Update app.py

app.py

@@ -19,6 +19,8 @@ import json
 cache_dir = '/data/KB'
 os.makedirs(cache_dir, exist_ok=True)
 
+
+
 @spaces.GPU
 def weighted_mean_pooling(hidden, attention_mask):
     attention_mask_ = attention_mask * attention_mask.cumsum(dim=1)
@@ -64,15 +66,12 @@ def calculate_md5_from_binary(binary_data):
     return hash_md5.hexdigest()
 
 @spaces.GPU
-def add_pdf_gradio(pdf_file_binary, progress=gr.Progress()):
+def add_pdf_gradio(pdf_file_path, progress=gr.Progress()):
     global model, tokenizer
     model.eval()
-
-    print(f"type pdf_file_binary: {type(pdf_file_binary)}")
-    print(f"len(pdf_file_binary): {len(pdf_file_binary)}")
-    print(f"type pdf_file_binary[0]: {type(pdf_file_binary[0])}")
+    print(f"pdf_file_path: {pdf_file_path}")
     
-    knowledge_base_name = calculate_md5_from_binary(pdf_file_binary)
+    knowledge_base_name = calculate_md5_from_binary(pdf_file_path)
     
     this_cache_dir = os.path.join(cache_dir, knowledge_base_name)
     os.makedirs(this_cache_dir, exist_ok=True)
@@ -261,7 +260,7 @@ with gr.Blocks() as app:
     gr.Markdown("- Currently online demo support PDF document with less than 50 pages due to GPU time limit. Deploy on your own machine for longer PDFs and books.")
     
     with gr.Row():
-        file_input = gr.File(file_types=["pdf"], file_count="multiple", type="binary", label="Step 1: Upload PDF")
+        file_input = gr.File(file_types=["pdf"], label="Step 1: Upload PDF")
         file_result = gr.Text(label="Knowledge Base ID (remember it, it is re-usable!)")
         process_button = gr.Button("Process PDF (Don't click until PDF upload success)")