Row():
-        analysis_count_display = gr.Textbox(label="Nombre total d'analyses", interactive=False)
-
-    download_file = gr.File(label="Téléchargement du CSV")
-
-    # Suggestions dynamiques du modèle
-    input_text.change(
-        lambda text: gr.update(value=suggest_model(text)), 
-        inputs=[input_text],
-        outputs=[model_selector]
-    ).then(
-        lambda text: gr.update(value=f"💡 Suggestion IA : {suggest_model(text)}"),
-        inputs=[input_text],
-        outputs=[model_suggestion]
-    )
-
-    # Actions boutons
-    analyze_btn.click(
-        lambda: gr.update(visible=True),
-        outputs=[loading_text]
-    ).then(
-        full_analysis,
-        inputs=[input_text, model_selector, count, history],
-        outputs=[sentiment_output, explanation_output_en, explanation_output_fr, count, history, download_file]
-    ).then(
-        lambda c: gr.update(value=f"{c} analyses réalisées"),
-        inputs=[count],
-        outputs=[analysis_count_display]
-    ).then(
-        lambda: gr.update(visible=False),
-        outputs=[loading_text]
-    )
-
-    download_btn.click(
-        download_history,
-        inputs=[history],
-        outputs=[download_file]
-    )
-
-iface.launch()
+
+        download_btn.click(
+            download_history,
+            inputs=[history],
+            outputs=[download_file]
+        )
+
+    iface.launch()
+
+if __name__ == "__main__":
+    launch_app()