feat(llm): timely execute vid embedding & enhance some HTTP logic

hugegraph-llm/requirements.txt

@@ -14,4 +14,4 @@ python-dotenv>=1.0.1
 pyarrow~=17.0.0 # TODO: a temporary dependency for pandas, figure out why ImportError
 pandas~=2.2.2
 openpyxl~=3.1.5
-pydantic-settings~=2.6.1
+pydantic-settings~=2.6.1

hugegraph-llm/src/hugegraph_llm/demo/rag_demo/app.py

@@ -17,6 +17,8 @@
 
 
 import argparse
+import asyncio
+from contextlib import asynccontextmanager
 
 import gradio as gr
 import uvicorn
@@ -40,10 +42,10 @@
 from hugegraph_llm.demo.rag_demo.vector_graph_block import create_vector_graph_block
 from hugegraph_llm.resources.demo.css import CSS
 from hugegraph_llm.utils.log import log
+from hugegraph_llm.utils.graph_index_utils import fit_vid_index
 
 sec = HTTPBearer()
 
-
 def authenticate(credentials: HTTPAuthorizationCredentials = Depends(sec)):
     correct_token = admin_settings.user_token
     if credentials.credentials != correct_token:
@@ -55,6 +57,29 @@ def authenticate(credentials: HTTPAuthorizationCredentials = Depends(sec)):
             headers={"WWW-Authenticate": "Bearer"},
         )
 
+async def schedule_fit_vid_index():
+    try:
+        while True:
+            log.info("Executing fit_vid_index function...")
+            await asyncio.to_thread(fit_vid_index)
+            log.info("fit_vid_index function executed successfully.")
+            await asyncio.sleep(3600)
+    except Exception as e: #pylint: disable=W0718
+        log.error(e)
+
+@asynccontextmanager
+async def lifespan(app: FastAPI): #pylint: disable=W0621
+    log.info("Starting periodic task...")
+    task = asyncio.create_task(schedule_fit_vid_index())
+
+    yield
+
+    log.info("Stopping periodic task...")
+    task.cancel()
+    try:
+        await task
+    except asyncio.CancelledError:
+        log.info("Periodic task has been cancelled.")
 
 # pylint: disable=C0301
 def init_rag_ui() -> gr.Interface:
@@ -158,7 +183,7 @@ def refresh_ui_config_prompt() -> tuple:
     parser.add_argument("--host", type=str, default="0.0.0.0", help="host")
     parser.add_argument("--port", type=int, default=8001, help="port")
     args = parser.parse_args()
-    app = FastAPI()
+    app = FastAPI(lifespan=lifespan)
 
     # we don't need to manually check the env now
     # settings.check_env()

hugegraph-llm/src/hugegraph_llm/demo/rag_demo/vector_graph_block.py

@@ -85,9 +85,9 @@ def create_vector_graph_block():
                 graph_index_btn1 = gr.Button("Clear Graph Data & Index", size="sm")
 
         vector_import_bt = gr.Button("Import into Vector", variant="primary")
-        graph_index_rebuild_bt = gr.Button("Rebuild vid Index")
         graph_extract_bt = gr.Button("Extract Graph Data (1)", variant="primary")
         graph_loading_bt = gr.Button("Load into GraphDB (2)", interactive=True)
+        graph_index_rebuild_bt = gr.Button("Rebuild vid Index")
 
     vector_index_btn0.click(get_vector_index_info, outputs=out).then(
         store_prompt,
@@ -122,6 +122,9 @@ def create_vector_graph_block():
     graph_loading_bt.click(import_graph_data, inputs=[out, input_schema], outputs=[out]).then(
         store_prompt,
         inputs=[input_schema, info_extract_template],
+    ).then(fit_vid_index).then(
+        store_prompt,
+        inputs=[input_schema, info_extract_template],
     )
 
     def on_tab_select(input_f, input_t, evt: gr.SelectData):