[refactor] try mount the gradio app within a fastapi to prepare logging the session id

app.py

@@ -5,6 +5,11 @@ import sys
 import logging
 from typing import Callable
 
+from fastapi import FastAPI, File, UploadFile, Request
+from fastapi.responses import HTMLResponse, RedirectResponse
+from fastapi.staticfiles import StaticFiles
+from fastapi.templating import Jinja2Templates
+
 import cv2
 import gradio as gr
 import nh3
@@ -21,6 +26,51 @@ from utils.utils import (DEFAULT_IM_END_TOKEN, DEFAULT_IM_START_TOKEN,
                          DEFAULT_IMAGE_TOKEN, IMAGE_TOKEN_INDEX)
 
 
+CUSTOM_GRADIO_PATH = "/gradio"
+app = FastAPI()
+os.makedirs("static", exist_ok=True)
+app.mount("/static", StaticFiles(directory="static"), name="static")
+templates = Jinja2Templates(directory="templates")
+
+
+def get_cleaned_input(input_str):
+    input_str = nh3.clean(
+        input_str,
+        tags={
+            "a",
+            "abbr",
+            "acronym",
+            "b",
+            "blockquote",
+            "code",
+            "em",
+            "i",
+            "li",
+            "ol",
+            "strong",
+            "ul",
+        },
+        attributes={
+            "a": {"href", "title"},
+            "abbr": {"title"},
+            "acronym": {"title"},
+        },
+        url_schemes={"http", "https", "mailto"},
+        link_rel=None,
+    )
+    return input_str
+
+
+@app.get("/", response_class=HTMLResponse)
+async def home(request: Request):
+    logging.info(f"Request raw: {request}.")
+    clean_request = get_cleaned_input(str(request))
+    logging.info(f"clean_request: {request}.")
+    return templates.TemplateResponse(
+        "home.html", {"clean_request": clean_request}
+    )
+
+
 # Gradio
 examples = [
     [
@@ -214,30 +264,7 @@ def get_inference_model_by_args(args_to_parse):
     def inference(input_str, input_image):
         ## filter out special chars
 
-        input_str = nh3.clean(
-            input_str,
-            tags={
-                "a",
-                "abbr",
-                "acronym",
-                "b",
-                "blockquote",
-                "code",
-                "em",
-                "i",
-                "li",
-                "ol",
-                "strong",
-                "ul",
-            },
-            attributes={
-                "a": {"href", "title"},
-                "abbr": {"title"},
-                "acronym": {"title"},
-            },
-            url_schemes={"http", "https", "mailto"},
-            link_rel=None,
-        )
+        input_str = get_cleaned_input(input_str)
         logging.info(f"input_str type: {type(input_str)}, input_image type: {type(input_image)}.")
         logging.info(f"input_str: {input_str}.")
 
@@ -334,12 +361,10 @@ def get_inference_model_by_args(args_to_parse):
     return inference
 
 
-def server_runner(
-        fn_inference: Callable,
-        debug: bool = False,
-        server_name: str = "0.0.0.0"
+def get_gradio_interface(
+        fn_inference: Callable
     ):
-    inference_app = gr.Interface(
+    return gr.Interface(
         fn_inference,
         inputs=[
             gr.Textbox(lines=1, placeholder=None, label="Text Instruction"),
@@ -356,19 +381,9 @@ def server_runner(
         allow_flagging="auto",
     )
 
-    inference_app.queue()
-    inference_app.launch(
-        share=False,
-        debug=debug,
-        server_name=server_name
-    )
-
 
 if __name__ == '__main__':
     args = parse_args(sys.argv[1:])
     inference_fn = get_inference_model_by_args(args)
-    server_runner(
-        inference_fn,
-        debug=True,
-        server_name="0.0.0.0"
-    )
+    io = get_gradio_interface(inference_fn)
+    app = gr.mount_gradio_app(app, io, path=CUSTOM_GRADIO_PATH)