Compatibilidad Gradio 3.0

app.py

@@ -1,69 +1,75 @@
 import gradio as gr
 from transformers import pipeline
-from PyPDF2 import PdfReader
+import PyPDF2
 
-# Mostrar la versión de Gradio
-print(f"Versión de Gradio instalada: {gr.__version__}")
+# Cargar el modelo de análisis de sentimientos
+sentiment_analysis = pipeline("sentiment-analysis", model="nlptown/bert-base-multilingual-uncased-sentiment")
 
-# Cargar el modelo de DistilBETO
-sentiment_analysis = pipeline("sentiment-analysis", 
-                              model="nlptown/bert-base-multilingual-uncased-sentiment")
-
-# Función para analizar texto
-def analyze_text_sentiment(text):
-    result = sentiment_analysis(text)[0]
-    score = result['score']
+def analyze_text(input_text):
+    result = sentiment_analysis(input_text)[0]
     label = result['label']
-    
-    # Convertir la etiqueta de sentimiento en un valor para el slider y su literal
+    score = round(result['score'] * 100, 2)  # Convertir a porcentaje
+
+    # Definir la categoría de sentimiento según la etiqueta
     if label == "1 star":
-        slider_value = 0
-        literal = "Muy negativo"
+        sentiment_label = "Muy negativo"
     elif label == "2 stars":
-        slider_value = 25
-        literal = "Negativo"
+        sentiment_label = "Negativo"
     elif label == "3 stars":
-        slider_value = 50
-        literal = "Neutro"
+        sentiment_label = "Neutro"
     elif label == "4 stars":
-        slider_value = 75
-        literal = "Positivo"
-    elif label == "5 stars":
-        slider_value = 100
-        literal = "Muy positivo"
+        sentiment_label = "Positivo"
+    else:
+        sentiment_label = "Muy positivo"
 
-    # Añadir el nivel de confianza al literal
-    confidence = round(score * 100, 2)
-    literal_with_confidence = f"{literal} (Confianza: {confidence}%)"
+    # Devolver el resultado con el nivel de confianza
+    return f"{sentiment_label} ({score}%)", score
 
-    return slider_value, literal_with_confidence
-
-# Función para extraer texto de PDF y analizar
-def analyze_pdf_sentiment(pdf_file):
-    pdf_reader = PdfReader(pdf_file.name)
+def analyze_pdf(pdf_file):
+    pdf_reader = PyPDF2.PdfReader(pdf_file)
     text = ""
     for page in pdf_reader.pages:
         text += page.extract_text()
-    
-    return analyze_text_sentiment(text)
+
+    return analyze_text(text)
 
 # Interfaz de Gradio
-with gr.Blocks() as demo:
-    with gr.Tabs():
-        with gr.Tab("Texto"):
-            text_input = gr.Textbox(label="Ingrese su texto aquí", lines=5)
-            analyze_button = gr.Button("Analizar")
-            slider_output = gr.Slider(label="Nivel de Sentimiento", minimum=0, maximum=100, step=1, interactive=False)
-            label_output = gr.Label()
+text_input = gr.Textbox(label="Ingrese texto para análisis")
+pdf_input = gr.File(label="Subir PDF para análisis", file_types=[".pdf"])
+
+text_output = gr.Label(label="Resultado del análisis")
+slider_output = gr.Slider(label="Nivel de sentimiento", minimum=0, maximum=100, interactive=False)
+
+text_button = gr.Button("Analizar texto")
+pdf_button = gr.Button("Analizar PDF")
+
+def text_analysis(input_text):
+    sentiment, score = analyze_text(input_text)
+    return sentiment, score
+
+def pdf_analysis(pdf_file):
+    sentiment, score = analyze_pdf(pdf_file)
+    return sentiment, score
+
+with gr.Blocks() as app:
+    gr.Markdown("## Análisis de Sentimientos con DistilBETO")
+    
+    with gr.Column():
+        gr.Markdown("### Análisis de Texto")
+        text_input.render()
+        text_button.render()
+        text_output.render()
+        slider_output.render()
 
-            analyze_button.click(analyze_text_sentiment, inputs=text_input, outputs=[slider_output, label_output])
+        text_button.click(text_analysis, inputs=text_input, outputs=[text_output, slider_output])
 
-        with gr.Tab("PDF"):
-            pdf_input = gr.File(label="Subir PDF", file_types=[".pdf"])
-            analyze_button = gr.Button("Analizar")
-            slider_output = gr.Slider(label="Nivel de Sentimiento", minimum=0, maximum=100, step=1, interactive=False)
-            label_output = gr.Label()
+        gr.Markdown("### Análisis de PDF")
+        pdf_input.render()
+        pdf_button.render()
+        text_output.render()
+        slider_output.render()
 
-            analyze_button.click(analyze_pdf_sentiment, inputs=pdf_input, outputs=[slider_output, label_output])
+        pdf_button.click(pdf_analysis, inputs=pdf_input, outputs=[text_output, slider_output])
 
-demo.launch()
+# Ejecutar la aplicación
+app.launch()

requirements.txt

@@ -1,7 +1,4 @@
-gradio==3.44.4
+gradio>=3.0
 transformers
 torch
-PyPDF2
-fastapi==0.100.0
-httpcore==0.17.3
-starlette==0.27.0
+PyPDF2