Update app.py

app.py

@@ -1,50 +1,44 @@
 import gradio as gr
 from docling.document_converter import DocumentConverter
-from docling_core.transforms.chunker import HierarchicalChunker
 import spaces
 
 
-def convert_document(file, output_format):
-    # Load document and convert it using Docling
-    converter = DocumentConverter()
-    result = converter.convert(file.name)
-
-    # Check available attributes in DoclingDocument
-    available_attributes = dir(result.document)
-    document = result.document
-
-
-    # Choose the output format
-    if output_format == "Markdown":
+def convert_document(file, method):
+    if method == "Docling":
+        # Load document and convert it using Docling
+        converter = DocumentConverter()
+        result = converter.convert(file.name)
+    
+        # Check available attributes in DoclingDocument
+        available_attributes = dir(result.document)
+        document = result.document
+    
+    
+        # Output
         converted_text = result.document.export_to_markdown()
-    elif output_format == "JSON":
-        converted_text = result.document.export_to_json()
+        metadata = {
+            "Available Attributes": available_attributes
+        }
+    
+        return converted_text, metadata
+    elif method == "Marker":
+        return 'unsupported method', ''
     else:
-        converted_text = "Unsupported format"
-
-    # Placeholder metadata extraction based on available attributes
-    metadata = {
-        "Available Attributes": available_attributes,
-        "Chunked": list(HierarchicalChunker().chunk(document))
-    }
-
-    return converted_text, metadata
+        return 'unknown method', ''
 
-# Define the Gradio interface using the new component syntax
 with gr.Blocks() as app:
-    gr.Markdown("# Document Converter with Docling")
-    gr.Markdown("Upload a document, choose the output format, and get the converted text with metadata.")
+    gr.Markdown("# Document Converter")
+    gr.Markdown("Upload a document, choose the backend, and get the converted text with metadata.")
 
     file_input = gr.File(label="Upload Document")
-    format_input = gr.Radio(["Markdown", "JSON"], label="Choose Output Format")
+    method_input = gr.Radio(["Docling", "Marker"], label="Choose Conversion Backend")
     output_text = gr.Textbox(label="Converted Document")
     output_metadata = gr.JSON(label="Metadata")
 
-    # Define the process button and bind it to the function
     convert_button = gr.Button("Convert")
     convert_button.click(
         convert_document,
-        inputs=[file_input, format_input],
+        inputs=[file_input, method_input],
         outputs=[output_text, output_metadata]
     )