feat: add virtual assistant with speech and text interaction

app.py

@@ -0,0 +1,168 @@
+import os
+import sys
+import tempfile
+import time
+import urllib.parse
+import webbrowser
+import gradio as gr
+from dotenv import load_dotenv
+
+# Patch para compatibilidade com Python 3.13+
+if sys.version_info >= (3, 13):
+    import types
+    sys.modules['aifc'] = types.ModuleType('aifc')
+    sys.modules['audioop'] = types.ModuleType('audioop')
+
+# Lazy imports para o Whisper e gTTS
+whisper_model = None
+
+def get_whisper_model():
+    global whisper_model
+    if whisper_model is None:
+        import whisper
+        print("Carregando modelo Whisper (Local e Gratuito)...")
+        whisper_model = whisper.load_model("base")
+    return whisper_model
+
+def get_chatgpt_response(text, api_key):
+    if not api_key or api_key == "sua_chave_api_aqui":
+        return "Erro: Chave API da OpenAI não configurada no arquivo .env."
+    
+    try:
+        import openai
+        client = openai.OpenAI(api_key=api_key)
+        response = client.chat.completions.create(
+            model="gpt-3.5-turbo",
+            messages=[
+                {"role": "system", "content": "Você é um assistente virtual útil e conciso. Responda em português."},
+                {"role": "user", "content": text}
+            ]
+        )
+        return response.choices[0].message.content
+    except Exception as e:
+        return f"Erro na IA: {str(e)}"
+
+def try_local_commands(text):
+    s = (text or "").lower()
+    if "wikipedia" in s:
+        query = s.replace("wikipedia", "").replace("pesquisar", "").strip()
+        if not query:
+            return "O que devo pesquisar na Wikipedia?"
+        url = "https://pt.wikipedia.org/wiki/Special:Search?search=" + urllib.parse.quote_plus(query)
+        webbrowser.open(url)
+        return f"Pesquisando '{query}' na Wikipedia."
+        
+    if "youtube" in s or "vídeo" in s or "video" in s:
+        query = s.replace("youtube", "").replace("vídeo", "").replace("video", "").replace("pesquisar", "").strip()
+        if not query:
+            return "O que devo pesquisar no YouTube?"
+        url = "https://www.youtube.com/results?search_query=" + urllib.parse.quote_plus(query)
+        webbrowser.open(url)
+        return f"Pesquisando '{query}' no YouTube."
+        
+    if "farmácia" in s or "farmacia" in s:
+        webbrowser.open("https://www.google.com/maps/search/farmacia+perto+de+mim")
+        return "Abrindo mapa de farmácias próximas."
+
+    return None
+
+def text_to_speech(text):
+    try:
+        from gtts import gTTS
+        tts = gTTS(text=text, lang='pt')
+        temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp3")
+        tts.save(temp_file.name)
+        return temp_file.name
+    except Exception as e:
+        print(f"Erro TTS: {e}")
+        return None
+
+def process_interaction(audio_path, text_input, history, api_key):
+    # Inicializar histórico se for None
+    if history is None:
+        history = []
+        
+    # Determinar a entrada (áudio ou texto)
+    input_text = ""
+    
+    try:
+        if audio_path:
+            print(f"Processando áudio de: {audio_path}")
+            model = get_whisper_model()
+            result = model.transcribe(audio_path, language="pt", fp16=False)
+            input_text = result["text"].strip()
+            print(f"Transcrição Whisper: {input_text}")
+        elif text_input:
+            input_text = text_input
+            print(f"Entrada de texto: {input_text}")
+        
+        if not input_text:
+            return history, "", gr.update()
+
+        # Processar comando local primeiro
+        response_text = try_local_commands(input_text)
+        
+        # Se não for comando local, usar IA
+        if response_text is None:
+            print("Usando IA para responder...")
+            response_text = get_chatgpt_response(input_text, api_key)
+        
+        print(f"Resposta: {response_text}")
+
+        # Gerar áudio
+        audio_response = text_to_speech(response_text)
+        
+        # Atualizar histórico (Gradio 4+ prefere lista de dicts ou lista de listas)
+        history.append({"role": "user", "content": input_text})
+        history.append({"role": "assistant", "content": response_text})
+        
+        return history, "", audio_response if audio_response else gr.update()
+
+    except Exception as e:
+        error_msg = f"Erro no processamento: {str(e)}"
+        print(error_msg)
+        history.append({"role": "user", "content": input_text if input_text else "???"})
+        history.append({"role": "assistant", "content": error_msg})
+        return history, "", gr.update()
+
+def main():
+    load_dotenv()
+    api_key = os.getenv("OPENAI_API_KEY", "")
+    
+    with gr.Blocks(title="Assistente Virtual IA") as demo:
+        gr.Markdown("# 🤖 Assistente Virtual com IA")
+        gr.Markdown("Fale com o assistente ou digite um comando. Ele pode pesquisar no Wikipedia, YouTube ou conversar via ChatGPT.")
+        
+        with gr.Row():
+            with gr.Column(scale=2):
+                chatbot = gr.Chatbot(label="Conversa")
+                audio_output = gr.Audio(label="Resposta em Áudio", autoplay=True)
+            
+            with gr.Column(scale=1):
+                audio_input = gr.Audio(label="Fale aqui", type="filepath")
+                text_input = gr.Textbox(label="Ou digite aqui", placeholder="Ex: Pesquisar Wikipedia sobre Python")
+                btn_send = gr.Button("Enviar", variant="primary")
+                btn_clear = gr.Button("Limpar Conversa")
+
+        # Estado para a chave API (pega do .env inicialmente)
+        api_key_state = gr.State(value=api_key)
+
+        # Eventos
+        btn_send.click(
+            process_interaction, 
+            inputs=[audio_input, text_input, chatbot, api_key_state], 
+            outputs=[chatbot, text_input, audio_output]
+        )
+        
+        text_input.submit(
+            process_interaction, 
+            inputs=[audio_input, text_input, chatbot, api_key_state], 
+            outputs=[chatbot, text_input, audio_output]
+        )
+
+        btn_clear.click(lambda: ([], "", gr.update(value=None)), None, [chatbot, text_input, audio_output])
+
+    demo.launch(share=True)
+
+if __name__ == "__main__":
+    main()

requirements.txt

@@ -0,0 +1,14 @@
+pyttsx3==2.90
+python-dotenv==1.0.1
+pytest==8.3.3
+jupytext==1.16.2
+openai>=1.0.0
+openai-whisper
+gTTS
+pygame
+pydub
+numpy
+requests
+sounddevice
+scipy
+gradio