Spaces:
Running
on
Zero
Running
on
Zero
Update app.py
Browse files
app.py
CHANGED
@@ -1,15 +1,14 @@
|
|
1 |
import gradio as gr
|
2 |
from docling.document_converter import DocumentConverter
|
3 |
-
import spaces
|
4 |
|
5 |
-
|
6 |
-
|
7 |
-
@spaces.GPU
|
8 |
def convert_document(file, output_format):
|
9 |
# Load document and convert it using Docling
|
10 |
converter = DocumentConverter()
|
11 |
result = converter.convert(file.name)
|
12 |
-
|
|
|
|
|
|
|
13 |
# Choose the output format
|
14 |
if output_format == "Markdown":
|
15 |
converted_text = result.document.export_to_markdown()
|
@@ -18,17 +17,14 @@ def convert_document(file, output_format):
|
|
18 |
else:
|
19 |
converted_text = "Unsupported format"
|
20 |
|
21 |
-
#
|
22 |
metadata = {
|
23 |
-
"
|
24 |
-
"Author": result.document.author,
|
25 |
-
"Language": result.document.language,
|
26 |
-
"References": result.document.references
|
27 |
}
|
28 |
|
29 |
return converted_text, metadata
|
30 |
|
31 |
-
# Define the Gradio interface using new component syntax
|
32 |
with gr.Blocks() as app:
|
33 |
gr.Markdown("# Document Converter with Docling")
|
34 |
gr.Markdown("Upload a document, choose the output format, and get the converted text with metadata.")
|
|
|
1 |
import gradio as gr
|
2 |
from docling.document_converter import DocumentConverter
|
|
|
3 |
|
|
|
|
|
|
|
4 |
def convert_document(file, output_format):
|
5 |
# Load document and convert it using Docling
|
6 |
converter = DocumentConverter()
|
7 |
result = converter.convert(file.name)
|
8 |
+
|
9 |
+
# Check available attributes in DoclingDocument
|
10 |
+
available_attributes = dir(result.document)
|
11 |
+
|
12 |
# Choose the output format
|
13 |
if output_format == "Markdown":
|
14 |
converted_text = result.document.export_to_markdown()
|
|
|
17 |
else:
|
18 |
converted_text = "Unsupported format"
|
19 |
|
20 |
+
# Placeholder metadata extraction based on available attributes
|
21 |
metadata = {
|
22 |
+
"Available Attributes": available_attributes
|
|
|
|
|
|
|
23 |
}
|
24 |
|
25 |
return converted_text, metadata
|
26 |
|
27 |
+
# Define the Gradio interface using the new component syntax
|
28 |
with gr.Blocks() as app:
|
29 |
gr.Markdown("# Document Converter with Docling")
|
30 |
gr.Markdown("Upload a document, choose the output format, and get the converted text with metadata.")
|