Update app.py

app.py

@@ -5,6 +5,8 @@ import spaces
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 import os
 from threading import Thread
+from fastapi import FastAPI
+import uvicorn
 
 import pymupdf
 import docx
@@ -33,15 +35,11 @@ h1 {
 }
 """
 
-
 tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, trust_remote_code=True)
 
-
-
 def extract_text(path):
     return open(path, 'r').read()
 
-
 def extract_pdf(path):
     doc = pymupdf.open(path)
     text = ""
@@ -49,7 +47,6 @@ def extract_pdf(path):
         text += page.get_text()
     return text
 
-
 def extract_docx(path):
     doc = docx.Document(path)
     data = []
@@ -58,7 +55,6 @@ def extract_docx(path):
     content = '\n\n'.join(data)
     return content
 
-
 def extract_pptx(path):
     prs = Presentation(path)
     text = ""
@@ -68,7 +64,6 @@ def extract_pptx(path):
                 text += shape.text + "\n"
     return text
 
-
 def mode_load(path):
     choice = ""
     file_type = path.split(".")[-1]
@@ -85,20 +80,15 @@ def mode_load(path):
         choice = "doc"
         print(content[:100])
         return choice, content[:5000]
-
-
     elif file_type in ["png", "jpg", "jpeg", "bmp", "tiff", "webp"]:
         content = Image.open(path).convert('RGB')
         choice = "image"
         return choice, content
-
     else:
         raise gr.Error("Oops, unsupported files.")
 
-
 @spaces.GPU()
 def stream_chat(message, history: list, temperature: float, max_length: int, top_p: float, top_k: int, penalty: float):
-    
     model = AutoModelForCausalLM.from_pretrained(
         MODEL_ID,
         torch_dtype=torch.bfloat16,
@@ -136,7 +126,6 @@ def stream_chat(message, history: list, temperature: float, max_length: int, top
                 choice = ""
                 conversation.append({"role": "user", "image": "", "content": message['text']})
 
-
             if choice == "image":
                 conversation.append({"role": "user", "image": contents, "content": message['text']})
             elif choice == "doc":
@@ -168,18 +157,11 @@ def stream_chat(message, history: list, temperature: float, max_length: int, top
             buffer += new_text
             yield buffer
 
-
-chatbot = gr.Chatbot(
-    #rtl=True,
-)
+chatbot = gr.Chatbot()
 chat_input = gr.MultimodalTextbox(
     interactive=True,
     placeholder="Enter message or upload a file ...",
     show_label=False,
-    #rtl=True,
-
-
-
 )
 
 EXAMPLES = [
@@ -189,14 +171,21 @@ EXAMPLES = [
     [{"text": "Quiero armar un JSON, solo el JSON sin texto, que contenga los datos de la primera mitad de la tabla de la imagen (las primeras 10 jurisdicciones 901-910). Ten en cuenta que los valores numéricos son decimales de cuatro dígitos. La tabla contiene las siguientes columnas: Codigo, Nombre, Fecha Inicio, Fecha Cese, Coeficiente Ingresos, Coeficiente Gastos y Coeficiente Unificado. La tabla puede contener valores vacíos, en ese caso dejarlos como null. Cada fila de la tabla representa una jurisdicción con sus respectivos valores.", }]
 ]
 
+app = FastAPI()
+
+def test():
+    return "Función test llamada con éxito"
+
+@app.get("/test")
+def call_test():
+    return {"message": test()}
+
 with gr.Blocks(css=CSS, theme="soft", fill_height=True) as demo:
     gr.HTML(TITLE)
     gr.HTML(DESCRIPTION)
     gr.ChatInterface(
         fn=stream_chat,
         multimodal=True,
-
-
         textbox=chat_input,
         chatbot=chatbot,
         fill_height=True,
@@ -247,5 +236,17 @@ with gr.Blocks(css=CSS, theme="soft", fill_height=True) as demo:
     gr.Examples(EXAMPLES, [chat_input])
 
 if __name__ == "__main__":
-    
-    demo.queue(api_open=False).launch(show_api=False, share=False, )#server_name="0.0.0.0", )
+    def run_fastapi():
+        uvicorn.run(app, host="0.0.0.0", port=8000)
+
+    def run_gradio():
+        demo.queue(api_open=False).launch(show_api=False, share=False)
+
+    fastapi_thread = Thread(target=run_fastapi)
+    fastapi_thread.start()
+
+    gradio_thread = Thread(target=run_gradio)
+    gradio_thread.start()
+
+    fastapi_thread.join()
+    gradio_thread.join()